the epidemiology of non-typhoidal salmonella enterica in

The epidemiology of non-typhoidal Salmonella enterica in Australia and the impact of whole genome sequencing on public health surveillance and outbreak investigation

Laura Ford

Research School of Population Health

A thesis submitted for the degree of

Doctor of Philosophy

of The Australian National University

May 2019

© Copyright by Laura Ford 2019

All Rights Reserved

i

Statement of Contribution

This thesis is submitted as a Thesis by Compilation in accordance with

https://policies.anu.edu.au/ppl/document/ANUP_003405.

I declare that the research presented in this Thesis represents original work that I carried out

during my candidature at the Australian National University, except for contributions to multi-

author papers incorporated in the Thesis where my contributors are specified in this Statement

of Contribution.

Title: Increasing incidence of Salmonella in Australia, 2000-2013

Authors: Ford L, Glass K, Veitch M, Wardell R, Polkinghorne B, Dobbins T, Lal A, Kirk MD

Publication outlet: PLoS One

Current status of paper: Published

Contribution to paper: LF, KG, MV, RW, BP, and MDK conceived this paper. LF conducted

the data analysis with assistance from KG, RW, and AL. LF drafted the manuscript with

assistance from RW. LF coordinated co-author input and revised the manuscript as required.

Senior author’s endorsement:

Title: The epidemiology of Salmonella outbreaks in Australia, 2001-2016

Authors: Ford L, Moffatt C, Fearnley E, Sloan-Gardner T, Miller M, Polkinghorne B, Franklin

N, Williamson DA, Glass K, Kirk MD

Publication outlet: Frontiers in Sustainable Food Systems


Contribution to paper: LF conceived this paper with guidance from MDK and KG. LF, CM,

TS, BP, and MD decided on the methodology. LF conducted the data analysis, drafted the

manuscript, coordinated co-author input, and revised the manuscript as required.


Title: Cost of Salmonella infections in Australia, 2015

Authors: Ford L, Haywood P, Kirk MD, Lancsar E, Williamson DA, Glass K

Publication outlet: Journal of Food Protection


Contribution to paper: LF conceived this paper with guidance from MDK and KG. LF decided

on the methodology with PH, MDK, and KG. LF conducted the data analysis with assistance

from PH and KG. LF drafted the manuscript, coordinated co-author input, and revised the

manuscript as required.

ii


Title: Cost of whole genome sequencing for non-typhoidal Salmonella enterica

Authors: Ford L, Glass K, Williamson, DA, Sintchenko V, Robson, JMB, Stafford, R, Kirk, MD

Current status of paper: To be submitted

Contribution to paper: LF conceived this paper with guidance from MDK, KG, and DAW. LF

decided on methodology with KG. LF conducted the data analysis with assistance from KG.

LF drafted the manuscript, coordinated co-author input, and revised the manuscript as

required.


Title: Incorporating whole-genome sequencing into public health surveillance: lessons from

prospective sequencing of Salmonella Typhimurium in Australia

Authors: Ford L, Carter GP, Wang Q, Seemann T, Sintchenko V, Glass K, Williamson DA,

Howard P, Valcanis M, Castillo CFS, Sait M, Howden BP, Kirk MD

Publication outlet: Foodborne Pathogens and Disease


Contribution to paper: LF conceived the study with support from GPC, QW and MDK. LF,

GPC, QW, TS, VS, DAW, PH, MV, CFSC, MS, and BPH generated the data. LF conducted

data analysis, drafted the manuscript, coordinated co-author input, and revised the manuscript

as required.


Title: Seven Salmonella Typhimurium outbreaks in Australia linked by trace-back and whole-

genome sequencing

Authors: Ford L, Wang Q, Stafford R, Ressler KA, Norton S, Shadbolt C, Hope K, Franklin N,

Krsteski R, Carswell A, Carter GP, Seemann T, Howard P, Valcanis M, Castillo CFS, Bates J,

Glass K, Williamson DA, Sintchenko V, Howden BP, Kirk MD

Publication outlet: Foodborne Pathogens and Disease


Contribution to paper: LF conceived this study with support from RS, KAR, SN, CS, and KH.

LF, RS, KAR, SN, CS, KH, NF, RK, AC, QW, GPC, TS, PH, MV, CFS, and JB generated the

data. LF conducted data analysis. LF drafted the manuscript with direct input from QW, RS,

KAR, and SN. LF coordinated all co-author input and revised the manuscript as required.


iii

Title: Whole-genome sequencing of Salmonella Mississippi and Typhimurium Definitive Type

160, Australia and New Zealand

Authors: Ford L, Ingle D, Glass K, Veitch M, Williamson DA, Harlock M, Gregory J, Stafford

R, French N, Bloomfield S, Grange Z, Conway ML, Kirk MD

Publication outlet: Emerging Infectious Diseases


Contribution to paper: LF conceived this paper with guidance from MDK, KG, DAW, and

MV. LF collated data from DI, MH, SB, and ZG for further analysis. LF drafted the manuscript

with direct input from DI. LF coordinated all co-author input and revised the manuscript as

required.


Laura Ford ____________ _________________________________ 30/04/19

Candidate – Print Name Signature Date

Endorsed

Kathryn Glass _________ _________________________________ _______

Primary Supervisor – Print Name Signature Date

06/05/19

iv

Acknowledgments

I thank the many people and organizations who provided the data used in this thesis. In

particular, I thank the OzFoodNet Network, which is funded by the Australian Government

Department of Health. I thank the public health laboratories that performed serotyping, phage

typing, MLVA, and WGS of Salmonella isolates, including the Microbiological Diagnostic Unit

Public Health Laboratory, the Institute for Clinical Pathology and Medical Research –

Pathology West, and Queensland Health Forensic and Scientific Services. I would also like to

thank the Communicable Disease Network of Australia, state and territory health departments,

and state and territory public health officers. I particularly give thanks to ACT Health and the

Tasmanian Department of Health and Human Services, as well as their communicable disease

control teams.

I am very grateful for the wonderful supervision I received throughout my PhD candidature. I

thank the chair of my panel Martyn Kirk, my primary supervisor Kathryn Glass, and my

supervisor Deborah Williamson for their guidance, support, and expertise. I am exceedingly

thankful for their generosity and encouragement throughout this research. They each took the

time to assist with everything from big-picture ideas, to the methodology of my research, to

improving my writing. I appreciate their unique contributions and I thoroughly enjoyed working

with them.

I received funding support through an Australian Government Research Training Program

(RTP) Scholarship throughout my PhD candidature. I also thank the ANU, the Research

School of Population Health, and the National Centre for Epidemiology and Population Health

for additional funding support and other opportunities. In particular, I am grateful for receiving

the Peter Baume Travel Grant and the Vice Chancellor’s Higher Degree by Research Travel

Grant to travel to an international conference to present on research from this thesis.

Finally, I thank my friends and colleagues at the Research School of Population Health for their

wonderful support, kindness, and advice over the last few years. This experience would have

been much harder without them. I thank my family in the United States for providing me with

every opportunity that has led me here. I am forever grateful to my husband Ashley for his

unending support, and for taking wonderful care of our daughter Georgia.

v

Abstract

In Australia, the rates of reported non-typhoidal Salmonella enterica infections are amongst

the highest in the industrialized world. Public health laboratories are increasingly adopting

whole genome sequencing (WGS) of non-typhoidal Salmonella enterica isolates to

characterize isolates for surveillance. In this thesis, I contribute to understanding the

epidemiology of salmonellosis and provide evidence on the use of phylogenomic salmonellosis

data for public health in Australia by examining trends in and sources of infection; costs-of-

illness and costs of WGS; and impacts of WGS on surveillance and outbreak investigation.

I analysed national surveillance data using negative binomial regression to show that most

states and territories had significantly increasing trends of Salmonella Typhimurium

notifications between 2000 and 2013. Geographic and age-specific analyses indicated the

importance of foodborne transmission for Salmonella Typhimurium, while suggesting that other

transmission pathways may be more important for non-Typhimurium serotypes. I found that

79% of the 990 Salmonella outbreaks in Australia between 2001 and 2016 were suspected or

confirmed to be transmitted through contaminated food. Most outbreaks were due to

Salmonella Typhimurium, and eggs or egg-containing foods were the most frequently identified

food vehicles.

I estimated the health care, lost productivity, and premature mortality costs of non-typhoidal

Salmonella enterica and sequelae irritable bowel syndrome (IBS) and reactive arthritis (ReA)

in Australia. Circa 2015, I estimated 90,833 salmonellosis cases, 4,312 hospitalizations, and

19 deaths at a cost of AUD 124.4 million (90% credible intervals 107.4-143.1 million) and AUD

146.8 million (90% CrI 127.8-167.9 million) when IBS and ReA were included. While WGS is

more expensive than traditional Salmonella typing methods and polymerase chain reaction

testing, it may result in cost savings if it can reduce case numbers through early detection of

outbreaks and sources of infection.

To examine the impacts of WGS on public health surveillance, prevention, and control, I

conducted two studies. Firstly, concurrent WGS and multiple locus variable-number-tandem

repeat analysis (MLVA) was performed on Salmonella Typhimurium isolated from residents of

the Australian Capital Territory for a 5 month period. Compared to MLVA, I found that WGS

was more sensitive, linking an additional 9% of isolates to at least one other isolate, and linking

cases from seven outbreaks occurring over 5 months in three Australian states and territories.

Secondly, I used phylogenomic and epidemiological risk factor data to characterize the

epidemiology and disease reservoirs for Salmonella Mississippi and Salmonella Typhimurium

definitive type 160 (DT160). Sequence and epidemiological data for isolates from humans,

animal, and environmental sources identified plausible sources of human infection from wildlife

vi

and environmental reservoirs and showed Australian and New Zealand isolates in distinct

clades for both serotypes. While the genetic relatedness of DT160 isolates suggested a

common reservoir, source attribution studies would be helpful for hyper-endemic serotypes

such as Salmonella Mississippi.

In this thesis, I demonstrate that the burden of Salmonella enterica in Australia is high, and

provide evidence for the formation of targeted policies and interventions to prevent infection.

My findings on using phylogenomic data will be instrumental in implementing WGS for routine

surveillance and outbreak investigations.

vii

Conference or workshop presentations arising from this thesis

1. Ford L. WGS and Salmonella outbreak investigations. Invited presentation at Food

Standards Australia New Zealand’s Microbial Genomics in Food Regulation workshop;

2016 November 14; Melbourne, Vic, Australia.

2. Ford L, Carter G, Wang Q, Seemann T, Stafford R, Hope K, Sintchenko V, Williamson

DA, Glass K, Kirk MD. Implementing whole genome sequencing for the surveillance of

Salmonella in the ACT. Oral presentation at The Communicable Disease Control

Conference; 2017 June 27-28; Melbourne, Vic, Australia.

3. Ford L, Ingle D, Gonçalves da Silva A, Harlock M, Veitch M, Williamson D, Glass K,

Kirk M. The emergence of Salmonella Typhimurium DT 160 in Australia among humans

and sparrows. Poster presentation at The International Conference on Emerging

Infectious Diseases; 2018 August 26-29; Atlanta, GA, USA.

4. Ford L, Ingle D, Harlock M, Veitch M, Glass K, Williamson D, Gregory J, Stafford R,

Kirk M. Genomic epidemiology of Salmonella Mississippi in Australia. Poster

presentation at The International Conference on Emerging Infectious Diseases; 2018

August 26-29; Atlanta, GA, USA.

5. Ford L, Haywood P, Kirk M, Williamson D, Glass K. The costs of non-typhoidal

Salmonella in Australia. Oral presentation at The Research School of Population Health

Student Conference, Australian National University; 2018 October 11; Canberra, ACT,

Australia.

viii

Table of Contents Chapter 1. Introduction ............................................................................................................. 1

1.1 Research motivation .................................................................................................. 1

1.2 Research themes ....................................................................................................... 3

Chapter 2. Background ............................................................................................................ 5

2.1 Salmonellosis .................................................................................................................. 5

2.2 Global burden of non-typhoidal Salmonella .................................................................... 5

2.3 Non-typhoidal Salmonella in Australia ............................................................................ 6

2.3.1 Burden ...................................................................................................................... 6

2.3.2 Cost .......................................................................................................................... 7

2.3.3 Public health surveillance of Salmonella .................................................................. 8

2.4 The impact of new Salmonella testing and typing technologies on public health

surveillance ......................................................................................................................... 12

2.4.1 Culture-independent diagnostic testing .................................................................. 12

2.4.2 Whole genome sequencing .................................................................................... 13

2.5 Conclusion .................................................................................................................... 15

Chapter 3. Trends of non-typhoidal Salmonella enterica in Australia .................................... 16

3.1 Introduction ................................................................................................................... 16

3.2 Paper ............................................................................................................................ 16

Chapter 4. Sources of non-typhoidal Salmonella enterica in Australia .................................. 28

4.1 Introduction ................................................................................................................... 28

4.2 Paper ............................................................................................................................ 28

Chapter 5. Costs of non-typhoidal Salmonella enterica in Australia ...................................... 37

5.1 Introduction ................................................................................................................... 37

5.2 Papers .......................................................................................................................... 37

Chapter 6. Impact of whole-genome sequencing on public health surveillance of Salmonella

Typhimurium ........................................................................................................................... 59

6.1 Introduction ................................................................................................................... 59

6.2 Papers .......................................................................................................................... 59

ix

Chapter 7. Impact of whole-genome sequencing on public health surveillance two non-

typhoidal Salmonella enterica serotypes ................................................................................ 75

7.1 Introduction ................................................................................................................... 75

7.2 Paper ............................................................................................................................ 75

Chapter 8. Discussion and conclusion ................................................................................... 84

8.1 Findings and implications ............................................................................................. 84

8.2 The future of Salmonella surveillance and research in Australia .................................. 90

8.3 Conclusion .................................................................................................................... 92

References ............................................................................................................................. 93

Appendix 1. Supplementary materials for chapter 3 ............................................................ 105





1

Chapter 1. Introduction

Non-typhoidal Salmonella enterica (Salmonella) infection causes acute gastroenteritis, along

with severe outcomes that occur more rarely including septicaemia, irritable bowel syndrome

(IBS), reactive arthritis (ReA) and death. Salmonellosis is an important cause of morbidity and

mortality from foodborne disease globally (Havelaar et al., 2015) and can be costly to health

care systems and society (Scharff et al., 2016). In Australia, rates of notified cases of

Salmonella are high and have more than doubled since notifiable disease surveillance began

in 1991 (Department of Health, 2019). Rates of Salmonella in Australia are much higher than

other high-income countries including the United States of America, Canada, the United

Kingdom, and much of the European Union (CDC, 2017; Government of Canada, 2019; EFSA

and ECDC, 2018). High rates and foodborne disease outbreaks have led Salmonella to be

targeted as one of two priority pathogens for attention in Australia’s Foodborne Illness

Reduction Strategy 2018–2021+ (Food Regulation, 2018).

With no human vaccine for non-typhoidal Salmonella, and treatment generally not

recommended, routine surveillance and outbreak investigation are key to reducing the burden

of illness. In Australia, an estimated 72% of Salmonella infections are transmitted through

contaminated food and outbreaks of foodborne salmonellosis are relatively common

(OzFoodNet Working Group, 2018; Valley et al., 2014). Public health surveillance and outbreak

investigation can lead to the identification of contaminated foods, as well as the identification

of persistent causes of food contamination. Epidemiological data of Salmonella transmitted

through contaminated food and other pathways is essential for the implementation of effective

prevention and control measures.

1.1 Research motivation

Understanding the epidemiology, costs, and impacts of laboratory typing for Salmonella will

help to inform public health surveillance and food safety policy, with the ultimate aim of

reducing the burden of illness in Australia. Analysing routine Salmonella surveillance and

outbreak investigation data helps to determine the magnitude of the public health problem

(Borgdoff & Motarjemi, 1997) and assists with pathogen and serotype prioritization for public

health resources. Monitoring trends through routine surveillance and outbreak data is

necessary for the early detection and response to emerging strains and sources of illness

(Groseclose et al., 2010). Examining trends of reported infection in Australia is needed to help

identify geographical differences, at risk populations, priority serotypes and sources and

reservoirs of infection, providing evidence for targeted interventions and policies for the

2

prevention and control of infection. Also, a comprehensive review of all reported foodborne

Salmonella outbreaks and their associated food vehicles is needed to provide evidence for

food safety agencies to implement standards and policies across the food chain.

Estimating the costs of illness can also inform food safety policy from farm to retail and provide

inputs for subsequent cost effectiveness of new policies and interventions that aim to reduce

contamination and therefore the burden of illness. As one of the leading causes of foodborne

gastroenteritis-associated hospitalizations and deaths in Australia, and a cause of IBS and

ReA, the burden of Salmonella is high (Kirk et al., 2014; Ford et al., 2014). While the overall

cost of foodborne disease in Australia circa 2000 was estimated at AUD 1.2 billion (Abelson et

al., 2006), the present-day costs of Salmonella and its sequelae in Australia have not been

estimated. Internationally, studies have demonstrated that Salmonella has a substantial

economic burden (Hoffmann et al., 2012, Santos et al., 2011, Scharff et al., 2016, Sundstrom,

2018) and that estimates of the cost of illness can be used to compare the burden of pathogens

for public health prioritization, and to evaluate the effectiveness of programs, policies, or

interventions for surveillance, prevention, and control of Salmonella (Gavin et al., 2018, Jain

et al., 2019, Scharff et al., 2016).

In Australia, Salmonella is routinely subtyped for public health surveillance when an isolate is

cultured from a human or non-human sample. Serotyping and molecular subtyping of

Salmonella isolates help distinguish between outbreak-related and sporadic isolates, including

closely related isolates, as well as link human cases to sources of infection (Bender et al.,

2001; Ross et al., 2011). Subtyping allows for the identification of outbreaks and non-human

sources for investigation, which eventually can result in the implementation of prevention and

control measures by food and environmental safety agencies.

The gold standard for Salmonella subtyping has been serotyping in Australia, and when the

isolate is serotyped as Salmonella Typhimurium multiple locus variable number tandem repeat

analysis is also conducted. While pulsed field gel electrophoresis (PFGE) is an important

Salmonella subtyping method globally and considered the gold standard by the PulseNet

International network (Camarda et al., 2015; WHO and FAO, 2009), it is not routinely

performed in Australian reference laboratories. Recent advances in laboratory technology has

resulted in whole genome sequencing (WGS) being used as a subtyping method for the

epidemiologic investigation and surveillance of Salmonella and other foodborne bacterial

pathogens (Deng et al., 2016). As WGS of Salmonella isolates generates the entire genetic

information of the organisms, WGS data provides higher discriminatory power to differentiate

closely related outbreaks, assist in tracking epidemiological trends, and monitor antimicrobial

resistances (Deng et al., 2016). While microbiology laboratories increasingly began using this

3

technology for Salmonella, it was unknown how public health surveillance was going to

incorporate highly technical and apparently costly data generated through WGS into

surveillance and control efforts.

Phylogenetic data generated from WGS has been used internationally to help investigate

outbreaks and during routine surveillance typing of Salmonella (den Bakker et al., 2014;

Leekitcharoenphon et al., 2014; Ashton et al., 2016; Inns et al., 2017). In addition, prospective

national surveillance of Listeria monocytogenes in Australia and the USA has shown the

benefits of using WGS for epidemiologic surveillance of a foodborne bacterial pathogen

(Jackson et al., 2016; Kwong et al., 2016). However, sequencing of Salmonella isolates is not

yet routine in all Australian public health laboratories for prospective public health surveillance.

Consequently, the impacts of whole genome sequencing on public health surveillance and

outbreak investigation of Salmonella have not been well established.

1.2 Research themes

The aim of this thesis is to understand the epidemiology of Salmonella in Australia, as well as

the impacts of whole genome sequencing on the public health surveillance and outbreak

investigation of Salmonella. I examined three research topics relating to Salmonella in

Australia in this thesis:

1. Trends and sources of infection;

2. Cost-of-illness and costs of WGS; and

3. Impacts of WGS on surveillance and outbreak control.

The thesis is a compilation of seven scientific manuscripts that address these research

themes relating to Salmonella in Australia, six of which have been published and one which

will be submitted to an international peer review journal. Papers are grouped into

chapters based on the research themes, along with chapters for background and

discussion and conclusion, which are unpublished.

Research theme 1: Trends and sources of infection

In this research theme, I investigated the trends and sources of Salmonella infection in

Australia to better understand the epidemiology of Salmonella and provide evidence of key

serotypes and geographical niches. I analysed national Salmonella surveillance data using

negative binomial regression to investigate trends, including geographic and age-specific

trends, of Salmonella Typhimurium and non-Typhimurium Salmonella separately. I used

descriptive analysis to identify trends of outbreaks and identify important food vehicles

associated with outbreaks from a national register of salmonellosis outbreaks in Australia. In

addition, I analysed phylogenomic data from WGS, investigating sources of infection of

4

Salmonella Typhimurium, Salmonella Typhimurium DT 160 (DT160), and Salmonella

Mississippi.

Chapters 3 and 4 address this research theme, with papers in chapters 6 and 7 also

investigating trends and sources of infection using data generated through WGS.

Research theme 2: Cost-of-illness and costs of WGS

The goal of this research theme was to estimate how much Salmonella infections and their

sequelae cost Australian society. This estimate of cost will inform food safety policy and

provide inputs for subsequent cost effectiveness analyses of new policies and interventions. I

used complementary datasets to estimate the burden and costs of health care usage, lost

productivity, and premature mortality from Salmonella and sequelae IBS and ReA illness circa

2015. I used Monte Carlo simulation to provide 90% credible intervals around point estimates.

The second goal of this theme was to estimate the costs of WGS for Salmonella to the

Australian society, compared to costs of the current laboratory testing and subtyping practices,

to provide evidence of cost-savings as Australia moves towards routine WGS for Salmonella.

I first aimed to determine how effective the laboratory methods would have to be at preventing

illnesses to be cost neutral. I used data generated from the Salmonella cost-of-illness

estimates and laboratory method costs in a mathematical equation with Monte Carlo simulation

to determine the median threshold with 90% credible intervals. I then examined the impact on

costs if using WGS during different outbreak scenarios, modelling if a product recall or

intervention occurred at earlier time points along the epidemiological curve.

Papers in chapter 5 address this research theme.

Research theme 3: Impacts of WGS on surveillance and outbreak investigation

In this research theme I investigated the impacts of using phylogenomic data from WGS of

Salmonella isolates from humans, food, animals, and the environment for public health

surveillance and outbreak investigation to inform the use of phylogenomic Salmonella data by

epidemiologists in Australia. I conducted a prospective trial of sequencing all Salmonella

Typhimurium isolates in the Australian Capital Territory over a period of 5 months, and

evaluated WGS performance for routine public health surveillance. I reported on the

epidemiological and environmental investigations, and the use of phylogenomic data for seven

linked Salmonella Typhimurium outbreaks occurring across Australia during the WGS trial. In

addition, I integrated epidemiological and phylogenomic data to better understand

geographical niche and transmission pathways of two Salmonella strains: serotypes

Typhimurium DT160 and Mississippi. Papers in chapters 6 and 7 address this research theme.

5

Chapter 2. Background

2.1 Salmonellosis

Salmonellosis is an acute gastroenteritis illness, with symptoms including diarrhoea,

abdominal pain, nausea, and fever. Symptoms usually start 6 to 72 hours after infection and

last for 2 to 7 days (Heymann, 2015). While most people with salmonellosis recover without

treatment, dehydration or invasive infection can lead to hospitalisation or death, particularly in

high risk groups including young children, the elderly, and the immunocompromised (CDC,

2019a). In addition, salmonellosis can result in severe and disabling sequelae, as

approximately 8.8% of people with salmonellosis develop IBS and 8.5% develop reactive

arthritis ReA following their acute gastroenteritis illness (Ford et al., 2014; Haagsma et al.,

2010).

Salmonellosis is caused by Salmonella enterica, a gram-negative bacteria that can infect

humans through the ingestion of contaminated food or water, direct contact with the

environment, and direct contact with infected animals and humans (Giannella, 1996; Vally et

al., 2014). There are more than 2,500 serotypes of Salmonella enterica. Typhoidal Salmonella

serotypes are human-adapted serotypes causing septicaemia and are uncommon in

industrialized countries. Most cases of S. Typhi and S. Paratyphi in Australia are associated

with international travel (Jones et al., 2008; The OzFoodNet Working Group, 2018). Although

all non-typhoidal Salmonella serotypes can cause disease, they have a variety of hosts and

reservoirs and can differ in pathogenic potential (Jones et al., 2008; WHO 2018). In

industrialized countries, two Salmonella serotypes: Enteritidis and Typhimurium cause a

significant number of sporadic illnesses and outbreaks (WHO 2018).

2.2 Global burden of non-typhoidal Salmonella

Salmonella is an important illness globally, with the World Health Organization (WHO)

estimating that in 2010, it resulted in 153 million illnesses and 57,000 deaths worldwide

(Havelaar et al., 2015; Kirk et al., 2015). Approximately 52% of Salmonella illnesses globally

were attributed to contaminated food, with the estimated proportion of foodborne salmonellosis

varying from 46% in areas of Africa with high child and very high adult mortality to 76% in areas

of Europe with very low child and adult mortality (Hald et al., 2016; Kirk et al., 2015). Although

only about half of infections were attributed to contaminated food overall, the burden of

foodborne Salmonella from invasive and non-invasive infection on human health is high and

6

results in the most disability adjusted life years (DALYs) of the 22 foodborne bacterial,

protozoal, and viral diseases included in the WHO estimates (Kirk et al., 2015).

The costs of Salmonella to the health care system and to the community have been estimated

in some countries. In the United States of America (USA), the mean cost of illness due to

Salmonella was USD 3.67 billion in 2013 (ERS, 2013). In addition, in the USA, Salmonella had

the highest number of estimated quality-adjusted life year (QALY) loss of 14 foodborne disease

pathogens at 16,782 QALYs lost per year (Batz et al., 2014). In Sweden, while the overall cost

of foodborne Campylobacter was higher than the cost for foodborne Salmonella (€97.7 million

vs €25.3 million) due to different rates of disease, the cost per case of salmonellosis was higher

(€979 vs €1,374). A study in the United Kingdom (UK) found that the cost of two most common

serotypes of Salmonella (Typhimurium and Enteritidis) in 2008 was £6.5 million. Costs per

case were higher for S. Typhimurium (£1282) than for S. Enteritidis (£993), due to higher

spending on health service resources and more time off work for S. Typhimurium cases

(Santos et al., 2011). A study in The Netherlands estimated the costs of illness for one

extensive food-related salmonellosis outbreak at €6.8 million, or €7.5 million including outbreak

control costs (Suijkerbuijk et al., 2017). Quantifying the costs of foodborne illness and of

salmonellosis is important to prioritise pathogens for intervention and evaluate the impact of

interventions on reducing costs and disease burden (Berends et al., 1998; Duff et al., 2013;

Korsgaard et al., 2009; Lawson et al., 2009; Sundstrom et al., 2014).

2.3 Non-typhoidal Salmonella in Australia

2.3.1 Burden

In Australia, salmonellosis is an important and increasing cause of gastroenteritis, particularly

foodborne gastroenteritis. Over the last two decades, the rate of salmonellosis notified to public

health departments has more than doubled, reaching a peak of 74.7 cases per 100,000

population in 2016 (Figure 1) (Department of Health, 2019). While the rate of Salmonella

decreased in 2017 and 2018, at 57.6 cases per 100,000 in 2018, the rate was still higher than

in other high-income countries, with a notification rate of 16.67 cases per 100,000 in the USA

(CDC, 2017), 15.5 cases per 100,000 in the UK, and 19.7 cases per 100,000 in the European

Union in 2017 (EFSA and ECDC, 2018). In 2016, Canada reported a notification rate of 21.03

per 100,000 (Government of Canada, 2019) and New Zealand reported a notification rate of

23.2 cases per 100,000 (ESR, 2017). Rates of non-typhoidal Salmonella in 2017 in Europe

ranged from 4.5 cases to 108.5 cases per 100,000, with a higher rate than Australia only in

Slovakia and the Czech Republic (EFSA and ECDC, 2018).

7

Figure 1: The notification rate of non-typhoidal Salmonella spp. per 100,000 population in

Australia, 1991-2018. Based on data from the National Notifiable Disease Surveillance System

(Department of Health, 2019).

The number of notified cases of salmonellosis is believed to be only a fraction of the number

of cases in the community, as not all cases will seek medical attention and not all who seek

medical attention will be tested (Hall et al., 2008). An Australian study estimated that circa

2010, there were 56,000 cases of Salmonella in the community, approximately 40,000 of which

were caused by contaminated food (Kirk et al., 2014). This equated to a rate of 185 cases of

foodborne salmonellosis per 100,000 population, a 24% increase since circa 2000 (Kirk et al.,

2014). A proportion of people infected with Salmonella have severe or continuing illness and

circa 2010 there were an estimated 2,100 hospitalisations for foodborne salmonellosis and

6,750 sequelae illnesses of post-infectious IBS and ReA resulting from foodborne

salmonellosis (Ford et al., 2014; Kirk et al., 2014). Additionally, Salmonella was one of the

leading causes of deaths from foodborne illness in Australia, with an estimated 15 deaths circa

2010 with salmonellosis as the primary or contributing cause of death (Kirk et al., 2014). As

Salmonella is a preventable infection, the burden on the Australian community is unnecessarily

high.

2.3.2 Cost

In 2006, a report was published estimating the total cost of foodborne illness in Australia at

AUD 1.25 billion per annum (Abelson et al., 2006). This report used data from foodborne illness

estimates circa 2000 and cost estimates from 2002 and 2004, so the estimate may no longer

be accurate today in 2019. The cost estimate was for all foodborne illness and the report does

0

10

20

30

40

50

60

70

80

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

Rat

e p

er 1

00,0

00 p

op

ula

tio

n

Year

8

not break down the costs for each gastrointestinal pathogen (Abelson et al., 2006). The costs

of salmonellosis are likely to be substantial, with one hospital outbreak in Queensland in 1996

affecting 52 people estimated to cost the hospital more than AUD 120,000 (USD 95,000) with

additional considerable indirect costs (Searing et al., 2000). Determining the cost of Salmonella

in Australia would not only quantify the burden of illness in monetary terms, but also provide

an opportunity to evaluate the cost-effectiveness of current and future interventions to prevent

disease.

2.3.3 Public health surveillance of Salmonella

As Salmonella has a propensity to cause outbreaks of foodborne disease, public health

surveillance and outbreak investigation have been undertaken to identify food sources linked

to illness (The OzFoodNet Working Group, 2018), potentially indicating a wider problem with

the food supply. In Australian states and territories, public health surveillance of non-typhoidal

Salmonella generally have followed a series of steps shown in Figure 2 and explained in detail

below.

2.3.3.1 Notification

Surveillance starts when a notification of laboratory-confirmed Salmonella in a person’s faecal,

urine, or blood specimen is received by the local public health unit, and then reported nationally

under public health legislation (Department of Health, 2018).

2.3.3.2 Serotyping and MLVA

Cultured Salmonella isolates are forwarded from a primary pathology laboratory to a state

reference laboratory for serotyping. This involves identifying the antigenic formula of the

forwarded isolates and classifying the antigenic formula according to the White-Kauffmann-Le

Minor scheme to give each isolate a name (Issenhuth-Jeanjean et al., 2014). While there are

over 2,500 different S. enterica serotypes, less than 100 account for most human infections

(CDC, 2019b). While S. Enteritidis is the most common serotype in most countries, in Australia,

S. Typhimurium was the most frequently notified Salmonella serotype, accounting for between

33% and 48% of notifications annually over the 5-year period 2013–2017 (Hendriksen et al.,

2011; The Department of Health, 2019). Because S. Typhimurium was so common, further

subtyping was performed to differentiate isolates. Phage typing or pulse field gel

electrophoresis (PFGE) previously occurred routinely depending on the reference lab (The

OzFoodNet Working Group, 2015). As of 2016, all reference laboratories in Australia routinely

performed Multiple Locus Variable-number tandem repeat Analysis (MLVA) when isolates

were serotyped as S. Typhimurium, however, not all of these laboratories followed the same

interpretive scheme. While S. Typhimurium was responsible for around 44% of notifications

nationally, the proportion of S. Typhimurium ranged by state and territory, making up only 12%

9

Figure 2: Flow chart describing Salmonella surveillance in Australia

Patient visits a doctor

Doctor requests patient to submit a sample for testing

Patient submits a sample to a clinical pathology laboratory

Salmonella is detected in the sample at a clinical pathology laboratory

The clinical pathology

lab notifies the public

health department

The clinical pathology lab forwards

the Salmonella isolate to a

reference laboratory for subtyping

The public health

department conducts follow

up with the patient/case

Patient becomes ill

The public health department

analyses Salmonella isolate

subtyping data

A common exposure with

other cases is identified

No common exposure is

identified

Outbreak investigation

initiated

Continued investigation or

case classified as sporadic

The public health laboratory

performs serotyping and MLVA if

the case is serotyped as

Typhimurium

The public health laboratory

notifies the public health

department of subtyping results

Pat

ien

t an

d d

oct

or

Clin

ical

pat

ho

log

y

lab

ora

tory

Pu

blic

hea

lth

dep

artm

ent

Pu

blic

hea

lth

ref

eren

ce

lab

ora

tory

The clinical pathology

lab notifies

requesting doctor

10

of notifications in the Northern Territory to 67% in the Australian Capital Territory in 2012

(Department of Health, 2019; The OzFoodNet Working Group, 2018). States and territories

with a lower proportion of S. Typhimurium notifications than nationally (Northern Territory,

Queensland, Western Australia, and Tasmania), had a higher proportion of other serotypes

including S. Saintpaul, S. Virchow, S. Enteritidis, and S. Mississippi (Ashbolt & Kirk, 2006; The

OzFoodNet Working Group 2015, 2018). Although S. Enteritidis was one of the top 5 reported

serotypes in Australia in 2012, accounting for 7% of all Salmonella notifications, only 8% of

these infections were thought to be locally acquired, with 86% acquired overseas and 6% with

unknown acquisition (The OzFoodNet Working Group, 2018).

Serotyping and MLVA can help in both the detection of outbreaks and provide important

information during the investigation of outbreaks (CDC, 2019b; Ross et al., 2011). When there

is an increase in cases of a serotype, health departments often initiate an investigation (CDC,

2019a). In addition, knowing what serotypes predominate in what region may assist in

hypothesis generation regarding the origin of infection (Hendriksen et al., 2011). Highly

discriminatory typing is also important. A study on using MLVA for the investigation of S.

Typhimurium clusters in South Australia found that MLVA provided different but complimentary

information by subdividing phenotypically closely-related isolates for investigation (Ross et al.,

2011). Serotyping and MLVA for public health surveillance must be performed and rapidly

reported to epidemiologists to enable timely outbreak detection and implementation of

interventions to prevent further illness (Ethelberg et al., 2007; Hendriksen et al., 2011).

2.3.3.3 Case follow up and analysis of subtyping data

Epidemiological information is obtained by interviewing all cases, or cases with a serotype or

subtype of interest, through case follow up. Gathering case exposure data is key to successful

investigations and interventions. Serotyping and MLVA data from public health reference

laboratories is analysed to identify types that are above baseline levels. Cases with these types

were often prioritised for follow up. Health department interviews of infected cases captured

information about potential exposures, such as: occupation, travel, environmental exposures,

animal exposure, and food consumption history. The risk of salmonellosis has been found to

be higher in certain groups including those living in rural or remote areas, those taking proton

pump inhibitors, those reporting chicken/poultry intake at least seven times per week (Chen et

al., 2016), children under 5 years of age (CDC, 2014), and those with weakened immune

systems (CDC, 2019). There are also specific risk factors for certain Salmonella serotypes

depending on the reservoir (Doorduyn et al., 2006; Funke et al., 2017; Marcus et al., 2007;

Meyer Sauteur et al., 2013).

11

2.3.3.4 Identifying common exposures

An outbreak investigation is often initiated if a common exposure was identified among cases

or an increase in a specific subtype is observed. If no common exposure is identified among

cases with a specific subtype that is above baseline levels, further questions may be asked of

the case to generate a hypothesis for an outbreak investigation. Cases with no common

exposures, links, and/or subtypes to other cases are classified as sporadic.

2.3.3.5 Outbreak investigations

In Australia, OzFoodNet is a collaborative network of epidemiologists within state and territory

health authorities and the Commonwealth Department of Health. OzFoodNet epidemiologists

have investigated and reported on foodborne or suspected foodborne disease due to

Salmonella spp. since 2001 (Department of Health, 2017). Annual reports from 2001 to 2012,

report between 26 and 66 foodborne salmonellosis outbreaks in Australia each year (Table 1)

(Department of Health, 2018a). S. Typhimurium was the causative agent in most salmonellosis

outbreaks, although the proportion of salmonellosis outbreaks due to S. Typhimurium has

ranged from 59% in 2001 to 92% in 2011 (Table 1) (Ashbolt 2002; The OzFoodNet Working

Group, 2015). Salmonellosis outbreaks have been associated with a variety of foods including

eggs, chicken and other poultry, pork, beef, lamb, fish, fresh produce, nuts, pureed/vitamised

food, and dips (Department of Health, 2018a). Outbreaks associated with the use of raw eggs

have been particularly common. One study found that the proportion of foodborne Salmonella

outbreaks linked to eggs significantly increased in Australia between 2001 and 2011, and 90%

of these outbreaks were caused by S. Typhimurium (Moffatt et al., 2016). There were only 6

outbreaks of salmonellosis associated with drinking water reported in Australia between 2001

and 2012 (The OzFoodNet Working Group, 2006, 2007, 2015).

Table 1: Number of foodborne and suspected foodborne outbreaks due to all non-typhoidal

Salmonella and due to Salmonella Typhimurium, Australia, 2001–2012 (Department of Health,

2018a).

Year Number of outbreaks due

to Salmonella

Number (%) of outbreaks due

to Salmonella Typhimurium

2001 27 16 (59%)

2002 26 21 (81%)

2003 31 25 (81%)

2004 36 29 (81%)

2005 33 26 (79%)

2006 41 25 (61%)

2007 50 39 (78%)

12

2008 35 31 (89%)

2009 59 47 (80%)

2010 58 53 (91%)

2011 61 56 (92%)

2012 66 58 (88%)

2.4 The impact of new Salmonella testing and typing technologies on

public health surveillance

In Australia, new technologies for testing and typing Salmonella spp. are moving in divergent

directions. On one hand, the emergence of commercially produced polymerase chain reaction

(PCR) kits for the diagnosis of salmonellosis allows for quick and sensitive Salmonella

detection without having to culture an isolate (also known as culture-independent diagnostic

testing). However, as an isolate is not grown, PCR tests provide limited information for public

health surveillance. On the other hand, using whole genome sequencing (WGS) for Salmonella

subtyping provides a large amount of highly discriminatory data on Salmonella isolates to

compare isolates for surveillance.

2.4.1 Culture-independent diagnostic testing

Although culture has been the foundation of clinical diagnostic testing for Salmonella, primary

pathology laboratories are increasingly adopting culture-independent diagnostic testing

(CIDT). CIDT uses methods including nucleic acid amplification tests such as PCR and

antigen-based methods to detect Salmonella DNA in a specimen (Cronquist et al., 2012).

Using CIDT, the laboratory does not need to grow a culture of Salmonella, which yields results

much quicker, is more sensitive, and is less expensive (Cronquist et al., 2012). However, as

an isolate is not grown, the Salmonella cannot be further typed. While there may be many

advantages of using CIDT from a clinical perspective, from a public health surveillance

perspective, there are more challenges than advantages, including a loss of subtyping for

outbreak detection and decreased ability to monitor disease trends (Cronquist et al., 2012).

In Australia, CIDT using a multiplex PCR test began to be widely used by primary pathology

laboratories from late 2013 (May et al., 2017). Notification rates of non-typhoidal Salmonella

increased by 22% nationally from 2013 to 2014, which was the largest annual recorded

increase in rate since the National Notifiable Disease Surveillance System began (Department

of Health, 2019). The extent that CIDT had on this rate increase is unknown. A survey in the

United States of FoodNet clinical laboratories between 2012 and 2014 found that only a small

proportion (1.3%) of laboratories were using CIDT, but in those laboratories, 60% of specimens

were not being cultured and 3% did not yield a pathogen when cultured (Iwamoto et al., 2015).

13

The number of laboratories using CIDT increased between 2013 and 2016 from <1% of

laboratories to 14% of laboratories, resulting in 7.6% of infections in 10 USA FoodNet sites

being CIDT positive-only in 2016 (Marder et al., 2017). CIDT without culture will result in a loss

of Salmonella serotyping and other typing information, which presents challenges to public

health surveillance.

2.4.2 Whole genome sequencing

Traditional Salmonella typing methods, such as serotyping and MLVA, have been successfully

used to differentiate salmonellosis cases for public health surveillance and outbreak

investigation, helping to save lives and reduce costs to human health and industry (Scharff et

al., 2016). WGS has emerged as an alternative to these traditional typing methods (Deng et

al., 2016). WGS is a group of laboratory techniques that are used to read and order nucleotides

within a genome (Nature Education, 2014). As sequences contain the entirety of the genetic

information of an organism, WGS provides highly discriminatory data to compare bacterial

foodborne disease strains, such as Salmonella, as well as providing information on

evolutionary context, virulence and antibiotic resistance markers, and in silico prediction of

phenotypic traits (Deng et al., 2016; Jenkins, 2015; Kwong et al., 2009; Public Health England,

2014).

In September 2013, the USA began to use WGS to characterise foodborne disease pathogens,

starting with Listeria monocytogenes (CDC, 2016; Jackson et al., 2016). The US Centers for

Disease Control and Prevention (CDC) found that WGS allowed scientists to detect more

clusters of L. monocytogenes illness, solve more L. monocytogenes outbreaks while they are

still small, link cases to likely food sources, and identify new food sources of infection (CDC,

2016). The US CDC is expanding their WGS program to characterise other foodborne disease

pathogens including Campylobacter, Shiga toxin-producing Escherichia coli (STEC) and

Salmonella (CDC, 2016). They now use WGS for almost all investigations of outbreaks caused

by foodborne bacteria (CDC, 2019c). Like the US CDC, Public Health England has also

implemented WGS as a routine typing tool for foodborne disease pathogens, including

routinely sequencing all Salmonella isolates from April 2015 (Ashton et al., 2016). In addition,

the European Centre for Disease Prevention and Control has committed to moving towards

using WGS as the choice method for typing microbial pathogens over the next five years

(ECDC, 2015, 2016). By mid-2017, while the vast majority of national public health reference

laboratories in EU countries have access to whole genome sequencing, challenges with the

costs and lack of expertise may limit it uses (ECDC 2018). Australia began to use WGS for

national surveillance of L. monocytogenes in 2014 (Kwong et al., 2016) and has used WGS

for Salmonella to investigate outbreaks and for research purposes (Arnott et al., 2018; Graham

et al., 2018; Lindsay et al., 2018; Thompson et al., 2017).

14

As WGS provides a large amount of highly discriminatory data, the use of WGS to differentiate

Salmonella and other foodborne disease pathogens will affect how data are analysed and

interpreted for public health surveillance and public health action. However, while several

public health agencies are using WGS for foodborne disease characterisation and outbreak

investigation, the adoption of WGS has not been systematic.

One challenge for implementing WGS is that there is limited evidence on the effective and

actionable use of WGS data within a public health unit for surveillance. The generation and

use of WGS data for public health action requires the close collaboration between

microbiologists, bioinformaticians and epidemiologists, and using new skills and tools for the

integration of WGS, epidemiological, and clinical data (WHO, 2018a). The receipt of timely and

usable data is also important for public health action.

An advantage of using WGS for surveillance is that it provides more detailed evolutionary data

on foodborne disease isolates than other molecular methods (Ju et al., 2012; Rahaman et al.,

2015). Examining evolutionary framework, dynamics and population structure of pathogens

can help increase the understanding of trends in distribution and infection (Deng et al., 2015),

estimate how fast the disease is spreading, help predict its spread (Neher and Bedford, 2018),

lead to better control measures (Bergholz et al., 2018), and assist with detailed epidemiological

surveillance (Holmes et al., 2015). However, the public health workforce needs the skills and

tools to use evolutionary data acquired from WGS and a sufficient number of isolates need to

be sequenced to provide contextual information in order to understand disease trends for

public health action. While microbiologists and bioinformaticians are best placed to provide

training and create or adopt tools for using WGS data, it must be done in consultation with

epidemiologists and other public health staff to ensure that the training and tools are what is

needed to provide evidence for public health action.

Additionally, WGS offers advantages over traditional typing methods for identifying outbreaks.

A number of studies have shown that WGS offers improved discrimination compared to

traditional typing methods so that outbreaks of foodborne disease can be detected more easily

(Ashton et al., 2015; Inns et al., 2015, 2017; Jackson et al., 2016). Using WGS has shown to

increase confidence that isolates included in genetically related clusters are cases and shown

to provide more precise data to direct public health investigation and response resources

(Holmes et al., 2015; Jensen et al., 2016). Sequence data from food and environmental

isolates can also be very useful in outbreak investigations. The improved discriminatory power

of WGS can provide a higher level of confidence either to implement control and prevention

measures for a food source or point of contamination, or to rule out associations with food

sources in order to prevent unnecessary public health action (Byrne et al., 2014; Dallman et

15

al., 2015; Deng et al., 2016; Jensen et al., 2016). For the most effective use of WGS to detect

outbreaks, timely sequencing data is required (Lindsay et al., 2018; Neher and Bedford, 2018).

The development of clear definitions of clusters and triggers for response is also necessary.

While the highly discriminatory nature of WGS data suggests that using WGS will improve

Salmonella surveillance and outbreak detection, another challenge to turning data into public

health action arises when that data is not standardized or shared across public health

jurisdictions. In Australia, Salmonella surveillance and outbreak investigation is largely

conducted at or below the state and territory level, with national OzFoodNet coordination when

a multi-jurisdictional outbreak is detected. If Salmonella subtyping data are not standardized

and shared, it will be difficult to connect cases across jurisdictions and link cases with

Salmonella isolates detected in the food supply. WGS data could also be shared internationally

if standardized, however this requires data storage capabilities and agreement from the global

community to share WGS and metadata.

There is a general consensus in the literature that the costs of WGS are decreasing, however

it is usually more expensive than other Salmonella subtyping methods. In addition, the initial

equipment costs and continuing reagent and technician salary costs are prohibitive for many

laboratories and unable to be adopted independently of new funding (Reuter et al., 2013, Allard

et al., 2016, Taylor et al., 2015). WGS also incurs costs for storage, quality checks, and data

analysis, which may be covered by NCBI if countries choose to use it (Allard et al., 2016);

however the need for local storage and analysis capacity is likely. While it is believed that

eventually the cost of WGS will be uniformly affordable (Rahaman et al., 2015), additional start-

up funding and sustainable funding in the long term are needed for the data to be collected for

public health surveillance. If the highly discriminatory WGS data can drive public health action

to prevent illness, it may be cost-competitive with traditional typing methods, and even CIDT.

2.5 Conclusion

Non-typhoidal Salmonella is an important foodborne illness globally and in Australia. As the

rate of non-typhoidal Salmonella in Australia is increasing and is one of the highest rates in the

industrialized world, public health surveillance, outbreak investigation, and interventions for

prevention and control of illness are important to reduce the burden in the Australian

community. Understanding the trends of Salmonella serotypes and sources of infection can

help to develop more targeted interventions. In addition, quantifying the cost of illness can help

in determining the cost effectiveness of surveillance systems and interventions. Finally, with

the emergence of CIDT and WGS, understanding the impacts of these technologies will help

with effective implementation into public health surveillance systems.

16

Chapter 3. Trends of non-typhoidal Salmonella enterica in

Australia

3.1 Introduction

In this chapter, I investigate the trends of non-typhoidal Salmonella enterica notifications in

Australia between 2000 and 2013. Salmonella is a significant cause of morbidity in Australia,

and unlike in many other places in the world, Salmonella Typhimurium the most commonly

notified serotype and etiological agent in outbreaks in Australia. While publically available data

from the National Notifiable Diseases Surveillance System (NNDSS) suggested that the rate

of Salmonella infections in the Australian population had increased over 2000-2013, trends in

state, age, and state and territory by S. Typhimurium and non-Typhimurium Salmonella

notifications had not previously been examined. I found that almost all states and territories

had significantly increasing trends, with annual increases of notification rates as high as 12%

(95% confidence interval 10-14%) for S. Typhimurium in the Australian Capital Territory and

6% (95% CI 5-7%) for non-Typhimurium Salmonella in Victoria. The reasons for this increase

are unknown, with no known significant changes to the surveillance system during the time

period, culture-independent diagnostic testing was not used until the end of 2013, and The

Primary Production and Processing Standard for Poultry commenced in 2012.

This work was published as an open access research article in PLoS One. The paper

contributes to an improved understanding of the epidemiology of Salmonella in Australia by

investigating geographic differences in rates, trends and serotypes, and generating

hypotheses on transmission pathways of certain serotypes for further investigation.

Supplementary tables and figures published online as a supplement to the paper can be found

in Appendix 1.

3.2 Paper

Ford L, Glass K, Veitch M, Wardell R, Polkinghorne B, Dobbins T, Lal A, Kirk MD. Increasing

incidence of Salmonella in Australia, 2000-2013. PLoS One. 2016;11(10): e0163989, doi:

10.1371/journal.pone.0163989.

RESEARCH ARTICLE

Increasing Incidence of Salmonella inAustralia, 2000-2013Laura Ford1*, Kathryn Glass1, Mark Veitch2, Rebecca Wardell1, Ben Polkinghorne3,

Timothy Dobbins1, Aparna Lal1, Martyn D. Kirk1

1 National Centre for Epidemiology and Population Health, The Australian National University, Canberra,

Australian Capital Territory (ACT), Australia, 2 Department of Health and Human Services, Hobart,

Tasmania (Tas.), Australia, 3 OzFoodNet, Australian Government Department of Health, Canberra,

Australian Capital Territory (ACT), Australia

* [email protected]

AbstractSalmonella is a key cause of foodborne gastroenteritis in Australia and case numbers are

increasing. We used negative binomial regression to analyze national surveillance data for

2000–2013, for Salmonella Typhimurium and non-Typhimurium Salmonella serovars. We

estimated incidence rate ratios adjusted for sex and age to show trends over time. Almost

all states and territories had significantly increasing trends of reported infection for S. Typhi-

murium, with states and territories reporting annual increases as high as 12% (95% confi-

dence interval 10–14%) for S. Typhimurium in the Australian Capital Territory and 6% (95%

CI 5–7%) for non-Typhimurium Salmonella in Victoria. S. Typhimurium notification rates

were higher than non-Typhimurium Salmonella rates in most age groups in the south east-

ern states of Australia, while non-Typhimurium rates were higher in most age groups else-

where. The S. Typhimurium notification rate peaked at 12–23 months of age and the non-

Typhimurium Salmonella notification rate peaked at 0–11 months of age. The age-specific

pattern of S. Typhimurium cases suggests a foodborne origin, while the age and geo-

graphic pattern for non-Typhimurium may indicate that other transmission routes play a key

role for these serovars.

Introduction

Salmonella enterica is transmitted via food, the environment, water, people and animals, andoften causes gastroenteritis in humans [1, 2]. Worldwide, Salmonella infections, excludingthose caused by S. Typhi and S. Paratyphi, were estimated in a paper published in 2010 tocause 93.8 million (90% credible interval 61.8–131.6 million) cases of gastroenteritis per year,80.3 million of which are considered foodborne [3]. Approximately 72% of salmonellosis inAustralia is estimated to be transmitted through contaminated food [1]. Common foods associ-ated with salmonellosis in outbreak investigations and source attribution studies include eggs,poultrymeat, pork, beef, dairy products, nuts, and fresh produce [4, 5, 6, 7].

PLOS ONE | DOI:10.1371/journal.pone.0163989 October 12, 2016 1 / 11

a11111

OPENACCESS

Citation: Ford L, Glass K, Veitch M, Wardell R,

Polkinghorne B, Dobbins T, et al. (2016) Increasing

Incidence of Salmonella in Australia, 2000-2013.

PLoS ONE 11(10): e0163989. doi:10.1371/journal.

pone.0163989

Editor: Dipshikha Chakravortty, Indian Institute of

Science, INDIA

Received: May 11, 2016

Accepted: August 27, 2016

Published: October 12, 2016

Copyright:© 2016 Ford et al. This is an open

access article distributed under the terms of the

Creative Commons Attribution License, which

permits unrestricted use, distribution, and

reproduction in any medium, provided the original

author and source are credited.

Data Availability Statement: The data that

underlies the findings of this study is National

Notifiable Disease Surveillance System data. We

are unable to upload the minimal data set used in

this study, as we obtained the data from a third

party. Australian Salmonella notification data, as

was used in this study, can be requested from the

Communicable Disease Network Australia. Data

requests can be sent to [email protected].

Funding: This work was funded by the Australian

Research Council (ARC Linkage Grant

LP110200431).

17

There are over 2,500 different serovars of Salmonella [8]. The most common serovar in Aus-tralia is S. Typhimurium, which is also the most commonly identified etiological agent in out-breaks [7, 9]. S. Enteritidis is not endemic in Australian poultry layer flocks and most humaninfections with S. Enteritidis are acquired overseas [7]. Many other serovars occupy distinctecological niches and epidemiological foci in Australia, as it is a large country with wide cli-matic and geo-physical variation [7,10].

In Australia, all laboratory confirmed Salmonella infections are reported to state and terri-tory health departments, and subsequently to the National Notifiable Diseases SurveillanceSystem (NNDSS) [11]. Surveillance data are an under representation of the total burden of sal-monellosis cases, with an estimated 7 salmonellosis cases (95% CI 4–16) occurring in the com-munity for every 1 notification to health departments [12]. Circa 2010, there were an estimated40,000 salmonellosis cases attributable to contaminated food in the Australian communityeach year [13]. In addition, there were an estimated 2,100 hospitalizations, 15 deaths and 6,750complications from contaminated food [13,14].

National surveillance figures over the last decade suggest the rate of Salmonella infectionshas been increasing [15]. We compared the rates of Salmonella infections between sex, agegroups, and Australian states and territories over 2000–2013 in order to examine trends in thereported incidence of infection and identify differences between states and territories.

Materials and Methods

In this study, we used national Australian human salmonellosis notification data to analyze dis-ease trends by state and territory from 2000 to 2013. In Australia, there are six states: NewSouthWales (NSW), Queensland (Qld), South Australia (SA), Tasmania (Tas.), Victoria(Vic.), and Western Australia (WA); and two territories: the Australian Capital Territory(ACT) and the Northern Territory (NT) (Fig 1). The climate and environment varies widelyboth across and within these states and territories.We chose to examine Salmonella trends at astate and territory level due to the availability of data at that level, and evidence that suggeststhat the frequency of Salmonella serovars differs by state and territory [7].

Ethics approval for this study was granted by the Australian National University HumanResearch Ethics Committee [protocol 2012/412].

Data sources

All states and territories have public health legislation that requires doctors and/or pathologylaboratories to report any confirmed cases of salmonellosis [17]. A confirmed case requiresdefinitive laboratory evidence of the isolation or detection of Salmonella species, excluding S.Typhi, which is notified separately [18]. State-based surveillance systems have collected dataand entered all confirmed salmonellosis cases into NNDSS since 1991. We requested de-identi-fied Salmonella spp. NNDSS notifications (including ‘notification receive date’, ‘true onsetdate’, ‘diagnosis date’, ‘age at onset’, ‘sex’, ‘organism’, and ‘serogroup’ fields) for each state andterritory for 1991 to 2013 from the Communicable Disease Network of Australia (CDNA). Weused ‘diagnosis date’ for all analyses, which is defined as the date a person’s illness began(onset), or where onset date is unknown, the earliest of the specimen collection date, the datethe health professional signed the notification form or the laboratory issued the results, or thedate the notification is received by the communicable diseases section of the health authority.Due to incompleteness of serotyping data in the NNDSS in the 1990s, we restricted the data foranalysis to cases with diagnosis dates from 2000 to 2013 inclusive.

Increasing Salmonellosis in Australia, 2000-2013


Competing Interests: The authors have declared

that no competing interests exist.

18

Rates of illness per 100,000 population were calculated using the estimated resident popula-tion by age and sex for each state and territory as of the June quarter for each year between2000 and 2013 from the Australian Bureau of Statistics (ABS) [19].

Analysis

The primary aim of the analysis was to examine the trends in reported incidence over the timeperiod.We excluded all Salmonella cases where the serovar, age, or sex was missing.We alsoexcluded infections due to S. Paratyphi A, S. Paratyphi B (except for bioser Java), and S. Paraty-phi C, as they predominantly result in enteric fever and are acquired while traveling overseas.S. Typhimurium and non-Typhimurium Salmonella were analyzed separately to examine thetrends of Australia’s most common serovar, S. Typhimurium, compared with trends of allother types of non-typhoidal Salmonella notified in Australia. Less than 0.1% of notificationswere typed as S. subspecies enterica, which were grouped with S. Typhimurium if they had anH = i in the antigenic formula or a known Typhimurium phage type (commonly known asmonophasic S. Typhimurium). We used negative binomial regression to estimate incidencerate ratios (IRR) by sex, age, and state and territory. We also included an interaction term inthe model to estimate IRRs for trend over time by state and territory. ABS population numbersby age, sex and state and territory were used as an offset to standardize the incidence rates to

Fig 1. Map of Australian States and Territories showing the crude notification rate of salmonellosis for 2013 after excluding cases with

missing data on serovar, age, or sex. Administrative boundaries from the Australian Bureau of Statistics [16].

doi:10.1371/journal.pone.0163989.g001



19

the population.We used the interaction between state/territory and year to produce a trendover time for each state and territory, with year defined as the year of diagnosis and treated as acontinuous variable. A state-by-state analysis showed that treating year as a continuous vari-able was appropriate for states and territory data (NT excepted—see S1 Fig). Age was catego-rized in one-year age groups from 0 to 4 and in 5-year age groups until 85 years and over.Analysis was performedwith Stata statistical package 12.1 (www.stata.com/) and graphs weremade using Stata and Microsoft Excel 2010. ArcGIS v10.3 (http://www.esri.com/software/arcgis) was used to create a map of the crude annual salmonellosis notification rate in 2013.

Sensitivity analysis

NNDSS data does not allow us to distinguish between sporadic and outbreak cases. To test theeffect of years with large outbreaks or a large number of sporadic cases on our rate ratio esti-mates, we removed outlier years of 2009 in ACT, 2011 in SA, and 2005 in Tas. from the S.Typhimurium analysis. An outlier year was defined as any year where there was an absolutedifference of 10 per 100,000 or greater between the crude rate and the predicted rate.

Results

There were 127,195 cases of salmonellosis reported to NNDSS with a diagnosis date between2000 and 2013. Of these notifications, 97.7% (124,235/127,195) included serovar data (S1Table) and of those, 99.6% (123,762/124,235) included age and sex data. Nationally, the crudeannual rate was lowest in 2000 (30.6 per 100,000) and increased to 53.0 per 100,000 in 2013 (S2Table). Fig 1 shows the geography of the Australian states and territories, together with thecrude notification rate of salmonellosis for 2013 by state.

Rates of both S. Typhimurium and non-Typhimurium Salmonella increased from 2000 to2013 with an IRR of 1.06 (95% CI 1.06–1.07) for S. Typhimurium and 1.03 (95% CI 1.02–1.03)for non-Typhimurium Salmonella (Fig 2). S. Typhimurium was the most frequently reportedserovar and was responsible for 43.9% (54,313/123,762) of notifications over the time period. S.Enteritidis was responsible for 5.7% (7,001/123,762) of notifications. The proportion of notifi-cations of S. Typhimurium and common non-Typhimurium serovars varied by state and terri-tory (S3 Table).

Of the salmonellosis notifications included in this analysis, 49.7% (61,557/123,762) were inmales. In the state and territorymodel, there was no significant difference between the sexesfor infectionwith non-Typhimurium Salmonella, but we found differences for S. Typhimuriuminfectionwith lower rates in males (Table 1; IRR 0.93; 95% CI 0.91–0.95). The median age at

Fig 2. Crude notification rates (dots) and negative binomial regression margins plot (lines with 95% CI) of S.

Typhimurium and non-Typhimurium Salmonella notification rates, Australia 2000–2013.




20

onset of salmonellosis cases included in this study was 22 years old (range<1–108 years).Non-Typhimurium Salmonella incidence rates peaked at 0–11 months of age and S. Typhi-murium incidence rates peaked at 1–2 years of age, with a relative increase in the IRR for S.Typhimurium in those 80 years and older (Fig 3). The age distributions of cases by states andterritories were consistent with this national age distribution (S2 Fig) although there were dif-ferences in the proportion of all salmonellosis that is due to S. Typhimurium. S. Typhimuriumrates were higher than non-Typhimurium rates in most age groups in the south eastern conti-nental states (ACT, NSW, Vic., and SA), while non-Typhimurium Salmonella rates were higher

Table 1. Incident rate ratios calculated using negative binomial regression of S. Typhimurium and non-Typhimurium Salmonella by gender, age,

state and time, 2000–2013.

Typhimurium Non-Typhimurium

IRRa 95% CI b P-value IRRa 95% CIb P-value

Sex (reference = female)

Male 0.93 0.91–0.95 <0.001 1.01 0.98–1.03 0.60

Age groups (reference = <1)

1 1.28 1.18–1.39 <0.001 0.77 0.72–0.83 <0.001

2 1.04 0.96–1.13 0.35 0.33 0.31–0.36 <0.001

3 0.89 0.82–0.97 0.01 0.19 0.18–0.21 <0.001

4 0.69 0.64–0.76 <0.001 0.13 0.12–0.14 <0.001

5–9 0.41 0.38–0.45 <0.001 0.08 0.07–0.08 <0.001

10–14 0.26 0.24–0.28 <0.001 0.05 0.05–0.06 <0.001

15–19 0.27 0.25–0.29 <0.001 0.06 0.06–0.07 <0.001

20–24 0.33 0.30–0.35 <0.001 0.10 0.09–0.11 <0.001

25–29 0.29 0.27–0.31 <0.001 0.10 0.09–0.10 <0.001

30–34 0.23 0.21–0.24 <0.001 0.08 0.07–0.08 <0.001

35–39 0.17 0.16–0.19 <0.001 0.06 0.05–0.06 <0.001

40–44 0.16 0.14–0.17 <0.001 0.06 0.05–0.06 <0.001

45–49 0.14 0.13–0.16 <0.001 0.07 0.06–0.07 <0.001

50–54 0.15 0.13–0.16 <0.001 0.07 0.06–0.07 <0.001

55–59 0.14 0.13–0.16 <0.001 0.07 0.06–0.07 <0.001

60–64 0.15 0.13–0.16 <0.001 0.08 0.07–0.08 <0.001

65–69 0.16 0.14–0.17 <0.001 0.08 0.07–0.08 <0.001

70–74 0.17 0.15–0.18 <0.001 0.08 0.07–0.09 <0.001

75–79 0.17 0.15–0.19 <0.001 0.08 0.07–0.08 <0.001

80–84 0.20 0.18–0.23 <0.001 0.09 0.08–0.10 <0.001

85+ 0.22 0.20–0.24 <0.001 0.08 0.07–0.09 <0.001

Trend over time by state and territory (2000–2013)

ACT 1.12 1.10–1.14 <0.001 1.03 1.01–1.05 <0.01

NSW 1.07 1.07–1.08 <0.001 1.04 1.04–1.05 <0.001

NT 1.03 1.01–1.05 <0.01 1.02 1.00–1.03 <0.01

QLD 1.04 1.03–1.05 <0.001 0.99 0.98–1.00 0.01

SA 1.05 1.04–1.06 <0.001 1.05 1.04–1.06 <0.001

TAS 1.04 1.02–1.06 <0.001 1.04 1.02–1.05 <0.001

VIC 1.08 1.08–1.09 <0.001 1.06 1.05–1.07 <0.001

WA 1.01 1.00–1.02 0.02 1.04 1.03–1.05 <0.001

aIncidence rate ratiobConfidence interval

doi:10.1371/journal.pone.0163989.t001



21

in most age groups elsewhere. The two northern-most Australian states (NT and Qld.) hadhigher rates of infection in 0–4 year old children than other states and territories (S2 Fig).

At the start of the period, the incidence rate of S. Typhimurium was similar across the east-ern states and territories but was significantly higher in the western states—WA, NT and SA(Fig 4). Similarly, non-Typhimurium Salmonella rates were significantly higher in the northernand western states and territories (NT, Qld, WA, and SA) and in Tasmania than in the southeastern continental states and territories. From 2000 to 2013, the incidence rates of S. Typhi-murium significantly increased in all states and territories with the highest incidence rate ratiosin the ACT (IRR 1.12; 95% CI 1.10–1.14) and Vic. (IRR 1.08; 95% CI 1.08–1.09). During thesame time period, incidence rates of non-Typhimurium Salmonella significantly increased inall states and territories, except Queensland (IRR 0.99; 95% CI 0.98–1.00).While significant,the regression model for non-Typhimurium Salmonella in the NT did not fit the data well, andsuggests a more complex pattern over time in this territory (S1 Fig). Trends in ACT, SA, andTas. remained significant when outlier years were removed from the analysis. Individual stateand territory regression lines plotted against the crude rates for both S. Typhimurium andnon-Typhimurium Salmonella can be found in S1 Fig.

Discussion

Rates of reported salmonellosis have increased significantly in Australia during the last decadeand are at unprecedented levels. In particular, S. Typhimurium was responsible for over 40% ofnotifications and is increasing in all states and territories. There is a need for Australian healthauthorities to identify the key sources of salmonellosis serovars in different states and territo-ries to identify effectiveways to substantially reduce infection and improve control of thisemerging infection.

Fig 3. Salmonella Typhimurium and non-Typhimurium predicted notification rates with 95% CI per 100,000 by

age group, Australia, 2000–2013. Note: the first five age groups are single years to highlight the pattern of Salmonella in

young children.




22

The relative importance of S. Typhimurium varies by state and territory, with higher ratesof non-Typhimurium Salmonella in tropical areas. In the Northern Territory, where the major-ity of people live in a tropical climatic zone, the rate of non-Typhimurium Salmonella wasabout 10 times higher than in most other states and territories. Queensland andWestern Aus-tralia—the only other states to have regions with tropical climatic zones—had the next highestrates of non-Typhimurium Salmonella. Analyzing specific non-Typhimurium serovars in thesestates and territories over the time periodmay help to elucidate this pattern and assist withidentifying priority serovars for investigation.

The higher incidence of S. Typhimurium in the 1–2 year old age group, compared to thoseunder 1, may indicate the importance of foodborne transmission for S. Typhimurium. This isconsistent with the results of foodborne outbreak investigations, with 92% (56/61) of Salmo-nella foodborne outbreaks attributed to S. Typhimurium in Australia in 2011 [7]. Of these, aspecific food vehicle was identified in 73% (41/56) outbreaks and 71% (29/41) were associatedwith the consumption of eggs and egg-baseddishes [7]. In addition, over the time period ofthis study, there has been a significant increase in the number of Salmonella outbreaks associ-ated with eggs, with S. Typhimurium responsible for nearly all egg-related outbreaks [20]. Ofthe 5 non-Typhimurium Salmonella outbreaks in 2011, 80% (4/5) had a known food vehicle[7]. Poultry was the responsible vehicle in 75% (3/4) of these outbreaks and the remaining out-break with a known food vehicle was associated with fruit [7]. The higher incidence of non-

Fig 4. Negative binomial regression margins plot of S. Typhimurium and non-Typhimurium

predicted notification rates by state and territory, Australia 2000–2013.




23

Typhimurium Salmonella in the 0–11 month age group than in any other age group suggeststhat these serovars may have other transmission pathways, such as environmental, waterborneand zoonotic routes. An Australian expert elicitation estimated that 15% of all Salmonella ill-ness is due to environmental sources, 5% is transmitted through contaminated water, and 4%is zoonotic [1]. For example, a case control study of S. Mississippi in Tasmania found indirectcontact with native animal birds, untreated drinkingwater and travel within the state as signifi-cant predictors of infection [10] and a study in the Northern Territory found a number of non-Typhimuirum Salmonella serovars in household environments, including in animal faeces,soil, and vaccum cleaner contents [21].

Unlike in the US, Canada, China, and most of the European Union (EU), S. Enteritidis isless common than S. Typhimurium in Australia and only makes up a small proportion of noti-fications (5.7%) [22–25]. The increasing trend of salmonellosis seen in Australian states andterritories also differs from the US and European experience.Although there was no change inthe overall rate of Salmonella in 2012, compared to 1996–1998 and 2006–2008, the US saw adecrease in S. Typhimurium during this time period, as well as a decrease in the overall rate ofSalmonella in 2013 compared to 2010–2012 [22,26]. The EU has shown a significantly decreas-ing trend of Salmonella notifications from 2008–2012. Crude notification rates of salmonellosisare higher in Australia, with a rate of 49.5 per 100,000 in 2012, compared to 22.2 (range 1.8–97.5) cases per 100,000 in the EU in 2012 and 15.9 cases per 100,000 in the US in 2013 [22, 24].While Canada has seen increases in their overall rate of Salmonella between 2003 and 2009,rates were lower (16.3–18 per 100,000) than in Australia and the increase has largely been dueto S. Enteritidis [23].

There were several limitations to our analysis. We were unable to remove travel associatedcases from the data; however a previous study of data from circa 2010 found that approxi-mately 15% of Salmonella notifications were travel associated [13]. Our unit of analysis wasstate and territory, so we were unable to examine trends at a finer scale or by climatic zones. Inaddition, we could not distinguish between outbreak and sporadic cases.We could not accountfor health-seeking behavior or increases in testing, which may have influenced the IRRs of S.Typhimurium and non-Typhimurium Salmonella. Methods for further characterization of S.Typhimurium isolates vary across states and territories and including Pulse-Field Gel electro-phoresis (PFGE), multilocus variable number tandem repeat analysis (MLVA), and phage typ-ing [27–29]. Two percent of all Salmonella notifications were monophasic S. subspecies Iisolates, some of which were classifiedwith S. Typhimurium based on their serotyping andphage typing characteristics. If phage typing was not performed this may have resulted in asmall number of misclassifications in the likely serovar category. Although the method of test-ing for Salmonella remained relatively consistent over the study time period [30], we wereunable to account for any changes that may have occurred in the rate of testing.

In Australia, culture has been the mainstay of clinical diagnostic testing for Salmonella;however, there has been increased adoption of culture-independent testing in diagnostic labs.Subsequent to this analysis, in 2014, the first full year that culture-independent diagnostic test-ing (CIDT) was widely used in Australian diagnostic labs, notification rates of salmonellosisincreased by 22% nationally from 2013 to 69.7 notifications per 100,000 [15]. This is the largestrecorded annual increase in rate since notifications began and the extent of the impact CIDThad on this rate is unknown. Although CIDT provides quicker results and can be cheaper thanculture, an isolate is needed for further characterization [31]. A survey of FoodNet clinical lab-oratories in the US found that while only a small proportion of laboratories surveyed (1.3%)were using CIDT, a concerning amount of specimens were either not being cultured (60%) orculture did not yield a pathogen (3%) [32]. As whole genome sequencing (WGS), which offershighly discriminatorymolecularmarkers for cluster detection [33], becomes available for



24

routine Salmonella typing, it will be important that isolates continue to be cultured and consis-tent characterizationmethods are used. CIDT without culture results in a loss of data regardingthe Salmonella serovar and other typing information, which presents challenges for outbreakidentification and monitoring disease burden and trends. Further research onWGS in the Aus-tralian setting will help determine whetherWGS can contribute to effectively detecting Salmo-nella clusters for investigation.

This study provides important insights into the epidemiology of Salmonella infections inAustralian states and territories.We observed sustained increases in both S. Typhimurium andnon-Typhimurium Salmonella between 2000 and 2013, with geographic differences in bothrates and trends. With the increasing use of CIDT, meaningful comparisons in disease ratesover timemay becomemore difficult; however novel typing methods such as whole genomesequencing offers the potential for a richer understanding of salmonellosis in Australia. Thisimproved understanding is needed to inform the control of salmonellosis in Australia.

Supporting Information

S1 Fig. State and territory crude (dots) and predicted (lines with 95% CI) notification ratesper 100,000 persons, Australia 2000–2013.(PDF)

S2 Fig. Salmonella Typhimurium and non-Typhimurium predictednotification rates per100,000 (with 95% CI) by age group for each State and Territory, 2000–2013.(PDF)

S1 Table. Number and proportion of Salmonella notificationswithout serovar data by stateand territory, Australia 2000–2013.(DOCX)

S2 Table. Salmonella spp. cases each year, the proportion of cases excluded due to missingserovar, age or sex data, crude notification rate after exclusions, S. Typhimurium notifica-tion rate, and Non-Typhimurium notification rate (per 100,000 persons, Australia 2000–2013).(DOCX)

S3 Table. Proportion (%) of S. Typhimurium and the 20 most notified non-TyphimuriumSalmonella serovars of total notifications included in this study for each state and territory,Australia, 2000–2013.(DOCX)

Acknowledgments

We would like to thank the Australian Research Council Linkage Project Team: Food Stan-dards Australia New Zealand, New SouthWales Food Authority, South Australia Health,South Australia Pathology, MicrobiologicalDiagnostic Unit Public Health Laboratory, Biose-curity South Australia, Tasmania Department of Health and Human Services, and HunterNew England Area Health.

We would also like to thank the laboratories that performed serotyping and phage typing ofSalmonella, along with the Communicable Disease Network of Australia and federal, State andTerritory health departments.

Finally, we would like to thankMary Valcanis, MicrobiologicalDiagnostic Unit PublicHealth Laboratory for her advice and expertise on grouping serovars.



25

Author Contributions

Conceptualization:LF KG MV RW BPMK.

Formal analysis: LF KG RW AL.

Methodology:LF KG MV TDMK.

Writing – original draft: LF RW.

Writing – review& editing: LF KG MV RW BP TD ALMK.

References1. Vally H, Glass K, Ford L, Hall G, Kirk MD, Shadbolt C, et al. Proportion of illness acquired by foodborne

transmission for nine enteric pathogens in Australia: an expert elicitation. Foodborne Pathog Dis.

2014; 11(9):727–33. doi: 10.1089/fpd.2014.1746 PMID: 25072416

2. Fierer J, Guiney DG. Diverse virulence traits underlying different clinical outcomes of Salmonella infec-

tion. J Clin Invest. 2001; 107(7):775–80. doi: 10.1172/JCI12561 PMID: 11285291

3. Majowicz SE, Musto J, Scallan E, Angulo FJ, Kirk M, O’Brien SJ, et al. The global burden of nontyphoi-

dal Salmonella gastroenteritis. Clin Infect Dis. 2010; 50(6):882–9. doi: 10.1086/650733 PMID:

20158401

4. Glass K, Fearnley E, Hocking H, Raupach J, Veitch M, Ford L, et al. Baysian source attribution of sal-

monellosis in South Australia. Risk Anal. 2016; 36(3):561–70. doi: 10.1111/risa.12444 PMID:

26133008

5. Mullner P, Jones G, Nobel A, Spencer SE, Hathaway S, French NP. Source attribution of food-borne

zoonoses in New Zealand: a modified Hald model. Risk Anal. 2009; 29(7):970–84. doi: 10.1111/j.

1539-6924.2009.01224.x PMID: 19486473

6. Hald T, Vose D, Wegener HC, Koupeev T. A Bayesian approach to quantify the contribution of animal-

food sources to human salmonellosis. Risk Anal. 2004; 24(1):255–69. doi: 10.1111/j.0272-4332.2004.

00427.x PMID: 15028016

7. The OzFoodNet Working Group. Monitoring the incidence and causes of diseases potentially transmit-

ted by food in Australia: annual report of the OzFoodNet network, 2011. Commun Dis Intell Q Rep.

2015; 39(2):E236–64. PMID: 26234259

8. Jones TF, Ingram LA, Cieslak PR, Vugia DJ, Tobin-D’Angelo M, Hurd S, et al. Salmonellosis outcomes

differ substantially by serotype. J Infect Dis. 2008; 198(1):109–14. doi: 10.1086/588823 PMID:

18462137

9. The OzFoodNet Working Group. Monitoring the incidence and causes of diseases potentially transmit-

ted by food in Australia: annual report of the OzFoodNet network, 2010. Commun Dis Intell Q Rep.

2012; 36(3):E213–41. PMID: 23186234

10. Ashbolt R, Kirk MD. Salmonella Mississippi infections in Tasmania: the role of native Australian ani-

mals and untreated drinking water. Epidemiol Infect. 2006; 134(6):1257–65. doi: 10.1017/

S0950268806006224 PMID: 16672107

11. The Department of Health. Australian national notifiable diseases and case definitions [Internet]. Can-

berra (ACT): Commonwealth of Australia; [updated 2015 Jul 1; cited 2015 Jul 6]. Available: http://

www.health.gov.au/casedefinitions.

12. Hall G, Yohannes K, Raupach J, Becker N, Kirk M. Estimating community incidence of Salmonella,

Campylobacter, and Shiga toxin-producing Escherichia coli infections, Australia. Emerg Infect Dis

2008; 14(10):1601–9. doi: 10.3201/eid1410.071042 PMID: 18826825

13. Kirk M, Ford L, Glass K, Hall G. Foodborne illness, Australia, circa 2000 and circa 2010. Emerg Infect

Dis. 2014; 20(11):1857–64. doi: 10.3201/eid2011.131315 PMID: 25340705

14. Ford L, Kirk M, Glass K, Hall G. Sequelae of foodborne illness caused by 5 pathogens, Australia, circa

2010. Emerg Infect Dis. 2014; 20(11):1865–71. doi: 10.3201/eid2011.131316 PMID: 25340885

15. The Department of Health. National Notifiable Diseases Surveillance System [Internet]. Canberra

(ACT): Commonwealth of Australia; [updated 2015 Jul 6; cited 2015 Jul 6]; Available: http://www9.

health.gov.au/cda/source/cda-index.cfm.

16. Australian Bureau of Statistics [Internet]. 1259.0.30.002 –Statistical Geography—Australian Standard

Geographical Classification (ASGC), Digital Boundaries, 2006. Canberra (ACT): Commonwealth of

Australia 2013; [2006 Jul 14; cited 2016 Jul 28]; Available: https://www.abs.gov.au/AUSSTATS/abs@.

nsf/Lookup/1259.0.30.002Main+Features12006?OpenDocument.



26

17. The Department of Health. Links to State and Territory Public Health Legislation, the Quarantine Act,

and the National Health Security Act 2007 [Internet]. Canberra (ACT): Commonwealth of Australia;

[updated 2009 Feb 12; cited 2015 Jan 6]; Available: http://www.health.gov.au/internet/main/

publishing.nsf/Content/cda-state-legislation-links.htm.

18. The Department of Health. Salmonellosis case definition [Internet]. Canberra (ACT): Commonwealth

of Australia; [updated 2004; cited 2015 Jan 6]; Available: http://www.health.gov.au/internet/main/

publishing.nsf/Content/cda-surveil-nndss-casedefs-cd_salmon.htm.

19. Australian Bureau of Statistics [Internet]. 3101.0—Australian Demographic Statistics, Dec 2013. Can-

berra (ACT): Commonwealth of Australia 2013; [2014 Jun 19; cited 2014 Sep 23]; Available: http://www.

abs.gov.au/AUSSTATS/[email protected]/allprimarymainfeatures/FA627CA7C5708380CA257D5D0015EB95?

opendocument.

20. Moffatt CRM, Musto J, Pingault N, Miller M, Stafford R, Gregory J et al. Salmonella Typhimurium and

outbreaks of egg-associated disease in Australia, 2001 to 2011. Foodborne Pathog Dis. 2016 Mar 30

[Epub ahead of print]. doi: 10.1089/fpd.2015.2110 PMID: 27028267

21. Williams S, Patel M, Markey P, Muller R, Benedict S, Ross I, Heuzenroeder M, Davos D, Cameron S,

Krause V. Salmonella in the tropical household environment—Everyday everywhere. J Infect. 2015;

71(6): 642–8. doi: 10.1016/j.jinf.2015.09.011 PMID: 26416474

22. Crim SM, Iwamoto M, Huang JY, Griffin PM, Gilliss D, Cronquist AB, et al. Incidence and trends of

infection with pathogens transmitted commonly through food—Foodborne Diseases Active Surveil-

lance Network, 10 U.S. sites, 2006–2013. MMWR Morb Mortal Wkly Rep. 2014; 63(15):328–32.

PMID: 24739341

23. Nesbitt A, Ravel A, Murray R, McCormick R, Savelli C, Finley R, et al. Integrated surveillance and

potential sources of Salmonella enteritidis in human cases in Canada from 2003 to 2009. Epidemiol

Infect. 2012; 140(10):1757–72. doi: 10.1017/S0950268811002548 PMID: 22166269

24. European Food Safety Authority, European Centre for Disease Prevention and Control. The European

Union Summary Report on Trends and Sources of Zoonoses, Zoonotic Agents and Food-borne Out-

breaks in 2012. EFSA Journal. 2014; 12(2):3547. doi: 10.2903/j.efsa.2014.3547

25. Ran L, Wu S, Gao Y, Zhang X, Feng Z, Wang Z, et al. Laboratory-based surveillance of nontyphoidal

Salmonella infections in China. Foodborne Pathog Dis. 2011; 8(8):921–7. doi: 10.1089/fpd.2010.0827

PMID: 21492026

26. Centers for Disease Control and Prevention (CDC). Incidence and trends of infection with pathogens

transmitted commonly through food—foodborne diseases active surveillance network, 10 U.S. sites,

1996–2012. MMWR Morb Mortal Wkly Rep. 2013; 62(15):283–7. PMID: 23594684

27. Fullerton K, The OzFoodNet Working Group. Monitoring the incidence and causes of diseases poten-

tially transmitted by food in Australia: annual report of the OzFoodNet Network, 2007. Commun Dis

Intell Q Rep. 2008; 32(4):400–24. PMID: 19378479

28. Gilbert GL. Using MLVA to type strains of Salmonella Typhimurium in New South Wales. NSW Public

Health Bull. 2008; 19(1–2):29–31. doi: 10.1071/NB07116 PMID: 18361866

29. Queensland OzFoodNet Site. OzFoodNet—Enhancing Foodborne Disease Surveillance Across Aus-

tralia. Annual Report 2010: Queensland. Brisbane (Qld): Queensland Health; 2010.

30. The Department of Health. Salmonella laboratory case definition (LCD) [Internet]. Canberra (ACT):

Commonwealth of Australia; [updated 2005; cited 2016 Aug 8]; Available: http://www.health.gov.au/

internet/main/publishing/nsf/Content/cda-phlncd-salmonella.htm.

31. Cronquist AB, Mody RK, Atkinson R, Besser J, Tobin D’Angelo M, Hurd S, et al. Impacts of culture-

independent diagnositic practices on public health surveillance for bacterial enteric pathogens. Clin

Infect Dis. 2012; 54(Suppl 5):S432–9. doi: 10.1093/cid/cis267 PMID: 22572666

32. Iwamoto M, Huang JY, Cronquist AB, Medus C, Hurd S, Zansky, et al. Bacterial enteric infections

detected by culture-independent diagnositic tests—FoodNet, United States 2012–2014. MMWR Morb

Mortal Wkly Rep. 2015; 64(9):252–7. PMID: 25763878

33. Leekitcharoenphon P, Nielsen EM, Kaas RS, Lund O, Aarestrup FM. Evaluation of whole genome

sequencing for outbreak detection of Salmonella enterica. PLoS One 2014; 9(2):e87991. doi: 10.1371/

journal.pone.0087991 PMID: 24505344



27

28

Chapter 4. Sources of non-typhoidal Salmonella enterica

in Australia

4.1 Introduction

This chapter documents the trends and sources of infection in Salmonella outbreaks in

Australia between 2001 and 2016. Over this time period, 79% of outbreaks were due to, or

suspected to be due to contaminated foods, with eggs or egg-containing foods as the most

common food vehicle in outbreaks. While few outbreaks were associated with fresh produce

(such as fruits, sprouts, some salad items, and nuts) they were generally larger than outbreaks

associated with other food vehicles.

This work contributes to an improved understanding of the epidemiology of Salmonella in

Australia by reviewing the transmission pathways, settings, serotypes, and food vehicles of

outbreaks. Identifying Salmonella serotypes and food vehicles associated with outbreaks is

important for providing evidence for sources of infection of Salmonella, as well as for prioritising

effective prevention and control measures. The paper was published in the open access

journal Frontiers in Sustainable Food Systems. Supplementary materials published online as

a supplement to the paper can be found in Appendix 2.

4.2 Paper

Ford L, Moffatt C, Fearnley E, Sloan-Gardner T, Miller M, Polkinghorne B, Franklin N,

Williamson DA, Glass K, Kirk MD. The epidemiology of Salmonella outbreaks in Australia,

2001-2016. Frontiers in Sustainable Food Systems. 2018;2:86, doi:

10.3389/fsufs.2018.00086.

ORIGINAL RESEARCHpublished: 12 December 2018doi: 10.3389/fsufs.2018.00086

Frontiers in Sustainable Food Systems | www.frontiersin.org 1 December 2018 | Volume 2 | Article 86

Edited by:

Joshua B. Gurtler,

Agricultural Research Service,

United States Department of

Agriculture, United States

Reviewed by:

Aparna Tatavarthy,

United States Food and Drug

Administration, United States

M. Leonor Faleiro,

University of Algarve, Portugal

Cheleste Thorpe,

Tufts University School of Medicine,

United States

*Correspondence:

Martyn D. Kirk

[email protected]

Specialty section:

This article was submitted to

Agro-Food Safety,

a section of the journal

Frontiers in Sustainable Food Systems

Received: 18 September 2018

Accepted: 27 November 2018

Published: 12 December 2018

Citation:

Ford L, Moffatt CRM, Fearnley E,

Miller M, Gregory J,

Sloan-Gardner TS, Polkinghorne BG,

Bell R, Franklin N, Williamson DA,

Glass K and Kirk MD (2018) The

Epidemiology of Salmonella enterica

Outbreaks in Australia, 2001–2016.

Front. Sustain. Food Syst. 2:86.

doi: 10.3389/fsufs.2018.00086

The Epidemiology of Salmonellaenterica Outbreaks in Australia,2001–2016Laura Ford 1, Cameron R. M. Moffatt 1, Emily Fearnley 1,2, Megge Miller 2, Joy Gregory 3,

Timothy S. Sloan-Gardner 4, Benjamin G. Polkinghorne 1, Robert Bell 5, Neil Franklin 1,6,

Deborah A. Williamson 7, Kathryn Glass 1 and Martyn D. Kirk 1*

1National Centre for Epidemiology and Population Health, Research School of Population Health, The Australian National

University, Canberra, ACT, Australia, 2OzFoodNet, South Australian Department for Health and Wellbeing, Adelaide, SA,

Australia, 3OzFoodNet, Department of Health and Human Services Victoria, Melbourne, VIC, Australia, 4OzFoodNet, Health

Protection Service, ACT Health, Canberra, ACT, Australia, 5OzFoodNet, Queensland Health, Brisbane, QLD, Australia,6OzFoodNet, NSW Ministry of Health, Sydney, NSW, Australia, 7Microbiological Diagnostic Unit Public Health Laboratory,

Department of Microbiology and Immunology, The University of Melbourne at The Peter Doherty Institute for Infection and

Immunity, Melbourne, VIC, Australia

Salmonella enterica is an important cause of foodborne illness in Australia, regularly

causing high-profile outbreaks involving commercially-available foods. We used the

national register of foodborne outbreaks to review the transmission pathways, settings,

serotypes, and food vehicles of Salmonella outbreaks in Australia between 2001 and

2016. We examined trends over time of implicated food vehicles in outbreaks where there

was statistical, microbiological, or descriptive evidence. Of the 990 Salmonella outbreaks

reported, 79% (778/990) were suspected or confirmed to have been transmitted through

contaminated food. Of these, 61% (472/778) occurred in food premises and 84%

(656/778) were caused by Salmonella Typhimurium. Eggs and egg-containing foods

were themost frequently identified food vehicle. Outbreaks due to egg-based sauces and

Vietnamese style sandwiches, which often contain pâté and raw egg butter, increased,

while outbreaks due to poultry meat, beef, pork, other sandwiches, and other desserts

had a decreasing trend from 2001 to 2016. Identifying food vehicles and the Salmonella

serotypes causing outbreaks in Australia provides important evidence for food regulation

strategies and control measures.

Keywords: disease outbreaks, eggs, Australia, foodborne disease, Salmonella Typhimurium

INTRODUCTION

Non-typhoidal Salmonella enterica spp. infection from contaminated food is an important causeof both sporadic gastroenteritis and outbreaks internationally. In Australia, the incidence ofinfection due to Salmonella spp. in the community is estimated to be 185 infections per 100,000population per year (Kirk et al., 2014). While only a small proportion of Salmonella cases areepidemiologically linked to outbreaks [6.6% of notifications in 2011 (OzFoodNet Working Group,2015)], outbreaks can be widespread and expensive for regulators and industry (Scharff et al., 2016).OzFoodNet—Australia’s enhanced foodborne disease surveillance network, has published annualsurveillance reports between 2001 and 2011, which report an annual median of 36 (range 26–61)salmonellosis outbreaks nationally that are confirmed or suspected to be caused by contaminatedfood, with themajority due to SalmonellaTyphimurium (OzFoodNetWorking Group, 2003, 2015).

29

Ford et al. Salmonella Outbreaks in Australia

The investigation of outbreaks to identify and control thesource of infection is an essential public health action toprevent further cases. Review of the Salmonella serotypesand associated food vehicles identified during outbreakscan assist with the development of targeted population-levelinterventions and prevention strategies to reduce the incidenceof Salmonella serotypes associated with these food vehicles.International studies have attributed Salmonella outbreaks tofood commodities, finding that eggs were the most commonlyimplicated food vehicle in the United States of America (USA)(28% of outbreaks), poultry in the United Kingdom (UK) (20.5%of outbreaks), and all meats in Latin America (24% of outbreaks)(Gormley et al., 2011; Pires et al., 2012; Jackson et al., 2013;Painter et al., 2013). In Australia, outbreaks of salmonellosishave been associated with a variety of foods including eggs,chicken and other poultry, pork, beef, lamb, fish, fresh produce,and nuts (Ashbolt et al., 2002; OzFoodNet Working Group,2012). While there is evidence that outbreaks associated witheggs in Australia have increased (Moffatt et al., 2016), the foodvehicles responsible for outbreaks of salmonellosis have not beensystematically assessed before.

In this study, we used the national foodborne diseaseoutbreak register data to describe the epidemiology of foodborneSalmonella outbreaks in Australia between 2001 and 2016.

MATERIALS AND METHODS

We used the OzFoodNet outbreak register, the national registerof foodborne disease outbreaks in Australia, to examine allreported outbreaks from 1 January 2001 to 31 December 2016where non-typhoidal Salmonella spp. was listed as the etiologicalagent. Australian states and territories who investigate theoutbreaks enter standardized outbreak data into this nationalrepository, including information on: the setting of the outbreak;where the food was prepared (if applicable and defined inSupplementary Information 1); number of symptomatic cases;number of cases with laboratory-confirmed Salmonella spp.,number of cases hospitalized, number who died during theoutbreak (the relative contribution of illness due to salmonellosisto each death is unknown); median incubation period ofcases; median duration of illness; geographical location ofexposures (local government area, multiple local governmentareas, multiple health department regions, state-wide, multi-state); mode of transmission; food vehicle; type of investigation(case series, cohort, case-control) and evidence (statistical,microbiological, descriptive); and Salmonella spp. serotypeclassified in accordance with the White-Kauffmann-Le Minorscheme (Grimont and Weill, 2007).

An outbreak was defined as ≥2 cases of Salmonella spp.orientated by person, place or time, or an increase in the numberof salmonellosis cases above what is normally expected and wherean investigation was undertaken to try to determine the sourceof illness. A foodborne or suspected foodborne outbreak wasdefined as an outbreak where cases had consumed a commonfood or meal that was implicated in causing their illness.We defined statistical, microbiological, and descriptive evidenceas in Moffatt et al. (2016). If an analytical epidemiologicalstudy was undertaken with a statistically significant association

observed between a food vehicle and illness, the food vehiclewas considered to have statistical evidence. Microbiologicalevidence was obtained when Salmonella was detected or culturedfrom the implicated food vehicle, food premises, processing, orprimary production environment. Microbiological investigationsand analytical studies were not undertaken in all outbreaks.An outbreak was considered to have descriptive evidence ifthere was information collected from the epidemiological and/orthe environmental health investigations that implicated thefood vehicle, which the outbreak investigation team consideredcompelling. This would include, for example, if an unsafe foodpreparation practice of a food reportedly eaten by at least someoutbreak cases was observed during the environmental healthinvestigation.

Outbreaks where <2 people had laboratory-confirmedSalmonella spp. infection and there was no microbiologicalevidence implicating a food were excluded (n = 35). Duplicateentries of multi-state outbreaks using OzFoodNet outbreakreports (Ashbolt et al., 2002; OzFoodNet Working Group,2007) (n = 2) and outbreaks where another pathogen besidesSalmonella spp. was detected or isolated in cases (n = 3) or inthe food vehicle (n = 2) were excluded. There was one outbreakin the outbreak register included in our analysis where a serotypewas unable to be determined.

We compared the number of outbreaks per year to thenational salmonellosis notification rate using data from theNational Notifiable Disease Surveillance System (Department ofHealth, 2018). Food vehicles were grouped into food categories(Supplementary Information 1). As the food vehicle field onlycontained the name of the implicated food or meal, ingredientlists were obtained through internet-based recipe searches. Ourfood categories included egg-based sauce, desserts containingraw or lightly cooked eggs, other desserts, eggs (other), poultry,beef, pork, lamb, fish, crustacean/mollusc, fruits, vegetables,nuts, sprouts, bánh mì (Vietnamese style sandwiches), othersandwiches, salads, sushi/kimbap/gimbap, tahini or helva, dips,and mixed dishes. If there were insufficient details in thefood vehicle field to categorize the food, we categorized it asundetermined. Food vehicle categories were examined by (i)evidence type and (ii) trend over time.

Descriptive analyses were performed in Stata SE 14 (StataCorp2014) and graphs were created in Microsoft Excel 2013. Dueto the importance of S. Typhimurium in Australia, we usedchi-square tests for homogeneity to assess differences betweenoutbreaks due to S. Typhimurium and non-Typhimuriumserotypes. We used the ptrend command to calculate a chi-square statistic for trend to assess trends in food vehicles overtime. Permission for data access was granted by OzFoodNet andethics approval was obtained through the Australian NationalUniversity Human Research Ethics Committee (2017/494).

RESULTS

From 1 January 2001 to 31 December 2016, there were 990outbreaks due to Salmonella spp. reported by Australian state andterritory health authorities. The number of outbreaks reportedper year ranged from 29 outbreaks in 2001 to 116 outbreaks in2014 (Figure 1).


30


FIGURE 1 | Salmonellosis outbreaks reported in the OzFoodNet outbreak register and National Notifiable Disease Surveillance System Salmonella notification rate,

Australia, 2001–2016.

TransmissionMost Salmonella spp. outbreaks were of foodborne or suspectedfoodborne transmission (79%; 778/990) (Table 1). In these 778outbreaks, a total of 14,708 people were reported to be ill,with a median of 9 people per outbreak (range 2–442). Overall,66% (9,683/14,708) of people affected in these outbreaks hadlaboratory-confirmed Salmonella spp., 15% (2,191/14,708) werehospitalized, and 0.3% (48/14,708) died. The median incubationperiod was available for 64% (500/778) of outbreaks, with amedian of 24 h (range of medians 7–192 h). Median illnessduration was available for 72% (560/778), with a median of 168 h(range of medians 10–504 h).

SettingImplicatedmeals were prepared in food premises for the majorityof foodborne or suspected foodborne outbreaks (Table 2),including 40% (314/778) in restaurants, 8% (63/778) in take-away stores, 7% (51/778) in bakeries, 4% (29/778) in commercialcaterers, 1% (8/778) in national franchised fast food stores,0.5% (4/778) in grocery stores/delicatessens, and 0.4% (3/778)on cruises or airlines. Home kitchens (18%, 142/778) were thenext most common setting. In 10% (80/778) of outbreaks, foodwas prepared in institutional settings, including 5% (39/778) inaged care facilities, 0.8% (6/778) in schools, 0.8% (6/778) inhospitals, 0.6% (5/778) in child care centers, 0.6% (5/778) incamps, and 2% (19/778) in other institutional settings such ascorrectional or military facilities. In the remaining 11% (84/778)of outbreaks, food was prepared at a market, fair/festival, ormobile service; commercially manufactured food; imported food;primary produce; prepared in other settings; or prepared inunknown settings (Table 2).

SerotypeThirty-nine different non-typhoidal Salmonella serotypes wereidentified among the 778 foodborne or suspected foodborne

outbreaks. Most outbreaks were due to Salmonella Typhimurium(84%, 656/778). Salmonella Saintpaul was the next most commoncause of outbreaks (2%, 15/778), followed by SalmonellaVirchow(2%, 12/778), Salmonella Singapore (1%, 9/778), Salmonellasubsp I (1%, 8/778), and Salmonella Infantis (1%, 8/778)(Table 3). No other serotype caused more than five outbreaksover the 16-year period assessed.

A higher proportion of outbreaks where food was preparedin food premises were due to S. Typhimurium (63%) thanto non-Typhimurium Salmonella (p < 0.01). There was nodifference in the reported number of S. Typhimurium and non-Typhimurium Salmonella outbreaks where food was preparedin private kitchens or in institutions. There was no differencein the number of people ill, the number of people hospitalized,the male-to-female ratio, or the median age of those affectedin foodborne or suspected foodborne outbreaks caused by S.Typhimurium compared to non-Typhimurium Salmonella in allsettings.

The number of S. Typhimurium outbreaks generally increasedover time, with 15 foodborne or suspected foodborne S.Typhimurium outbreaks reported in 2001, peaking at 88outbreaks reported in 2014. In comparison to the previousyear, in 2016, the number of S. Typhimurium outbreaks andthe associated number of cases in these outbreaks decreased,whereas, the number of associated cases in non-TyphimuriumSalmonella outbreaks increased. This was largely due to fourmulti-state non-Typhimurium Salmonella outbreaks in 2016(Figure 2).

Food VehiclesA food vehicle was listed in the OzFoodNet outbreak register for69% (537/778) of foodborne or suspected foodborne outbreaks.Of these, there was statistical and/or laboratory evidence tosupport the food vehicle as the cause of the outbreak for50% (271/537) of outbreaks. Where there was no statistical


31


TABLE 1 | Number of reported Salmonella spp. outbreaks, ill persons, laboratory-confirmed infections, hospitalizations, and deaths by transmission pathway, Australia,

2001–2016.

Transmission Number of outbreaks

(% of total)

Number ill

(% of total)

Number laboratory

confirmed (% of ill)

Number hospitalized

(% of ill)

Number died (%

of ill)

Foodborne or suspected foodborne 778 (79) 14,708 (88) 9,683 (66) 2,191 (15) 48 (0.3)

Waterborne or suspected waterborne 13 (1) 164 (1) 101 (62) 11 (7) 1 (0.6)

Person-to-person 30 (3) 259 (2) 134 (52) 45 (17) 4 (2)

Animal-to-person 4 (0.4) 34 (0.2) 28 (82) 0 (0) 0 (0)

Unknown 165 (17) 1,585 (9) 1,440 (91) 362 (23) 9 (0.6)

Total 990 (100) 16,750 (100) 11,386 (68) 2,609 (16) 62 (0.4)

TABLE 2 | Number of reported foodborne or suspected foodborne Salmonella spp. outbreaks, ill persons, laboratory-confirmed infections, hospitalizations, and deaths

by setting food prepared, Australia, 2001-2016.

Setting Number of outbreaks

(% of total)

Number ill

(% of total)

Number lab

confirmed (% of ill)

Number hospitalized

(% of ill)

Number died (%

of ill)

Food premises 472 (61) 9,505 (65) 5,919 (62) 1,448 (15) 10 (0.1)

Home kitchens 142 (18) 1,150 (8) 664 (58) 261 (23) 2 (0.2)

Institutions 80 (10) 1,323 (9) 615 (47) 185 (14.0) 31 (2)

Market, fair/festival, or mobile service 9 (1) 102 (0.7) 64 (63) 23 (23) 0 (0)

Commercially manufactured food 10 (1) 420 (3) 412 (98) 42 (10) 4 (1)

Imported food 2 (0.3) 48 (0.3) 48 (100) 11 (23) 0 (0)

Primary produce 12 (2) 427 (3) 425 (99) 75 (18) 1 (0.2)

Other 18 (2) 500 (3) 353 (71) 43 (9) 0 (0)

Unknown 33 (4) 1,233 (8) 1183 (96) 103 (8) 0 (0)

Total 778 (100) 14,708 (100) 9,683 (66) 2,191 (15) 48 (0.3)

FIGURE 2 | Number of outbreaks and outbreak cases due to Salmonella Typhimurium and non-Typhimurium Salmonella reported by state and territory health

authorities, Australia, 2001–2016.

or laboratory evidence, the food vehicle was supported bydescriptive evidence in 38% (205/537) of outbreaks, and therewas no evidence to support the food vehicle in the remaining11% (61/537) of outbreaks. We excluded the 61 outbreaks with

no evidence from further analysis, leaving 476 outbreaks with afood vehicle listed.

In our food categories, frequently identified food vehicles inoutbreaks were eggs and egg-containing foods (Table 4). Eggs,


32


TABLE 3 | Number of reported foodborne or suspected foodborne Salmonella spp. outbreaks, ill persons, hospitalizations, sex, and median of outbreak median age by

serotype, Australia, 2001-2016.

Serotype Number of

outbreaks (%)

Number ill

(% of total)

Number hospitalized

(% of ill)

Sex (% of ill) Median age

Male Female Unknown

Typhimurium 656 (84) 12,550 (85) 1,885 (15) 5360 (43) 6094 (48) 1096 (9) 31

Saintpaul 15 (2) 197 (1) 30 (15) 110 (56) 87 (44) 0 (0) 32

Virchow 12 (2) 127 (0.9) 21 (17) 59 (47) 66 (52) 2 (1) 28

Singapore 9 (1) 102 (0.7) 6 (6) 48 (47) 54 (53) 0 (0) 37

Infantis 8 (1) 67 (0.5) 15 (22) 26 (39) 41 (61) 0 (0) 55

Subsp I 8 (1) 91 (0.6) 15 (16) 54 (60) 37 (40) 0 (0) 28

Anatum 5 (0.6) 339 (2) 26 (8) 140 (41) 199 (59) 0 (0) 39

Bovismorbificans 4 (0.5) 138 (0.9) 34 (25) 53 (38) 38 (27) 47 (34) 34

Litchfield 4 (0.6) 108 (0.7) 8 (7) 86 (80) 22 (20) 0 (0) 46

Montevideo 4 (0.6) 62 (0.4) 10 (16) 29 (47) 33 (53) 0 (0) 27

Potsdam 4 (0.6) 101 (0.7) 15 (15) 37 (36) 64 (64) 0 (0) 34

Other 49 (6) 826 (6) 126 (15) 319 (39) 483 (58) 24 (3) 32

Total 778 (100) 14,708 (100) 2,191 (14.9) 6321 (43) 7218 (49) 1169 (8) 31

egg-based sauces (e.g., mayonnaise, aioli, hollandaise, tartare),desserts containing raw or lightly cooked eggs (e.g., tiramisu,fried ice cream, ice cream, mousse, custard), and fresh pasta eatenlightly cooked or with a lightly cooked egg based sauce, werethe identified food vehicle in 238/476 (50%) of Salmonella spp.outbreaks. S. Typhimurium was the responsible serotype in 95%(226/238) of these outbreaks (see Supplementary Information 2

for all serotypes for all food categories). Chicken and otherpoultry were the implicated food vehicle for 34/476 (7%)and pork for 15/476 (3%) outbreaks (Table 4), with 28/34(82%) and 9/15 (60%) due to S. Typhimurium, respectively(Supplementary Information 2). Other food source animals,including beef, lamb, fish, and crustaceans/molluscs, were eachidentified as the food vehicle for 1% of these outbreaks. Fruits,vegetables and sprouts were each responsible for around 1%of outbreaks, but unlike outbreaks involving animal-derivedfood vehicles, these were more common for non-Typhimuriumserotypes (p < 0.001) (Table 4; Supplementary Information 2).

Outbreaks due to sprouts, salads, Vietnamese sandwiches,fruits, nuts, tahini/helva, and dips tended to affect more people,with a median outbreak size of more than 15 cases (Table 4).Between 2001 and 2016, there was an increase in the proportionof outbreaks due to egg-based sauces (p< 0.001) and Vietnamesestyle sandwiches (p < 0.001), while poultry (p = 0.033), beef(p= 0.046), pork (p= 0.047), other sandwiches (p= 0.048), andother desserts (p = 0.017) had a decreasing trend over this timeperiod (Supplementary Information 3).

DISCUSSION

Outbreaks of non-typhoidal Salmonella spp. are a significantand high-profile cause of foodborne illness in Australia. Wefound an increasing trend of outbreaks in Australia, particularlybetween 2008 and 2014, beginning before the introduction ofculture-independent diagnostic testing (May et al., 2017). Thisis similar to the trend in the USA, where there was an increase

in the number of outbreaks reported between 2008 and 2013(Centers for Disease Control Prevention, 2018). The increasingtrend of outbreaks in Australia, together with comparativelyhigh and increasing Salmonella spp. notification rates (Fordet al., 2016), emphasizes the continued importance of identifyingfood vehicles in outbreaks and implementing control strategiesthroughout the food chain to prevent illness.

Similar to the USA and Canada, eggs and foods containing

eggs were themost commonly reported food vehicle in Australianfoodborne or suspected foodborne Salmonella spp. outbreaks

(Jackson et al., 2013; Belanger et al., 2015). Eggs were the

responsible food vehicle in 50% of these outbreaks in Australia,compared to 28% in the USA (Jackson et al., 2013) and 39% inCanada (Belanger et al., 2015). We took a conservative approachin attributing outbreaks to eggs, as eggs may have also beenthe responsible food vehicle where there was more than onehigh risk ingredient (e.g., Vietnamese style rolls, sandwiches,and salads). While the proportion of outbreaks associated withpoultry meat, beef and pork in Australia declined from 2001to 2016, egg-based sauces, desserts containing raw or lightlycooked eggs, and Vietnamese style sandwiches, which usuallycontain a raw-egg butter and/or pork or chicken liver pâté, wereincreasingly associated with Salmonella outbreaks over the timeperiod.

As most foodborne and suspected foodborne Salmonellaoutbreaks, including most S. Typhimurium outbreaks werelinked to commercial food premises, additional interventionstargeted at Salmonella control measures in these settings arerequired, particularly around the preparation of foods containingraw or lightly cooked eggs (Moffatt et al., 2016). While thereare state-based guidelines, there are no rules or restrictions onthe use of raw eggs in ready-to-eat foods in Australia (Moffattet al., 2016). As Australian states and territories have oversightof food safety regulation, some states have implemented controlmeasures across the supply chain to try to reduce the burdenof egg-related salmonellosis, including targeted communication


33


TABLE 4 | Median number of cases affected and number of outbreaks by type of evidence due to food vehicles, Australia, 2001–2016.

Median number of

outbreak cases (range)

Outbreaks with statistical

or laboratory evidence

Outbreaks with

descriptive evidence

Total

Number % Number % Number %

Raw egg desserts* 10 (2–140) 58/106 55 48/106 45 106/476 22

Egg-based sauce 14 (2–319) 43/79 54 36/79 46 79/476 17

Eggs, other 10 (2–143) 20/47 43 27/47 57 47/476 10

Mixed dishes 10 (2–90) 24/36 67 12/36 33 36/476 8

Poultry 11 (2–391) 20/34 59 14/34 41 34/476 7

Sandwiches, other 10 (3–213) 9/24 38 15/24 63 24/476 5

Desserts, other 13 (4–202) 18/23 78 5/23 22 23/476 5

Vietnamese sandwiches 17 (2–85) 7/17 41 10/17 59 17/476 4

Pork 10 (3–27) 10/15 67 5/15 33 15/476 3

Salads 18 (3–350) 8/10 80 2/10 20 10/476 2

Sushi 9.5 (3–85) 5/8 63 3/8 37 8/476 1

Beef 8.5 (4–15) 2/6 33 4/6 67 6/476 1

Pasta 6 (3–23) 2/6 33 4/6 67 6/476 1

Lamb 6 (3–43) 3/5 60 2/5 40 5/476 1

Fish 3 (3–14) 2/5 40 3/5 60 5/476 1

Sprouts 17 (4–126) 4/5 80 1/5 10 5/476 1

Crustacean/ mollusc 4 (2–23) 2/5 40 3/5 60 5/476 1

Fruits 29 (17–144) 4/4 100 0/4 0 4/476 0.8

Vegetables 10 (5–311) 3/4 75 1/4 25 4/476 0.8

Tahini or helva 24 (3–51) 3/3 100 0/3 0 3/476 0.6

Dips 33 (2–442) 2/3 67 1/3 33 3/476 0.6

Nuts 24 (19–43) 3/3 100 0/3 0 3/476 0.6

Undetermined 18 (2–50) 19/28 68 9/28 32 28/476 6

*Desserts containing raw or lightly cooked eggs (see Supplementary Information 1 for more information)

and education for bakeries, mandatory training at retail level,industry food safety plans, and the vaccination of many layingflocks against S. Typhimurium (NSW Food Authority, 2007;Groves et al., 2016; Ford et al., 2018).

Unlike in other countries where S. Enteritidis is moreprevalent in eggs, S. Typhimurium caused most (84%) foodborneor suspected foodborne Salmonella spp. outbreaks between 2001and 2016 (Gormley et al., 2011; Pires et al., 2012; Jacksonet al., 2013; Belanger et al., 2015), with only 3 outbreaks of S.Enteritidis reported during this time period. The majority offresh produce outbreaks associated with fruits, vegetables andsprouts were caused by non-Typhimurium serotypes, with S.Saintpaul, S. Litchfield, and S. Oranienburg causing more thanhalf of the outbreaks associated with these food vehicles. Whilefew in number, outbreaks associated with sprouts, nuts, fruitand some fresh salad produce, which are usually consumed raw,were generally large due to the wide distribution of these foodproducts. Food vehicle identification is often long and difficultfor these types of fresh produce outbreaks as these foods arefrequently poorly recalled and reported by outbreak cases duringfood history interviews. Also, as most horticulture products arenot packaged and labeled and there is high product turnover, foodtrace-back investigations are difficult and associated public healthinterventions can be delayed for these reasons (Munnoch et al.,

2009). In Australia, there is currently no national horticultureprimary production and processing standard, except for seedsprouts (Food Standards Australia New Zealand, 2011), soit is important to identify interventions to target these foodvehicles.

While there have been multiple cohort or case-control studiesundertaken during outbreaks to identify the food vehicle causingillness, there have been few published case-control studies ofsporadic Salmonella illness in Australia. Without this evidence,it is difficult to compare whether the food vehicles that causeoutbreaks identified in this study are the same as those thatcause sporadic illness. Also, while individual risk factors, suchas age and sex, have been identified to be important risk factorsfor sporadic illness in Australia (Ford et al., 2016), in thisstudy, there was little difference in demographic characteristicsof cases in foodborne or suspected foodborne S. Typhimuriumand non-Typhimurium Salmonella outbreaks. This suggests thatthe individual risk factors of age and sex are not as importantin outbreaks as they are with sporadic illness. While therewas no data on the immune status of people exposed tooutbreaks, the highest number of deaths occurred in outbreaksin institutional settings, including aged care facilities andhospitals where large numbers of immunocompromised peopleare expected.


34


A limitation of this study is that while OzFoodNetcommenced in 2001, not all states and territories were routinelyreporting data into the outbreak register until 2003, so someoutbreaks may have been missed prior to 2003. In this study,17% of Salmonella spp. outbreaks had an unknown modeof transmission. While it is likely that illness in many ofthese outbreaks was transmitted through contaminated food,there was insufficient evidence available to conclude this,so data presented in this paper is likely to under-representthe true number of foodborne outbreaks of salmonellosisin Australia. In addition to the 17% of outbreaks with anunknown mode of transmission there were: 25% of outbreaksthat had a suspected foodborne/foodborne mode of transmissionidentified with an unknown food vehicle; 6% of suspectedfoodborne/foodborne outbreaks that had no evidence to supportthe food vehicle; and the food vehicle in an additional 6%of suspected foodborne/foodborne outbreaks could not becategorized because the meal listed lacked sufficient details(undetermined) or contained multiple foods. In addition,38% of outbreaks were attributed to a food vehicle withdescriptive evidence alone. This may be because not all outbreaksare conducive to an analytical investigation (Moffatt et al.,2016), food or environmental sampling is not always routinelyperformed, or because cross-contamination can make it difficultto identify a single food vehicle. While cross-contamination islikely to be an important aspect in many outbreaks (OzFoodNetWorking Group, 2015; Osimani et al., 2016), another limitationof this study is that there was insufficient data to quantifythe importance of cross-contamination in these Australianoutbreaks.

As Australia moves toward a national strategy to reducefoodborne illness from Salmonella, data on Salmonella outbreaks,significant serotypes, and the food vehicles that cause illnessare important for monitoring and surveillance, which can beused to provide evidence for effective control measures (FoodRegulation Secretariat, 2018). Consistent with other studies,eggs and egg-containing foods were the most common causeof outbreaks in Australia over the period 2001–2016 causingsignificant morbidity in the population. Control measures have

been, and continue to be implemented around preparation ofegg-containing foods, particularly in food premises. Additionalinterventions focused on fresh produce items, such as ahorticulture primary production and processing standard, couldbe strengthened since these vehicles can cause larger outbreaks.While collecting evidence about the food vehicles that causeoutbreaks is important in the implementation of controlmeasures, it can be challenging and further research on causesof sporadic illness and the importance of cross-contaminationfor Salmonella outbreaks in Australia could identify importantpublic health interventions.

AUTHOR CONTRIBUTIONS

LF conceived the paper with feedback from all authors andconducted the analysis and drafted the paper. All authors madecontributions to the final manuscript.

FUNDING

LF is supported by an Australian Government Research TrainingProgram (RTP) Scholarship. MK is supported by a NationalHealth & Medical Research Council fellowship (APP1145997).DW is supported by a National Health & Medical ResearchCouncil fellowship (APP1123854).

ACKNOWLEDGMENTS

We thank the OzFoodNet Network, which is funded bythe Australian Government Department of Health. We thankstate public health reference laboratories who performed theSalmonella serotyping and state public health officers whoinvestigated these outbreaks.

SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be foundonline at: https://www.frontiersin.org/articles/10.3389/fsufs.2018.00086/full#supplementary-material

REFERENCES

Ashbolt, R., Givney, R., Gregory, J. E., Hall, G., Hundy, R., Kirk, M., et al.

(2002). Enhancing foodborne disease surveillance across Australia in 2001: the

OzFoodNet working group. Commun. Dis. Intell. Q. Rep. 26, 375–406.

Belanger, P., Tanguay, F., Hamel, M., and Phypers, M. (2015). An overview of

foodborne outbreaks in Canada reported through outbreak summaries: 2008-

2014. Canada Commun. Dis. Rep. 41, 254–262. doi: 10.14745/ccdr.v41i11a01

Centers for Disease Control and Prevention (2018). National Outbreak Reporting

System (NORS) Dashboard [Online]. Atlanta, GA: CDC. Availableonline at:

https://wwwn.cdc.gov/norsdashboard/ (Accessed August 25, 2018).

Department of Health (2018). National Notifiable Disease Surveillance System.

Notification Rate of Salmonellosis, Received From State and Territory Health

Authorities in the Period of 1991 to 2017 [Online]. Canberra, ACT: Australian

Government. Available online at: http://www9.health.gov.au/cda/source/cda-

index.cfm (Accessed July 5, 2018).

Food Regulation Secretariat (2018).Australia’s Foodborne Illness Reduction Strategy

2018-2021+ [Online]. Canberra ACT: Australian Government Department

of Health. Available online at: http://health.gov.au/internet/fr/publishing.

nsf/Content/australia-foodborne-illness-reduction-strategy-2018%E2%80

%932021 (Accessed June 18, 2018).

Food Standards Australia New Zealand (2011). Proposal P1004 - Primary

Production & Processing Standard for Seed Sprouts [Online]. Available

online at: http://www.foodstandards.gov.au/code/proposals/pages/

proposalp1004primary4361.aspx (Accessed November 10, 2018).

Ford, L., Glass, K., Veitch, M., Wardell, R., Polkinghorne, B., Dobbins, T., et al.

(2016). Increasing incidence of Salmonella in Australia, 2000-2013. PLoS ONE

11:e0163989. doi: 10.1371/journal.pone.0163989

Ford, L., Wang, Q., Stafford, R., Ressler, K. A., Norton, S., Shadbolt, C., et al.

(2018). Seven Salmonella Typhimurium outbreaks in Australia linked by

trace-Back and whole genome sequencing. Foodborne Path. Dis. 15, 285–292.

doi: 10.1089/fpd.2017.2353

Gormley, F. J., Little, C. L., Rawal, N., Gillespie, I. A., Lebaigue, S., and Adak, G.

K. (2011). A 17-year review of foodborne outbreaks: describing the continuing

decline in England and Wales (1992-2008). Epidemiol. Infect. 139, 688–699.

doi: 10.1017/S0950268810001858


35


Grimont, P. A., and Weill, F.-X. (2007). Antigenic Formulae of the Salmonella

serovars, 9th edn. Paris: WHOCollaborating Centre for Reference and Research

on Salmonella.

Groves, P. J., Sharpe, S. M., Muir, W. I., Pavic, A., and Cox, J. M. (2016). Live

and inactivated vaccine regimens against caecal Salmonella Typhimurium

colonisation in laying hens. Aust. Vet. J. 94, 387–393. doi: 10.1111/avj.12490

Jackson, B. R., Griffin, P. M., Cole, D., Walsh, K. A., and Chai, S.

J. (2013). Outbreak-associated Salmonella enterica serotypes and food

commodities, United States, 1998-2008. Emerg. Infect. Dis. 19, 1239–1244.

doi: 10.3201/eid1908.121511

Kirk, M., Ford, L., Glass, K., and Hall, G. (2014). Foodborne illness,

Australia, circa 2000 and circa 2010. Emerg. Infect. Dis. 20, 1857–1864.

doi: 10.3201/eid2011.131315

May, F. J., Stafford, R. J., Carroll, H., Robson, J. M., Vohra, R., Nimmo, G. R., et al.

(2017). The effects of culture independent diagnostic testing on the diagnosis

and reporting of enteric bacterial pathogens in Queensland, 2010 to 2014.

Commun Dis Intell Q Rep. 41, E223–E230.

Moffatt, C. R., Musto, J., Pingault, N., Miller, M., Stafford, R., Gregory, J.,

et al. (2016). Salmonella Typhimurium and outbreaks of egg-associated

disease in Australia, 2001 to 2011. Foodb. Path. Dis. 13, 379–385.

doi: 10.1089/fpd.2015.2110

Munnoch, S. A., Ward, K., Sheridan, S., Fitzsimmons, G. J., Shadbolt, C. T.,

Piispanen, J. P., et al. (2009). A multi-state outbreak of Salmonella Saintpaul

in Australia associated with cantaloupe consumption. Epidemiol. Infect. 137,

367–374. doi: 10.1017/S.0950268808000861

NSWFood Authority (2007).Microbiological Quality of High Risk Bakery Products.

A Survey to Determine the Microbiological Quality of Bakery Products Sold in

NSW. Sydney, NSW: NSW Food Authority.

Osimani, A., Aquilanti, L., and Clementi, F. (2016). Salmonellosis associated

with mass catering: a survey of European Union cases over a 15-year period.

Epidemiol. Infect. 144, 3000–3012. doi: 10.1017/S0950268816001540

OzFoodNet Working Group (2003). Foodborne disease in Australia: incidence,

notifications and outbreaks. Annual report of the OzFoodNet network, 2002.

Commun. Dis. Intell. Q. Rep. 27, 209–243.

OzFoodNet Working Group (2007). Monitoring the incidence and causes of

diseases potentially transmitted by food in Australia: annual report of the

Ozfoodnet network, 2006. Commun. Dis. Intell. Q. Rep. 31, 345–365.



OzFoodNet network, 2010. Commun. Dis. Intell. Q. Rep. 36, E213–E241.



OzFoodNet network, 2011. Commun. Dis. Intell. Q. Rep. 39, E236–E264.

Painter, J. A., Hoekstra, R. M., Ayers, T., Tauxe, R. V., Braden, C. R., Angulo, F.

J., et al. (2013). Attribution of foodborne illnesses, hospitalizations, and deaths

to food commodities by using outbreak data, United States, 1998-2008. Emerg.

Infect. Dis. 19, 407–415. doi: 10.3201/eid1903.111866

Pires, S. M., Vieira, A. R., Perez, E., Lo Fo Wong, D., and Hald, T. (2012).

Attributing human foodborne illness to food sources and water in Latin

America and the Caribbean using data from outbreak investigations. Int. J.

Food Microbiol. 152, 129–138. doi: 10.1016/j.ijfoodmicro.2011.04.018

Scharff, R. L., Besser, J., Sharp, D. J., Jones, T. F., Peter, G. S., and

Hedberg, C. W. (2016). An economic evaluation of PulseNet: a network for

foodborne disease surveillance. Am. J. Prevent. Med. 50(5 Suppl 1), S66–S73.

doi: 10.1016/j.amepre.2015.09.018

Conflict of Interest Statement: The authors declare that the research was

conducted in the absence of any commercial or financial relationships that could

be construed as a potential conflict of interest.

Copyright © 2018 Ford, Moffatt, Fearnley, Miller, Gregory, Sloan-Gardner,

Polkinghorne, Bell, Franklin, Williamson, Glass and Kirk. This is an open-access

article distributed under the terms of the Creative Commons Attribution License (CC

BY). The use, distribution or reproduction in other forums is permitted, provided

the original author(s) and the copyright owner(s) are credited and that the original

publication in this journal is cited, in accordance with accepted academic practice.

No use, distribution or reproduction is permitted which does not comply with these

terms.


36

37

Chapter 5. Costs of non-typhoidal Salmonella enterica in

Australia

5.1 Introduction

In this chapter, I investigate the cost of non-typhoidal Salmonella in Australia, and the costs of

PCR and WGS for Salmonella in terms of how these laboratory methods assist public health

action. The chapter includes two papers. The first is a cost-of-illness study describing the costs

of non-typhoidal Salmonella enterica and its sequelae illnesses from health care usage, lost

productivity, and premature mortality in a typical year circa 2015. I estimate a total cost of AUD

146.8 million (90% credible intervals 127.8-167.9 million), with 104.8 million (90% 75.5-132.3

million) from foodborne infections. Quantifying the costs of illness in Australia will assist with

prioritization for public health interventions and demonstrates that targeted interventions

across the food chain could reduce costs and societal impact. This paper is published in the

Journal of Food Protection.

In the second paper, I compare the costs of current Salmonella testing and subtyping methods

to PCR and WGS. There is evidence that WGS can enable earlier detection of outbreaks,

allowing for quicker public health action to implement control measures and prevent further

cases. Therefore, I calculated how many cases would have to be prevented using WGS before

it would cost no more than the status quo or than PCR, and examined the costs of using current

testing and subtyping methods compared with WGS in three different outbreak scenarios. The

findings suggest that WGS will result in cost savings compared to both current testing and

typing methods and PCR if its use leads to the prevention of cases. This paper will be

submitted to an international peer-review journal.

Supplementary materials for these two papers can be found in Appendix 3.

5.2 Papers

Ford L, Haywood P, Kirk MD, Lancsar E, Williamson DA, Glass K. Cost of Salmonella

infections in Australia, 2015. Journal of Food Protection. 2019;82(9):1607-1614, doi:

10.4315/0362-028X.JFP-19-105..

Ford L, Glass K, Williamson DA, Sintchenko V, Robson JMB, Stafford R, Kirk MD. Cost of

whole genome sequencing for non-typhoidal Salmonella enterica. To be submitted 2019.

Research Paper

Cost of Salmonella Infections in Australia, 2015

LAURA FORD,1 PHILIP HAYWOOD,2 MARTYN D. KIRK,1 EMILY LANCSAR,3 DEBORAH A. WILLIAMSON,4 AND

KATHRYN GLASS1*

1National Centre for Epidemiology and Population Health and 3Department of Health Services Research and Policy, Research School of PopulationHealth, The Australian National University, Canberra, Australian Capital Territory 2601, Australia (ORCID: https://orcid.org/0000-0002-6253-9672[L.F.]); 2Centre for Health Economics Research and Evaluation, University of Technology Sydney, P.O. Box 123, Broadway, New South Wales 2007,

Australia; and 4Microbiological Diagnostic Unit Public Health Laboratory, The University of Melbourne at The Peter Doherty Institute for Infection andImmunity, Parkville, Victoria 3010, Australia

MS 19-105: Received 28 February 2019/Accepted 20 May 2019/Published Online 22 August 2019

ABSTRACT

Gastroenteritis caused from infections with Salmonella enterica (salmonellosis) causes significant morbidity in Australia.In addition to acute gastroenteritis, approximately 8.8% of people develop irritable bowel syndrome (IBS) and 8.5% of peopledevelop reactive arthritis (ReA). We estimated the economic cost of salmonellosis and associated sequel illnesses in Australia ina typical year circa 2015. We estimated incidence, hospitalizations, other health care usage, absenteeism, and prematuremortality for four age groups using a variety of complementary data sets. We calculated direct costs (health care) and indirectcosts (lost productivity and premature mortality) by using Monte Carlo simulation to estimate 90% credible intervals (CrI)around our point estimates. We estimated that 90,833 cases, 4,312 hospitalizations, and 19 deaths occurred from salmonellosis inAustralia circa 2015 at a direct cost of AUD 23.8 million (90% CrI, 19.3 to 28.9 million) and a total cost of AUD 124.4 million(90% CrI, 107.4 to 143.1 million). When IBS and ReAwere included, the estimated direct cost was 35.7 million (90% CrI, 29.9to 42.7 million) and the total cost was AUD 146.8 million (90% CrI, 127.8 to 167.9 million). Foodborne infections wereresponsible for AUD 88.9 million (90% CrI, 63.9 to 112.4 million) from acute salmonellosis and AUD 104.8 million (90% CrI,75.5 to 132.3 million) when IBS and ReA were included. Targeted interventions to prevent illness could considerably reducecosts and societal impact from Salmonella infections and sequel illnesses in Australia.

HIGHLIGHTS

� The rate of salmonellosis in Australia is high, but the costs have not previously been assessed.� Salmonellosis illness and sequelae cost Australia AUD 146.8 million circa 2015.� Foodborne Salmonella infections and sequelae cost AUD 104.8 million.� Quantifying costs helps prioritize interventions across the food chain to reduce societal impact.

Key words: Cost of illness; Health care costs; Incidence; Monte Carlo methods; Salmonella infections

Acute gastroenteritis from nontyphoidal Salmonellaenterica (NTS) infection (salmonellosis) causes significantmorbidity. Globally, the World Health Organization hasestimated that NTS is one of the highest burden foodbornepathogens, responsible for approximately 78.7 millionillnesses, 59 thousand deaths, and 4 million disabilityadjusted life years in 2010 (44). In Australia, there were anestimated 262 cases per 100,000 population circa 2010, with72% of cases transmitted from contaminated food (28, 43).The reported incidence of salmonellosis in Australia hasbeen increasing and is higher than in the United States,Canada, the United Kingdom, and New Zealand (7, 12, 20,27, 35).

NTS illness is one of the leading causes of foodbornegastroenteritis-associated hospitalizations and deaths in

Australia. In addition to salmonellosis, it has been estimatedthat approximately 8.8% of people infected with NTSsubsequently develop irritable bowel syndrome (IBS) and8.5% develop reactive arthritis (ReA) (18). These chronicsequelae can be severe and require additional medical tests;treatments; visits to primary care, specialists, or hospitals;and time off work.

As the burden of salmonellosis and its sequelae arehigh, estimated costs of illness are needed to better informfood safety policy and provide inputs for calculating thecost effectiveness of new policies and interventions.Internationally, there have been several studies estimatingthe cost of or loss of health-related quality of life from NTSillness, demonstrating that NTS has a substantial economicburden (25, 29, 38, 40). Although the overall cost offoodborne disease in Australia circa 2000 was estimated atAUD 1.2 billion (1), the costs were not assessed at apathogen level. Given the burden of salmonellosis in

* Author for correspondence. Tel: þ61 2 6125 2468; Fax: þ61 2 61250270; E-mail: [email protected].

1607

Journal of Food Protection, Vol. 82, No. 9, 2019, Pages 1607–1614https://doi.org/10.4315/0362-028X.JFP-19-105Copyright �, International Association for Food Protection

38

Australia, our aim was to provide contemporary estimatesof the cost of illness associated with NTS and resultingsequelae (IBS and ReA) in Australia circa 2015.

MATERIALS AND METHODS

We estimated costs of NTS illness and resulting sequelaeillnesses in @Risk (http://www.palisade.com) by using directhealth care costs, productivity loss, and the value of prematuremortality for acute and ongoing illness. The estimated directand indirect costs were based on the incidence of NTS, IBS,and ReA illness and ongoing illness from IBS and ReA in atypical year circa 2015. We estimated incidence and costs byfour age groups: 0 to 4 years, 5 to 19 years, 20 to 64 years, and65þ years. Data sources and model structure are describedbelow with additional information in Supplemental Material S1and S2.

Data sources for incidence burden estimates. We used datafrom the National Notifiable Disease Surveillance System by sexand 5-year age groups from 2013 to 2015 (12). Notificationswhere age, sex, or both were unknown were excluded (0.16%).Denominator data were based on the Australian populationprovided by the Australian Bureau of Statistics for that year (2).As IBS and ReA are rare in children ,5 years old (23, 39), weused the ratio of incidence to hospitalizations for IBS and ReA inthose aged 5 to 64 years to estimate incidence in children aged ,5years.

Data sources for health care usage burden estimates. ForNTS, we used data from the 2008 to 2009 National GastroenteritisSurvey II, a retrospective cross-sectional survey of the populationburden of infectious gastroenteritis in that year (8), to estimate theproportion of incident cases that consulted a general practitioner(GP), visited the emergency department, or took medication(antibiotics, diarrhea relief, pain relief, nausea relief, cramp relief).We weighted each of these proportions by the severity of diarrhea,with 1 to 2 days of diarrhea classified as mild disease, 3 to 4 daysas moderate, and 5 or more days as severe (Supplemental MaterialS1 and S2), and we assumed that 3% of NTS cases were mild,14% were moderate, and 83% were severe as in Hall et al. (21).The number of stool microscopy, culture, and sensitivity tests,stool PCR tests, or both were estimated from the median numberof notifications by age group in the National Disease SurveillanceSystem (12) from 2013 to 2015, all of which were confirmed bylaboratory testing (Supplemental Material S1). For IBS and ReA,we estimated the proportion of cases that see a GP, visit aspecialist, receive treatment, or undergo testing from the literatureor from expert clinical opinion (for details, see SupplementalMaterial S1 and S2). We made an a priori assumption there wouldbe no emergency department visits for IBS or ReA followingSalmonella infection.

We used hospital separation statistics and average length ofstay data by principal diagnoses for the financial years 2011 to2012, 2012 to 2013, and 2013 to 2014 from the AustralianInstitute of Health and Welfare to estimate NTS, IBS, and ReAhospitalizations (5, 6). Diagnostic codes were based on theAustralian modification of the 10th International Classification ofDiseases and are detailed in Technical Appendix 3 of Kirk et al.(28) and Technical Appendix 4 of Ford et al. (18). Becauseprincipal diagnosis data only are available online from theAustralian Institute of Health and Welfare, we imputed thenumber of additional diagnoses for NTS, IBS, and ReA based onKirk et al. (28) and Ford et al. (18).

Data sources for lost productivity burden estimates. Daysof lost paid work for illness and caring were estimated from theNational Gastroenteritis Survey II for NTS. Using the sameapproach taken for GP and emergency department visits, weadjusted data on the number of reported days of paid work misseddue to gastroenteritis illness or caring for someone with agastroenteritis illness for the severity of diarrhea and by age group.We followed Abelson et al. (1) in estimating days of lost paidwork as days in hospital plus 0.5 day for visiting the GP for IBSand data from Townes et al. (42) to estimate a distribution of 3.37days (90% credible interval [CrI], 2.59 to 4.3) for ReA.Confidence intervals (CI) for days of lost paid work due toNTS, IBS, and ReA illness were generated using cii means with aPoisson distribution in Stata SE 14.2 (StataCorp LLC, CollegeStation, TX).

Data sources for premature mortality burden estimates.We used data from the Australian Bureau of Statistics onunderlying or contributing causes of death in males and femalesaged 0 to 14, 15 to 64, and 65þ years from 2001 to 2010. Thediagnostic code used was A02, based on the 10th InternationalClassification of Diseases as detailed in Technical Appendix 3 ofKirk et al. (28). We estimated the number of deaths due to NTSinfection for our age groups by using a statistical model withmultipliers for underreporting, domestic acquisition, and food-borne proportion, as in Kirk et al. (28). We assumed no prematuremortality due to IBS or ReA.

Data sources for ongoing illness burden estimates. Weused the literature and built on previous work (1) that wasreviewed and updated by an expert clinician, to estimate theproportion of cases that have ongoing illness from incident IBS orReA following Salmonella infection in a typical year circa 2015.For IBS, we estimated that 42.9% (95% CI, 21.8 to 66%) of caseswould have continuing symptoms at 12 months (31). For ReA, weestimated that 50% (95% CI, 23 to 77%) would have continuingsymptoms at 3 months (23, 30).

Data sources for health care usage cost estimates. Weassumed costs for GP and specialist visits based on the MedicareBenefits Schedule (9). We used an estimate from 2009 to 2010 inNew South Wales for emergency department triage costs (36),and hospital cost data were extracted as Australian RefinedDiagnosis Related Group costs published by the IndependentHospital Pricing Authority for 2013 to 2014 (26). Becausepathology costs are generally paid for through national healthinsurance, we calculated the costs by dividing total benefits(monetary value spent) by services (number billed) fromMedicare item reports (13) for Medicare Benefits Schedule (9)for relevant item numbers. Costs were extracted from thePharmaceutical Benefits Schedule by dividing total benefits byservices (pharmaceuticals) from Pharmaceutical Benefits Sched-ule item reports (14) for Pharmaceutical Benefits Schedule itemnumbers (10) for the calendar year 2015. If the medication ortreatment was not listed on the Pharmaceutical Benefits Scheduleor may commonly be bought over the counter, we extracted costsfrom an Australian pharmaceutical store (https://www.chemistwarehouse.com.au). See Supplementary Material S2 fordetails.

Data sources for lost productivity cost estimates. We useda human capital approach to calculate the cost of lost productivity(34) and tested lost productivity in sensitivity analysis. We usedthe participation rate in the workforce by age group for June 2015

1608 FORD ET AL. J. Food Prot., Vol. 82, No. 9

39

extracted from the Australian Bureau of Statistics (79.2% for 20 to64 years and 12.2% for 65 years and older) (3) and multiplied thisby the estimated days of lost paid work and by the daily wage rate.As the Australian Bureau of Statistics provided a participation ratefor 15- to 19-year-olds (52.6%), we weighted this across our 5- to19-year-old age group to produce a participation rate of 17.5%.We used the participation rate for 20- to 64-year-olds (79.2%) fordays of paid work missed by carers reported in the NationalGastroenteritis Survey II for the age groups 0 to 4 years, 5 to 19years, and 20 to 64 years and the participation rate for 65 years andolder (12.2%) for the age group 65þ years. To calculate the dailywage rate, we divided the average weekly earnings for May 2015extracted from the Australian Bureau of Statistics (4) by 5 toestimate a daily productivity cost of AUD 227.38.

Data sources for premature mortality cost estimates. Tocalculate mortality costs, we used a value of statistical life: ‘‘thefinancial value society places on reducing the average number ofdeaths by one’’ (24, 33). We used a 2014 value of statistical lifefrom the Australian Office of Best Practice Regulation (33),adjusted to 2015 dollars by using the Reserve Bank of Australiainflation calculator (37), of AUD 4,263,351.50.

Data sources for new ongoing illness cost estimates. Tocalculate the health care costs of new ongoing illness over 1 year,we used medication costs and specialist visit costs as describedabove. To calculate lost productivity costs from ongoing illnessover 1 year, we used the lost productivity cost as described above.

Burden estimates. We used the same estimation approach,distributions, and multipliers to estimate foodborne NTS, IBS, andReA incidences, hospitalizations, and deaths as in the foodbornedisease estimation study circa 2010 (18, 28). We used simulationtechniques in @Risk for each estimate using multiple inputs, eachwith a level of uncertainty for our four age groups. The finaloutput was a distribution, from which we extracted a medianestimate of incidence, hospitalizations, and deaths together with90% CrI. Because mortality data were only available for 2000 to2010, we added an additional step to these models that multipliedthe estimated median rate of foodborne illness for NTS by thepopulation for 2013 to 2015. This adjusted the estimate forchanges in the population since 2000 to 2010, although it does notallow us to detect any change in the death rate since 2010. Moreinformation about distributions and outputs can be found inSupplemental Material S1.

Cost estimates. In @Risk, we multiplied burden estimates bycosts for health care, lost productivity, and valuation of prematuremortality for each age group and for all cases to calculate costs inAUD value for health care usage, the loss to productivity, and theimplied value of life. From the resulting distribution, we extracteda median, mean, and 90% CrI. We used an expert elicitation of theproportion of foodborne NTS infections (43) to estimate the costof NTS, IBS, and ReA due to contaminated food.

Sensitivity analyses. We used a probabilistic simulationapproach by calculating 90% CrI for key parameters known tovary in our burden estimates. We also conducted a one-waysensitivity analysis by halving and doubling individually firsthealth care usage costs and then productivity costs and finally thevalue of statistical life values (15). In addition, because estimatingthe net present value of production changes is difficult (11), wetested the impact of removing all productivity losses from acuteillness and used an alternative participation rate to model the

probability that both parents of 0- to 4-year-olds and 5- to 19-year-olds are working (0.627) in calculating lost productivity due tocaring. Because approximately 75% of 15- to 19-year-olds are inthe labor force part-time (2) and a day of missed paid work maynot be the equivalent of a full-time day, we also performed asensitivity analysis of halving the daily wage rate for this group.

RESULTS

Estimated burden of NTS in Australia. We estimated90,833 (90% CrI, 51,583 to 158,265) cases, 4,312 (90% CrI,3,335 to 11,091) hospitalizations, and 19 (90% CrI, 15 to22) deaths from salmonellosis in Australia circa 2015 (Table1). Of these, 64,000 (90% CrI, 34,000 to 117,000) cases,3,100 (90% CrI, 1,829 to 4,786) hospitalizations, and 13(90% CrI, 10 to 17) deaths were estimated to be due tocontaminated food.

Estimated costs of NTS disease. We estimate thatNTS cost a median of AUD 124.4 million and a mean of124.7 million (90% CrI, 107.4 to 143.1 million) circa 2015,with a median of 23.7 million and a mean of 23.8 million(90% CrI, 19.3 to 28.9 million) from health care usage, amedian of 21.3 million and a mean of 22.0 million (90%CrI, 13.7 to 32.6 million) from lost productivity, and amedian and mean of 79.0 million (90% CrI, 66.0 to 92.1million) from premature mortality. When acute andongoing illness from IBS and ReA following Salmonellainfection were included, the estimated cost was a median ofAUD 146.8 million and a mean of 147.2 million (90% CrI,127.8 to 167.9 million) (Table 2 and Supplemental MaterialS3). NTS illness due to contaminated food was responsiblefor a median of AUD 88.9 million and a mean of 88.7million (90% CrI, 63.9 to 112.4 million) of the total cost ofNTS illness and a median of AUD 104.8 million and amean of 104.6 (90% CrI, 75.5 to 132.3 million) when IBSand ReA were included. Premature mortality from NTS asthe underlying or contributing cause of death was thehighest contributor to the total cost, accounting for 63% ofthe NTS cost and 54% of the total cost. Lost productivitydue to acute NTS infection accounted for 17% of the NTScost, lost productivity due to acute and ongoing IBSaccounted for 39% of the IBS cost, and lost productivityfrom acute and ongoing ReA accounted for 59% of the ReAcost.

The total cost of illness and the total cost per case washighest in the 65þ years age group (Tables 3 and 4;Supplemental Material S3 for means). Overall, we estimat-ed that the cost of acute salmonellosis was a median ofAUD 1,322 per case and AUD 1,559 per case when IBS andReA were considered (Table 4; Supplemental Material S3for means).

Sensitivity analysis. In probabilistic simulation of theburden estimate parameters, uncertainty was greatest inpremature mortality, followed by estimates of lost produc-tivity (Fig. 1).

In one-way sensitivity analyses of costs, varying valueof statistical life for premature mortality costs by halvingand doubling costs had the largest impact on cost values

J. Food Prot., Vol. 82, No. 9 COST OF SALMONELLA IN AUSTRALIA 1609

40

(Fig. 2). Additional sensitivity analysis showed thatremoving costs associated with short-term productivity lossfrom acute NTS, IBS, and ReA illness resulted in an AUD21.1 million drop in the NTS cost to 102.7 million (90%CrI, 89.0 to 116.8 million) and an AUD 26.5 million drop inthe total cost estimate to 119.8 million (90% CrI, 104.7 to135.7 million). Adjusting the participation rate of carers for0- to 4-year-olds and for 5- to 19-year-olds, and halving thedaily wage for 5- to 19-year-olds had a minimal impact onthe final cost estimates, reducing the estimate by less than2% and less than 1%, respectively.

DISCUSSION

We estimated that the burden of salmonellosis and itssequelae in Australia are substantial, costing a median ofAUD 146.8 million in a typical year circa 2015. Of the totalcost, 35.7 million (24%) was due to health care costs, 31.7million (22%) was due to costs associated with lostproductivity, and 79.0 million (54%) was due to prematuremortality. Quantifying these costs allows the burden of NTSto be compared against other illnesses, assisting withprioritization of public health interventions for policy-making.

TABLE 1. Estimated median number of cases, hospitalizations, and deaths from NTS and resulting IBS and ReA circa 2015, Australia

Median no. (90% CrI)

NTS IBS ReA

Cases

0–4 yr 20,924 (12,059–36,250) 1.46 (0.82–2.57) 341 (56–898)5–19 yr 14,617 (8,408–25,384) 1,284 (714–2,308) 1,206 (290–3,125)20–64 yr 44,525 (25,012–77,489) 3,906 (2,104–7,077) 3,652 (865–9,626)65þ yr 10,767 (6,104–19,042) 942 (516–1,703) 881 (209–2,310)

Total 90,833 (51,583–158,165) 6,133 (3,335–11,091) 6,080 (1,420–15,959)

Hospitalizations

0–4 yr 984 (661–1,346) 0.29 (0–0.23) 1.84 (1.07–3.52)5–19 yr 607 (391–912) 16 (9–25) 4 (3–6)20–64 yr 1,814 (1,162–2,750) 326 (186–517) 20 (13–26)65þ yr 907 (570–1,382) 67 (39–104) 2.5 (2–4)

Total 4,312 (2,784–6,390) 409 (234–647) 28 (19–40)

Deaths

0–4 yr 0.65 (0.5–0.82) —a —5–19 yr 1.7 (1.34–2.07) — —20–64 yr 4 (2.71–4.38) — —65þ yr 13 (10–15) — —

Total 19 (15–22) — —

a —, not applicable.

TABLE 2. Estimated annual cost of illness of health care usage, lost productivity, and premature mortality, for acute and ongoing illness,circa 2015, Australia

Median cost in millions of AUD (90% CrI)

NTS IBS ReA Totala

Health care usage

Acute illness 23.7 (19.3–28.9) 5.6 (4.0–8.0) 1.9 (1.0–3.6) 31.4 (26.5–37.2)Ongoing illness —b 3.2 (1.8–6.9) 0.77 (0.34–1.7) 4.2 (2.7–6.5)

Lost productivity

Acute illness 21.3 (13.7–32.6) 2.0 (1.3–3.3) 2.7 (1.0–6.5) 26.5 (18.5–38.4)Ongoing illness — 3.6 (1.8–6.9) 1.0 (0.37–2.8) 4.8 (2.7–8.4)

Premature mortality

Acute illness 79.0 (66.0–92.1) — — 79.0 (66.0–92.1)

Total costs

Acute illness 124.4 (107.4–143.1) 7.6 (5.3–11.3) 4.6 (2.0–10.0) 137.7 (119.7–157.0)Ongoing illness — 6.8 (3.9–12.3) 1.8 (0.73–4.4) 9.0 (5.5–14.8)

Total 124.7 (107.4–143.1) 14.5 (9.5–23.0) 6.5 (2.8–14.2) 146.8 (127.8–167.9)

a Numbers may not sum due to simulation and rounding.b —, not applicable.


41

Because NTS is mostly transmitted through contami-nated food, targeted interventions across the food chaincould help to prevent acute and ongoing illness. Australia’sFoodborne Illness Reduction Strategy 2018 to 2021þ aimsto reduce the number of illnesses from foodborne

salmonellosis (17). Although the national strategy doesnot set a specific reduction target, a 10% reduction in acutefoodborne NTS illness incidence would equate to more thanAUD 8 million immediate reduction in costs, withadditional reductions from sequel illnesses expected as well.

Because of the higher proportion of people aged 65þyears with salmonellosis taking medications, being admittedto the hospital, and dying compared with other age groups,the cost per case was highest in the 65þ age group.Therefore, interventions targeted at preventing illness in thisage group would be most effective in reducing all costs. Our

TABLE 3. Estimated annual cost of acute and ongoing illness in 1 yr by age group, circa 2015, Australia

Median cost in millions of AUD (90% CrI)

NTS IBS ReA Totala

Health care usage

0–4 yr 4.2 (3.0–5.9) 0.004 (0.002–0.006) 0.14 (0.04–0.37) 4.4 (3.2–6.1)5–19 yr 2.8 (2.0–4.0) 1.6 (0.90–3.0) 0.48 (0.13–1.3) 5.1 (3.7–6.9)20–64 yr 8.6 (5.9–12.4) 5.7 (3.3–9.8) 1.5 (0.46–4.0) 16.3 (11.9–22.0)65þ yr 7.6 (5.1–11.0) 1.4 (0.80–2.4) 0.34 (0.10–0.93) 9.4 (6.8–13.0)

Lost productivity

0–4 yr 4.9 (2.3–10.0) 0.002 (0.001–0.004) 0.29 (0.07–0.78) 5.2 (2.6–10.3)5–19 yr 1.5 (0.46–4.4) 0.34 (0.17–0.65) 0.23 (0.05–0.61) 2.2 (1.0–5.0)20–64 yr 12.6 (6.7–22.8) 5.0 (2.7–9.4) 3.1 (0.73–8.4) 21.7 (13.9–33.6)65þ yr 1.2 (0.33–3.0) 0.18 (0.10–0.35) 0.11 (0.03–0.31) 1.5 (0.64–3.3)

Premature mortality

0–4 yr 2.8 (2.1–3.5) —b — 2.8 (2.1–3.5)5–19 yr 7.2 (5.7–8.8) — — 7.2 (5.7–8.8)20–64 yr 15.0 (11.6–18.7) — — 15.0 (11.6–18.7)65þ yr 53.8 (41.6–66.3) — — 53.8 (41.6–66.3)

Total costs

0–4 yr 11.9 (8.5–18.1) 0.006 (0.004–0.009) 0.43 (0.11–1.1) 12.4 (8.9–18.6)5–19 yr 11.8 (9.4–15.3) 2.0 (1.1–3.6) 0.70 (0.09–1.9) 14.7 (11.9–18.6)20–64 yr 36.5 (27.7–49.8) 10.7 (6.0–19.1) 4.6 (1.2–12.2) 53.3 (41.2–70.0)65þ yr 62.9 (50.3–75.9) 1.5 (0.9–2.7) 0.46 (0.12–1.2) 65.2 (52.5–78.2)

a Numbers may not sum due to simulation and rounding.b —, not applicable.

TABLE 4. Estimated median annual cost of acute and ongoingillness per case by age group, circa 2015, Australia

Median cost (AUD)

NTS IBS ReA Total

Health care usage only

0–4 yr 200.78 854.05 302.90 208.655–19 yr 190.85 434.31 287.48 344.3320–64 yr 192.17 491.17 301.1 363.4965þ yr 691.93 491.80 284.09 866.14

Total 250.02 476.64 293.32 377.97

Health care usage with lost productivity

0–4 yr 443.64 1,398.50 920.24 467.785–19 yr 307.88 525.02 423.95 507.5620–64 yr 480.79 930.31 916.21 845.7865þ yr 814.92 558.66 378.88 1,020.85

Total 479.99 783.99 718.10 713.85

Health care usage, lost productivity, and premature mortality

0–4 yr 579.88 1,398.50 920.24 603.975–19 yr 807.77 525.02 423.95 1,003.8820–64 yr 818.47 930.31 916.21 1,190.3565þ yr 5,798.29 558.66 378.88 6,003.72

Total 1,322.14 783.36 718.96 1,558.53

FIGURE 1. Median cost of acute and ongoing illness from NTS,IBS, and ReA from health care usage, lost productivity, andpremature mortality, circa 2015, Australia. Ranges reflect 90%credible intervals for burden estimates.


42

total cost per case for NTS of AUD 1,322 (~USD 995) isgenerally lower than in studies from the United States from2013 and 2015 (USD 3,568 (16), USD 5,335 (32)), althoughhigher than an estimate of E691 (~USD 765) by Sundstrom(40) for Sweden in 2016 dollars and the estimate of NZD220 (~USD 151) by Lake et al. (29) for New Zealand in2006 and 2007. The estimates from these studies are notdirectly comparable due to differing years, variablesestimated, and estimation methods.

Because premature mortality from NTS as the under-lying or contributing cause of death was the largestcontributor to our cost estimates, accounting for 63% oftotal NTS cost and 54% of NTS and sequelae cost, themethod used to value a lost life is very influential. In one-way sensitivity analysis, premature mortality costs had thelargest range of uncertainty, compared with health careusage and lost productivity costs. Methods and values forlost life are also influential when comparing estimatesbetween cost-of-illness studies. For example, although thetotal cost per NTS case estimated by the EconomicResearch Service for the United States is higher than ourestimate at USD 3,568, when only costs from health careusage and lost productivity are taken into account, the costestimates are similar at USD 383.55 for the United Statescompared with our estimate of AUD 480 (~USD 361) (16).

Costs associated with IBS and ReA following Salmo-nella were also a large contributor to our total cost estimateof AUD 146.8 million, accounting for 14% of the total cost(10% from IBS and 4% from ReA). Although we didaccount for these sequelae costs in our estimates, we havenot included costs of NTS to industry and public healthagencies for surveillance and regulation. Because NTS canbe expensive for the food industry, including costsassociated with everyday prevention measures, productrecalls, and liability, we have likely underestimated thecosts to society.

Although we estimated lost productivity costs in linewith previously published methods (1), these methods mayoverestimate costs as we have not considered that either

production will be made upon return to work or thatemployers will have excess capacity in the labor force tocover absenteeism for short-term absences (11). Inaddition, a weakness of the human capital approach isthat it overestimates productivity loss compared with thefriction cost approach (34). We restricted lost productivityestimates to paid work only, not including the value of losttime for those not employed, such as unpaid carers orhousework, or the loss of leisure time. One-way sensitivityanalysis of removing productivity loss from acute illnessresulted in a AUD 21.1 million reduction in the estimatefor NTS and a AUD 26.5 million reduction for NTS, IBS,and ReA.

A limitation of this study is that due to a lack of data forIBS and ReA, we relied on expert opinion from a physicianto review and update the proportion of cases receivingcertain medications, pathology tests, and IBS cases seeingspecialists from previous work and the literature. We useduncertainty intervals around these estimates to account forthe uncertainty in the data. There is also limited data onduration of ongoing ReA symptoms, with some studiesfinding that symptoms had ceased in most patients in lessthan 12 months (22, 23, 30), whereas others found caseswith long-lasting symptoms (19, 41). We used a conserva-tive estimate of the proportion of ReA cases with ongoingsymptoms in 1 year based on Hannu et al. (23) andLeirisalo-Repo et al. (30). Although we were not able toinclude the decrease in quality of life to calculate qualityadjusted life years, we have captured the decrease inquantity of life through the value of statistical life. Althoughour cost of illness estimates provide a baseline, a fulleconomic evaluation that compares costs and benefits ofalternative policy interventions is needed to identify howeffective new programs and policies would be at reducingburden. Identifying and implementing targeted preventionmeasures in the food supply chain could considerablyreduce overall NTS costs.

In summary, the clinical and economic burden ofsalmonellosis and its sequelae are high. The cost of illnessestimates in this article will inform food safety policy andcan be used in subsequent analysis of cost effectiveness ofnew policies and interventions aimed at the prevention andcontrol of NTS infection in Australia.

ACKNOWLEDGMENTS

L. Ford is supported by an Australian Government Research TrainingProgram scholarship. M. D. Kirk is supported by a National Health andMedical Research Council fellowship (APP1145997). D. A. Williamson issupported by a National Health & Medical Research Council fellowship(APP1123854).

SUPPLEMENTAL MATERIAL

Supplemental material associated with this article can befound online at: https://doi.org/10.4315/0362-028X.JFP-19-105.s1https://doi.org/10.4315/0362-028X.JFP-19-105.s2, and https://doi.org/10.4315/0362-028X.JFP-19-105.s3

REFERENCES

1. Abelson, P., M. P. Forbes, and G. Hall. 2006. The annual cost offoodborne illness in Australia. Australian Government Department ofHealth and Ageing, Canberra, Australia.

FIGURE 2. One-way sensitivity analysis of health care usage, lostproductivity, and premature mortality for NTS, IBS, and ReA costof illness circa 2015, Australia. Low values represent when costswere halved, and high values when costs were doubled around ourcost estimate.


43

2. Australian Bureau of Statistics. 2016. 3101.0 - Australian demo-graphic statistics, Dec 2015. Table 59. Estimated resident populationby single year of age, Australia. Available at: http://www.abs.gov.au/AUSSTATS/[email protected] /Deta i l sPage/3101.0Dec%202015?OpenDocument. Accessed 20 June 2017.

3. Australian Bureau of Statistics. 2018. 6202.0 - Labour Force,Australia, May 2018. GM1 - Labour force status and gross change(flows) by age, sex, state and territory, February 1991 onwards.Available at: http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/6202.0May%202018?OpenDocument. Accessed 3 July2018.

4. Australian Bureau of Statistics. 2018. 6302.0 - Average weeklyearnings, Australia, Nov 2017. Available at: http://www.abs.gov.au/AUSSTATS/[email protected] /Deta i l sPage/6302.0Nov%202017?OpenDocument. Accessed 7 June 2018.

5. Australian Institute of Health and Welfare. 2014. Separation statisticsby principal diagnosis (ICD-10-AM 7th edition), Australia, 2011–12to 2012–13. Available at: http://www.aihw.gov.au/hospitals-data/principal-diagnosis-data-cubes/. Accessed 11 June 2016.

6. Australian Institute of Health and Welfare. 2016. Separation statisticsby principal diagnosis (ICD-10-AM 7th edition), Australia, 2013–14to 2014–15. Available at: http://www.aihw.gov.au/hospitals-data/principal-diagnosis-data-cubes/. Accessed 11 June 2016.

7. Centers for Disease Control and Prevention. 2016. NationalSalmonella surveillance annual report, 2013. U.S. Department ofHealth and Human Services, Centers for Disease Control andPrevention, Atlanta.

8. Chen, Y., L. Ford, G. Hall, T. Dobbins, and M. Kirk. 2016.Healthcare utilization and lost productivity due to infectiousgastroenteritis, results from a national cross-sectional surveyAustralia 2008–2009. Epidemiol Infect. 144:241–246. doi:10.1017/s0950268815001375.

9. Department of Health. 2018. MBS online. Available at: http://www.mbsonline.gov.au/internet/mbsonline/publishing.nsf/Content/Home.Accessed 7 June 2018.

10. Department of Health. The pharmaceutical benefits scheme.Available at: http://www.pbs.gov.au/. Accessed 8 June 2018.

11. Department of Health. 2016. Guidelines for preparing a submissionto the Pharmaceutical Benefits Advisory Committee. AustralianGovernment, Canberra, Australia.

12. Department of Health, National Notifiable Disease SurveillanceSystem. n.d. Notifications of a selected disease by age group, sex andyear. Available at: http://www9.health.gov.au/cda/source/rpt_5_sel.cfm. Accessed 11 June 2016.

13. Department of Human Services. n.d. Medicare item reports.Available at: http://medicarestatistics.humanservices.gov.au/statistics/mbs_item.jsp. Accessed 7 June 2018.

14. Department of Human Services. n.d. Pharmaceutical benefitsschedule item reports. Available at: http://medicarestatistics.humanservices.gov.au/statistics/pbs_item.jsp. Accessed 8 June 2018.

15. Drummond, M., M. J. Sculpher, K. Claxton, G. L. Stoddart, and G.W. Torrance. 2015. Methods for the economic evaluation of healthcare programs, 4th ed. Oxford University Press, Oxford.

16. Economic Research Service. 2014. Cost estimates of foodborneillness. Available at: https://www.ers.usda.gov/data-products/cost-estimates-of-foodborne-illnesses/. Accessed 18 May 2017.

17. Food Standards Australia New Zealand. 2018. Australia’s foodborneillness reduction strategy 2018–2021þ. Australian Government,Canberra, Australia.

18. Ford, L., M. Kirk, K. Glass, and G. Hall. 2014. Sequelae offoodborne illness caused by 5 pathogens, Australia, circa 2010.Emerg. Infect. Dis. 20:1865–1871. doi:10.3201/eid2011.131316.

19. Fox, R., A. Calin, R. C. Gerber, and D. Gibson. 1979. The chronicityof symptoms and disability in Reiter’s syndrome. An analysis of 131consecutive patients. Ann. Intern. Med. 91:190–193.

20. Government of Canada. Reported cases from 1924 to 2016 in Canada- notifiable diseases on-line. Available at: http://diseases.canada.ca/ndis/charts.php?c¼plt. Accessed 1 January 2019.

21. Hall, G., K. Yohannes, J. Raupach, N. Becker, and M. Kirk. 2008.Estimating community incidence of Salmonella, Campylobacter, and

Shiga toxin–producing Escherichia coli infections, Australia. Emerg.Infect. Dis. 14:1601–1609. doi:10.3201/eid1410.071042.

22. Hannu, T., L. Mattila, H. Rautelin, P. Pelkonen, A. Siitonen, and M.Leirisalo-Repo. 2002. Campylobacter-triggered reactive arthritis: apopulation-based study. Rheumatology 41:312–318.

23. Hannu, T., L. Mattila, A. Siitonen, and M. Leirisalo-Repo. 2005.Reactive arthritis attributable to Shigella infection: a clinical andepidemiological nationwide study. Ann. Rheum. Dis. 64:594–598.doi:10.1136/ard.2004.027524.

24. HM Treasury. 2018. The green book. Central government guidanceon appraisal and evaluation. UK Government, London.

25. Hoffmann, S., M. B. Batz, and J. G. Morris, Jr. 2012. Annual cost ofillness and quality-adjusted life year losses in the United States dueto 14 foodborne pathogens. J. Food Prot. 75:1292–1302. doi:10.4315/0362-028X.JFP-11-417.

26. Independent Hospital Pricing Authority. 2016. National hospital costdata collection, Australian Public Hospitals Cost Report, Round 18(financial year 2013–14). National Hospital Cost Data Collection,Darlinghurst, Australia.

27. The Institute of Environmental Science and Research Ltd. 2017.Notifiable diseases in New Zealand: Annual report 2016. Institute ofEnvironmental Science and Research Ltd., Porirua, New Zealand.

28. Kirk, M., L. Ford, K. Glass, and G. Hall. 2014. Foodborne illness,Australia, circa 2000 and circa 2010. Emerg. Infect. Dis. 20:1857–1864. doi:10.3201/eid2011.131315.

29. Lake, R. J., P. J. Cressey, D. M. Campbell, and E. Oakley. 2010. Riskranking for foodborne microbial hazards in New Zealand: burden ofdisease estimates. Risk Anal. 30:743–752. doi:10.1111/j.1539-6924.2009.01269.x.

30. Leirisalo-Repo, M., P. Helenius, T. Hannu, A. Lehtinen, J. Kreula, M.Taavitsainen, and S. Koskimies. 1997. Long-term prognosis ofreactive salmonella arthritis. Ann. Rheum. Dis. 56:516–520.

31. Marshall, J. K., M. Thabane, M. R. Borgaonkar, and C. James. 2007.Postinfectious irritable bowel syndrome after a food-borne outbreakof acute gastroenteritis attributed to a viral pathogen. Clin. Gastro-enterol. Hepatol. 5:457–460. doi:10.1016/j.cgh.2006.11.025.

32. Minor, T., A. Lasher, K. Klontz, B. Brown, C. Nardinelli, and D.Zorn. 2015. The per case and total annual costs of foodborne illnessin the United States. Risk Anal. 35:1125–1139. doi:10.1111/risa.12316.

33. Office of Best Practice Regulation. 2014. Best practice regulationguidance note: value of statistical life. Department of the PrimeMinister and Cabinet, Australian Government, Canberra, Australia.

34. Pike, J., and S. D. Grosse. 2018. Friction cost estimates ofproductivity costs in cost-of-illness studies in comparison withhuman capital estimates: a review. Appl. Health Econ. Health Policy16:765–778. doi:10.1007/s40258-018-0416-4.

35. Public Health England. 2017. Salmonella data 2006 to 2015.November 2016. National laboratory data for residents of Englandand Wales. Public Health England, London.

36. Reeve, R., and M. Haas. 2013. Estimating the cost of emergencydepartment presentations in NSW. Ultimo NSW: Centre for HealthEconomics Research and Evaluation. University of Technology,Sydney, Australia.

37. Reserve Bank of Australia. n.d. Inflation calculator. Available at:https://www.rba.gov.au/calculator/. Accessed 17 July 2018.

38. Santos, A. C., J. A. Roberts, A. J. Cook, R. Simons, R. Sheehan, C.Lane, G. K. Adak, F. A. Clifton-Hadley, and L. C. Rodrigues. 2011.Salmonella Typhimurium and Salmonella Enteritidis in England:costs to patients, their families, and primary and community healthservices of the NHS. Epidemiol. Infect. 139:742–753. doi:10.1017/s0950268810001615.

39. Schonberg-Norio, D., L. Mattila, A. Lauhio, M. L. Katila, S. S.Kaukoranta, M. Koskela, S. Pajarre, J. Uksila, E. Eerola, S. Sarna,and H. Rautelin. 2010. Patient-reported complications associatedwith Campylobacter jejuni infection. Epidemiol. Infect. 138:1004–1011. doi:10.1017/s0950268809991099.

40. Sundstrom, K. 2018. Cost of illness for five major foodborneillnesses and sequelae in Sweden. Appl. Health Econ. Health Policy16:243–257. doi:10.1007/s40258-017-0369-z.


44

41. Thompson, G. T., D. A. DeRubeis, M. A. Hodge, C. Rajanayagam,

and R. D. Inman. 1995. Post-Salmonella reactive arthritis: late

clinical sequelae in a point source cohort. Am. J. Med. 98:13–21.

42. Townes, J. M., A. A. Deodhar, E. S. Laine, K. Smith, H. E. Krug, A.

Barkhuizen, M. E. Thompson, P. R. Cieslak, and J. Sobel. 2008.

Reactive arthritis following culture-confirmed infections with

bacterial enteric pathogens in Minnesota and Oregon: a population-

based study. Ann Rheum. Dis. 67:1689–1696. doi:10.1136/ard.2007.

083451.

43. Valley, H., K. Glass, L. Ford, G. Hall, M. D. Kirk, C. Shadbolt, M.Veitch, K. E. Fullerton, J. Musto, and N. Becker. 2014. Proportion ofillness acquired by foodborne transmission for nine enteric pathogensin Australia: an expert elicitation. Foodborne Pathog. Dis. 11:727–733. doi:10.1089/fpd.2014.1746.

44. World Health Organization. 2015. WHO estimates of the globalburden of foodborne diseases. Foodborne Diseases Burden Epide-miology Reference Group 2007–2015. Available at: http://www.who.int/foodsafety/areas_work/foodborne-diseases/ferg/en/. Accessed 18May 2017.


45

46

Cost of whole genome sequencing for non-typhoidal Salmonella enterica

Laura Ford, Kathryn Glass, Deborah A. Williamson, Vitali Sintchenko, Jennifer M.B. Robson,

Russell Stafford, Martyn D. Kirk

ABSTRACT

Rapidly changing technology for testing and typing non-typhoidal Salmonella enterica has

significant implications for public health surveillance. We examined the costs of whole genome

sequencing (WGS) in Australia compared to traditional testing, which is required to assist

public health action to prevent illness. We used laboratory testing and Salmonella case costs

to determine how many cases the use of WGS data would need to prevent to be cost-equal to

serotyping and multiple locus variable number tandem repeat analysis, or culture independent

testing based on polymerase chain reaction (PCR). We then examined the costs and cost-

savings of using current typing methods compared with WGS in outbreak scenarios. A median

of 275 (90% CrI -55-775) or 1.9% (90% CrI -0.4%-5.4%) of notified serotyped Salmonella

cases would need to be prevented for WGS to be cost-equal to current typing methods and

1,550 (90% CrI 820-2,725) or 9.6% of all notified Salmonella cases would need to be prevented

to be cost-equal to PCR. WGS has different cost effects in different outbreak scenarios, but

WGS data would result in cost savings in prolonged outbreaks if it helped public health action

occur earlier. Despite currently having a higher cost per isolate, routine WGS of Salmonella

will be no more expensive than existing typing methods or PCR if it can help avert illness.

INTRODUCTION

Culturing human samples for non-typhoidal Salmonella enterica informs both clinical diagnosis

and public health surveillance. Routine epidemiological typing of cultured Salmonella isolates

allows health agencies to identify outbreaks and sources of infection, and implement control

measures to prevent further illness. Conventional typing methods, such as serotyping, pulse

field gel electrophoresis (PFGE), and multiple locus variable-number tandem repeat analysis

(MLVA) may differentiate Salmonella cases for surveillance and outbreak detection, helping to

save lives and reduce costs to human health and industry (Ross et al. 2011; Scharff et al.

2016).

New technologies for laboratory testing and typing Salmonella are moving in divergent

directions. The emergence of culture-independent diagnostic testing through Polymerase

Chain Reaction (PCR) assays allows for quick, sensitive and inexpensive Salmonella

detection, compared with conventional culture on selective media. However, unless reflexive

culture (i.e. culture on PCR-positive samples) is performed, no isolate is grown and available

47

for typing. PCR tests alone provide limited information for public health surveillance. This

delays the recognition of outbreaks, potentially leading to larger outbreaks and increased

societal costs (Cronquist et al. 2012).

If an isolate is cultured, whole genome sequencing (WGS) has emerged as an alternative to

conventional typing methods such as serotyping and MLVA. WGS generates highly

discriminatory data on Salmonella isolates for surveillance and outbreak detection (Deng et al.

2016). WGS has been shown to help detect outbreaks while they are still small and to link food

sources to outbreaks, allowing for quick and effective intervention and control (Jackson et al.

2016). However, at present, public health and laboratory infrastructure required to adopt WGS

and sequencing costs can be high.

Although sequencing costs are declining, WGS of foodborne bacterial pathogens is currently

still more expensive than PCR and the cost differential with conventional typing methods

varies. In Australia, once a patient submits a clinical sample to a laboratory, Salmonella can

either be detected (with PCR) or isolated from that sample. If isolated, several typing methods

can be employed to characterise the isolate for public health surveillance and outbreak

investigation (Figure 1). We argue that WGS could reduce the societal costs of Salmonella if

it can reduce the number of people affected, and the costs of these affected cases are

considered. We examined the societal costs of three Salmonella testing and typing approaches

in terms of how they assist public action to detect outbreaks, implement control measures, and

prevent cases in Australia: (1) the current standard of culture, serotyping and MLVA; (2) culture

and WGS; and (3) PCR testing.

METHODS

Figure 1 highlights the three methods for which we have examined costs. We have not costed

PCR followed by reflexive culture, as this will always be more expensive than PCR or culture

alone. Costs and cost savings are reported in 2018 US dollars (USD), with Australian dollars

(AUD) converted to USD using the average monthly exchange rate from Jan-Jun 2018 of

0.7676 from the Reserve Bank of Australia (Reserve Bank of Australia 2018a) and inflation

adjusted for using the Reserve Bank of Australia’s inflation calculator (Reserve Bank of

Australia 2018b). To assess costs, we first determined how many cases must be prevented

for each method to be cost-equal to serotyping and MLVA, and then modelled outbreak

scenarios with different intervention points based on test data to assess cost savings.

Data sources

We used a cost per Salmonella case circa 2015 from Ford et al (forthcoming) including direct

and indirect costs of health care usage, lost productivity, and premature mortality from acute

48

infection, adjusted for inflation to 2018. We removed testing costs from the cost per case to

enable comparison of the three testing and typing regimes. Salmonella notification numbers

from 2017 were obtained from the National Notifiable Diseases Surveillance System (NNDSS)

(Department of Health 2018b). As 35% of notifications in Australia are serotyped as S.

Typhimurium (Supplementary information) and subsequently undergo multiple variable

number tandem repeat analysis (MLVA) (Ford et al. 2016), we separated Salmonella

Typhimurium and non-Typhimurium Salmonella. We used the mean and range of costs for

serotyping, MLVA (S. Typhimurium), and WGS (where available) collected from each of the 5

Australian state public health reference laboratories in mid-2018. For culture and PCR test

costs, we calculated cost per test by dividing the total benefits (monetary value spent) by the

number of services (number billed) of item reports 69345 and 69496 respectively in the

Medicare Benefits Schedule for the financial year 2017/18 (Department of Health 2018a;

Department of Human Services 2018). This resulted in a cost of USD 34.70 for PCR and USD

40.61 for culture. To model cost savings in an outbreak scenario, we used simulated data

based on Australian Salmonella outbreak data (Denehy et al. 2011; Todd 2018).

Figure 1: Flow chart of testing and typing methods for Salmonella in Australia

Analysis

Cases prevented

We calculated how many cases WGS data would have to prevent before it would cost the

Australian society no more than serotyping and MLVA. As it is difficult to detect outbreaks with

PCR, leading to delays in implementing control measures and potentially larger outbreaks

(Cronquist et al. 2012), we also used the same analysis to calculate how many cases WGS

49

and the status quo would have to prevent before they cost no more than PCR. The equation

to determine the threshold is given by:

In addition, to examine the effects of changes in WGS costs, we calculated how much the price

would have to drop before WGS was cost equal with (1) serotyping and (2) PCR. We performed

these analyses in @Risk version 6 (http://www.palisade.com). Uncertainty intervals for

Australia were generated from the minimum and maximum cost estimate for serotyping, MLVA,

and WGS from the reference laboratories. We used a PERT distribution for cost estimates with

uncertainty to generate median and 90% credible intervals for the number and proportion of

cases that need to be prevented, as well as the WGS cost decrease needed, for tests to be

cost neutral.

Outbreak scenario

The costs of using the testing and typing regimes were examined in three simulated

community outbreak scenarios: (1) a point source outbreak, (2) a prolonged outbreak with no

peak, and (3) a prolonged outbreak with a late peak. We sourced mean case numbers from

Australian Salmonella outbreak data (Denehy et al. 2011; Todd 2018) to generate random

daily case numbers sampled from a Poisson distribution for each outbreak scenario. We

estimated the cost of the outbreak using the cost per case and the costs per test as described

above. With Microsoft Excel and @Risk, we calculated the cost differences if WGS was used

in all three scenarios, and if a product recall or intervention occurred at an earlier time point

along the epidemiological curve (30, 60, and 90 days earlier) of the prolonged outbreak

scenarios using WGS. While it is unlikely that outbreaks would have the same epidemiological

curve if all human samples were only tested by PCR, we have considered outbreak costs using

PCR testing in these scenarios for comparison. We compared characteristics of our outbreak

scenario models with real Australian salmonellosis outbreak data from Queensland Health

(see Supplementary Information).

RESULTS

In 2018 in Australia, reference laboratories reported that the mean cost for serotyping

Salmonella was USD 42.37 (range USD 13.82-75.22), while MLVA cost USD 52.66 (range

USD 24.56-95.95) and WGS cost USD 83.15 (range USD 72.92-95.95). The cost per case of

Salmonella spp. infection was USD 1,098 (90% CrI USD 623-1,963).

50

We estimated that approximately 365 (90% CrI 145-775) or 4.2% of non-Typhimurium

Salmonella spp. cases needed to be prevented in Australia in a year for WGS to be cost-equal

to serotyping, at current costs (Table 1). Serotyping and MLVA for S. Typhimurium was more

expensive than WGS, meaning that WGS is already less expensive than current methods for

S. Typhimurium.

Table 1: Number of cases that need to be prevented in order for WGS and PCR methods to be cost-equal to current methods, Australia, 2018

Salmonella Typhimurium Non-Typhimurium Salmonella Median cases

(90% CrI) %

(90% CrI) Median cases

(90% CrI) %

(90% CrI) WGS vs current methods*

-80 (-325-100) -1.4 (-5.6-1.7) 365 (145-775) 365 (145-775)

Current methods* vs PCR

390 (170-790) 6.8 (3.0-13.9) 80 (-115-340) 0.9 (-1.3-3.9)

*Current methods are serotyping and MLVA (if serotyped as Salmonella Typhimurium)

Combining S. Typhimurium and non-Typhimurium cases, 275 (90% CrI -55-775) or 1.9% (90%

CrI -0.4%-5.4%) of all notified serotyped Salmonella cases needed to be prevented for WGS

to be cost-equal to serotyping and MLVA. For WGS to be cost-equal to PCR, 1,550 (90% CrI

820-2,725) or 9.6% (90% CrI 5.1%-17%) of Salmonella spp. cases need to be prevented in a

year.

If WGS costs dropped a median of USD 40.65 (95% CrI USD 20.00-60.73), or approximately

33%, then WGS would be cost-equal to serotyping with current case numbers. The costs would

need to drop by a median of USD 89.39 (95% CrI USD 82.53-96.79), or approximately 72% to

be cost-equal to PCR with current case numbers.

Outbreak scenarios

Point source outbreak

As exposure during a point source salmonellosis outbreak tends to occur over a relatively short

period (e.g. 1 meal), we used a conservative assumption that interventions will have no effect

on reducing case numbers in that outbreak. Therefore, detecting the outbreak sooner through

WGS would have no effect on reducing those outbreak costs. In our point source outbreak

scenario, there were 31 cases, with illness onsets occurring over 7 days (Figure 2). We

estimated the cumulative costs of the outbreak with PCR at USD 35,102 (90% CrI 20,222-

61,847), culture and serotyping at USD 36,648 (90% CrI 21,750-63,489), with WGS at USD

37,851 (90% CrI 23,019-64,699) (USD 1,203 more than culture and serotyping), and with

51

culture, serotyping, and MLVA at USD 38,369 (90% CrI 23,416-65,082) (USD 518 more than

WGS) (Figure 2).

Prolonged outbreak, no peak

We simulated a prolonged salmonellosis outbreak lasting 150 days, with a mean daily case

number of 1.5 cases for the first 120 days, and assuming an intervention was put in place at

120 days with a mean daily case number of 0.2 cases for the last 30 days. In this outbreak

scenario, there were 174 cases, with a cumulative cost of USD 205,808 (90% CrI 121,842-

352,912) using culture and serotyping, USD 212,807 (90% CrI 129,134-360,199) using WGS

(Figure 2), and USD 215,587 (90% CrI 131,564-362,438) using culture, serotyping and MLVA.

While an outbreak like this may not be detected if all samples were PCR-only, if it were

detected with the same epidemiological curve, we estimated a cumulative cost of USD 197,117

(113,315-344,546) using PCR (Figure 2). If WGS enabled early detection such that the

intervention was put in place 30 days earlier, this would result in a savings of USD 34,359

(90% CrI 14,606-68,792) over PCR or USD 42,814 (90% CrI 22,756-77,399) over culture and

serotyping, rising to a savings of USD 97,112 (90% CrI 52,167-174,489) over PCR or USD

105,057 (90% CrI 60,323-183,673) over culture and serotyping if the intervention was 60 days

earlier, and USD 138,783 (90% CrI 77,197-244,690) over PCR or USD 146,711 (90% CrI

85,408-254,676) over culture and serotyping if the intervention was 90 days earlier (Figure 2).

Prolonged outbreak, late peak

We simulated a prolonged salmonellosis outbreak based on real outbreak data lasting 180

days, with a 1 week peak occurring approximately 120 days after the start of the outbreak

(Todd 2018). We assumed a mean daily case number of 1.35 cases for the first 120 days, 4.5

cases for 7 days, 13.23 cases for 7 days, 2.25 cases for 7 days, and 0.14 cases for the last 39

days. We assumed an intervention was put in place at 140 days, following the peak. In this

outbreak scenario, there were 322 cases, with a cumulative cost USD 329,186 (90% CrI

194,382-575,442) using culture and serotyping, USD 340,759 (90% CrI 206,014-586,903)

using WGS (Figure 2), and USD 344,359 (90% CrI 209,629-590,760) using culture, serotyping

and MLVA. Again, while an outbreak like this may not be detected if all samples were PCR-

only, with the same epidemiological curve, we estimated a cumulative cost of USD 315,846

(90% CrI 181,015-561,523) using PCR (Figure 2). If WGS were able to detect the outbreak or

food vehicle earlier, this would result in a savings of USD 119,224 (90% CrI 62,199-223,053)

52

Figure 2: Epidemiological curve and cumulative outbreak case costs in simulated point-source

outbreak. Confidence intervals omitted for visual clarity.

53

over PCR or USD 132,739 (90% CrI over culture and serotyping if the intervention as put in

place 30 days earlier, USD 171,771 (90% CrI 93,859-313,571) over PCR or USD 185,042

(90% CrI 107,353-326,844) over culture and serotyping if the intervention was put in place 60

days earlier, and USD 225,469 (90% CrI 126,201-405,909) over PCR or USD 238,848 (90%

CrI 139,689-419,376) over culture and serotyping if the intervention was put in place 90 days

earlier (Figure 2).

DISCUSSION

In this study, we examined costs of three Salmonella testing and typing systems in Australia

in 2018 in terms of how they assist public health action. We found that WGS data needs to

prevent approximately 2% of all notifications currently serotyped, or approximately 10% of all

notifications if they were only tested through PCR to be cost-equal to current testing and typing

methods and PCR respectively. WGS could also significantly reduce costs in prolonged

outbreaks if the data helps public health officials to implement interventions earlier. Even in

point source outbreaks, where WGS is potentially least effective at reducing costs, by linking

multiple point source outbreaks or linking a specific food or food preparation practice to illness,

WGS data could help to avert future cases and reduce costs. Our findings present a compelling

case for widespread adoption of WGS in public health reference laboratories in Australia for

Salmonella surveillance and investigation, and can inform on the capacity of WGS for public

health globally.

As there are few examples in the literature of prospective use of WGS for surveillance and

outbreak detection, the effectiveness of preventing foodborne illness and the average time

from outbreak detection to the implementation of control measures is largely unknown.

However, there is evidence that compared to current Salmonella typing methods, WGS data

is more sensitive and specific in linking salmonellosis cases, helps link cases over wide

geographical areas and long time frames, reveals geographically distinct clusters,

differentiates cases not in an outbreak, results in better detection of Salmonella in food sources

and faster source tracking, and can provide evidence for the implementation of Salmonella

control plans in the food industry (Allard et al. 2018; Angelo et al. 2015; Bell et al. 2016; Ford

et al. 2018a; Ford et al. 2018b; Inns et al. 2017; Luna et al. 2018; Siira et al. 2019; Ung et al.

2019; Waldram et al. 2018). All of these advancements could lead to earlier intervention and

support more targeted use of public health resources, therefore reducing costs. In the USA,

PulseNet, a molecular subtyping network of federal, state, and local public health laboratories

has demonstrated significant economic and public health benefits through averting foodborne

illness with PFGE (Scharff et al. 2016), benefits which will likely be increased through the use

of WGS. A recent Canadian study estimated that WGS will result in a net benefit of $5.21

54

million for reported salmonellosis cases, $64.98 million if there is a 50% reduction of illness, or

$90.25 million if there is a 70% reduction in illness through the reduction of direct and indirect

salmonellosis costs from contamination in fresh produce, poultry, and eggs (Jain et al. 2019).

We have not addressed the costs of PCR and reflexive culture for all human samples in this

study, instead examining only costs associated with Salmonella positive samples first detected

either by PCR or culture. While PCR-only testing remains the cheapest option at present in

terms of raw costs, it results in a loss of subtyping capability (Cronquist et al. 2012), making

outbreak detection and linking human Salmonella infections to specific sources impossible.

Therefore, while we estimated PCR costs in our outbreak scenarios, it is likely that outbreak

epidemiological curves would be much larger, resulting in higher costs over other testing and

typing methods. In Australia, the large majority of Salmonella human isolates are still cultured,

either in the first instance or reflexively. Since the introduction of the multiplex PCR test in late

2013, the proportion of notifications without a serotype has increased from around 2% annually

(2009–2012) to 10% in 2017 (Department of Health 2018b). Although some of these may have

been PCR detections that were not reflexively cultured, it may also be a reflection of the

increased sensitivity of PCR over culture (Vohra 2019). A study in one Australian state found

that in 2014, while 6% of Salmonella notifications were diagnosed only by PCR test, a further

12% were PCR positive, but culture negative (May et al. 2017).

In this study, we assumed that WGS data will be received prospectively by the public health

departments in a timely, and standardized format. The harmonisation of WGS and

bioinformatic methods, terminology, and reporting across public health laboratories is essential

for public health surveillance and outbreak detection nationally (Williamson et al. 2019; World

Health Organization 2018). The ability to link Salmonella isolates from humans, foods, water,

animals, and the environment through standardized WGS data across states, regions, and

even globally will be key to using WGS data effectively to prevent cases.

Several limitations of this study have to be acknowledged. First, only per-isolate sequencing

costs were taken into account; we did not consider laboratory expenditures associated with

the transition to WGS. However, WGS costs have declined over time (World Health

Organization 2018) and this trend will likely continue as WGS replaces existing typing

techniques for foodborne pathogens. The Salmonella case cost estimate used here focuses

on healthcare expenditure and lost productivity, and does not include costs of Salmonella for

industry. The rapid detection of outbreaks can reduce costs for industry and trade, which we

have not accounted for in this study. Second, our analysis also does not take into account the

number of isolates required to fill a flow cell on the instrument and maximise the efficiency of

sequencing runs from a laboratory perspective. In lesser populated areas, waiting for a

55

sufficient number of isolates to complete a sequencing run or sending isolates away to other

public health reference laboratories may result in a delay in the public health department

receiving WGS data results, impacting the ability to implement timely public health action.

While laboratory subtyping data is an important component of Salmonella surveillance and

outbreak investigation, epidemiological and environmental investigation are also necessary for

the implementation of successful interventions. We acknowledge that there are other factors

in outbreaks besides laboratory typing data that may affect turn-around-time of WGS reporting.

Our approach could be applied to other pathogens to ascertain how many cases of infection

would need to be prevented before WGS was cost-equal to other methods. The models reflect

the monetary and non-monetary costs of disease outbreaks including the opportunity costs of

preventive interventions at the point of emergence. In the era of global trade and health

reforms, the added value of public health extends beyond national borders and strengthens

the economic imperatives for public health surveillance. Even though WGS currently has a

higher cost per isolate than serotyping for Salmonella in 2018 in Australia, WGS will be more

cost-effective if the genomic information can help prevent human infections by creating

opportunities for earlier public health action and identification of vehicles of infection. Currently

WGS for public health surveillance of Salmonella is dependent on culture either initially or

reflexively. However, there is significant research into metagenomics direct from faecal

specimens (Andersen and Hoorfar 2018; Besser 2018). Metagenomics has even been used

in a limited way to investigate foodborne disease outbreaks (Besser 2018). While waiting for

the costs of sequencing to decline and research in metagenomics to progress, the

effectiveness of WGS may actually reduce costs.

ACKNOWLEDGMENTS

We thank the state public health reference laboratories, Katherine Todd, and Queensland

Health for providing data. We also thank Emily Lancsar for her contribution.

Laura Ford is supported by an Australian Government Research Training Program (RTP)

Scholarship. Martyn D. Kirk is supported by a National Health & Medical Research Council

fellowship (APP1145997). Deborah A. Williamson is supported by a National Health & Medical

Research Council fellowship (APP1123854).

56

REFERENCES

Allard MW, Bell R, Ferreira CM, Gonzalez-Escalona N, Hoffmann M, Muruvanda T, Ottesen

A, Ramachandran P, Reed E, Sharma S, Stevens E, Timme R, Zheng J,Brown EW. Genomics

of foodborne pathogens for microbial food safety. Curr Opin Biotechnol 2018; 49:224-229.

Andersen SC,Hoorfar J. Surveillance of Foodborne Pathogens: Towards Diagnostic

Metagenomics of Fecal Samples. Genes 2018; 9:14.

Angelo KM, Chu A, Anand M, Nguyen TA, Bottichio L, Wise M, Williams I, Seelman S, Bell R,

Fatica M, Lance S, Baldwin D, Shannon K, Lee H, Trees E, Strain E,Gieraltowski L. Outbreak

of Salmonella Newport infections linked to cucumbers--United States, 2014. MMWR Morb

Mortal Wkly Rep 2015; 64:144-7.

Bell RL, Jarvis KG, Ottesen AR, McFarland MA,Brown EW. Recent and emerging innovations

in Salmonella detection: a food and environmental perspective. Microb Biotechnol 2016; 9:279-

92.

Besser JM. Salmonella epidemiology: A whirlwind of change. Food Microbiol 2018; 71:55-59.

Cronquist AB, Mody RK, Atkinson R, Besser J, Tobin D'Angelo M, Hurd S, Robinson T,

Nicholson C,Mahon BE. Impacts of culture-independent diagnostic practices on public health

surveillance for bacterial enteric pathogens. Clin Infect Dis 2012; 54 Suppl 5:S432-9.

Denehy EJ, Raupach JC, Cameron SA, Lokuge KM,Koehler AP. Outbreak of Salmonella

typhimurium phage type 44 infection among attendees of a wedding reception, April 2009.

Commun Dis Intell Q Rep 2011; 35:192-6.

Deng X, den Bakker HC,Hendriksen RS. Genomic Epidemiology: Whole-Genome-

Sequencing-Powered Surveillance and Outbreak Investigation of Foodborne Bacterial

Pathogens. Annu Rev Food Sci Technol 2016; 7:353-74.

Department of Health. MBS Online. 2018a. Available

at:http://www.mbsonline.gov.au/internet/mbsonline/publishing.nsf/Content/Home, accessed 7

Jun, 2018

Department of Health. Salmonella public dataset. 2018b. Available

at:http://www9.health.gov.au/cda/source/pub_salmo.cfm, accessed 21 Nov, 2018

Department of Human Services. Medicare Item Reports. 2018. Available

at:http://medicarestatistics.humanservices.gov.au/statistics/mbs_item.jsp, accessed 7 Jun,

2018

57

Ford L, Carter GP, Wang Q, Seemann T, Sintchenko V, Glass K, Williamson DA, Howard P,

Valcanis M, Castillo CFS, Sait M, Howden BP,Kirk MD. Incorporating Whole-Genome

Sequencing into Public Health Surveillance: Lessons from Prospective Sequencing of

Salmonella Typhimurium in Australia. Foodborne Pathog Dis 2018a; 15:161-167.

Ford L, Glass K, Veitch M, Wardell R, Polkinghorne B, Dobbins T, Lal A,Kirk MD. Increasing

Incidence of Salmonella in Australia, 2000-2013. PLoS One 2016; 11:e0163989.

Ford L, Wang Q, Stafford R, Ressler KA, Norton S, Shadbolt C, Hope K, Franklin N, Krsteski

R, Carswell A, Carter GP, Seemann T, Howard P, Valcanis M, Castillo CFS, Bates J, Glass K,

Williamson DA, Sintchenko V, Howden BP,Kirk MD. Seven Salmonella Typhimurium

Outbreaks in Australia Linked by Trace-Back and Whole Genome Sequencing. Foodborne

Pathog Dis 2018b; 15:285-292.

Inns T, Ashton PM, Herrera-Leon S, Lighthill J, Foulkes S, Jombart T, Rehman Y, Fox A,

Dallman T, E DEP, Browning L, Coia JE, Edeghere O,Vivancos R. Prospective use of whole

genome sequencing (WGS) detected a multi-country outbreak of Salmonella Enteritidis.

Epidemiol Infect 2017; 145:289-298.

Jackson BR, Tarr C, Strain E, Jackson KA, Conrad A, Carleton H, Katz LS, Stroika S, Gould

LH, Mody RK, Silk BJ, Beal J, Chen Y, Timme R, Doyle M, Fields A, Wise M, Tillman G,

Defibaugh-Chavez S, Kucerova Z, Sabol A, Roache K, Trees E, Simmons M, Wasilenko J,

Kubota K, Pouseele H, Klimke W, Besser J, Brown E, Allard M,Gerner-Smidt P.

Implementation of Nationwide Real-time Whole-genome Sequencing to Enhance Listeriosis

Outbreak Detection and Investigation. Clin Infect Dis 2016.

Jain S, Mukhopadhyay K,Thomassin PJ. An economic analysis of salmonella detection in fresh

produce, poultry, and eggs using whole genome sequencing technology in Canada. Food Res

Int 2019; 116:802-809.

Luna S, Taylor M, Galanis E, Asplin R, Huffman J, Wagner D, Hoang L, Paccagnella A, Shelton

S, Ladd-Wilson S, Seelman S, Whitney B, Elliot E, Atkinson R, Marshall K,Basler C. Outbreak

of Salmonella Chailey Infections Linked To Precut Coconut Pieces - United States and

Canada, 2017. MMWR Morb Mortal Wkly Rep 2018; 67:1098-1100.

May FJ, Stafford RJ, Carroll H, Robson JM, Vohra R, Nimmo GR, Bates J, Kirk MD, Fearnley

EJ,Polkinghorne BG. The effects of culture independent diagnostic testing on the diagnosis

and reporting of enteric bacterial pathogens in Queensland, 2010 to 2014. Commun Dis Intell

Q Rep 2017; 41:E223-e230.

58

Reserve Bank of Australia. 2018a. Exchange Rates - Monthly - January 2010 to latest

complete month of current year. f11hist, ed., https://www.rba.gov.au/statistics/historical-

data.html#exchange-rates.

Reserve Bank of Australia. Inflation Calculator. 2018b. Available

at:https://www.rba.gov.au/calculator/, accessed 17 July, 2018

Ross IL, Davos DE, Mwanri L, Raupach J,Heuzenroeder MW. MLVA and phage typing as

complementary tools in the epidemiological investigation of Salmonella enterica serovar

Typhimurium clusters. Curr Microbiol 2011; 62:1034-8.

Scharff RL, Besser J, Sharp DJ, Jones TF, Peter GS,Hedberg CW. An Economic Evaluation

of PulseNet: A Network for Foodborne Disease Surveillance. Am J Prev Med 2016; 50:S66-

73.

Siira L, Naseer U, Alfsnes K, Hermansen NO, Lange H,Brandal LT. Whole genome sequencing

of Salmonella Chester reveals geographically distinct clusters, Norway, 2000 to 2016. Euro

Surveill 2019; 24.

Todd K. 2018. 2016 Australian Salmonella Sainpaul Outbreak.

Ung A, Baidjoe AY, Van Cauteren D, Fawal N, Fabre L, Guerrisi C, Danis K, Morand A, Donguy

MP, Lucas E, Rossignol L, Lefevre S, Vignaud ML, Cadel-Six S, Lailler R, Jourdan-Da Silva

N,Le Hello S. Disentangling a complex nationwide Salmonella Dublin outbreak associated with

raw-milk cheese consumption, France, 2015 to 2016. Euro Surveill 2019; 24.

Vohra R. Introducing the new Multiplex Faecal PCR in the algorithm for diagnosis of enteric

pathogens causing gastroenteritis. 2019. Available

at:http://www.qml.com.au/IamaDoctor/News/001.aspx, accessed 5 April, 2019

Waldram A, Dolan G, Ashton PM, Jenkins C,Dallman TJ. Epidemiological analysis of

Salmonella clusters identified by whole genome sequencing, England and Wales 2014. Food

Microbiology 2018; 71:39-45.

Williamson DA, Kirk MD, Sintchenko V,Howden BP. The importance of public health genomics

for ensuring health security for Australia. Med J Aust 2019; 210:295-297.e1.

World Health Organization. Whole genome sequencing for foodborne disease surveillance:

landscape paper. WHO. Geneva. 2018.

59

Chapter 6. Impact of whole-genome sequencing on public

health surveillance of Salmonella Typhimurium

6.1 Introduction

In this chapter, I analyse data collected during 5 months of prospective sequencing of

Salmonella Typhimurium isolates from humans in the Australian Capital Territory. I investigate

the impact of phylogenomic Salmonella Typhimurium data on public health surveillance. The

chapter includes two peer-review papers, both published in Foodborne Pathogens and

Diseases. The first paper evaluates the performance of WGS for routine public health

surveillance, highlighting some challenges that should be addressed before routine

implementation. The findings of the evaluation show that phylogenomic data offers higher

discriminatory S. Typhimurium laboratory surveillance than existing methods.

The second paper reports on seven outbreaks across three states and territories that were

detected during the 5 months of sequencing of S. Typhimurium isolates in the ACT.

Phylogenomic data assisted in showing that isolates (both human and non-human) from the

seven outbreaks were related. This finding suggests that other outbreaks of S. Typhimurium

occurring across the country may be related and highlights the need for integrated national

surveillance as laboratories move towards routine WGS for non-typhoidal Salmonella

surveillance.

These two papers contribute to an improved understanding of the sources of infection of S.

Typhimurium and the implications of WGS for public health surveillance of non-typhoidal

Salmonella. Supplementary materials for these papers can be found in Appendix 4.

6.2 Papers


Valcanis M, Castillo CFS, Sait M, Howden BP, Kirk MD. Incorporating whole-genome

sequencing into public health surveillance: lessons from prospective sequencing of Salmonella

Typhimurium in Australia. Foodborne Pathogens and Disease. 2018;15(3): 161-167, doi:

10.10189/fpd.2017.2352.



Williamson DA, Sintchenko V, Howden BP, Kirk MD. Seven Salmonella Typhimurium

outbreaks in Australia linked by trace-back and whole-genome sequencing. Foodborne

Pathogens and Disease. 2018;15(5):285-292, doi: 10.1089/fpd.2017.2353

Incorporating Whole-Genome Sequencinginto Public Health Surveillance:

Lessons from Prospective Sequencingof Salmonella Typhimurium in Australia

Laura Ford,1,2 Glen P. Carter,3 Qinning Wang,4 Torsten Seemann,3 Vitali Sintchenko,4,5

Kathryn Glass,1 Deborah A. Williamson,3,6 Peter Howard,4 Mary Valcanis,6

Cristina Fabiola Sotomayor Castillo,4,5,7,8 Michelle Sait,5

Benjamin P. Howden,3,6,9 and Martyn D. Kirk1

Abstract

In Australia, the incidence of Salmonella Typhimurium has increased dramatically over the past decade. Whole-genomesequencing (WGS) is transforming public health microbiology, but poses challenges for surveillance. To compare WGS-based approaches with conventional typing for Salmonella surveillance, we performed concurrent WGS and multilocusvariable-number tandem-repeat analysis (MLVA) of Salmonella Typhimurium isolates from the Australian CapitalTerritory (ACT) for a period of 5 months. We exchanged data via a central shared virtual machine and performedcomparative genomic analyses. Epidemiological evidence was integrated with WGS-derived data to identify relatedisolates and sources of infection, and we compared WGS data for surveillance with findings from MLVA typing. Wefound that WGS data combined with epidemiological data linked an additional 9% of isolates to at least one other isolate inthestudy incontrast toMLVAandepidemiologicaldata, and 19%more isolates thanepidemiologicaldataalone.Analysisof risk factors showed that in one WGS-defined cluster, human cases had higher odds of purchasing a single egg brand.While WGS was more sensitive and specific than conventional typing methods, we identified barriers to uptake ofgenomic surveillance around complexity of reporting of WGS results, timeliness, acceptability, and stability. In con-clusion, WGS offers higher resolution of Salmonella Typhimurium laboratory surveillance than existing methods and canprovide further evidence on sources of infection in case and outbreak investigations for public health action. However,there are several challenges that need to be addressed for effective implementation of genomic surveillance in Australia.

Keywords: Salmonella Typhimurium, whole-genome sequencing, MLVA, public health, surveillance

Introduction

Nontyphoidal Salmonella (NTS) enterica causessignificant morbidity and mortality in Australia (Kirk

et al., 2014). Rates of NTS, particularly Salmonella Typhi-

murium, have been increasing and are higher in Australia thanother developed nations, including the United States and theEuropean Union (Ford et al., 2016). While NTS is transmittedthrough several routes, approximately three-quarters of salmo-nellosis cases are estimated to be transmitted via contaminated

1National Centre for Epidemiology and Population Health, Research School of Population Health, The Australian National University,Canberra, Australia.

2OzFoodNet, Health Protection Service, Population Health Protection and Prevention, ACT Health, Canberra, Australia.3Doherty Applied Microbial Genomics, Department of Microbiology and Immunology, The University of Melbourne at The Peter

Doherty Institute for Infection and Immunity, Melbourne, Australia.4Centre for Infectious Diseases and Microbiology Laboratory Services, Pathology West—Institute of Clinical Pathology and Medical

Research, Sydney, Australia.5Centre for Infectious Diseases and Microbiology—Public Health, Marie Bashir Institute for Infectious Diseases and Biosecurity,

The University of Sydney, Sydney, Australia.6Microbiological Diagnostic Unit Public Health Laboratory, Department of Microbiology and Immunology, The University of Melbourne

at The Peter Doherty Institute for Infection and Immunity, Melbourne, Australia.7Sydney Medical School—Westmead, The University of Sydney, Sydney, Australia.8Instituto de Salud Publica, Facultad de Medicina, Universidad Austral de Chile, Valdivia, Chile.9Infectious Diseases Department, Austin Health, Heidelberg, Australia.

FOODBORNE PATHOGENS AND DISEASEVolume 15, Number 3, 2018ª Mary Ann Liebert, Inc.DOI: 10.1089/fpd.2017.2352

161

Dow

nloa

ded

by A

ustr

alia

n N

atio

nal U

nive

rsity

fro

m w

ww

.lieb

ertp

ub.c

om a

t 02/

17/1

9. F

or p

erso

nal u

se o

nly.

60

food (Vally et al., 2014). Most foodborne NTS outbreaks inAustralia are due to Salmonella Typhimurium and have beenassociated with a variety of sources, although eggs and poultryare the most common (OzFoodNet Working Group, 2015). Inparticular, the number of Salmonella Typhimurium outbreaksassociated with eggs in Australia has increased from 2001 to2011 (Moffatt et al., 2016a).

Whole-genome sequencing (WGS) using high-throughputnext-generation sequencing offers an alternative to tradi-tional subtyping methods for NTS and other foodbornepathogens (Deng et al., 2016). WGS has been shown to bemore discriminatory and sometimes less expensive than tra-ditional typing methods (Deng et al., 2016). Internationally,there have been several examples of the application of WGSto investigate outbreaks and for routine surveillance typing ofNTS (Byrne et al., 2014; den Bakker et al., 2014; Leekitch-aroenphon et al., 2014; Angelo et al., 2015; Ashton et al.,2015, 2016; Inns et al., 2017). However, there is limitedevidence on the effective and actionable use of WGS datawithin a public health unit. To examine the potential benefitsof WGS within a public health unit, we conducted a pro-spective trial of sequencing all Salmonella Typhimuriumisolates in the Australian Capital Territory (ACT) over aperiod of 5 months. We evaluated WGS performance forroutine public health surveillance using previously publishedguidelines as described by German et al. (2001).

Methods

Study design

In line with routine surveillance procedures, isolates cul-tured from human fecal, urine, or blood samples of ACTresidents were notified by the diagnostic laboratory to thepublic health unit and forwarded to either the Micro-biological Diagnostic Unit Public Health Laboratory (MDUPHL) in Melbourne or the Institute for Clinical Pathology andMedical Research (ICPMR)—Pathology West in Sydney forserotyping, and multilocus variable-number tandem-repeatanalysis (MLVA) if Salmonella Typhimurium. All isolatescollected between January 1, 2016, and June 2, 2016, ser-otyped as Salmonella Typhimurium, were included in thisstudy and were sequenced on a fortnightly basis either atMDU PHL or at ICPMR. The WGS inclusion criteria wereexpanded to include human isolates from two outbreaks inQueensland and food/environmental isolates from one out-break in New South Wales where epidemiological trace-backindicated a common source to outbreaks in the ACT.

Ethics approval for this project was granted by the ACTHealth Human Research Ethics Committee (16.215) and theAustralian National University Human Research EthicsCommittee (2016/528).

Phenotypic serotyping

The antigenic formulae of Salmonella isolates were de-termined using antisera that agglutinated with specific O(somatic) and H (flagellar) antigens and were classified intoserovars in accordance with the White–Kauffmann–LeMinor scheme (Issenhuth-Jeanjean et al., 2014). Isolatesserotyped as Salmonella Typhimurium were submitted forMLVA and WGS.

Multilocus variable-number tandem-repeat analysis

MDU PHL and ICPMR performed MLVA on all Salmo-nella Typhimurium isolates as previously described (Lind-stedt et al., 2004).

Whole-genome sequencing

Study isolates at either MDU PHL or ICPMR were sub-cultured for purity on nutrient agar twice before genomicDNA extractions from single colonies were performed usinga JANUS automated workstation (PerkinElmer) and an au-tomated DNA extraction instrument, Chemagic Prepito-D�

(PerkinElmer) based on magnetic particle separation enablingisolation of high-quality DNA. Extracted genomic DNA wasdiluted and DNA libraries prepared using the Illumina NexteraXT kit according to the manufacturer’s instructions (Illumina,Inc.). Following DNA Library preparation, quality control wasperformed using a PerkinElmer LabChip GX Touch bioana-lyzer and Thermo Fisher Qubit fluorometer to accuratelyquantitate the amount of DNA in each indexed library and todetermine the average size and DNA fragment distributionwithin each library. After normalization, indexed librarieswere mixed in the calculated ratios and WGS performed usingthe Illumina NextSeq 500 platform with 2 · 150 bp paired-endchemistry.

Data sharing and bioinformatic analysis

MDU PHL and ICPMR exchanged data via a dedicatedshared virtual machine running on the Australian NationalResearch Cloud (National eResearch Collaboration Toolsand Resources project or NECTAR). Each participatinglaboratory was provided with an account on the remotecomputer with a Secure Shell login. This allowed us to uploadsequencing data files over an encrypted channel, as well asbrowse and download the files uploaded by others. Noidentifiable patient data were uploaded to the cloud. Fol-lowing the study, the data were deleted, the virtual machinewas shut down, and its resources returned to the NECTARcloud.

We used the ‘‘Nullarbor’’ pipeline (https://github.com/tseemann/nullarbor) to de novo assemble and align short readdata against the Salmonella Typhimurium LT2 reference ge-nome (GenBank accession Nos. AE006468 and AE006471).Similarity between isolates was assessed by the identificationand alignment of core genome single-nucleotide polymor-phisms (SNPs). The core genome comprised 4248 genesthat are present in all isolates following comparison of thegenomes. Maximum likelihood phylogenetic trees were esti-mated with FastTree software using the general time revers-ibility model, and SNP distances were calculated with Snippyv3.1 (http://github.com/tseemann/snippy).

Public health follow-up

Where possible, we interviewed all notified cases of NTSusing a standardized questionnaire to obtain informationabout potential food and environmental exposures in the 7 dbefore illness onset under The Public Health Act 1997 inACT. If the case was identified as part of an outbreak, weused additional questions specific to the outbreak, or onlyinterviewed with an outbreak-specific questionnaire.

162 FORD ET AL.

Dow

nloa

ded

by A

ustr

alia

n N

atio

nal U

nive

rsity

fro

m w

ww

.lieb

ertp

ub.c

om a

t 02/

17/1

9. F

or p

erso

nal u

se o

nly.

61

Definitions

For the purposes of this study, a case was defined as anindividual with Salmonella Typhimurium notified to thepublic health unit, with a specimen collection date betweenJanuary 1 and June 2, 2016. Epidemiological evidence of alink was defined as in the 7 d before illness onset: havinghousehold contact with a known case; reporting eating at thesame commercial food premises as another case; buying thesame egg brand; swimming in the same pool or lake as an-other case; or reporting a similar environmental contact asanother case (e.g., handled the same pet food, attended thesame petting zoo). Isolation of Salmonella Typhimuriumfrom food or the environment in a food premises linked to aknown outbreak was considered an epidemiological link fornonhuman samples. Cases who reported travel outside ofAustralia or within Australia but outside the ACT in the 7 dbefore illness onset were classified as travelling overseas orinterstate, respectively.

Isolates were defined as linked by MLVA if they had anidentical MLVA profile, and linked by WGS if they were £8SNPs from any other isolate in the cluster. This SNP cutoffwas chosen from observed SNPs in known outbreaks in thesedata. This cutoff will need to be further evaluated for futureanalyses.

An outbreak was defined as two or more cases who con-sumed a common food, or food from a common commercialfood premises, and epidemiological and/or microbiologicalevidence implicated the food or premises as the source ofillness. A cluster was defined as fully supported if all cases inthe cluster were linked by epidemiological evidence, anddefined as partially supported if two or more, but not all casesin the cluster, were linked by epidemiological evidence.

Analysis

We conducted descriptive analysis on deidentified data ofall NTS cases in the ACT during the study period to deter-mine the proportion of NTS isolates cultured and forwardedto a reference laboratory, as well as the number of isolatessequenced and the number in an epidemiological, MLVA, orWGS-defined cluster.

We used a case/case analysis for one WGS-identifiedcluster, where cases were isolates in the cluster, and controlswere randomly selected from both Salmonella Typhimuriumcases outside the cluster and non-Typhimurium NTS casesduring the study period. Questionnaire data were combinedwith cluster variables for MLVA and WGS and analyzedusing Microsoft Excel 2013 and Stata SE 14. Univariableanalysis with a Fisher’s exact test was used to generate anodds ratio and p-value.

Evaluation of WGS for surveillance

We qualitatively evaluated the implementation of WGS asa surveillance system based on its performance, particularlythe sensitivity and specificity, positive predictive value,simplicity, timeliness, acceptability, and stability (Germanet al., 2001).

Results

Of the 108 Salmonella Typhimurium isolates sequenced,there were 92 human isolates from 90 ACT residents, 13

human isolates from Queensland residents, 2 food isolates,and 1 environmental isolate from a food premises. Overall,MDU PHL sequenced 80% (86/108) and ICPMR sequenced20% (22/108) of the Salmonella Typhimurium isolates.Where two people provided two stool specimens on differentdates (10 and 1 d apart, respectively), there were no SNPdifferences found between sequential isolates.

WGS data identified 14 clusters of 2 or more isolates with£8 SNPs (median cluster size 3, range 2–32 isolates), whileMLVA data only identified 11 clusters with the same MLVAprofile (median cluster size 5, range 2–31 isolates). Epide-miological evidence fully supported 43% (6/14) and partiallysupported a further 43% (6/14) of the WGS clusters, com-pared to fully supporting 18% (2/11) and partially supportinga further 73% (8/11) of the MLVA clusters (Table 1).Overall, WGS with epidemiological data linked 9% (10/108)more isolates to at least one other isolate in the study thanMLVA and epidemiological data, and 19% (21/108) moreisolates than epidemiological data alone.

Of isolates linked by WGS, 24% (21/86) did not report anillness onset within 2 weeks of another isolate in the cluster oroutbreak occurrence. While some of the WGS clusters re-presented already recognized outbreaks (Ford et al., 2017)(Supplementary Data and Supplementary Figure 1; Supple-mentary Data are available online at www.liebertpub.com/fpd),exposure data in some of these WGS clusters were not directlycompared before receiving the WGS data due to differences inMLVA profiles or time of illness onsets. Epidemiological dataoverlaid with WGS data showed that cases in WGS clusters, whohad not previously been linked by epidemiological data, oftenreported buying the same egg brand (Fig. 1). A case/case anal-ysis of one of these clusters showed that cases within the WGScluster had a higher odds of reporting egg brand D comparedwith cases outside the cluster (odds ratio 45, 95% confidenceinterval 1.34–2507.21, p = 0.01).

Evaluation

Sensitivity and specificity

While culture-independent diagnostic testing has in-creased recently (Iwamoto et al., 2015), pathology providersin the ACT still culture most stool specimens from patientswith diarrhea for NTS (R Hundy, personal communication,

Table 1. Number and Percent of Salmonella

Typhimurium Isolates Linked to at Least One

Other Isolate in the Study Through Epidemiological

Data, Multilocus Variable-Number Tandem-Repeat

Analysis, and Whole-Genome Sequencing Data,

Australian Capital Territory, January to June 2016

Linked by

Isolates (n = 108)

N %

Epidemiological data alone 48 44MLVA alone 89 82WGS alone 86 80MLVA and epidemiological data 59 55WGS and epidemiological data 69 64

MLVA, multilocus variable-number tandem-repeat analysis;WGS, whole-genome sequencing.

INCORPORATING WGS INTO SALMONELLA SURVEILLANCE 163

Dow

nloa

ded

by A

ustr

alia

n N

atio

nal U

nive

rsity

fro

m w

ww

.lieb

ertp

ub.c

om a

t 02/

17/1

9. F

or p

erso

nal u

se o

nly.

62

2017). During the study period, 7% of NTS notifications werederived from nucleic acid testing only with no isolatesavailable, and 2% of notifications were cultured but isolateswere not forwarded to the reference laboratory for furthercharacterization.

WGS data were sensitive to detect outbreaks, identifyingmore genetically linked cases than epidemiological data andMLVA alone. WGS data were also specific by showing ge-netic differences within an MLVA profile and thus excludingsome cases from an MLVA-defined cluster.

Predictive value positive

All isolates (100%) linked by both epidemiological dataand MLVA were also linked by WGS, including cases indifferent states and territories. WGS data showed close ge-netic relatedness between sporadic cases (some occurringmonths after an outbreak) to point-source outbreaks or toother isolates of suspected sporadic infection.

Simplicity and timeliness

Reference laboratories perform and report NTS typingresults independently of each other; however, in this study,sequencing data needed to be shared for a combined analy-sis and report for effective surveillance of all SalmonellaTyphimurium in the ACT. A virtual machine was establishedto share data between two laboratories and bioinformaticexpertise was required for the analysis and report of WGSdata. Public health unit staff needed training in WGS datainterpretation. With current systems, the WGS analysis wasmore time-consuming than the analysis of MLVA profiles,even for a relatively small number of isolates.

As seen for Listeria monocytogenes in the United States( Jackson et al., 2016), due to the highly discriminatory powerof WGS, less data (fewer isolates) should be required toidentify an outbreak, facilitating timely intervention. In thisstudy, traditional typing methods were performed and theirresults reported before sequencing, and WGS data were notavailable early enough to be used to first identify an outbreak.

Acceptability and stability

The largest barriers to acceptability from the public healthunit were the capacity to understand and use the data, and thecost of sequencing. While the costs of WGS have been de-creasing and are likely to continue to decrease (Wang et al.,2015; Deng et al., 2016), at the time of this study, the cost ofWGS to the public health unit was higher than the combinedcost of serotyping and MLVA. Using WGS for surveillancerelies on the expertise and communication among microbi-ologists, bioinformaticians, and epidemiologists, and is notflexible to the loss of personnel who can sequence and in-terpret the data.

Discussion

Our findings demonstrated that even with a relatively smallnumber of culture-confirmed cases of Salmonella Typhi-murium gastroenteritis, WGS of Salmonella Typhimuriumwas more discriminatory than MLVA. WGS with epidemi-ological data was both sensitive (linking 9% of cases thattraditional typing had not linked) and specific (differentiatingisolates with the same MLVA profile that were not part of anoutbreak). WGS improved source attribution of SalmonellaTyphimurium cases by linking food and environmental iso-lates in an outbreak, and by identifying clusters with higherprecision for risk factor analysis—in this study, identifyingegg brands for further investigation. This high discriminationis invaluable at the public health unit level to better targetedepidemiological investigations.

Our results are consistent with reports from Denmark andthe United States, which identified benefits of WGS for Sal-monella Enteritidis compared to pulsed-field gel electropho-resis (den Bakker et al., 2014; Leekitcharoenphon et al.,2014).Several other recent studies have used WGS to help retro-spectively investigate NTS outbreaks in the United States,United Kingdom, and Europe (Byrne et al., 2014; Ashtonet al., 2015; Inns et al., 2015; Taylor et al., 2015; Hoffmannet al., 2016). Characterizing NTS by WGS is becoming morecommon, with Public Health England now routinely usingWGS for prospective characterization of all NTS (Ashtonet al., 2016). The improved sensitivity and specificity of WGScompared to phage typing likely increased the speed at which amulticountry outbreak of NTS in the United Kingdom andSpain in 2015 was recognized (Inns et al., 2017).

Several studies have suggested that WGS can also be usedto genetically link NTS isolates from food or environmentalsources to isolates from human cases, which can help torapidly identify the source of an outbreak (Yokoyama et al.,2014; Ashton et al., 2015; Inns et al., 2015). However, for thisto occur, NTS isolates from food and environmental sourcesneed to be sequenced, as, for example, has been done with theU.S. Food and Drug Administration’s GenomeTrakr (Allardet al., 2016). A database of NTS sequences from local sourcesmight have assisted this ACT project to identify sources ofinfection. Salmonella Typhimurium was only isolated fromfood or the environment in one outbreak, with the isolatesbeing highly related to those of the human cases from thatoutbreak. While risk factor information from cases providedsome evidence of food or environmental sources of infection,this evidence would be strengthened if isolates from thesesources could be linked through WGS.

We identified a number of implementation challenges ofWGS surveillance within the public health unit. While our useof SNP differences to assess relatedness between cases pro-vided high discrimination, they may be dependent on the se-lection of reference genomes, can change as the number ofisolates in the analysis increases, and require a national database

FIG. 1. Maximum likelihood core genome SNP phylogeny of Salmonella Typhimurium isolates with epidemiologicaldata, ACT, January to June 2016. Clusters identified by WGS are highlighted in the tree. Figure created with InteractiveTree of Life (https://itol.embl.de). Note: Where there was epidemiological or environmental evidence that an outbreak wasassociated with eggs, all cases in the outbreak were assigned that egg brand. Outbreak-associated case isolates were isolatesepidemiologically linked with a known outbreak. ACT, Australian Capital Territory; SNP, single-nucleotide polymorphism;WGS, whole-genome sequencing. Color images available online at www.liebertpub.com/fpd

‰

164 FORD ET AL.

Dow

nloa

ded

by A

ustr

alia

n N

atio

nal U

nive

rsity

fro

m w

ww

.lieb

ertp

ub.c

om a

t 02/

17/1

9. F

or p

erso

nal u

se o

nly.

63

165

Dow

nloa

ded

by A

ustr

alia

n N

atio

nal U

nive

rsity

fro

m w

ww

.lieb

ertp

ub.c

om a

t 02/

17/1

9. F

or p

erso

nal u

se o

nly.

64

for data sharing and analysis. This highlighted the need for anew system to integrate WGS results into current systems, asWGS data and epidemiological data needed to be manuallyoverlaid for analysis. Unlike a Campylobacter outbreak in theACT, where there were large SNP distances between epide-miologically linked cases (Moffatt et al., 2016b), we did notfind any cases where epidemiological evidence linked them, butWGS data did not. However, WGS data suggested links be-tween cases where there was no supporting epidemiologicalevidence. It is unclear how much evidence is needed to triggerfurther investigation or how resource intensive that investigationshould be. Finally, routine WGS for Salmonella Typhimuriumsurveillance remains costly, and the current turn-around-time ofWGS exceeds that of traditional methods. Using WGS as asurveillance system is unstable without sustainable funding forthe costs of WGS. Investment into continued training and in-frastructure to share data nationally and combine epidemio-logical data with WGS data is needed. While WGS is highlydiscriminatory, if the data are not received in a timely manner,this limits the utility for public health surveillance. We wereunable to measure timeliness of WGS in this study as traditionaltyping methods were performed before sequencing began.

Our study had a relatively small number of isolates andsmall cluster sizes, which limited our power to detect sig-nificant associations between risk factors and illness withinclusters. Despite this, our study showed the potential of WGSto link ‘‘sporadic’’ cases, which could be increased if otherAustralian states and territories contributed data.

In conclusion, WGS of Salmonella Typhimurium presentsa more discriminatory approach to laboratory surveillancethan MLVA and has the potential to identify outbreaks andsources of infection investigated by a small public health unit.If sequenced and reported in a timely manner, this couldincrease the speed of preventative action and reduce thenumber of illnesses. Several challenges need to be addressedbefore WGS can be used routinely in a public health unit,including reporting, triggers for investigation, and sustain-able funding and resources. This study facilitated close col-laboration among epidemiologists, microbiologists, andbioinformaticians, a key requirement for an effective transi-tion to WGS for NTS.

Acknowledgments

The authors thank ACT Health and OzFoodNet. They alsothank the laboratories that performed the serotyping, MLVA,and WGS, including the MDU PHL, the ICPMR, andQueensland Health Forensic and Scientific Services (FSS).Doherty Applied Microbial Genomics is funded by the De-partment of Microbiology and Immunology at The University ofMelbourne. The National Health and Medical Research Coun-cil, Australia, funded a Practitioner Fellowship GNT1105905 toB.P.H. and Project Grant GNT1129770 to B.P.H., D.A.W, andM.D.K. Finally, they thank Milica Stefanovic, Russel Stafford,Kirsty Hope, Craig Shadbolt, and John Bates for their con-tribution to this project. This research is supported by anAustralian Government Research Training Program (RTP)Scholarship.

Disclosure Statement

No competing financial interests exist.

References

Allard MW, Strain E, Melka D, Bunning K, Musser SM, BrownEW, Timme R. The practical value of foodborne pathogentraceability through building a whole genome sequencingnetwork and database. J Clin Microbiol 2016;54:1975–1983.

Angelo KM, Chu A, Anand M, Nguyen TA, Bottichio L, WiseM, Williams I, Seelman S, Bell R, Fatica M, Lance S,Baldwin D, Shannon K, Lee H, Trees E, Strain E, Gier-altowski L; Centers for Disease Control and Prevention(CDC). Outbreak of Salmonella Newport infections linked tocucumbers—United States, 2014. MMWR Morb Wkly Rep2015;64:144–147.

Ashton PM, Nair S, Peters TM, Bale JA, Powell DG, PaintsetA, Tewold R, Schaefer U, Jenkins C, Dallman TJ, de PinnaEM, Grant KA; Salmonella Whole Genome Sequencing Im-plementation Group. Identification of Salmonella for publichealth surveillance using whole genome sequencing. PeerJ2016;4:e1752.

Ashton PM, Peters T, Ameh L, McAleer R, Petrie S, Nair S,Muscat I, de Pinna E, Dallman T. Whole genome sequencingfor the retrospective investigation of an outbreak of Salmo-nella Typhimurium DT 8. PLoS Curr 2015;7:pii:ecurrent-s.outbreaks.2c05a47d292f376afc5a6fcdd8a7a3b6.

Byrne L, Fisher I, Peters T, Mather A, Thomson N, Rosner B,Bernard H, McKeown P, Cormican M, Cowden J, Aiyedun V,Lane C; International Outbreak Control Team. A multi-country outbreak of Salmonella Newport gastroenteritis inEurope associated with watermelon from Brazil, confirmedby whole genome sequencing: October 2011 to January 2012.Euro Surveill 2014;19:6–13.

den Bakker HC, Allard MW, Bopp D, Brown EW, Fontana J,Igbal Z, Kinney A, Limberger R, Musser KA, Shudt M,Strain E, Wiedmann M, Wolfgang WJ. Rapid whole-genomesequencing for surveillance of Salmonella enterica serovarEnteritidis. Emerg Infect Dis 2014;20:1306–1341.

Deng X, den Bakker HC, Hendriksen RS. Genomic epidemi-ology: Whole-genome-sequencing-powered surveillance andoutbreak investigation of foodborne bacterial pathogens. AnnuRev Food Sci Technol 2016;7:353–374.

Ford L, Glass K, Veitch M, Wardell R, Polkinghorne B, Dob-bins T, Lal A, Kirk MD. Increasing incidence of Salmonellain Australia, 2000–2013. PLoS One 2016;11:e0163989.

Ford L, Wang Q, Stafford R, Ressler KA, Norton S, Shadbolt C,Hope K, Franklin N, Krsteski R, Carswell A, Carter GP,Seemann T, Howard P, Valcanis M, Castillo CFS, Bates J,Glass K, Williamson DA, Sintchenko V, Howden BP, KirkMD. Seven Salmonella Typhimurium MLVA type 03-26-13-08-523 outbreaks in Australia linked by trace-back and wholegenome sequencing. (in press).

German RR, Lee LM, Horan JM, Milstein RL, Pertowski CA,Waller MN; Guidelines Working Group Centers for DiseaseControl and Prevention (CDC). Updated guidelines for evalu-ating public health surveillance systems: Recommendationsfrom the Guidelines Working Group. MMWR Recomm Rep2001;50:1–35.

Hoffmann M, Luo Y, Monday SR, Gonzalez-Escalona N, OttesenAR, Muruvanda T, Wang C, Kastansis G, Keys C, Janies D,Senturk IF, Catalyurek UV, Wang H, Hammack TS, WolfgangWJ, Schoonmaker-Bopp D, Chu A, Myers R, Haendiges J,Evans PS, Meng J, Strain EA, Allard MW, Brown EW. Tracingorigins of the Salmonella Bareilly strain causing a food-borneoutbreak in the United States. J Infect Dis 2016;213:502–508.

Inns T, Ashton PM, Herrera-Leon S, Lighthill J, Foulkes S,Jombart T, Rehman Y, Fox A, Dallman T, de Pinna E,

166 FORD ET AL.

Dow

nloa

ded

by A

ustr

alia

n N

atio

nal U

nive

rsity

fro

m w

ww

.lieb

ertp

ub.c

om a

t 02/

17/1

9. F

or p

erso

nal u

se o

nly.

65

Browning L, Coia JE, Edeghere O, Vivancos R. Prospectiveuse of whole genome sequencing (WGS) detected a multi-country outbreak of Salmonella Enteritidis. Epidemiol Infect2017;145:289–298.

Inns T, Lane C, Peters T, Dallman T, Chatt C, McFarland N,Crook P, Bishop T, Edge J, Hawker J, Elson R, Neal K, AdakGK, Cleary P; Outbreak Control Team. A multi-countrySalmonella Enteritidis phage type 14b outbreak associatedwith eggs from a German producer: ‘‘Near real-time’’ ap-plication of whole genome sequencing and food chain in-vestigations, United Kingdom, May to September 2014. EuroSurveill 2015;20:pii:21098.

Issenhuth-Jeanjean S, Roggentin P, Mikoleit M, GuibourdenchM, de Pinna E, Nair S, Fields PL, Weill FX. Supplement2008–2010 (no. 48) to the White-Kauffmann-Le Minorscheme. Res Microbiol 2014;165:526–530.

Iwamoto M, Huang JY, Cronquist AB, Medus C, Hurd S,Zansky S, Dunn J, Woron AM, Oosmanally N, Griffin PM,Besser J, Henao OL; Centers for Disease Control and Pre-vention (CDC). Bacterial enteric infections detected byculture-independent diagnostic tests—FoodNet, UnitedStates, 2012–2014. MMWR Morb Mortal Wkly Rep 2015;64:252–257.

Jackson BR, Tarr C, Strain E, Jackson KA, Conrad A, CarletonH, Katz LS, Stroika S, Gould LH, Mody RK, Silk BJ, Beal J,Chen Y, Timme R, Doyle M, Fields A, Wise M, Tillman G,Defibaugh-Chavez S, Kucerova Z, Sabol A, Roache K, TreesE, Simmons M, Wasilenko J, Kubota K, Pouseele H, KlimkeW, Besser J, Brown E, Allard M, Gerner-Smidt P. Imple-mentation of nationwide real-time whole-genome sequencingto enhance listeriosis outbreak detection and investigation.Clin Infect Dis 2016;63:380–386.

Kirk M, Ford L, Glass K, Hall G. Foodborne illness, Australia,circa 2000 and circa 2010. Emerg Infect Dis 2014;20:1857–1864.

Leekitcharoenphon P, Nielsen EM, Kaas RS, Lund O, Aar-estrup FM. Evaluation of whole genome sequencing foroutbreak detection of Salmonella enterica. PLoS One 2014;9:e87991.

Lindstedt BA, Vardund T, Aas L, Kapperud G. Multiple-locusvariable-number tandem-repeat analysis of Salmonella en-terica subsp. enterica serovar Typhimurium using PCRmultiplexing and multicolor capillary electrophoresis. J Mi-crobiol Methods 2004;59:163–172.

Moffatt CR, Greig A, Valcanis M, Gao W, Seemann T, HowdenBP, Kirk MD. A large outbreak of Campylobacter jejuniinfection in a university college caused by chicken liver pate,Australia 2013. Epidemiol Infect 2016a;144:2971–2978.

Moffatt CR, Musto J, Pingault N, Miller M, Stafford R, GregoryJ, Polkinghorne BG, Kirk MD. Salmonella Typhimurium andoutbreaks of egg-associated disease in Australia, 2001 to2011. Foodborne Pathog Dis 2016b;13:379–385.

OzFoodNet Working Group. Monitoring the incidence andcauses of diseases potentially transmitted by food in Aus-tralia: Annual report of the OzFoodNet network, 2011.Commun Dis Intell Q Rep 2015;39:E236–E264.

Taylor AJ, Lappi V, Wolfgang WG, Lapierre P, Palumbo MJ,Medus C, Boxrud D. Characterization of foodborne outbreaksof Salmonella enterica serovar Enteritidis with whole-genome sequencing single nucleotide polymorphism-basedanalysis for surveillance and outbreak detection. J Clin Mi-crobiol 2015;53:334–3340.

Vally H, Glass K, Ford L, Hall G, Kirk MD, Shadbolt C, VeitchM, Fullerton KE, Musto J, Becker N. Proportion of illnessacquired by foodborne transmission for nine enteric patho-gens in Australia: An expert elicitation. Foodborne PathogDis 2014;11:727–733.

Wang Q, Holmes N, Martinez E, Howard P, Hill-Cawthorne G,Sintchenko V. It is not all about single nucleotide polymor-phisms: Comparison of mobile genetic elements and dele-tions in Listeria monocytogenes genomes links cases ofhospital-acquired listeriosis to the environmental source. JClin Microbiol 2015;53:3492–3500.

Yokoyama E, Murakami K, Shiwa Y, Ishige T, Ando N, Ki-kuchi T, Murakami S. Phylogenetic and population geneticanalysis of Salmonella enterica subsp. enterica serovar In-fantis strains isolated in Japan using whole genome se-quencing data. Infect Genet Evol 2014;27:62–68.

Address correspondence to:Laura Ford, MCHAM

National Centre for Epidemiology and Population HealthResearch School of Population Health

The Australian National UniversityCanberra ACT 2601

Australia

E-mail: [email protected]

INCORPORATING WGS INTO SALMONELLA SURVEILLANCE 167

Dow

nloa

ded

by A

ustr

alia

n N

atio

nal U

nive

rsity

fro

m w

ww

.lieb

ertp

ub.c

om a

t 02/

17/1

9. F

or p

erso

nal u

se o

nly.

66

Seven Salmonella Typhimurium Outbreaks in AustraliaLinked by Trace-Back and Whole Genome Sequencing

Laura Ford,1,2 Qinning Wang,3 Russell Stafford,4 Kelly-Anne Ressler,5 Sophie Norton,6 Craig Shadbolt,7

Kirsty Hope,8 Neil Franklin,8 Radomir Krsteski,2 Adrienne Carswell,2 Glen P. Carter,9 Torsten Seemann,9

Peter Howard,3 Mary Valcanis,10 Cristina Fabiola Sotomayor Castillo,3,11–13 John Bates,14 Kathryn Glass,1

Deborah A. Williamson,9,10 Vitali Sintchenko,3,13 Benjamin P. Howden,9,10,15 and Martyn D. Kirk1

Abstract

Salmonella Typhimurium is a common cause of foodborne illness in Australia. We report on seven outbreaksof Salmonella Typhimurium multilocus variable-number tandem-repeat analysis (MLVA) 03-26-13-08-523(European convention 2-24-12-7-0212) in three Australian states and territories investigated between November2015 and March 2016. We identified a common egg grading facility in five of the outbreaks. While noSalmonella Typhimurium was detected at the grading facility and eggs could not be traced back to a particularfarm, whole genome sequencing (WGS) of isolates from cases from all seven outbreaks indicated a commonsource. WGS was able to provide higher discriminatory power than MLVA and will likely link more SalmonellaTyphimurium cases between states and territories in the future. National harmonization of Salmonella surveillanceis important for effective implementation of WGS for Salmonella outbreak investigations.

Keywords: Salmonella Typhimurium, whole genome sequencing, outbreaks, trace-back, eggs

Introduction

The incidence of nontyphoidal Salmonella entericahas been increasing in Australia (Ford et al., 2016).

Unlike in the United States and Europe, Salmonella sero-type Typhimurium is the most common cause of humanSalmonella infection and outbreaks in Australia (Fordet al., 2016). Salmonella serotype Enteritidis is not en-demic in Australian egg-laying flocks and makes up only

about 6% of nontyphoidal S. enterica notifications of humaninfection nationally (OzFoodNet Working Group, 2015; Fordet al., 2016).

Although Salmonella Typhimurium is believed to be en-demic in layer flocks in Australia, there is no ongoing, sys-tematic national surveillance of Salmonella in poultry farms(Chousalkar et al., 2016). Commercial egg farms and gradingfacilities are subject to a number of regulatory controls, in-cluding audits and inspections, which aim to ensure adequate

1National Centre for Epidemiology and Population Health, Research School of Population Health, The Australian National University,Canberra, Australia.

2OzFoodNet, Health Protection Service, Population Health Protection and Prevention, ACT Health, Canberra, Australia.3Centre for Infectious Diseases and Microbiology Laboratory Services, Pathology West—Institute of Clinical Pathology and Medical

Research, Sydney, Australia.4Communicable Diseases Branch, Prevention Division, Queensland Health, Brisbane, Australia.5South Eastern Sydney Local Health District, NSW Health, Sydney, Australia.6Western Sydney Local Health District, NSW Health, Penrith, Australia.7New South Wales Food Authority, Sydney, Australia.8New South Wales Ministry of Health, Sydney, Australia.9Doherty Applied Microbial Genomics, Department of Microbiology and Immunology, The University of Melbourne at The Peter

Doherty Institute for Infection and Immunity, Melbourne, Australia.10Microbiological Diagnostic Unit Public Health Laboratory, Department of Microbiology and Immunology, The University of Melbourne

at The Peter Doherty Institute for Infection and Immunity, Melbourne, Australia.11Sydney Medical School—Westmead, The University of Sydney, Sydney, Australia.12Instituto de Salud Publica, Facultad de Medicina, Universidad Austral de Chile, Valdivia, Chile.13Centre for Infectious Diseases and Microbiology—Public Health, Marie Bashir Institute for Infectious Diseases and Biosecurity, The

University of Sydney, Sydney, Australia.14Public Health Microbiology, Public & Environmental Health, Forensic and Scientific Services, Health Support Queensland, Department

of Health, Coopers Plains, Australia.15Infectious Diseases Department, Austin Health, Heidelberg, Australia.

FOODBORNE PATHOGENS AND DISEASEVolume 15, Number 5, 2018ª Mary Ann Liebert, Inc.DOI: 10.1089/fpd.2017.2353

285

Dow

nloa

ded

by A

ustr

alia

n N

atio

nal U

nive

rsity

fro

m w

ww

.lieb

ertp

ub.c

om a

t 04/

29/1

9. F

or p

erso

nal u

se o

nly.

67

biosecurity and food safety accountability in egg productionto minimize Salmonella prevalence (Food Standards Aus-tralia New Zealand, 2012; Chousalkar et al., 2017). Addi-tional standards at retail level provide through chain foodsafety control of egg-based foods. Despite these controls,outbreaks associated with eggs and raw egg products haveincreased across Australia, and *37% of sporadic Salmo-nella cases and 59% of Salmonella outbreak cases have beenattributed to eggs (Glass et al., 2016; Moffatt et al., 2016;OzFoodNet Working Group, 2016a).

As Salmonella Typhimurium is the most common serotypein Australia, reference laboratories routinely perform multi-locus variable-number tandem-repeat analysis (MLVA) forSalmonella Typhimurium isolates to help detect and inves-tigate outbreaks. MLVA has been particularly useful in trace-back investigations of Salmonella Typhimurium outbreaksassociated with eggs (Chousalkar et al., 2017).

Whole genome sequencing (WGS) has also shown to en-able dramatic improvements in linking Salmonella cases re-lated to outbreaks, as well as attributing potential food orenvironmental sources and tracing back to production. WGShas recently been applied to several nontyphoidal S. entericaoutbreak investigations internationally (Angelo et al., 2015;Ashton et al., 2015; Inns et al., 2015, 2017). While not yetroutinely performed on all nontyphoidal S. enterica isolatesin Australia, WGS has been used in outbreaks and researchstudies.

Between November 2015 and March 2016, seven localizedoutbreaks of Salmonella Typhimurium MLVA type 03-26-13-08-523 (European convention 2-24-12-7-0212) wereidentified and investigated by Australian state and territoryagencies: three in New South Wales, two in the AustralianCapital Territory, and two in Queensland. Before these out-breaks, this MLVA profile was uncommon and had only oncebeen associated with a notified outbreak in Australia (Oz-FoodNet Working Group, 2016b). Investigations into theseseven outbreaks were initiated to identify the source of in-fection and implement control measures to prevent furthercases. In this report, we describe the epidemiological andenvironmental investigations of these outbreaks and examinerelatedness using WGS results, with the aim to provide fur-ther evidence of egg-associated salmonellosis in Australia.

Materials and Methods

Epidemiological investigations

Each outbreak was investigated by local public health of-ficials. We defined an outbreak as two or more cases ofSalmonella Typhimurium 03-26-13-08-523 who consumeda common food, or food from a common place. Outbreak-associated human cases were first identified locally fromthree sources: (1) following presentations of patients withgastroenteritis at a local emergency department, (2) fromnotifications of Salmonella infection either through routinecase interviews or MLVA profiles, or (3) through the sub-mission of a food complaint to food regulators.

Separate case definitions were generated for each of theoutbreaks (Supplementary Table S1; Supplementary Data areavailable online at www.liebertpub.com/fpd), but broadly in-cluded anyone who ate at the implicated food premises duringthe outbreak day or period and subsequently developed gas-trointestinal illness. Cases were interviewed with a similar

telephone-administered structured hypothesis generatingquestionnaire for Salmonella or local outbreak-specific ques-tionnaires to obtain information about potential exposures,including food eaten in the week before onset. We conductedcase series of affected persons in six outbreaks, and a cohortstudy to investigate the first outbreak.

Data entry, tabulation, and analysis were completed inMicrosoft Excel. Tables were constructed to compare theattack rates of gastroenteritis for persons exposed and notexposed to each food item, followed by the calculation ofunivariate relative risks (RR) and 95% confidence intervals(CIs) for individual exposures to illness. The public healthunit where the outbreak occurred led the individual outbreakinvestigations.

Food and environmental investigations

Food safety authorities in the affected states and territoryinspected the implicated food premises. Where possible,food or environmental samples were collected for testing.The inspections and sampling at these premises aimed toidentify any food safety hazards and detect contamination infood or the environment of the premises. We also conductedegg trace-back at the implicated food premises to identify ifthey were using a common egg supplier. We inspected an egggrading facility and the newest layer flock shed on the layerfarm next to the grading facility, where we collected samplesof chicken feces, feed, egg, and environmental swabs.

Microbiological investigation

The food and environmental samples collected from theimplicated food premises in each outbreak were tested inthe state or territory in which the outbreak occurred. Swabsamples from the egg grading facility were tested using amodified Australian Standard 5013.10 (Standards Australia,2009). This method incorporates nonselective liquid resus-citation and then screening for the presence of Salmonellausing a commercial polymerase chain reaction kit. Liquidresuscitation broths positive for Salmonella are confirmedculturally using liquid selective enrichment and solid selec-tive media for the isolation of Salmonella. Typical colonieswere inoculated onto plate count agar and individual coloniesconfirmed by matrix-assisted laser desorption/ionization timeof flight mass spectrometry. Confirmed Salmonella isolatesfrom the grading facility were sent to the MicrobiologicalDiagnostic Unit Public Health Laboratory (MDU PHL) forserotyping.

A number of isolates associated with these outbreakswere sequenced as part of a study to prospectively WGSSalmonella Typhimurium in the Australian Capital Terri-tory (Ford et al., 2018). Serotyping and MLVA of the isolateswere completed at MDU PHL in Victoria, the Institute forClinical Pathology and Medical Research (ICPMR) in NewSouth Wales, or Queensland Health Forensic and ScientificServices. Briefly, the antigenic formulae of isolates weredetermined using antisera and serotyped in accordance withthe White-Kauffman-Le Minor scheme (Issenhuth-Jeanjeanet al., 2014), and MLVA was performed as previouslydescribed (Lindstedt et al., 2004).

Sequenced isolates were then subcultured for purity, DNAextracted using Presto Mini gDNA Bacteria kit (GeneAid),and DNA libraries prepared using the Illumina Nextera XT

286 FORD ET AL.

Dow

nloa

ded

by A

ustr

alia

n N

atio

nal U

nive

rsity

fro

m w

ww

.lieb

ertp

ub.c

om a

t 04/

29/1

9. F

or p

erso

nal u

se o

nly.

68

kit (Illumina, Inc.). WGS was performed using the IlluminaNextSeq 500 platform with 2 · 150 bp paired—end chemis-try. The ‘‘Nullarbor’’ pipeline (https://github.com/tseeman/nullarbor) was used to trim the reads, check the sequence dataquality, and perform core genome single-nucleotide poly-morphisms (SNPs) by aligning short-read data against theSalmonella Typhimurium LT2 reference genome. The sig-nificant thresholds for SNP calling were set for a minimumcoverage at 30. The phylogenetic analysis was performed onthe generated SNP alignment file to infer core SNP phylog-eny using the maximum likelihood method at 100 bootstrapsby MEGA 7 (Kumar et al., 2016). We describe these methodsin more detail elsewhere (Ford et al., 2018).

While this study aimed to prospectively WGS all Salmo-nella Typhimurium 03-26-13-08-523 isolates from the Aus-tralian Capital Territory, sequencing was not performed inreal time. Isolates from outbreaks in Queensland were se-quenced at MDU PHL and added into the analysis after theoutbreaks had been identified and investigated. Additionalisolates from outbreaks in New South Wales were sequencedjust for this study at ICPMR and added into the analysisretrospectively.

Results

Epidemiological investigations

Between October 2015 and March 2016, there were 272cases of Salmonella Typhimurium 03-26-13-08-523 notifiedin the Australian Capital Territory, New South Wales, andQueensland (Fig. 1). Of these, 115 (42%) cases were linked to1 of 7 point-source outbreaks, with outbreak size ranging from

2 to 81 cases. In a cohort study of the first outbreak, personseating mayonnaise containing raw egg were 3.6 times (RR 3.6,95% CI 1.04-12.3) more likely to have developed illness thanthose who did not report eating the mayonnaise. In case in-terviews, cases linked to 3 of the other outbreaks also reportedeating foods containing eggs at the implicated food premises.We did not identify a single suspected food source throughcase interviews in the remaining 3 outbreaks (Table 1).

Food and environmental investigations

In 3 outbreaks, environmental investigations identifiedcontributing factors at the premises associated with the out-breaks, including the use of raw egg foods, poor hygiene,poor food storage, and a lack of food safety knowledge. In theother three outbreaks, no food safety compliance issues wereidentified in the environmental findings. Food samples werecollected in five outbreaks and environmental samples inthree outbreaks (Table 1 and Supplementary Table S2).

Trace-back investigation. We identified the eggs used atfood premises in five of the seven outbreaks (Table 1). In allfive, eggs were produced by company X. We used egg stampsand purchase invoices to trace back the eggs used in the fiveoutbreaks to the same grading facility. During the inspectionof the grading facility, no food safety compliance issues wereidentified. The grading facility processes *1 million cage,barn, and free-range eggs per day, including about 120,000from a farm onsite and 800,000 eggs from 15 different farms.As egg packaging with farm establishment number was notavailable from any of the outbreaks, and eggs from more thanone farm were processed at the same facility on the egg stamp

FIG. 1. Epidemic curve of sporadic and outbreak Salmonella Typhimurium 03-26-13-08-523 case notifications by weekin the Australian Capital Territory, New South Wales, and Queensland, Australia, October 2015 to June 2016. If illnessonset date was unknown, specimen collection date was used.

SALMONELLA TYPHIMURIUM OUTBREAKS LINKED BY TRACE-BACK AND WGS 287

Dow

nloa

ded

by A

ustr

alia

n N

atio

nal U

nive

rsity

fro

m w

ww

.lieb

ertp

ub.c

om a

t 04/

29/1

9. F

or p

erso

nal u

se o

nly.

69

date, trace-back to the individual farm level from egg stampsand invoices was not possible (Fig. 2).

Microbiological investigation

Salmonella was only detected from food (ready-to eatsalad items) and environmental samples collected from foodpremises in one outbreak (Table 1). Cross-contaminationwas suspected to be a contributing cause; so the 23 positivefood and environmental samples detected in this outbreakwere unable to indicate any one food as the source of con-tamination. A more detailed summary of each outbreak canbe found in the Supplementary Data and SupplementaryTables 1 and 2.

Environmental swabs were taken from a new layer shed onone of the farms supplying eggs to the grading facility andfrom the grading facility itself, from chicken cloacae, anesting rail, three egg conveyor belts, the preprocessed eggs,an egg pulp collection tub (prewash), and an egg pulp col-lection tub (postwash). Salmonella was cultured from theswab of the egg pulp collection tub (prewash). Of the threeisolates serotyped, two were Salmonella subsp. I 16:I,v:- andone was Salmonella Singapore.

In total, 37 isolates of the outbreak MLVA profile werewhole genome sequenced: 9 human isolates from the 2 out-breaks in the Australian Capital Territory, 13 human isolatesfrom the 2 outbreaks in Queensland, 12 human isolates fromthe 3 outbreaks in New South Wales, and 2 food isolates(cucumber and onion) and 1 environmental swab isolate(mixing bowl) from 1 of the outbreaks in New South Wales.The isolates were clustered together and highly related, with0–10 SNPs difference between all isolates tested (Fig. 3).

Discussion

WGS illustrates that human, food, and environmentalisolates from seven outbreaks of Salmonella Typhimurium03-26-13-08-523 across three Australian states and territoriesinvestigated between November and March 2016 were highlyrelated. While MLVA and epidemiological investigationsfirst identified the seven outbreaks and initiated investiga-tions, WGS was able to provide significantly more discrim-inatory detail to show that outbreaks across jurisdictions wererelated. While systemic retail level failures were importantcontributing factors in several of the outbreaks and we werenot able to definitively link the outbreaks through sampling,the WGS data and the food trace-back investigations suggestthat the outbreaks were linked to a common source, likely tobe eggs graded at the same facility. Salmonella Typhimurium03-26-13-08-523 was not isolated from any samples at theegg grading facility; however, the facility was only sampledon one occasion.

These seven Australian outbreaks demonstrate how WGScan help to definitively link cases over wide geographic areas.This suggests that these outbreaks, and many SalmonellaTyphimurium outbreaks occurring across Australia, (Oz-FoodNet Working Group, 2015) may not be isolated events,but associated with a common source. By linking more cases,WGS will help identify and prioritize these clusters for fur-ther investigation. If WGS is timely and cases can be linkedtogether soon after their onset of illness, an outbreak can bedetected and preventive measures taken sooner, ideally re-ducing the number of cases potentially affected. A limitation

Ta

ble

1.

Ou

tbrea

ks

of

Sa

lm

on

ella

Ty

ph

im

uriu

m03-2

6-1

3-0

8-5

23

in

th

eA

ustra

lia

nC

apita

lT

errito

ry

,N

ew

So

uth

Wa

les,

an

dQ

ueen

sla

nd

In

vestig

ated

Betw

een

No

vem

ber

2015

an

dM

arch

2016

Sta

teM

onth

and

year

of

inve

stig

ati

on

Set

ting

No.

of

case

sa

No.

of

labora

tory

-confirm

edca

ses

Susp

ecte

dfo

od

sourc

eF

ood

sam

pli

ng

Envi

ronm

enta

lsa

mpli

ng

Food/e

nvi

ronm

enta

lsa

mpli

ng

resu

lts

Egg

trace

-back

NS

WN

ovem

ber

2015

Res

taura

nt

40

7C

ori

ander

may

onnai

se,

conta

inin

gra

weg

gU

UN

oSalm

onel

ladet

ecte

dC

om

pan

yX

NS

WJa

nuar

y2016

Bak

ery

203

81

Mult

iple

food

item

s:cr

oss

-conta

min

atio

nsu

spec

ted

UU

Salm

onel

laT

yphim

uri

um

03-2

6-1

3-0

8-5

23

det

ecte

din

23

sam

ple

s

Com

pan

yX

NS

WJa

nuar

y2016

Aged

care

faci

lity

22

Unknow

n˚

˚—

No

trac

e-bac

kco

nduct

edA

CT

Feb

ruar

y2016

Res

taura

nt

54

Eggs

ben

edic

tU

˚N

oSalm

onel

ladet

ecte

dC

om

pan

yX

QL

DF

ebru

ary

2016

Mar

ket

12

6F

rench

crep

esU

UN

oSalm

onel

ladet

ecte

dC

om

pan

yX

AC

TF

ebru

ary

2016

Res

taura

nt

55

Egg

and

lett

uce

sandw

ich

U˚

No

Salm

onel

ladet

ecte

dC

om

pan

yX

QL

DM

arch

2016

Fes

tival

10

10

Unknow

n˚

˚—

No

trac

e-bac

kco

nduct

ed

aO

utb

reak

case

defi

nit

ions

can

be

found

inth

eS

upple

men

tary

Dat

a.A

CT

,A

ust

rali

anC

apit

alT

erri

tory

;N

SW

,N

ewS

outh

Wal

es;

QL

D,

Quee

nsl

and.

288 FORD ET AL.

Dow

nloa

ded

by A

ustr

alia

n N

atio

nal U

nive

rsity

fro

m w

ww

.lieb

ertp

ub.c

om a

t 04/

29/1

9. F

or p

erso

nal u

se o

nly.

70

of this Australian investigation is that WGS results were notavailable in real time and were not used in immediate publichealth action. As Australia moves toward surveillance ofnontyphoidal S. enterica using WGS, it will be importantthat Salmonella surveillance is harmonized nationally to ef-fectively detect multistate outbreaks and take rapid publichealth action.

In this study, at the retail food service level, interventionswere implemented in three outbreaks to address noncompli-ances with food safety standards (such as inadequate foodhandling practice) and to prevent further cases. Interventionssuch as these that target food handlers and the public to raiseawareness about safe handling of eggs and raw egg products

have helped to control Salmonella Typhimurium outbreaks(Stephens et al., 2007; Craig et al., 2013). At the productionlevel, no interventions occurred due to these outbreaks. Eggsused by the individual food premises in this study couldnot be traced back to a specific farm without egg cartons orpackaging. No specific reasons contributing to a higher prev-alence of Salmonella on eggs were identified and SalmonellaTyphimurium was not isolated at the grading facility.

Trace-back is often difficult, with a low prevalenceof Salmonella, particularly Salmonella Typhimurium, oncommercially produced eggs in Australia, and a failure torecover the outbreak strain on farms in 49% of SalmonellaTyphimurium outbreaks where testing occurred between

FIG. 2. Flowchart describing egg trace-back process with a large commercial egg producer.


Dow

nloa

ded

by A

ustr

alia

n N

atio

nal U

nive

rsity

fro

m w

ww

.lieb

ertp

ub.c

om a

t 04/

29/1

9. F

or p

erso

nal u

se o

nly.

71

2000 and 2011 (Daughtry et al., 2005; Chousalkar andRoberts, 2012; Moffatt et al., 2016). In addition, the eggdistribution network is complex. Reducing the burden ofegg-associated salmonellosis cases in Australia requirescontinued collaboration and communication between pub-lic health officers, food regulators, and industry groups tostrengthen control measures at the point of production, atretail and wholesale, and at the consumer level (Craig et al.,2013; Moffatt et al., 2016).

To prevent egg-associated outbreaks such as the onesdiscussed in this study, additional Salmonella control mea-sures across the supply chain, particularly at retail and onfarm, have been implemented over the last few years. Somefood safety regulators have implemented stronger require-ments around the production and service of raw egg foods,including minimum levels for pH and maximum length ofstorage. This has been complemented by additional mandatorytraining for retail food service in cleaning and sanitizing pro-cedures, use of raw egg foods, and general skills and knowledge.The Australian egg industry has increased the level of awarenessof human salmonellosis as a significant issue. Strengthenedindustry education and food safety plans have been im-plemented, along with many laying flocks now vaccinated forSalmonella Typhimurium (Groves et al., 2016). Continuedcontrol measures are important for further prevention.

A limitation of this Australian outbreak investigation isthat WGS was not performed on all human isolates duringthe time period with the outbreak MLVA and we could notexclude any outlier cases from the outbreaks. In addition,the impact of WGS data from food or environmental isolateson control measures at retail or production could not beevaluated, as it was not timely and no isolates from the egggrading facility were sequenced because none were typed asSalmonella Typhimurium.

In the United Kingdom, where isolates of Salmonellaare routinely sequenced, WGS has been used successfullyto investigate Salmonella outbreaks associated with eggs.Similar to the investigation described in this study, cases ofSalmonella Enteritidis PT14b across the United Kingdomand Europe were found to be related through WGS andtraced back to imported eggs from a German egg producer in2014 (Inns et al., 2015). In addition, WGS of human andfood isolates was used to retrospectively investigate anoutbreak of Salmonella Typhimurium DT8 in the UnitedKingdom associated with a raw egg mayonnaise (Ashtonet al., 2015). More recently, prospective WGS was used tohelp detect a Salmonella Enteritidis outbreak associatedwith eggs, including cases from the United Kingdom andSpain (Inns et al., 2017). In these examples and in the out-breaks described in this study, WGS has been a useful tool

FIG. 3. Maximum likelihood core genome SNP phylogeny of Salmonella Typhimurium 03-26-13-08-523 isolates fromoutbreaks in the ACT, QLD, and NSW, 2015–2016. Figure created with Interactive Tree of Life (iTOL) (https://itol.embl.de). ACT, Australian Capital Territory; SNP, single-nucleotide polymorphism.

290 FORD ET AL.

Dow

nloa

ded

by A

ustr

alia

n N

atio

nal U

nive

rsity

fro

m w

ww

.lieb

ertp

ub.c

om a

t 04/

29/1

9. F

or p

erso

nal u

se o

nly.

72

by providing improved discrimination, enhancing outbreakinvestigations occurring across states or countries, or as-sisting with trace-back and control measures.

While we were unable to link cases to eggs produced bycompany X through food and environmental testing at thegrading facility, epidemiological evidence, egg trace-back,and WGS data indicate a likely common source for the casesin multiple-point source outbreaks occurring over severalmonths in three Australian states and territories. The increasein egg-associated outbreaks and egg-associated salmonello-sis since 2000 remains a concern in Australia (Moffatt et al.,2016). WGS is a tool that will provide more evidence toimplement preventative measures at retail and at production,and will be most effective if WGS data are timely, nationallyharmonized, and integrated with food and environmentalisolates.

Acknowledgments

The authors would like to thank ACT Health, NSW Health,Queensland Health, and OzFoodNet. We would also liketo thank the laboratories that performed the serotyping,MLVA, and WGS, including the ICPMR, the MDU PHL,and Queensland Health Forensic and Scientific Services. Doh-erty Applied Microbial Genomics is funded by the Departmentof Microbiology and Immunology at The University of Mel-bourne. The National Health and Medical Research Council,Australia, funded a Practitioner Fellowship GNT1105905 toB.P.H. and Project Grant GNT1129770 to B.P.H., D.A.W., andM.D.K. Finally, we would like to thank Milica Stefanovic andSam McEwen for their contribution to this project. This researchis supported by an Australian Government Research TrainingProgram (RTP) Scholarship.

Disclosure Statement

No competing financial interests exist.

References

Angelo KM, Chu A, Anand M, Nguyen TA, Bottichio L, WiseM, Williams I, Seelman S, Bell R, Fatica M, Lance S,Baldwin D, Shannon K, Lee H, Trees E, Strain E, Gier-altowski L, Centre for Disease Control and Prevention(CDC). Outbreak of Salmonella Newport infections linkedto cucumbers—United States, 2014. MMWR Morb MortalWkly Rep 2015;64:144–147.

Ashton PM, Peters T, Ameh L, McAleer R, Petrie S, NairS, Muscat I, de Pinna E, Dallman T. Whole genome se-quencing for the retrospective investigation of an outbreakof Salmonella Typhimurium DT 8. PLoS Curr 2015;7:pii:ecurrents.outbreaks.2c05a47d292f376afc5a6fcdd8a7a3b6.

Chousalkar K, Gole V, Caraguel C, Rault JL. Chasing Salmo-nella Typhimurium in free range egg production system. VetMicrobiol 2016;192:67–72.

Chousalkar KK, Roberts JR. Recovery of Salmonella fromeggshell wash, eggshell crush, and egg internal contents ofunwashed commercial shell eggs in Australia. Poultry Sci2012;91:1739–1741.

Chousalkar KK, Sexton M, McWhorter A, Hewson K, MartinG, Shadbolt C, Goldsmith P. Salmonella Typhimurium inthe Australian egg industry: Multidisciplinary approach toaddressing the public health challenge and future directions.Crit Rev Food Sci Nutr 2017;57:2706–2711.

Craig AT, Musto JA, Carroll TF, Roberts-Witteveen AR,McCarthy RA, Wang Q. An investigation of SalmonellaTyphimuruim linked to contaminated eggs on the CentralCoast of NSW, 2008. NSW Public Health Bull 2013;24:99–101.

Daughtry B, Sumner J, Hooper G, Thomas C, Grimes T, HornR, Moses A, Pointon A. National food safety risk profileof eggs and egg products. North Sydney: Australian EggCorporation Limited, 2005. Available at: www.aecl.org/assets/www.aecl.org/outputs/National-Food-Safety-Risk-Profile-of-Eggs-and-Egg-Products-SAR-47A-Final-Report.pdf, accessedDecember 12, 2016.

Food Standards Australia New Zealand. Standard 4.2.5—Primary production and processing standard for eggs and eggproduct. Zealand. Canberra: Australian Government, 2012.Available at: www.legislation.gov.au/Details/F2011L00860,accessed December 12, 2016.

Ford L, Carter GP, Wang Q, Seemann T, Sintchenko V, Glass K,Williamson DA, Howard P, Valcanis M, Castillo CFS, Sait M,Howden BP, Kirk MD. Incorporating whole genome sequen-cing into public health surveillance: Lessons from prospectivesequencing of Salmonella Typhimurium in Australia. Food-borne Pathog Dis 2018. DOI: 10.1089/fpd.2017.2352. [Epubahead of print].

Ford L, Glass K, Veitch M, Wardell R, Polkinghorne B, DobbinsT, Lal A, Kirk MD. Increasing incidence of Salmonella inAustralia, 2000–2013. PLoS One 2016;11:e0163989.

Glass K, Fearnley E, Hocking H, Raupach J, Veitch M, Ford L,Kirk MD. Bayesian source attribution of salmonellosis inSouth Australia. Risk Anal 2016;36:561–570.

Groves PJ, Sharpe SM, Muir WI, Pavic A, Cox JM. Live andinactivated vaccine regimens against caecal Salmonella Ty-phimurium colonisation in laying hens. Aust Vet J 2016;94:387–393.

Inns T, Ashton PM, Herrera-Leon S, Lighthill J, FoulkesS, Jombart T, Rehman Y, Fox A, Dallman T, de Pinna E,Browning L, Coia JE, Edeghere O, Vivancos R. Prospectiveuse of whole genome sequencing (WGS) detected a multi-country outbreak of Salmonella Enteritidis. Epidemiol In-fect 2017;145:289–298.

Inns T, Lane C, Peters T, Dallman T, Chatt C, McFarland N,Crook P, Bishop T, Edge J, Hawker J, Elson R, Neal K,Adak GK, Cleary P, Outbreak Control Team. A multi-countrySalmonella Enteritidis phage type 14b outbreak associatedwith eggs from a German producer: ‘‘Near real-time’’ ap-plication of whole genome sequencing and food chain in-vestigations, United Kingdom, May to September 2014.Euro Surveill 2015;20:pii:21098.

Issenhuth-Jeanjean S, Roggentin P, Mikoleit M, GuibourdenchM, de Pinna E, Nair S, Fields Pl, Weill FX. Supplement 2008–2010 (no. 48) to the White-Kauffmann-Le Minor scheme. ResMicrobiol 2014;165:526–530.

Kumar S, Stecher G, Tamura K. MEGA7: Molecular evolu-tionary genetics analysis version 7.0 for bigger datasets. MolBiol Evol 2016;33:1870–1874.

Lindstedt BA, Vardund T, Aas L, Kapperud G. Multiple-locusvariable-number tandem-repeat analysis of Salmonella entericasubsp. enterica serovar Typhimurium using PCR multiplexingand multicolor capillary electrophoresis. J Microbiol Methods2004;59:163–172.

Moffatt CR, Musto J, Pingault N, Miller M, Stafford R, GregoryJ, Polkinghorne BG, Kirk MD. Salmonella Typhimurium andoutbreaks of egg-associated disease in Australia, 2001 to2011. Foodborne Pathog Dis 2016;13:379–385.


Dow

nloa

ded

by A

ustr

alia

n N

atio

nal U

nive

rsity

fro

m w

ww

.lieb

ertp

ub.c

om a

t 04/

29/1

9. F

or p

erso

nal u

se o

nly.

73

OzFoodNet Working Group. Monitoring the incidence and cau-ses of diseases potentially transmitted by food in Australia:Annual report of the OzFoodNet network, 2011. Commun DisIntell Q Rep 2015;39:E236–E264.

OzFoodNet Working Group. OzFoodNet quarterly report, 1April to 30 June 2014. Commun Dis Intell Q Rep 2016a;40:E290–E296.

OzFoodNet Working Group. OzFoodNet quarterly report, 1July to 30 September 2014. Commun Dis Intel Q Rep 2016b;40:E539–E544.

Standards Australia. 2009. AS 5013.10-2009. Microbiology offood and animal feeding stuffs—Horizontal method for thedetection of Salmonella spp. (ISO 6579:2002, MOD). Stan-dards Australia.

Stephens N, Sault C, Firestone SM, Lightfoot D, Bell C. Largeoutbreaks of Salmonella Typhimurium phage type 135 infectionsassociated with the consumption of products containing raw eggin Tasmania. Commun Dis Intell Q Rep 2007;31:118–124.

Address correspondence to:Laura Ford, MCHAM

National Centre for Epidemiology and Population HealthResearch School of Population Health

The Australian National UniversityCanberra, ACT 2601

Australia

E-mail: [email protected]

292 FORD ET AL.

Dow

nloa

ded

by A

ustr

alia

n N

atio

nal U

nive

rsity

fro

m w

ww

.lieb

ertp

ub.c

om a

t 04/

29/1

9. F

or p

erso

nal u

se o

nly.

74

75

Chapter 7. Impact of whole-genome sequencing on public

health surveillance two non-typhoidal Salmonella enterica

serotypes

7.1 Introduction

This chapter examines two case studies of how phylogenomic data from WGS can be used to

inform public health surveillance for zoonotic Salmonella infections. While most Salmonella

infections are transmitted through contaminated food, animals and contaminated

environmental sources are also important transmission pathways. In the rapidly developing

field of genomics, there is limited evidence on how to integrate epidemiological and genomic

data to better understand Salmonella serotypes for actionable public health prevention and

control strategies. This chapter aims to fill that gap, showing that phylogenomic data, combined

with epidemiological data, helps identify plausible sources of Salmonella infection from wildlife

and environmental reservoirs. The results provide a case for using phylogenomic analyses to

provide evidence for prevention and control strategies.

This chapter addresses the implications of whole genome sequencing for Salmonella

surveillance, prevention, and control. The paper has been published in Emerging Infectious

Diseases. Supplementary materials submitted with the paper can be found in Appendix 5.

7.2 Paper

Ford L, Ingle D, Glass K, Veitch M, Williamson DA, Harlock M, Gregory J, Stafford R, French

N, Bloomfield S, Grange Z, Conway ML, Kirk MD. A tale of two Salmonella serotypes in

Tasmania, Australia: how whole genome sequencing can inform prevention and control efforts.

Emerging Infectious Diseases. 2019;25(9):1690-1697, doi: 10.3201/eid2509.181811.

We used phylogenomic and risk factor data on isolates of Sal-monella enterica serovars Mississippi and Typhimurium de-finitive type 160 (DT160) collected from human, animal, and environmental sources to elucidate their epidemiology and disease reservoirs in Australia and New Zealand. Sequence data suggested wild birds as a likely reservoir for DT160; ani-mal and environmental sources varied more for Salmonella Mississippi than for Salmonella Typhimurium. Australia and New Zealand isolates sat in distinct clades for both serovars; the median single-nucleotide polymorphism distance for DT160 was 29 (range 8–66) and for Salmonella Mississippi, 619 (range 565–737). Phylogenomic data identified plausible sources of human infection from wildlife and environmental reservoirs and provided evidence supporting New Zealand–acquired DT160 in a group of travelers returning to Austra-lia. Wider use of real-time whole-genome sequencing in new locations and for other serovars may identify sources and routes of transmission, thereby aiding prevention and control.

Nontyphoidal Salmonella enterica subsp. enterica causes substantial illness and death throughout the

world (1,2). In Australia, rates of notified infection are

higher than in other high-income countries (3). Preventing infection by understanding sources and routes of transmis-sion and controlling outbreaks rapidly is key to reducing the rate of salmonellosis in Australia. Whole-genome se-quencing (WGS) is increasingly being used as a tool to help with prevention and control by investigating the relation-ship between isolates, sources of infection, and routes of transmission (4). Evidence shows that WGS is useful in foodborne nontyphoidal S. enterica outbreak detection and control (5–7).

On mainland Australia (Figure 1), Salmonella Ty-phimurium is the most commonly notified nontyphoidal S. enterica serovar. In contrast, Salmonella Mississippi is the most commonly notified nontyphoidal S. enterica se-rovar infecting residents of the island state of Tasmania, where 2.1% of the population of Australia resides (3,8). Most persons with Salmonella Mississippi who are resi-dents of mainland Australia have traveled to Tasmania or one of several Pacific Islands to which Salmonella Missis-sippi was endemic during their exposure period (9,10). In Tasmania, Salmonella Mississippi has been isolated from wildlife, and some evidence indicates that human infec-tions might result from environmental transmission more frequently than from foodborne transmission (9,11). A case–control study in Tasmania during 2001–2002 found that case-patients were more likely than controls to have had indirect contact with native birds, consumed untreated drinking water, and traveled within the state (9), although the sources of infection and vehicles of transmission are still largely unknown.

Salmonella Typhimurium definitive type 160 (DT160) has more recently emerged in Tasmania, while remaining rare in the rest of the country. In 2008, ten years after its emergence in humans in New Zealand, the first locally ac-quired case of DT160 was reported in Tasmania; unusual sparrow (Passer domesticus) deaths were observed in the same area in 2009, consistent with sparrow deaths in New

Whole-Genome Sequencing of Salmonella Mississippi and

Typhimurium Definitive Type 160, Australia and New Zealand

Laura Ford, Danielle Ingle, Kathryn Glass, Mark Veitch, Deborah A. Williamson, Michelle Harlock, Joy Gregory, Russell Stafford, Nigel French, Samuel Bloomfield,

Zoe Grange, Mary Lou Conway, Martyn D. Kirk

1690 Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 25, No. 9, September 2019

Author affiliations: The Australian National University, Acton, Australian Capital Territory, Australia (L. Ford, D. Ingle, K. Glass, M.D. Kirk); The University of Melbourne at the Peter Doherty Institute for Infection and Immunity, Melbourne, Victoria, Australia (D. Ingle, D.A. Williamson); Department of Health, Hobart, Tasmania, Australia (M. Veitch, M. Harlock); Victorian Department of Health and Human Services, Melbourne (J. Gregory); Queensland Department of Health, Brisbane, Queensland, Australia (R. Stafford); New Zealand Food Safety Science and Research Centre, Manawatu-Wanganui, New Zealand (N. French); Massey University, Manawatu-Wanganui (N. French); Quadram Institute, Norwich, UK (S. Bloomfield); University of California– Davis, Davis, California, USA (Z. Grange); Department of Primary Industries, Parks, Water and Environment, Hobart (M.L. Conway)

DOI: https://doi.org/10.3201/eid2509.181811

76

Whole-Genome Sequencing of Salmonella

Zealand in 2000 (12–15). Since then, DT160 has affected ≈50 Tasmania residents and ≈3,000 New Zealand residents and has been associated with wild bird deaths in both coun-tries (16–18). Although a case–control study conducted in 2001 in New Zealand suggested that handling of dead wild birds, contact with persons with diarrheal illness, and ingestion of fast food were associated with illness (12), the relationship between the Tasmania and New Zealand DT160 infections and the relationship between animal and human isolates in Tasmania is unknown. Accordingly, we aimed to use these 2 nontyphoidal S. enterica serovars as case studies to investigate how epidemiologic and genomic data can be integrated to better understand the geographic niche and transmission pathways of these organisms to subsequently improve prevention and control strategies.

Methods

Ethics Considerations and Data SourcesThe Australian National University Human Research Eth-ics Committee (2016/269) granted ethics approval for this project. We used data from the Australian National Notifi-able Diseases Surveillance System (18) and the New Zea-land Enteric Reference Laboratory (19) to examine trends in Salmonella Mississippi and DT160 in Australia and New Zealand. Population denominator data were obtained from the Australian Bureau of Statistics (8) and Statistics New Zealand (20). We used postcode of residence to determine Australian state or territory. Travel information was not available, so postcode might not represent place of acquisi-tion for all notifications.

We collated epidemiologic data obtained through en-hanced surveillance of cases of Salmonella Mississippi and

DT160 that were notified to the Department of Health and Human Services in Tasmania, the Department of Health and Human Services in Victoria, and Queensland Health. Case-patients were interviewed at the time of notification using a standardized questionnaire, and this information was collected under each jurisdiction’s public health leg-islation. For human isolates for which sequence data were available, we obtained the following case data fields from these questionnaires: type of case (sporadic, household, cluster, outbreak); hospitalization (yes/no); symptoms; travel; close contact with farm animals, native animals, birds, or pets (yes/no and type); lives on a rural property (yes/no); bushwalking or camping (yes/no); water source (public, private, or bottled); gardening (yes/no); swimming (yes/no); other risk factors (free text). Data on risk factors were collected for the week before illness onset.

Isolate SelectionFor Salmonella Mississippi, we selected 34 human iso-lates from Tasmania residents and 28 human isolates from residents of other states and territories in Australia that were referred for characterization to the Microbio-logical Diagnostic Unit Public Health Laboratory (MDU PHL) in Melbourne during January 1, 2011–December 31, 2015, for WGS (Appendix, https://wwwnc.cdc.gov/EID/article/25/9/18-1811-App1.pdf). We also selected all viable isolates with a recorded source from 42 animal sources and 18 environmental sources in the MDU PHL collection with an isolation date from January 1, 2000, through December 31, 2016; these isolates were all from Tasmania. For DT160, all viable isolates held in the MDU PHL collection at the beginning of 2016 were included in genomic analysis.

Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 25, No. 9, September 2019 1691

Figure 1. Relative locations of Australia, New Zealand, and Vanuatu.

77

RESEARCH

Sequencing and BioinformaticsMDU PHL performed DNA extraction and WGS for the Australia isolates. Sequence libraries were prepared us-ing NexteraXT and sequenced on the Illumina NextSeq500 platform (Illumina, https://www.illumina.com) with 150 bp paired-end reads. Reads are available from the National Cen-ter for Biotechnology Information Sequence Read Archive (PRJNA319593). Salmonella Typhimurium L2 (accession no. NC003197 [https://www.ncbi.nlm.nih.gov/nuccore/NC_003197.2]) was used as a reference for DT160, and because complete genomes were not publicly available, we used a local reference (AUSMDU00020775) for Salmonella Mississippi by assembling 1 of the isolates in this analysis (Appendix). We included 10 publicly available draft assem-blies from 2011 through 2013 from New Zealand and the United States (Appendix) in the Salmonella Mississippi anal-ysis (21,22) and 106 publicly available DT160 genomes from 1992 through 2012 from humans, wild birds, poultry, and bovine sources in New Zealand in the DT160 analysis (16).

Salmonella Mississippi and DT160 genomes were an-alyzed separately using Nullarbor version 2 (https://github.com/tseemann/nullarbor). Short-read data of the isolates were mapped to the reference using Snippy version 4.0-dev2 (https://github.com/tseemann/snippy). The 10 public-ly available Salmonella Mississippi draft assemblies were also mapped to the Salmonella Mississippi reference using Snippy , with the –ctgs parameter that enables mapping of assembly contigs to a reference. A core genome alignment was produced using Snippy-core (v4.0-dev 2), and the re-sulting full alignment was then filtered for recombination using Gubbins (23) using the weighted Robinson-Foulds method to estimate convergence with an initial 10 itera-tions. The resulting recombination-filtered core genome alignment of 8,573 bases for Salmonella Mississippi and 2,203 bases for DT160 was then passed to RAxML ver-sion 8.2.11 (24) to infer maximum-likelihood (ML) phylo-genetic trees, using the generalized time-reversible model with a γ-distribution to model site-specific rate variation and ascertain bias correction. For each analysis, we used

3 independent runs with 1,000 bootstrap pseudoreplicates to assess branch support, with the phylogenetic trees with the highest support across the 3 runs used as the final tree for each analysis. The pairwise single-nucleotide polymor-phism (SNP) distances between isolates were calculated from the recombination-filtered core genome alignment us-ing afa-pairwise.pl within Nullarbor. De novo genome as-semblies were generated using SPAdes version 3.12.0 (25), and the presence of known antimicrobial resistance genes was investigated using ABRicate (https://github.com/tsee-mann/abricate) in conjunction with the genome assemblies and the National Center for Biotechnology Information an-timicrobial resistance database with a minimum coverage of 90% and minimum identity of 90%.

Data AnalysisWe collated epidemiologic and SNP data in Microsoft Ex-cel 2013 (https://www.microsoft.com) and performed de-scriptive analyses in Stata SE 14 (https://www.stata.com). We used sequence data to explore hypotheses about the epidemiologic relatedness of isolates. We compared risk factors between DT160 and Salmonella Mississippi cases using a 2-sample test of proportions. Based on SNPs be-tween isolates of each serovar with known epidemiologic links (household or epidemiologic cluster), we considered isolates within 8 SNPs of each other for DT160 and 10 SNPs of each other for Salmonella Mississippi a putative phylogenetic cluster for further investigation.

Results

Salmonella MississippiDuring January 1, 2000–December 31, 2016, the me-dian annual notification rate of Salmonella Mississippi in Tasmania was 15 cases (range 12–24 cases) per 100,000 population, compared with a notification rate of 0.11 cases (range 0.007–0.16 cases) per 100,000 population on main-land Australia and 0.3 cases (range 0.16–0.47 cases) per 100,000 population in New Zealand (Figure 2). In Australia,


Figure 2. Notification rates for Salmonella enterica serovar Mississippi, Tasmania (A) and mainland Australia and New Zealand (B), 2000–2016. Rates are per 100,000 population.

78


934 (50.5%) of 1,851 notifications occurred in female pa-tients. Children 0–4 years of age were the most frequently notified age group (458 [24.7%]). Notification rates were higher in warmer months, similar to those for other non-Typhimurium serovars (26).

For sequenced Salmonella Mississippi isolates, the ML tree showed that the isolates from Vanuatu, United States, and New Zealand were distinctly different from the Austra-lia isolates (Figure 3); median SNP distance was 619 (range 565–737) between Australia and New Zealand isolates (Appendix Figure 1) and 1,625 (range 963–1,923) between Australia and Vanuatu or US isolates. An isolate from a 17-week-old mainland Australia resident with no history of overseas travel grouped on the phylogenetic tree with the isolates acquired in Vanuatu. Within the large Austra-lia group, the 114 isolates were diverse; median SNP dis-tance was 169 (range 3–649). Of these Australia isolates, 24 (21.1%) of 114 were within 10 SNPs of another isolate and grouped into 8 phylogenetic clusters. The Australia hu-man isolates grouped with Australia animal and environ-mental isolates over several years. We observed consider-able genetic diversity between isolates from various animal sources. For example, isolates from wombats were a me-dian of 87 (range 44–97) SNPs apart, and isolates from bo-vines were a median of 140 (range 10–201) SNPs apart. We did not detect any antimicrobial resistance genes in these

isolates, except for 1 human isolate from Tasmania that had the blaTEM-1 gene, which mediates resistance to ampicillin.

Travel data were available for 51 (82%) of 62 of case-patients in Australia residents for which an isolate was sequenced. Of these, 6 (12%) reported international travel to Vanuatu and 19 (37%) reported domestic travel during their incubation period; 8 traveled from mainland states to Tasmania, 9 traveled within Tasmania, 1 traveled from Queensland to South Australia, and 1 traveled from Victoria to Queensland. Three Tasmania isolates investi-gated as an epidemiologic cluster clustered genetically and temporally with an isolate from Victoria, for which no epi-demiologic data were available. Although 2 other phylo-genetic clusters included >1 human isolate, epidemiologic data were limited for these cases, and the infections were not clustered in time. Although not statistically significant, among case-patients who resided in or were known to have acquired infection in Tasmania and answered risk factor questions about exposures, a higher proportion of Salmo-nella Mississippi than DT160 case-patients reported drink-ing water from an untreated raw water source (i.e., tank, spring, or bore) (61% vs. 40%; p = 0.1) and camping (8% vs. 0%; p = 0.07) (Appendix Table 4). A similar propor-tion of DT160 and Salmonella Mississippi cases reported bushwalking (8% vs. 7.5%; p = 0.94), gardening (16% vs. 22%; p = 0.56), and swimming (16% vs. 18%; p = 0.84)


Figure 3. Maximum-likelihood phylogeny of 132 sequenced Salmonella enterica serovar Mississippi isolates from Australia and New Zealand and reference isolates, inferred from 8,573 core single-nucleotide polymorphisms. Nodes are labeled with isolation year, isolate source if nonhuman (all from Tasmania), and Australia state of acquisition or residence if human. Tree visualized with iTOL (https://itol.embl.de) and midpoint rooted. Scale bar indicates nucleotide substitutions per site. *State of residence was used instead of state of acquisition because no travel data were available. †Investigated as part of an epidemiologic cluster. A color version of this figure is available online (http://wwwnc.cdc.gov/EID/article/25/9/18-1811-F3.htm).

79

RESEARCH

exposures. Of the 13 case-patients who resided in or ac-quired their infection in an Australia state other than Tas-mania, 4 (31%) reported eating oysters during the expo-sure period, including 2 cases in a phylogenetic cluster; however, not all case-patients were asked specifically about oysters.

Salmonella Typhimurium DT160A total of 61 DT160 cases in Australia residents were noti-fied during 1999–2014. The 12 infections reported in Aus-tralia residents before 2008 were believed to be acquired overseas, including cases on an Australia–New Zealand cruise in 2003. Most Australia DT160 notifications had a postcode of residence in Tasmania, where the median an-nual rate per 100,000 population of DT160 in the 9 years from 1999 to 2007 was 0 cases and in the 7 years from 2008 to 2014 was 1.2, peaking at 2.8 in 2009, when 14 cases occurred. In New Zealand, the median annual rate per 100,000 population of DT160 in the 9 years from 1999 to 2007 was 6, peaking at 20.4 in 2001, when 791 cases occurred. In the 7 years from 2008 to 2014, the median annual rate was 1.6 per 100,000 population (Figure 4). Of all 61 Australia notifications during 1999–2014, a total of 34 (56%) occurred in females, and children aged 0–4 years were the most frequently notified age group (14 [23%]). Because of the small number of cases, we found no clear seasonal pattern.

Of Australia isolates, we sequenced 62 human and 30 animal isolates (20 from sparrows). The ML tree of Austra-lia and New Zealand isolates showed 2 distinct groups; 1 comprised isolates from humans and animals in Australia, and 1 comprised humans and animals from New Zealand and the 7 Australia residents who had reported travel to New Zealand (Figure 5). The median pairwise SNP dif-ference between the Australia and New Zealand groups was 29 (range 8–66), and the median pairwise SNP dif-ference within each group was 21 (Australia, range 2–56; New Zealand, range 0–55) (Appendix Figure 4). Within the Australia group, 45 (53%) of 85 isolates were within

8 SNPs of another isolate, and the isolates grouped into 8 phylogenetic clusters. Of these 8 phylogenetic clusters, 3 contained >2 isolates; isolates from humans and birds over several years clustered. No known antimicrobial resistance genes were detected among any of the isolates.

Epidemiologic risk factor data were available for 55 (90%) of the 61 DT160 cases in Australia residents from the 62 sequenced human isolates (1 person contributed 2 isolates, 13 days and 7 SNPs apart). Of these 55 persons, 7 (12%) of 59 acquired their infection in New Zealand, 6 in 2003 on an Australia–New Zealand cruise, and 1 in 2009. All others were residents of, or had traveled to, Tasmania, except for 1 case-patient, who was a resident of New South Wales and had no reported travel outside the state. Two separate household clusters were investigated in 2012 and were phylogenetically clustered. In 2015, five isolates were investigated as part of an epidemiologic cluster; however, no epidemiologic link was found, and they were subse-quently found not to be phylogenetically clustered (median SNPs 25.5, range 9–33). A higher proportion of DT160 case-patients than Salmonella Mississippi case-patients re-ported direct contact with wild or domestic animals (88% vs. 68%; p = 0.04) (Appendix Table 4).

DiscussionPhylogenomics plays a valuable role in identifying plau-sible sources of Salmonella infection from wildlife and environmental reservoirs. For both serovars considered in this article, the integration of clinical and genomic data enhanced existing evidence on source reservoirs by show-ing that human and animal or environmental isolates were genetically interspersed. Phylogenomic analysis revealed genetic diversity and persistence of Salmonella Mississippi strains in the environment and animals, suggesting it is en-demic with a broad range of host reservoirs in Tasmania that is persisting over time. As in other countries (16,27–29), genomic analysis provided evidence that wild birds are a source of human infection with DT160. No character-ized antimicrobial resistance genes were detected in any of


Figure 4. Salmonella enterica serovar Typhimurium definitive type 160 notification rate, Tasmania and mainland Australia (A) and New Zealand (B), 1999–2014. Rates are per 100,000 population.

80


the DT160 draft genome sequences, suggesting low or no antimicrobial resistance, consistent with other Salmonella strains found in wild birds (30,31).

In contrast to Salmonella Mississippi, Salmonella Ty-phimurium DT160 isolates from Australia and New Zea-land were similar, suggesting possible recent trans-Tasma-nia transmission of DT160 through wildlife, as well as its potential to spread from Tasmania to the Australia main-land. Further, the inferred population structure of these Australia strains provided evidence for this hypothesis of import and subsequent microevolution within the Australia strains. In New Zealand, DT160 has been transmitted be-tween multiple hosts, and humans have been infected from multiple sources (16). Other Salmonella serovars with wild bird reservoirs have been transmitted to cattle, pigs, sheep, and poultry (32). The incidence of human infection in Aus-tralia most likely would increase if DT160 were to become established in local animal food sources.

Epidemiologic evidence that 88% of DT160 case-patients had direct animal contact and the close genetic relatedness between Australia human and animal isolates suggest that DT160 in Tasmania is predominantly a lo-cally acquired zoonotic infection. Control measures should therefore focus on promoting hand hygiene after contact with wild birds and other animals, keeping food prepa-ration and eating areas free from birds, treating drinking water that is accessible to birds and other animals, and

appropriately cleaning and maintaining birdfeeders (32,33). In addition to monitoring the effect of such measures, con-tinued isolation, identification, and WGS of DT160 isolates from humans, animals, and the environment could be used to monitor emergence in other settings that pose particular risks to the food supply and provide early warning of the need for specific control measures.

Risk factors for Salmonella Mississippi are less evident, although the proportion of case-patients who reported drink-ing water from a private source (61%) was similar to that of an Australia case–control study of Salmonella Mississippi (63% of case-patients vs. 23% of controls reported drinking any untreated water; adjusted odds ratio 6.13, 95% CI 3.19–11.76) (9). The range of animal and environmental sources and the high genetic diversity make identifying control strat-egies difficult. Because genomic analyses of WGS data have detected clusters when epidemiologic links are obscured (7), prospective WGS of case isolates and integration of WGS data from human, animal, and environmental isolates could help to identify putative clusters for targeted epidemiologic investigation. Source attribution studies might enable quan-tification of the contribution of raw water as a vehicle for Salmonella Mississippi infection in Tasmania.

One limitation of this study is that our sampling frame for sequencing and analysis might not have produced a representative sample of all infections; however, we tried to maximize variability of sources and isolation dates.


Figure 5. Maximum-likelihood phylogeny of 198 sequenced Salmonella enterica serovar Typhimurium definitive type 160 isolates from Australia and New Zealand and reference isolates, inferred from 2,203 core single-nucleotide polymorphisms, Australia and New Zealand. Nodes are labeled with isolate type and isolation year. All Australian isolates are from Tasmania unless specified otherwise. Figure created with iTOL (https://itol.embl.de). Scale bar indicates nucleotide substitutions per site. *Specimens from the same person. †Investigated as part of an epidemiologic cluster. ‡Acquired in New South Wales. A color version of this figure is available online (http://wwwnc.cdc.gov/EID/article/25/9/18-1811-F5.htm).

81

RESEARCH

Although epidemiologic risk factor data were incomplete for sequenced human cases, sufficient data were available for us to generate hypotheses that could be further inves-tigated. Our case–case method is not as robust as a case–control study with a neutral control group, but we believe it is a reasonable point of comparison, emphasizing the difference in risk between these 2 serovars. Some more recent Australia DT160 notifications might not have been captured by the national notification system because phage typing for Salmonella Typhimurium is being phased out across Australia. However, we believe this omission would be a small number because phage typing continued in most of the country until 2016 (M. Valcanis, MDU PHL, pers. comm., 2018 Oct 10). Without phage typing, and as the use of WGS for Salmonella surveillance becomes more rou-tine, it will be difficult to compare new isolates with histori-cally phage-typed isolates that have not been sequenced.

We used SNP thresholds based on known epidemio-logic clusters to define putative phylogenetic clusters and to examine epidemiologic risk factors. Because SNPs de-pend on the reference genome and the isolates in the analy-sis, SNP thresholds for cluster analysis are likely to differ according to context. Local references were unavailable for both serovars in this study. We assembled a reference for Salmonella Mississippi using an isolate in this study, and the Salmonella Typhimurium isolate used was a median of 899 (range 883–921) SNPs from the DT160 isolates in our study. A closer reference for DT160 might have provided higher resolution of the relatedness of isolates. Few inter-national Salmonella Mississippi genomes were publicly available. Therefore, the relationship we found between Australia and New Zealand isolates might not be repre-sentative of all Salmonella Mississippi isolates in these 2 countries. Although beyond the scope of this study, iden-tifying the most recent common ancestor using Bayesian phylogeographic analyses would improve our understand-ing of endemic strains such as Salmonella Mississippi and the translocation of emerging strains such as DT160.

Wildlife can contribute to substantial rates of endemic and epidemic infection from Salmonella. For these 2 Sal-monella serovars with wild animal and environmental res-ervoirs in Australia and New Zealand, WGS combined with epidemiologic risk factor data provided some evidence for prevention and control efforts demonstrating the potential benefits of using WGS for prospective Salmonella surveil-lance. Real-time sequencing of these strains could help mon-itor emergence and identify clusters, enabling epidemiolo-gists to more accurately identify common risk factors and aid in source attribution. Local references, publicly available international genomes, phylogeographic analyses, additional tools to define a WGS cluster, and source-assigned case–control studies would improve our understanding of the epi-demiology of these 2 Salmonella serovars in this region.

AcknowledgmentsWe thank Kinley Wangdi for his assistance in generating the map in Figure 1. We also thank Anastasia Stylianopoulos, Marion Easton, Timothy Sloan-Gardner, and Robert Bell for their assistance in collecting case epidemiologic data. Finally, we thank John Bates, Mary Valcanis, Anders Gonçalves da Silva, Dieter Bulach, and Siobhan St. George for their assistance with sequence data and related metadata.

L.F. is supported by an Australian Government Research Training Program Scholarship. M.D.K. is supported by a National Health & Medical Research Council fellowship (APP1145997). D.A.W. is supported by a National Health & Medical Research Council fellowship (APP1123854).

About the AuthorMs. Ford is a PhD candidate in the Infectious Disease, Epidemiology, and Modelling group at the National Centre for Epidemiology and Population Health at the Australian National University. Her research interests include foodborne diseases and the application of sequencing for public health surveillance.

References 1. Majowicz SE, Musto J, Scallan E, Angulo FJ, Kirk M,

O’Brien SJ, et al.; International Collaboration on Enteric Disease ‘Burden of Illness’ Studies. The global burden of nontyphoidal Salmonella gastroenteritis. Clin Infect Dis. 2010;50:882–9. https://doi.org/10.1086/650733

2. Havelaar AH, Kirk MD, Torgerson PR, Gibb HJ, Hald T, Lake RJ, et al.; World Health Organization Foodborne Disease Burden Epidemiology Reference Group. World Health Organization global estimates and regional comparisons of the burden of foodborne disease in 2010. PLoS Med. 2015;12:e1001923. https://doi.org/10.1371/journal.pmed.1001923

3. Ford L, Glass K, Veitch M, Wardell R, Polkinghorne B, Dobbins T, et al. Increasing incidence of Salmonella in Australia, 2000–2013. PLoS One. 2016;11:e0163989. https://doi.org/10.1371/ journal.pone.0163989

4. Deng X, den Bakker HC, Hendriksen RS. Genomic epidemiology: whole-genome-sequencing-powered surveillance and outbreak investigation of foodborne bacterial pathogens. Annu Rev Food Sci Technol. 2016;7:353–74. https://doi.org/10.1146/ annurev-food-041715-033259

5. Simon S, Trost E, Bender J, Fuchs S, Malorny B, Rabsch W, et al. Evaluation of WGS based approaches for investigating a food-borne outbreak caused by Salmonella enterica serovar Derby in Germany. Food Microbiol. 2018;71:46–54. https://doi.org/10.1016/j.fm.2017.08.017

6. Leekitcharoenphon P, Nielsen EM, Kaas RS, Lund O, Aarestrup FM. Evaluation of whole genome sequencing for outbreak detection of Salmonella enterica. PLoS One. 2014;9:e87991. https://doi.org/10.1371/journal.pone.0087991

7. Waldram A, Dolan G, Ashton PM, Jenkins C, Dallman TJ. Epidemiological analysis of Salmonella clusters identified by whole genome sequencing, England and Wales 2014. Food Microbiol. 2018;71:39–45. https://doi.org/10.1016/ j.fm.2017.02.012

8. Australian Bureau of Statistics. 3101.0 Australian demographic statistics, Sep 2017. Table 4. Estimated resident population, states and territories (number) [dataset]. 2018 Mar 22 [cited



2018 Mar 28]. http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/3101.0Sep%202017?OpenDocument

9. Ashbolt R, Kirk MD. Salmonella Mississippi infections in Tasmania: the role of native Australian animals and untreated drinking water. Epidemiol Infect. 2006;134:1257–65. https://doi.org/10.1017/S0950268806006224

10. OzFoodNet Working Group. Monitoring the incidence and causes of diseases potentially transmitted by food in Australia: annual report of the OzFoodNet Network, 2009. Commun Dis Intell Q Rep. 2010;34:396–426.

11. Hall GV, D’Souza RM, Kirk MD. Foodborne disease in the new millennium: out of the frying pan and into the fire? Med J Aust. 2002;177:614–8.

12. Thornley CN, Simmons GC, Callaghan ML, Nicol CM, Baker MG, Gilmore KS, et al. First incursion of Salmonella enterica serotype typhimurium DT160 into New Zealand. Emerg Infect Dis. 2003;9:493–5. https://doi.org/10.3201/eid0904.020439

13. Alley MR, Connolly JH, Fenwick SG, Mackereth GF, Leyland MJ, Rogers LE, et al. An epidemic of salmonellosis caused by Salmonella Typhimurium DT160 in wild birds and humans in New Zealand. N Z Vet J. 2002;50:170–6. https://doi.org/ 10.1080/00480169.2002.36306

14. Wildlife Health Australia. WHA Fact sheet: Salmonella Typhimurium DT160 in house sparrows [cited 2018 Mar 28]. https://wildlifehealthaustralia.com.au/Portals/0/Documents/ FactSheets/Avian/Salmonella%20Typhimurium%20DT160%20in%20House%20Sparrows%20in%20Australia.pdf

15. Grillo T. Australian Wildlife Health Network. Animal Health Surveillance Quarterly Report. 2009;14(3):6–8 [cited 2018 Mar 29]. http://www.sciquest.org.nz/elibrary/edition/5546

16. Bloomfield SJ, Benschop J, Biggs PJ, Marshall JC, Hayman DTS, Carter PE, et al. Genomic analysis of Salmonella enterica serovar Typhimurium DT160 associated with a 14-year outbreak, New Zealand, 1998–2012. Emerg Infect Dis. 2017;23:906–13. https://doi.org/10.3201/eid2306.161934

17. Grillo T, Post L. Australian Wildlife Health Network. Animal Health Surveillance Quarterly Report. 2010;14(4):6–8 [cited 2018 Mar 29]. http://www.sciquest.org.nz/elibrary/edition/5545

18. Australian Government Department of Health. National Notifiable Diseases Surveillance System. 2018 [cited 2018 Aug 6]. http://www9.health.gov.au/cda/source/cda-index.cfm

19. New Zealand Ministry of Health. Enteric Reference Laboratory. Human Salmonella isolates. 2018 [cited 2018 Apr 20]. https://surv.esr.cri.nz/enteric_reference/human_salmonella.php

20. Statistics New Zealand Infoshare. Estimated resident population by age and sex (1991+) (Annual-Jun). 2018 [cited 2018 Jun 26]. http://archive.stats.govt.nz/infoshare/?url=/infoshare/ViewTable.aspx&pxID=d25f4bca-d895-4ef4-8cf5-c6368155c39e

21. Grange ZL, Biggs PJ, Rose SP, Gartrell BD, Nelson NJ, French NP. Genomic epidemiology and management of Salmonella in island ecosystems used for takahe conservation. Microb Ecol. 2017;74:735–44. https://doi.org/10.1007/s00248-017-0959-1

22. Gupta R, Schmidtke A, Sabol A, Castillo D, Ribot E, Trees E. Salmonella enterica subsp. enterica serovar Mississippi str. 2010K–1406, whole genome shotgun sequencing project. Accession ALPQ000000001: GenBank; 2012 [cited 2018 Apr 23]. https://www.ncbi.nlm.nih.gov/nuccore/ALPQ00000000.1

23. Croucher NJ, Page AJ, Connor TR, Delaney AJ, Keane JA, Bentley SD, et al. Rapid phylogenetic analysis of large samples of recombinant bacterial whole genome sequences using Gubbins. Nucleic Acids Res. 2015;43:e15. https://doi.org/10.1093/nar/gku1196

24. Stamatakis A. RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics. 2014;30:1312–3. https://doi.org/10.1093/bioinformatics/btu033

25. Bankevich A, Nurk S, Antipov D, Gurevich AA, Dvorkin M, Kulikov AS, et al. SPAdes: a new genome assembly algorithm and its applications to single-cell sequencing. J Comput Biol. 2012;19:455–77. https://doi.org/10.1089/cmb.2012.0021

26. Fearnley EJ, Lal A, Bates J, Stafford R, Kirk MD, Glass K. Salmonella source attribution in a subtropical state of Australia: capturing environmental reservoirs of infection. Epidemiol Infect. 2018;146:1903–8. https://doi.org/10.1017/S0950268818002224

27. Piccirillo A, Mazzariol S, Caliari D, Menandro ML. Salmonella Typhimurium phage type DT160 infection in two Moluccan cockatoos (Cacatua moluccensis): clinical presentation and pathology. Avian Dis. 2010;54:131–5. https://doi.org/10.1637/ 8969-062509-Case.1

28. Tizard IR, Fish NA, Harmeson J. Free flying sparrows as carriers of salmonellosis. Can Vet J. 1979;20:143–4.

29. Lawson B, Howard T, Kirkwood JK, Macgregor SK, Perkins M, Robinson RA, et al. Epidemiology of salmonellosis in garden birds in England and Wales, 1993 to 2003. EcoHealth. 2010;7:294–306. https://doi.org/10.1007/s10393-010-0349-3

30. Hughes LA, Shopland S, Wigley P, Bradon H, Leatherbarrow AH, Williams NJ, et al. Characterisation of Salmonella enterica serotype Typhimurium isolates from wild birds in northern England from 2005–2006. BMC Vet Res. 2008;4:4. https://doi.org/10.1186/1746-6148-4-4

31. Afema JA, Sischo WM. Salmonella in wild birds utilizing protected and human impacted habitats, Uganda. EcoHealth. 2016; 13:558–69. https://doi.org/10.1007/s10393-016-1149-1

32. Tizard I. Salmonellosis in wild birds. Seminars in Avian and Exotic Pet Medicine. 2004;13:50–66.

33. Friend M. Salmonellosis. In: Friend M, Franson J, editors. Field manual of wildlife diseases. General field procedures and diseases of birds. Washington (DC): US Geological Survey, US Department of the Interior; 1999. p. 99–110.

Address for correspondence: Martyn D. Kirk, National Centre for Epidemiology and Population Health, Research School of Population Health, The Australian National University, Canberra, Australian Capital Territory, 2601 Australia; email: [email protected]


84

Chapter 8. Discussion and conclusion

In this thesis I have presented the results of seven studies investigating the epidemiology of

Salmonella in Australia and the impacts of WGS on public health surveillance and outbreak

investigation. Here, I synthesize the findings and implications and explore the future of whole

genome sequencing for Salmonella surveillance in Australia.

8.1 Findings and implications

The studies in my thesis offer findings and implications for one or more of my three research

themes below.

Trends and sources of infection

A main finding of my thesis is that Salmonella infections in Australia have been increasing

over time. I found that the incidence of Salmonella notifications to public health departments

in Australia increased to 53.0 cases per 100,000 population over 2000–2013. There were

national annual increases of 6% (95% confidence intervals 6%-7%) for Salmonella

Typhimurium and 3% (95% CI 2%-3%) for non-Typhimurium Salmonella. I also found that

there was an increase in the number of reported outbreaks of Salmonella in Australia over

2001–2016, peaking at 116 outbreaks reported in 2014. These findings, together with

comparatively high rates of Salmonella compared to other high-income countries (CDC, 2017;

EFSA and ECDC 2018; Government of Canada, 2019), indicate that prevention and control

measures over these time periods have not been successful in Australia.

As well as examining the incidence of national notifications and outbreaks, I estimated the

burden of domestically acquired illness in the Australian community, taking into account

underreporting. Circa 2015, I estimated 90,833 (90% CrI 51,583-158,256) cases of non-

typhoidal Salmonella. This is a 49% increase in the estimated rate of salmonellosis cases

since 2010 due to all non-typhoidal Salmonella (Kirk et al., 2014), reflecting an increase in the

national number of notifications from 2013–2015, compared with 2006–2010 (Department of

Health, 2019). This confirms that the burden of Salmonella incidence the Australian community

is comparatively high and is increasing.

The community estimates also show that salmonellosis can be severe, resulting in 4,312 (90%

3,335-11,091) hospitalizations. This is a 38% increase in the rate of estimated hospitalizations

from circa 2010 (Kirk et al., 2014), reflecting an increase in the number of hospitalizations

reported by the Australian Institute of Health and Welfare from 2011–12 to 2014–15 (AIHW,

2016), with 2006–2010 hospitalization data. The estimate suggests that approximately 5% of

85

reported salmonellosis cases result in hospitalization, compared to my finding that 15% of

cases in salmonellosis outbreaks over 2001–2016 were hospitalized (with varying

hospitalization rates by serotype). This indicates that illness in outbreaks may be more severe

than sporadic illness, the lower proportion of hospitalizations in the community may reflect an

average proportion across different serotypes (whereas the outbreak proportion is largely

driven by the Salmonella Typhimurium hospitalization rate), or that the community estimates

may underestimate hospitalization numbers.

I estimated that there were 6,133 (90% CrI 3,335-11,091) cases of IBS and 6,080 (90% CrI

1,420-15,959) cases of ReA following salmonellosis circa 2015. Additionally, I estimated 19

(90% CrI 15-22) deaths due to Salmonella circa 2015, a conservative estimate that only

increased about 5% from circa 2010. This is due to the fact that I was unable to get updated

mortality data and had to adjust the 2001–2010 data for the 2015 population. Despite this

limitation, the number of hospitalizations, sequelae, and deaths from Salmonella in the

community circa 2015 are high.

In addition to finding that Salmonella is increasing, I showed that there are geographic

differences in the rates of infection, trends over time, and serotypes causing infection. While

states and territories with tropical climatic zones (Northern Territory, Queensland, and

Western Australia) had some of the lowest estimated annual increases, they had the highest

rates of infection. In addition, these states and territories, along with the island state Tasmania,

had a higher proportion of notifications that were non-Typhimurium Salmonella. This supports

research that environmental factors are important for survival of certain Salmonella serotypes

(Akil et.al. 2014; Finn et al., 2013; Martinez-Urtaza et al., 2004; Milazzo et al., 2016; Stephen

and Barnett, 2016) and implies that different interventions to prevent and control infection for

different geographic regions are likely important for reducing the rates of infection.

I found that 79% of reported salmonellosis outbreaks in Australia over 2000–2016 were of

foodborne or suspected foodborne transmission, with 84% of these due to Salmonella

Typhimurium. Salmonella Typhimurium was the most commonly notified serotype in Australia,

responsible for over 40% of notifications over 2000–2013. With 72% of Salmonella infections

(Vally et al., 2014) and 79% of outbreaks in Australia estimated to be transmitted through

contaminated food, prevention and control measures along the food chain need to be

improved to reduce rates of infection and outbreaks, particularly rates of Salmonella

Typhimurium.

From my work, it is clear that eggs are an important source of salmonellosis in Australia. In

investigating the food vehicles associated with Salmonella outbreaks in Australia, I found that

eggs and egg-containing foods were the responsible food vehicle in 50% of the 476 (61%)

86

foodborne or suspected foodborne outbreaks where a food vehicle was listed. While eggs

have also been identified as an important cause of Salmonella outbreaks in the USA and

Canada (Belanger et al., 2005; Jackson et al., 2013), the proportion of outbreaks where eggs

were the reported food vehicle was higher in Australia, with S. Typhimurium the responsible

serotype in 95% of these outbreaks. I also identified eggs as an important source of

Salmonella Typhimurium cases in the Australian Capital Territory through a case-case

analysis of WGS clusters and an investigation of several outbreaks of one MLVA pattern

associated with eggs. These findings support a previous study, which showed an increasing

proportion of foodborne Salmonella outbreaks linked to eggs over 2001–2011, with 90%

caused by Salmonella Typhimurium (Moffatt et al., 2016). While control measures across the

egg supply chain have been implemented in Australia over the last several years, the high

proportion of outbreaks due to eggs implies that more needs to be done to prevent egg-related

salmonellosis. There is hope that the recent introduction of the vaccination of laying flocks for

Salmonella Typhimurium will help to reduce the burden of egg-related illness (Groves et al.,

2016; McWhorter and Chousalkar, 2018).

In this thesis, I identified other key food vehicles of Salmonella infection. While these key food

vehicles may not be representative of the food sources in all foodborne outbreaks, and less-

commonly associated foods may not always be identified by investigators, identifying these

help to target food safety standards and policies. There were few outbreaks associated with

sprouts, nuts, fruit, and some fresh salad produce over 2001–2016, however these outbreaks

tended to be larger than outbreaks associated with many other food vehicles, likely due to the

wide distribution of these food products. This finding implies that food safety standards and

policies targeted at the foods causing large outbreaks may be important to prevent

salmonellosis associated with fresh produce.

Although most Salmonella infections are transmitted through contaminated food, I showed

that wildlife and the environment can be an important reservoir of infection for some serotypes.

I found that wild birds are a likely reservoir for Salmonella Typhimurium DT160 infection, while

Salmonella Mississippi is endemic with a broad range of host reservoirs in the Tasmanian

environment. Epidemiological investigations for these serotypes should be focused on

environmental and animal risk factors, in order to target specific sources for prevention and

policy measures.

Costs of illness and costs of WGS

A main finding of my thesis is that Salmonella illness results in substantial costs to the

Australian society. I estimated that circa 2015, Salmonella in Australia cost AUD 124.4 million

(90% CrI 107.4-143.1 million), or AUD 146.8 million (90% CrI 127.8-167.9 million) when IBS

87

and ReA following Salmonella infection were included. As 72% of salmonellosis is estimated

to be transmitted through contaminated food (Valley et al., 2014), the burden from foodborne

Salmonella illness alone is estimated to cost AUD 88.9 million (90% CrI 63.9-112.4 million)

and AUD 104.8 million (90% CrI 75.5-132.3 million) with costs for IBS and ReA. As costs

associated with foodborne illness make up most of the overall cost, targeted interventions

across the food chain to reduce salmonellosis would help to reduce costs.

I estimated the costs of illness as costs for health care usage, lost productivity, and premature

mortality by 4 age groups, showing the contribution of direct and indirect costs to the total cost.

Examining where costs are highest can indicate the most beneficial areas for potential

intervention. For example, the highest cost per case was for those in the 65 years and older

age group, particularly costs associated with health care use and premature mortality.

Therefore, reducing illness in this group would have a significant impact on reducing overall

costs. Reducing illness in those 65 years and older could include addressing identified risk

factors including proton pump inhibitor use, high chicken consumption, and environmental

exposures related to rural and remote living (Chen et al., 2016), as well as strengthening food

safety and infection control policies in aged care facilities (Kirk et al., 2008).

These much-needed estimates fill a gap in the literature by quantifying the costs of

salmonellosis and foodborne salmonellosis in Australia. As reducing Salmonella

contamination in foods, animals, and the environment can be costly, these estimates will help

to quantify the economic trade-offs for controlling salmonellosis in different environments

(McLinden et al., 2014; Roberts, 1988). In particular, those working on Australia’s Foodborne

Illness Reduction Strategy 2018–2021+ (Food Regulation, 2018), can use these estimates to

quantify the effects of certain actions at reducing illness, helping to prioritize interventions and

policies.

I used the costs of illness estimates in this thesis to assess the costs of WGS in terms of how

the technology may assist public health action. I found that WGS data would need to prevent

approximately 275 (2%) of all notified cultured Salmonella cases each year to be cost-equal

to traditional typing methods, or 1,550 (10%) of notified Salmonella cases to be cost-equal to

PCR in Australia in 2018. WGS has potential to save significant amounts of money in

prolonged outbreaks if data can be used to detect outbreaks and implement interventions

earlier. My findings are in line with a recent Canadian economic analysis of Salmonella

detection in fresh produce, poultry, and eggs using WGS, which found that WGS will result in

a net benefit of millions of dollars (Jain et al., 2019). My analysis is an important piece of

evidence as Australia and other countries consider transitioning completely to WGS for routine

surveillance of Salmonella and other foodborne disease pathogens.

88

Impacts of WGS on surveillance and outbreak detection

A key finding of this thesis is the value of WGS data in showing the relatedness of cases’

Salmonella isolates. WGS linked an additional 9% of Salmonella Typhimurium isolates in the

Australian Capital Territory thought to be sporadic to at least one other isolate, compared with

MLVA and epidemiological data. WGS data also showed the relatedness of Salmonella

Typhimurium DT160 and Salmonella Mississippi isolates in Tasmania and New Zealand and

confirmed that cases from seven point-source Salmonella Typhimurium outbreaks occurring

across a wide geographic area were related. The greater sensitivity and specificity of WGS

data compared to traditional typing methods increases the speed at which outbreaks can be

detected and provides more detail for the outbreaks to be understood (WHO, 2018).

Therefore, this accurate, robust, reliable and high throughput typing method (Ashton et al.,

2016) is useful in responding to threats quickly, with the ultimate aim of controlling and

preventing infection.

WGS can also be useful in detecting antimicrobial resistance in Salmonella isolates, tracking

changes in microbial populations, detecting resistant strains of public health importance,

informing clinical therapy decisions, guiding policy recommendations, and assessing the

impact of resistance containment interventions (McDermott et al., 2016; WHO, 2019). The

Salmonella isolates sequenced for my thesis had little antimicrobial resistance detected. There

was one Salmonella Mississippi isolate with blaTEM-1 gene that mediates resistance to

ampicillin. While there is an increasing incidence of antimicrobial resistance in Salmonella

isolates in Australia, the finding of low levels of resistance in this thesis is likely due to the low

prevalence of resistance in S. Typhimurium, the fact that many multi-drug resistant Salmonella

were related to overseas travel, and a conservative use of antibiotics in Australian food

animals (Australian Government, 2017; Williamson et al., 2018).

In this thesis, I show the value of linking human cases to food, environmental, and animal

sources. Not only can WGS show the relatedness of isolates from food vehicles to human

cases, but it can also provide more detail on human cases to help statistically associate food

vehicles and illness (Byrne et al., 2014; Inns et al., 2017; Waldram et al., 2018). Although no

Salmonella Typhimurium was detected on eggs, at the egg grading facility, or on an egg farm

to link it to the human cases in my study of seven point-source outbreaks, WGS did show that

isolates from foods positive for Salmonella in one of the outbreaks (likely contaminated

through cross contamination) were highly related to human outbreak cases. WGS data also

elucidated animal and environmental reservoirs for DT160 and Mississippi, particularly when

combined with epidemiological data. Human and sparrow DT160 isolates were genetically

related, with a high proportion of cases reporting direct contact with animals. The genetic

89

heterogeneity of Salmonella Mississippi isolates was mirrored in the fact that no risk factors

were significantly higher for Salmonella Mississippi cases compared to DT160 cases,

suggesting multiple potential host reservoirs. Real-time sequencing of Salmonella could help

monitor emergence and identify clusters, enabling epidemiologists to more accurately identify

common risk factors and aid in source attribution.

My thesis includes seminal research on the prospective use of WGS for surveillance of

Salmonella Typhimurium in a small public health unit. While WGS had been used for

surveillance and outbreak investigation prior to this study, nothing had been published on

using the data effectively at a local level for these purposes. I identified several issues to

consider when implementing routine WGS for Salmonella surveillance in a public health unit,

including timeliness, national harmonization, integration of sequence with epidemiological

data, and integration of food and environmental sequence data with human data. WGS will be

most effective if data are received by the public health unit in a timely manner, so that risk

factor data for related cases can be obtained and public health action can be implemented

when a source of illness or outbreak is detected.

National harmonization of sequence data and Salmonella surveillance are key to effective use

of WGS data for surveillance and outbreak investigations. As Australian public health

reference laboratories are state-based, this work was novel in that it required two state

reference laboratories to routinely share sequence data using a virtual machine and a single

bioinformatic pipeline for analysis. My work has highlighted the importance of a sustainable

system for the sharing of standardized data and analysis among public health laboratories and

state health departments nationally. As I found highly related Salmonella isolates across large

geographic areas in Australia, and highly related Salmonella isolates have been found across

countries (Inns et al., 2017), WGS data needs to be shared and surveillance harmonized in

order to detect and investigate these widespread outbreaks.

I identified difficulties in integrating WGS data with epidemiological data, as public health units’

surveillance databases did not have the capacity to store and use WGS data. Overlaying and

analysing these two data sets together was time consuming without a fit-for-purpose computer

program. Although I used programs like Phandango (https://jameshadfield.github.io/phandango/#/),

Figtree (http://tree.bio.ed.ac.uk/software/figtree/), Microreact (https://microreact.org/), and iTOL

(https://itol.embl.de/) to integrate and visualize the data, I was unable to identify one accessible

program that could easily integrate the data and visualize all risk factors of infection among

WGS-based clusters. In addition, public health unit staff required additional training to use and

understand WGS data to detect and investigate outbreaks. Ideally, resources should be devoted

to the development and training of an integrated surveillance system that can identify WGS

https://jameshadfield.github.io/phandango/#/

http://tree.bio.ed.ac.uk/software/figtree/

https://microreact.org/

https://itol.embl.de/

90

clusters of human and non-human isolates and allow for the input and visualization of

epidemiological information, including risk factor data.

As I have shown with Salmonella Typhimurium outbreaks, DT160, and Salmonella Mississippi,

not only is it important to integrate human sequence data and epidemiological data, but

integrating food, environmental, and animal sequence data can also help to elucidate food

vehicles and animal or environmental reservoirs of infection. While integrated surveillance is

resource-intensive, it provides comprehensive information for surveillance and greatly

improves the ability to rapidly trace sources of contamination and estimate the relative

contribution of different sources to human infection (WHO, 2018). The high discriminatory

power of WGS data will be extremely useful in integrated surveillance systems and particularly

helpful when WGS data detects small clusters where epidemiological links are obscured.

However, integrated surveillance requires the willingness and capacity for cross-sectoral data

sharing, which is something to work towards in Australia.

8.2 The future of Salmonella surveillance and research in Australia

Since the start of my PhD, the epidemiology of Salmonella in Australia has begun to change.

National notification rates, which steadily increased from 49.1 cases per 100,000 in 2012 to

74.7 cases per 100,000 in 2016, dropped to 66.7 cases and then to 57.6 cases per 100,000

in 2017 and 2018 respectively (Department of Health, 2019). Not only have rates dropped

since 2016, but the proportion of Salmonella Typhimurium notifications has also decreased

from between 44% and 48% over 2012–2015 to 33% in 2016 and 35% in 2017 (Department

of Health, 2019). As my findings show that eggs are the main food vehicle in Salmonella

Typhimurium outbreaks, the egg industry and some food regulators have attributed the

decrease in Salmonella Typhimurium rates to improvements along the egg food chain, in

particular administration of an aroA-deletion live Salmonella Typhimurium vaccine in layer

flocks and the Egg Standards of Australia quality assurance program (Druce, 2018; Groves et

al., 2016, 2017).

Alongside this decrease in notifications, and the proportion of notifications that are Salmonella

Typhimurium, there has recently been an increase in notifications of domestically-acquired

Salmonella Enteritidis (Australian Associated Press, 2019; “More egg recalls”, 2019; NSW

Health, 2019). While the New South Wales Department of Primary Industry states that the

Australian egg industry is free from Salmonella Enteritidis, between May 2018 and March

2019, a total of 149 New South Wales residents, and additional cases in Victoria, Queensland

and Tasmania, were reported to be unwell following a Salmonella Enteritidis outbreak linked

to contaminated eggs (FSANZ, 2019; NSW Department of Primary Industries, n.d.; NSW

Health, 2019). This may be an example of how controlling one Salmonella serotype can lead

91

to the emergence of another serotype to fill the niche left by those control measures (Cogan

and Humphrey, 2003). Continued monitoring and research on the emergence of Salmonella

Enteritidis in Australia will be necessary for understanding and controlling infection. WGS will

be an important tool in this monitoring to help determine lineage, detect outbreaks, and identify

sources of infection.

WGS is revolutionizing public health surveillance of Salmonella in Australia. The use of WGS

data requires changes, including new or adapted systems and tools for storing and analysing

information for surveillance purposes. WGS also requires collaboration unlike previous typing

methods have, with epidemiologists, microbiologists and bioinformaticians working closely

together, as well as increased communication and harmonization among public health

laboratories and public health departments across different jurisdictions. In addition,

communication and collaboration globally has become more important, as WGS can help to

break down borders making global surveillance of foodborne illness easier with standardized

and shared data. By offering highly discriminatory data for human cases, as well as for

reservoirs and vehicles of infection, real-time national and integrated WGS data will ultimately

improve the speed and effectiveness of public health action.

In my thesis, I have highlighted areas of future research that could assist in understanding the

epidemiology of Salmonella in Australia, with the ultimate aim to control and prevent illness.

Further research on the sources of Salmonella illness, particularly for emerging or high

incidence serotypes would be useful. I have described the vehicles of infection related to

Salmonella outbreaks, but it is unknown whether these are the same vehicles causing

sporadic infection. Serotype specific case control studies in the community would help to

identify vehicles for sporadic infection, providing evidence beyond outbreak studies for the

implementation of control measures. WGS strain-specific case control studies may also be

possible and useful in the future if a particular strain is widespread in the community.

My cost of illness estimates for Salmonella offer the opportunity for further research on the

cost effectiveness of different interventions, both along the food supply chain, and at the

surveillance level (e.g. WGS). The methods I used to calculate the number of cases that need

to be prevented for WGS to be cost-equal and the potential cost-savings using WGS in

outbreak scenarios can be applied to other countries and other foodborne pathogens as they

consider transitioning to WGS for surveillance. Future research on the costs of salmonellosis

to industry and trade would also be helpful in quantifying the full burden of Salmonella in

Australia.

For a genetically heterogeneous serotype such as Salmonella Mississippi, which likely has a

multiple potential host reservoirs, Bayesian source attribution research using WGS data could

92

prove helpful in identifying the relative importance of different host reservoirs and sources of

infection. While there have been few published Bayesian source attribution studies using

WGS, one study concludes that WGS is transforming source attribution studies and provides

guidance on the interpretation of WGS data (Pightling et al., 2018). A major source attribution

study for Campylobacter in New Zealand initiated a sentinel surveillance site and evidence for

the advocacy of more rigorous controls on foodborne pathways, helping to more than halve

notification rates (Sears et al., 2011). Access to farm or producer level Salmonella data for

source attribution studies would be the most useful for identifying sources of infection and

implementing prevention and control measures.

8.3 Conclusion

The findings in this thesis increases Australian agencies’ understanding of the epidemiology

of non-typhoidal Salmonella. My thesis has made a unique contribution to the knowledge

around effective and actionable use of WGS data for public health surveillance and outbreak

investigation of Salmonella. I have provided justification for routine use of sequencing by

finding that WGS data is sensitive, specific, can help detect and solve outbreaks, elucidates

sources of infection, and could result in significant cost-savings. I found that Salmonella,

particularly infections transmitted through contaminated food, have been an increasing and

costly problem in Australia. While recent control measures, including vaccination and changes

in the egg industry appear to have reduced rates of serotype Typhimurium, future efforts

should focus on control measures over multiple food vehicles, environmental sources, and

strains. Monitoring emerging strains will also be imperative. WGS is a revolutionary tool that

will enhance public health surveillance and outbreak investigations, hopefully reducing rates

of infection in Australia.

93

References Abelson, P., Forbes, M.P. and Hall, G. (2006). The annual cost of foodborne illness in Australia.

Canberra (Australia): Commonwealth Department of Health and Ageing.

AIHW (Australian Institute of Health and Welfare) (Accessed 11 June 2016). Separation

statistics by principal diagnosis (ICD-10-AM 7th edition), Australia, 2011-12 to 2012-13

and 2013-14 to 2014-15. Retrieved from: http://www.aihw.gov.au/hospitals-

data/principal-diagnosis-data-cubes/.

Australian Associated Press (2019, March 22). Eggs recalled in four states due to ‘exotic’

Salmonella strain. The Guardian. Retrieved from https://www.theguardian.com/.

Australian Government. (2017). Antimicrobial Resistance. AMR and animal health in Australia.

Retrieved March 27, 2019, from, https://www.amr.gov.au/about-amr/amr-

australia/amr-and-animal-health-australia.

Akil, L., Anwar Ahmad, H. and Reddy, R.S. (2014). Effects of climate change on Salmonella

infections. Foodborne Pathogens and Disease, 11(12):974–980.

Allard, M.W., Strain, E., Melka, D., Bunning, K., Musser, S.M., Brown, E.W. and Timme, R.

(2016). The practical value of food pathogen traceability through building a whole-

genome sequencing network and database. Journal of Clinical Microbiology,

54(8):1975–1983.

Arnott, A., Wang, Q., Bachmann, N., Sadsad, R., Biswas, C., Sotomayor, C., Howard, P.,

Rockett, R., Wiklendt, A., Iredell, J.R. and Sintchenko, V. Multidrug-resistant

Salmonella enterica 4,[5],12:i- sequence type 34, New South Wales, Australia, 2016–

2017. Emerging Infectious Diseases, 24(4):751–753.

Ashbolt, R., Givney, R., Gregory, J.E., Hall, G., Hundy, R., Kirk, M., McKay, I., Meuleners, L.,

Millard, G., Raupach, J., Roche, P., Prasopa-Plaizier, N., Sarna, M.K., Stafford, R.,

Tomaska, N., Unicomb, L., Williams, C. and the OzFoodNet Working Group. (2002).

Enhancing foodborne disease surveillance across Australia in 2001: the OzFoodNet

Working Group. Communicable Diseases Intelligence Quarterly Report, 26(3):375–

406.

Ashbolt, R. and Kirk, M.D. (2006). Salmonella Mississippi infections in Tasmania: the role of

native Australian animals and untreated drinking water. Epidemiology and Infection,

134(6):1257–1265.

Ashton, P.M., Nair, S., Peters, T.M., Bale, J.A., Powell, D.G., Painset, A., Tewolde, R.,

Schaefer, U., Jenkins, C., Dallman, T.J., de Pinna, E.M. and Grant, K.A. (2016).

Identification of Salmonella for public health surveillance using whole genome

sequencing. PeerJ, 4:e1752.

94

Ashton, P.M., Peters, T., Ameh, L., McAleer, R., Petrie, S., Nair, S., Muscat, I., de Pinna, E.

and Dallman, T. Whole genome sequencing for the retrospective investigation of an

outbreak of Salmonella Typhimurium DT 8. PLoS Current,

7:ecurrents.outbreaks.2c05a47d292f376afc5a6fcdd8a7a3b6.

Batz, M., Hoffmann, S., Morris, J.G. Jr. (2014). Disease-outcome trees, EQ-5D scores, and

estimated annual losses of quality-adjusted life years (QALYs) for 14 foodborne

pathogens in the United States. Foodborne Pathogens and Disease, 11(5):395-402.

Bender, J.B., Hedberg, C.W., Boxrud, D.J., Besser, J.M., Wicklund, J.H., Smith, K.E. and

Osterholm, M.T. (2001). Use of molecular subtyping in surveillance for Salmonella

enterica serotype Typhimurium. New England Journal of Medicine, 344:189–195.

Berends, B.R., Van Knapen, F., Mossel, D.A., Burt, S.A. and Snijders, J.M. (1998). Impact on

human health of Salmonella spp. on pork in The Netherlands and the anticipated

effects of some currently proposed control strategies. International Journal of Food

Microbiology, 44(3):219–229.

Bergholz, T.M., den Bakker, H.C., Katz, L.S., Silk, B.J., Jackson, K.A., Kucerova, Z., Joseph,

L.A., Turnsek, M., Gladney, L.M., Halpin, J.L., Xavier, K., Gossack, J., Ward, T.J.,

Frace, M. and Tarr, C.L. Determination of evolutionary relationships of outbreak-

associated Listeria monocytogenes strains of serotypes 1/2a and 1/2b by whole-

genome sequencing. Applied and Environmental Microbiology, 82(3):928–938.

Borgdorf, M.W. and Motarjemi, Y. (1997). Surveillance of foodborne diseases: What are the

options? World Health Statistics Quarterly, 50(1–2):12–23.

Byrne, L., Fisher, I., Peters, T., Mather, A., Thomson, N., Rosner, B., Bernard, H., McKeown,

P., Cormican, M., Cowden, J., Aiyedun, V. and Lane C. (2014). A multi-country

outbreak of Salmonella Newport gastroenteritis in Europe associated with watermelon

from Brazil, confirmed by whole genome sequencing: October 2011 to January 2012.

Euro Surveillance, 19(31):6–13.

Carmada, A., Circella, E., Pupillo, A., Legretto, M., Marino, M., Pugliese, N. (2015). Pulsed-

field gel electrophoresis of Salmonella enterica. In: Jordon K., Dalmasso M. (eds) Pulse

Field Gel Electrophoresis. Methods in Molecular Biology vol 1301. Humana Press, New

York, NY.

CDC (Centres for Disease Control and Prevention). (2014). Foodborne diseases active

surveillance network (FoodNet): FoodNet Surveillance Report for 2012 (Final Report).

Atlanta: U.S Department of Health and Human Services. Retrieved from

https://www.cdc.gov/foodnet/PDFs/2011_annual_report_508c.pdf.

CDC (2016). Whole genome sequencing (WGS). PulseNet. Retrieved March 13, 2017, from

https://www.cdc.gov/pulsenet/pathogens/wgs.html.

95

CDC. (2017). Annual tables of infectious disease data. National Notifiable Infectious Diseases

and Conditions: United States. Retrieved February 14, 2019, from

https://www.cdc.gov/nndss/infectious-tables.html.

CDC. (2019). Salmonella: Information for healthcare professionals and laboratories. Retrieved

March 2, 2017, from https://www.cdc.gov/Salmonella/general/technical.html.

CDC. (2019a). Salmonella: Questions and answers. Retrieved January 23, 2017, from

https://www.cdc.gov/Salmonella/general/index.html.

CDC. (2019b). Salmonella: Serotypes and the importance of serotyping Salmonella. Retrieved

February 23, 2017, from https://www.cdc.gov/Salmonella/reportspubs/Salmonella-

atlas/serotyping-importance.html.

CDC. (2019c). Whole Genome Sequencing. Division of Foodborne, Waterborne, and

Environmental Disease. Retrieved February 18, 2019, from

https://www.cdc.gov/ncezid/dfwed/keyprograms/tracking-foodborne-illness-wgs.html.

Chen, Y., Glass, K., Liu, B., Hope, K. and Kirk, M. (2016). Salmonella infection in middle-aged

and older adults: Incidence and risk factors from the 45 and up study. Foodborne

Pathogens and Disease, 13(12):689–694.

Cogan, T.A. and Humphrey, T.J. (2003). The rise and fall of Salmonella Enteritidis in the UK.

Journal of Applied Microbiology, 94:114S–119S.

Cronquist, A.B., Mody, R.K., Atkinson. R., Besser, J., Tobin D’Angelo, M., Hurd, S., Robinson,

T., Nicholson, C. and Mahon, B.E. (2012). Impacts of culture-independent diagnostic

practices on public health surveillance for bacterial enteric pathogens. Clinical

Infectious Diseases, 54(Suppl 5):S432–S439.

Dallman, T.J., Byrne, L., Launders, N., Glen, K., Grant, K.A. and Jenkins, C. The utility and

public health implications of PCR and whole genome sequencing for the detection and

investigation of an outbreak of Shiga toxin-producing Escherichia coli serotype group

O26:H11. Epidemiology and Infection, 143(8):1672–1680.

den Bakker, H.C., Allard, M.W., Bopp, D., Brown, E.W., Fontana, J., Iqbal, Z., Kinney, A.,

Limberger, R., Musser, K.A., Shudt, M., Strain, E., Wiedmann, M. and Wolfgang, W.J.

(2014). Rapid whole-genome sequencing for surveillance of Salmonella enterica

serovar Enteritidis. Emerging Infectious Diseases, 20(8):1306–1314.

Deng, X., den Bakker, H.C. and Hendriksen, R.S. (2016). Genomic epidemiology: Whole-

genome-sequencing-powered surveillance and outbreak investigation of foodborne

bacterial pathogens. Annual Review of Food Science and Technology, 7:353–374.

Deng, X., Shariat, N., Driebe, E.M., Roe, C.C., Tolar, B., Trees, E., Keim, P., Zhang, W.,

Dudley, E.G., Fields, P.I. and Engelthaler, D.M. (2015). Comparative analysis of

subtyping methods against a whole-genome-sequencing standard for Salmonella

enterica serotype Enteritidis. Journal of Clinical Microbiology, 53(1):212–218.

96

Department of Health. (2017). OzFoodNet – Enhancing surveillance for foodborne disease in

Australia. Retrieved February 17, 2017, from

http://www.health.gov.au/internet/main/publishing.nsf/Content/cdna-ozfoodnet.htm.

Department of Health. (2018). Links to State and Territory Public Health Legislation, the

Quarantine Act, and the National Health Security Act 2007. Retrieved February 11,

2017, from http://www.health.gov.au/internet/main/publishing.nsf/Content/cda-state-

legislation-links.htm.

Department of Health (2018a). OzFoodNet Reports. Retrieved February 18, 2019, from

http://www.health.gov.au/internet/main/publishing.nsf/Content/cdna-ozfoodnet-

reports.htm#annual.

Department of Health. (2019). National Notifiable Disease Surveillance System. Notifications

of a selected disease by age group, sex and year. Available at:

http://www9.health.gov.au/cda/source/rpt_4_sel.cfm.

Druce, A. (2018, July 4). Australian Eggs crack a smile as salmonellosis rates show continued

improvement. The Land. Retrieved from https://www.theland.com.au.

Duff, S.B., Scott, E.A., Mafilios, M.S., Todd, E.C., Krilov, L.R., Geddes, A.M. and Ackerman,

S.J. (2013). Cost-effectiveness of a targeted disinfection program in household

kitchens to prevent foodborne illness in the United States, Canada, and the United

Kingdom. Journal of Food Protection, 66(11):2103–2015.

ECDC (European Centre for Disease Prevention and Control). (2015). Technical consultation:

ECDC strategy to harness whole genome sequencing for cross-border outbreak and

surveillance. Retrieved March 13, 2017, from https://ecdc.europa.eu/en/news-

events/technical-consultation-ecdc-strategy-harness-whole-genome-sequencing-

cross-border.

ECDC. (2016). Expert opinion on whole genome sequencing for public health surveillance.

Stockholm: ECDC. Retrieved from,

https://ecdc.europa.eu/sites/portal/files/media/en/publications/Publications/whole-

genome-sequencing-for-public-health-surveillance.pdf.

ECDC. (2018). Monitoring the use of whole-genome sequencing in infectious disease

surveillance in Europe. Stockholm, ECDC. Retrieved from

https://ecdc.europa.eu/sites/portal/files/media/en/publications/Publications/whole-

genome-sequencing-for-public-health-surveillance.pdf.

EFSA (European Food Safety Authority) and ECDC. (2018). The European Union summary

report on trends and sources of zoonoses, zoonotic agents and food-borne outbreaks

in 2017. EFSA Journal, 16(12):5500.

97

ERS (Economic Research Service), U.S. Department of Agriculture (USDA) (2013). Cost

estimates of foodborne illness. Retrieved February, 3 2017, from

http://ers.usda.gov/data-products/cost-estimates-of-foodborne-illnesses.aspx.

ESR (The Institute of Environmental Science and Research Ltd). (2017). Notifiable diseases

in New Zealand Annual Report 2016. Porirua, NZ: ESR.

Ethelberg, S., Sørensen, G., Kristensen, B., Christensen, K., Krussell, L., Hempel-Jørgensen,

A., Perge, A. and Nielsen, E.M. (2007). Outbreak with multi-resistant Salmonella

Typhimurium DT104 linked to carpaccio, Denmark, 2005. Epidemiology and Infection,

135(6):900–907.

Finn, S., Condell, O., McClure, P., Amezquita, A. and Fanning S. (2013). Mechanisms of

survival, responses, and sources of Salmonella in low-moisture environments.

Frontiers in Microbiology, 4:331.

Food Regulation (2018). Australia’s Foodborne Illness Reduction Strategy 2018–2020+.

Retrieved March 15, 2019 from

http://foodregulation.gov.au/internet/fr/publishing.nsf/Content/aus-foodborne-illness-

reduction-strategy-2018-2021-Jun-2018.

FSANZ (Food Standards Australia New Zealand) (2019). Salmonella Enteritidis (SE) linked

to eggs. Retrieved March 29, 2019, from

http://www.foodstandards.gov.au/consumer/safety/Pages/Salmonella-Enteritidis-

linked-to-eggs.aspx.

Ford, L., Kirk, M., Glass, K. and Hall, G. (2015). Sequelae of foodborne illness caused by 5

pathogens, Australia, circa 2010. Emerging Infectious Diseases, 20(11):1865–1871.

Gavin, C., Simons, R.R.L., Berriman, A.D.C., Moorhouse, D., Snary, E.L., Smith, R.P. and Hill,

A.A. (2018). A cost-benefit assessment of Salmonella-control strategies in pigs reared

in the United Kingdom. Preventive Veterinary Medicine, 160:54–62.

Giannella, R.A. (1996). Chapter 21: Salmonella. In: Baron, S., editor, Medical Microbiology

(4th ed.). Galveston: University of Texas Medical Branch at Galveston.

Government of Canada. (2019). Reported cases from 1924 to 2016 in Canada – Notifiable

diseases online. Retrieved February 11, 2019, from http://dsol-smed.phac-

aspc.gc.ca/notifiable/charts?c=pl.

Graham, R.M.A., Hiley, L., Rathnayake, I.U. and Jennison, A.V. (2018). Comparative

genomics identifies distinct lineages of S. Enteritidis from Queensland, Australia. PLoS

One, 13(1): e0191042.

Groseclose, S.L., German, R.R. and Nsubuga, P. (2010). Evaluating public health surveillance.

In: Lee, L.M., Teutsch, S.M., Thacker, S.B., St. Louis, M.E., editors, Principles &

Practice of Public Health Surveillance (3rd ed.), 166-197. Oxford University Press, New

York.

98

Groves, P.J., Sharpe, S.M., Muir, W., Pavic A. and Cox, J.M. (2016). Live and inactivated

vaccine regimens against caecal Salmonella Typhimurium colonisation in laying hens.

Australian Veterinary Journal, 94(10):387–393.

Groves, P.J., Harris, T. and Sharpe, S.M. (2017). Administration of a live attenuated

Salmonella vaccine using an inactivated oil-emulsion vaccine as a vehicle for

commercial chicken flocks. Animal Production Science, 58(7):1316–1322.

Haagsma, J.A., Siersema, P.D., De Wit, N.J., Havelaar, A.H. (2010). Disease burden of post-

infectious irritable bowel syndrome in the Netherlands. Epidemiology and Infection,

138(11):1650-1656.

Hald, T., Aspinall, W., Devleesschauwer, B., Cooke, R., Corrigan, T., Havellar, A.H., Gibb,

H.J., Torgerson, P.R., Kirk, M.D., Angulo, F.J., Lake, R.J., Speybroeck, N. and

Hoffmann, S. (2016). World Health Organization estimates of the relative contributions

of food to the burden of disease due to selected foodborne hazards: A structured expert

elicitation. PLoS One, 19(11):e0145839.

Hall, G., Yohannes, K., Raupach, J., Becker, N. and Kirk, M. (2008). Estimating community

incidence of Salmonella, Campylobacter, and Shiga-toxin producing Escherichia

infections in Australia. Emerging Infectious Diseases, 14(10):1601–1609.

Havelaar, A.H., Kirk, M.D., Torgerson, P.R., Gibb, H.R., Hald, T., Lake, R.J., Praet, N.,

Bellinger, D.C., de Silva, N.R., Gargouri, N., Speybroeck, N., Cawthorne, A., Mathers,

C., Stein, C., Angulo, F.J. and Devleesschauwer, B., on behalf of World Health

Organization Foodborne Disease Burden Epidemiology Reference Group. (2015).

World Health Organization global estimates and regional comparisons of the burden of

foodborne disease in 2010. PLoS Medicine, 12(12):e1001923.

Hendriksen, R.S., Vieira, A.R., Karlsmose, S., Lo Fo Wong, D.M., Jensen, A.B., Wegner, H.C.

and Aarestrup, F.M. (2011). Global monitoring of Salmonella serovar distribution from

the World health Organization Global Foodborne Infections Network Country Data

Bank: results of quality assured laboratories from 2001 to 2007. Foodborne Pathogens

and Diseases, 8(8):887–900.

Heymann, D., (Ed.). (2015). Control of communicable disease manual (20th ed.). Washington

DC: American Public Health Association. p. 532-539.

Hoffmann, S., Batz, M.B. and Morris, J.G. Jr. (2012). Annual cost of illness and quality-adjusted

life year losses in the United States due to 14 foodborne pathogens. Journal of Food

Protection, 75(7):1292–1302.Defibaugh-Chavez, S., Kucerova, Z., Sabol, A., Roache,

K., Trees, E., Simmons, M., Wasilenko, J., Kubota, K., Pouseele, H., Klimke, W.,

Besser, J., Brown, E., Allard, M. and Gerner-Smidt, P. Implementation of nationwide

real-time whole-genome sequencing to enhance listeriosis outbreak detection and

investigation. Clinical Infectious Diseases, 63(3):380–386.

99

Holmes, A., Allison, L., Ward, M., Dallman, T.J., Clark, R., Fawkes, A., Murphy, L. and Hanson,

M. (2015). Utility of whole-genome sequencing of Escherichia coli O157 for outbreak

detection and epidemiological surveillance. Journal of Clinical Microbiology,

53(11):3565–3573.

Inns, T., Ashton, P.M., Herrera-Leon, S., Lighthill, J., Foulkes, S., Jombart, T., Rehman, Y.,

Fox, A., Dallman, T., Pinna E, D.E., Browning, L., Coia, J.E., Edeghere, O. and

Vivancos, R. (2017). Prospective use of whole genome sequencing (WGS) detected a

multi-country outbreak of Salmonella Enteritidis. Epidemiology and Infection,

145(2):289–298.

Inns, T., Lane, C., Peters, T., Dallman, T., Chatt, C., McFarland, N., Crook, P., Bishop, T.,

Edge, J., Hawker, J., Elson, R., Neal, K., Adak, G.K. and Cleary, P. (2015). A multi-

country Salmonella Enteritidis phage type 14b outbreak associated with eggs from a

German producer: ‘near real-time’ application of whole genome sequencing and food

chain investigations, United Kingdom, May to September 2014. Euro Surveillance,

20(16):21098.

Iwamoto, M., Huang, J.Y., Cronquist, A.B., Medus, C., Hurd, S., Zansky, S., Dunn, J., Woron,

A.M., Oosmanally, N., Griffin, P.M., Besser, J., Henao, O.L. and Centers for Disease

Control and Prevention. (2015). Bacterial enteric infections detected by culture-

independent diagnostic tests—Foodnet, United States, 2012-2014. MMWR Morbidity

and Mortality Weekly Report, 64(9):252–257.

Issenhuth-Jeanjean, S., Roggentin, P., Mikoleit, M., Guibourdenche, M., de Pinna, E., Nair, S.,

Fields, P.I. and Weill, F.X. (2014). Supplement 2008-2010 (no. 48) to the White-

Kauffmann-Le Minor scheme. Research in Microbiology, 165(7):526–530.

Jackson, B.R., Tarr, C., Strain, E., Jackson, K.A., Conrad, A., Carleton, H., Katz, L.S., Stroika,

S., Hannah Gould, L., Mody, R.K., Silk, B.J., Beal, J., Chen, Y., Timme, R., Doyle, M.,

Fields, A., Wise, M., Tillman, G., Defibaugh-Chavez, S., Kucerova, Z., Sabol, A.,

Roache, K., Trees, E., Simmons, M., Wasilenko, J., Kubota, K., Pouseele, H., Klimke,

W., Besser, J., Brown, E., Allard, M. and Gerner-Smidt, P. (2016). Implementation of

nationwide real-time whole-genome sequencing to enhance listeriosis outbreak

detection and investigation. Clinical Infectious Diseases, 63(3):380–386.

Jain, S., Mukhopadhyay, K. and Thomassin, P.J. (2019). An economic analysis of Salmonella

detection in fresh produce, poultry, and eggs using whole genome sequencing

technology in Canada. Food Research International, 116:802–809.

Jenkins, C. (2015). Whole-genome sequencing data for serotyping Escherichia coli-it’s time

for a change! Journal of Clinical Microbiology, 53(8):2402–2403.

Jensen, A., Nielsen, E.M., Bjorkman, J.T., Jensen, T., Muller, L., Persson, S., Bjerager, G.,

Perge, A., Krause, T.G., Kiil, K., Sorensen, G., Andersen, J.K., Molbak, K. and

100

Ethelberg, S. (2016). Whole-genome sequencing used to investigate a nationwide

outbreak of listeriosis caused by ready-to-eat delicatessen meat, Denmark, 2014.

Clinical Infectious Diseases, 63(1):64–70.

Jones, T.F., Ingram, L.A., Cieslak, P.R., Vugia, D.J., Tobin-D’Angelo, M., Hurd, S., Medus, C.,

Cronquist, A. and Angulo, F.J. (2008). Salmonellosis outcomes differ substantially by

serotype. The Journal of Infectious Diseases, 198(1):109-114.

Ju, W., Cao, G., Rump, L., Strain, E., Luo, Y., Timme, R., Allard, M., Zhao, S., Brown, E. and

Meng, J. (2012). Phylogenetic analysis of non-O157 Shiga-toxin producing Escherichia

coli strains by whole genome sequencing. Journal of Clinical Microbiology,

50(12):4123–4127.

Kirk, M.D., Roberts, L. and Horvath, J. (2008). Understanding gastroenteritis in elderly

residents of aged-care facilities. Medical Journal of Australia, 189(9):476–477.

Kirk, M., Ford, L., Glass, K. and Hall, G. (2014). Foodborne illness, Australia, circa 2000 and

circa 2010. Emerging Infectious Diseases, 20(11):1857–1864.

Kirk, M.D., Pires, S.M., Black, R.E., Caipo, M., Crump, J.A., Devleesschauwer, B., Dopfer, D.,

Fazil, A., Fischer-Walker, C.L., Hald, T., Hall, A.J., Keddy, K.H., Lake, R.J., Lanata,

C.F., Torgerson, P.R., Havelaar, A.H. and Angulo, F.J. (2015). World Health

Organization Estimates of the Global and Regional Disease Burden of 22 Foodborne

Bacterial, Protozoal, and Viral Diseases, 2010: A Data Synthesis. PLoS Medicine,

12(12):e1001921.

Korsgaard, H., Madsen, M., Feld, N.C., Mygind, J. and Hald. T. (2009). The effects, costs and

benefits of Salmonella control in the Danish table-egg sector. Epidemiology and

Infection, 137(6):828–836.

Kwong, J.C., Mercoulia, K., Tomita, T., Easton, M., Li, H.Y., Bulach, D.M., Stinear, T.P.,

Seemann, T. and Howden, B.P. (2016). Prospective whole-genome sequencing

enhances national surveillance of Listeria monocytogenes. Journal of Clinical

Microbiology, 54(2):333–342.

Lawson, L.G., Jensen, J.D., Christiansen, P. and Lund, M. (2009). Cost-effectiveness of

Salmonella reduction in Danish abattoirs. International Journal of Food Microbiology,

134(1–2):126–32.

Leekitcharoenphon, P., Nielsen, E.M., Kaas, R.S., Lund, O. and Aarestrup, F.M. (2014).

Evaluation of whole genome sequencing for outbreak detection of Salmonella enterica.

PLoS One, 9(2):e87991.

Lindsay, C., Flint, J., Lily, K., Hope, K., Wang, Q., Howard, P., Sintchenko, V. and Durrheim,

D.N. (2018). Retrospective use of whole genome sequencing to better understand an

outbreak of Salmonella enterica serovar Mbandaka in New South Wales, Australia.

Western Pacific Surveillance and Response Journal, 9(2):20–25.

101

Majowicz, S.E., Musto, J., Scallan, E., Angulo, F.J., Kirk, M., O’Brien, S.J., Jones, T.F., Fazil,

A., Hoekstra, R.M. and International Collaboration on Enteric Disease ‘Burden of

Illness’ Studies. (2010). The global burden of nontyphoidal Salmonella gastroenteritis.

Clinical Infectious Diseases, 50(6):882-889.

Marcus, R., Varma, J.K., Medus, C., Boothe, E.J., Anderson, B.J., Crume, T., Fullerton, K.E.,

Moore, M.R., White, P.L., Lyszkowicz, E., Voetsch, A.C., Angulo, F.J. and Emerging

Infections Program FoodNet Working Group. (2007). Re-assessment of risk factors for

sporadic Salmonella serotype Enteritidis infections: a case-control study in five

FoodNet Sites, 2002-2003. Epidemiology and Infection, 135(1):84–92.

Marder, E.P., Cieslak, P.R., Cronquist, A.B., Dunn, J., Lathrop, S., Rabatsky-Her, T., Ryan, P.,

Smith, K., Tobin-D’Angelo, M., Vugia, D.J., Zansky, S., Holt, K.G., Wolpert, B.J., Lynch,

M., Tauxe R. and Geissler, A.L. (2017). Incidence and trends of infection with

pathogens transmitted commonly through food and the effect of increasing use of

culture-independent diagnostic tests on surveillance – Foodborne Diseases Active

Surveillance Network, 10 U.S. sites, 2013–2016. MMWR Morbidity and Mortality

Weekly Report, 66(15):397–403.

Martinez-Urataza, J., Saco, M., de Novo, J., Perez-Pineiro, P., Peiteado, J., Lozano-Leon, A.

and Garcia-Martin, O. (2004). Influence of environmental factors and human activity on

the presence of Salmonella serovars in a Maine Environment. Applied and

Environmental Microbiology, 70(4):2089–2097.

May, F.J., Stafford, R.J., Carroll, H., Robson, J.M.B., Vohra, R., Nimmo, G.R., Bates, J., Kirk,

M.D., Fearnley E.J. and Polkinghorne B.G. (2017). The effects of culture independent

diagnostic testing on the diagnosis and reporting of enteric bacterial pathogens in

Queensland, 2010 to 2014. Communicable Diseases Intelligence, 41(3):E223–E230.

McDermott, P.F., Tyson, G.H., Kabera, C., Chen, Y., Li C., Folster, J.P., Ayers, S.L., Lam, C.,

Tate, H.P. and Zhao, S. (2016). Whole-genome sequencing for detecting antimicrobial

resistance in nontyphoidal Salmonella. Antimicrobial agents and Chemotherapy,

60(9):5515–5520.

McLinden, T., Sargeant, J.M., Thomas, M.K., Papadopoulos, A. and Fazil, A. (2014).

Component costs of foodborne illness: a scoping review. BMC Public Health, 14:509.

McWhorter, A.R. and Chousalkar, K.K. (2018). A long-term efficacy trial of a live, attenuated

Salmonella Typhimurium vaccine in layer hens. Frontiers in Microbiology, 9:1380.

Meyer Sauteur, P.M., Relly, C., Hug, M., Wittenbrink, M.M. and Berger, C. (2013). Risk factors

for invasive reptile-associated salmonellosis in children. Vector Borne and Zoonotic

Diseases, 13(6):419–421.

Milazzo, A., Giles, L.C., Zhang, Y., Koehler, A.P., Hiller, J.E. and Bi, P. (2016). The effect of

temperature on different Salmonella serotypes during warm seasons in a

102

Mediterranean climate city, Adelaide, Australia. Epidemiology and Infection,

144(6):1231–1240.

Moffatt, C.R.M., Musto, J., Pingault, N., Miller, M., Stafford, R., Gregory, J., Polkinghorne, B.G.

and Kirk, M.D. (2016). Salmonella Typhimurium and outbreaks of egg-associated

disease in Australia, 2001 to 2011. Foodborne Pathogens and Disease, 13(7):379–

385.

More egg recalls hit supermarkets due to Salmonella fears. (2019, March 28). The New Daily.

Retrieved from https://thenewdaily.com.au/money/consumer/2019/03/28/egg-

Salmonella-recall-nsw/.

Nature Education. (2014). The order of nucleotides in a gene is revealed by DNA sequencing.

Retrieved March 2, 2017, from http://www.nature.com/scitable/topicpage/the-order-of-

nucleotides-in-a-gene-6525806.

Neher, R.A. and Bedford, T. (2018). Real-time analysis and visualization of pathogen

sequence data. Journal of Clinical Microbiology, 56(11):e00480-18.

NSW Department of Primary Industries. National Salmonella Enteritidis monitoring &

accreditation program (NSEMAP). (n.d.). Retrieved March 29, 2019, from

https://www.dpi.nsw.gov.au/animals-and-livestock/poultry-and-birds/health-

disease/national-Salmonella-enteritidis-monitoring-and-accreditation-program.

NSW Health. (2019, March 27). Salmonella Enteritidis outbreak linked to contaminated eggs.

Retrieved from March 29, 2019, from

https://www.health.nsw.gov.au/Infectious/alerts/Pages/enteritidis-eggs-2018.aspx.

Pightling, A.W., Pettengill, J.B., Luo, Y., Baugher, J.D., Rand, H. and Strain E. Interpreting

whole-genome sequence analyses of foodborne bacteria for regulatory applications

and outbreak investigations. Frontiers in Microbiology, 9:1482.

Public Health England. (2014). Whole genome sequencing to revolutionise investigation of

outbreaks of infectious diseases. Retrieved March 2, 2017, from

https://www.gov.uk/government/news/whole-genome-sequencing-to-revolutionise-

investigation-of-outbreaks-of-infectious-disease.

Rahaman, M.H., Islam, T., Colwell, R.R. and Alam, M. (2015). Molecular tools in understanding

the evolution of Vibrio cholerae. Frontiers in Microbiology, 6:1040.

Reuter, S., Ellington, M.J., Cartwright, E.J., Koser, C.U., Torok, M.E., Gouliouris, T., Harris,

S.R., Brown, N.M., Holden, M.T., Quail, M., Parkhill, J., Smith, G.P., Bentley S.D. and

Peacock, S.J. (2013). Rapid bacterial whole-genome sequencing to enhance

diagnostic and public health microbiology. JAMA Internal Medicine, 173(15):1397–

1404.

Roberts, T. (1988). Salmonellosis control: estimated economic costs. Poultry Science,

67(6):936–943.

103

Ross, I.L., Davos, D.E., Mwanri, L., Raupach, J. and Heuzenroeder, M. (2011). MLVA and

phage typing as complementary tools in the epidemiological investigation of Salmonella

enterica serovar Typhimurium clusters. Current Microbiology, 62(3):1034–1038.

Santos, A.C., Roberts, J.A., Cook, A.J., Simons, R., Sheehan, R., Lane, C., Adak, G.K., Clifton-

Hadley, F.A. and Rodrigues, L.C. (2011). Salmonella Typhimurium and Salmonella

Enteritidis in England: costs to patients, their families, and primary and community

health services of the NHS. Epidemiology and Infection, 139(5):742–753.

Searing, N.M., Jensen, A., McCall, B.J., Neill, A.S. and McCormack, J.G. (2000). Direct costs

associated with a nosocomial outbreak of Salmonella infection: an ounce of prevention

is worth a pound of cure. American Journal of Infection Control, 28(1):54–57.

Sears, A., Baker, M.G., Wilson, N., Marshall, J.C., Muellner, P., Campbell, D.M., Lake, R.J.

and French, N. (2011). Marked campylobacteriosis decline after interventions aimed at

poultry, New Zealand. Emerging Infectious Diseases, 17(6):1007–1015.

Scharff, R.L., Besser, J., Sharp, D.J., Jones, T.F., Peter, G.S. and Hedberg, C.W. (2016). An

economic evaluation of PulseNet: A network for foodborne disease surveillance.

American Journal of Preventive Medicine, 50(5 Suppl 1):S66–73.

Stephen, D.M. and Barnett, A.G. (2016). Effect of temperature and precipitation on

salmonellosis cases in South-East Queensland, Australia: an observational study. BMJ

Open, 6(2):e010204.

Suijkerbuijk, A.W., Bouwknegt, M., Mangen, M.J., de Wit, G.A., van Pelt, W., Bijkerk, P. and

Friesema, I.H.M. (2017). The economic burden of a Salmonella Thompson outbreak

caused by smoked salmon in the Netherlands, 2012-2013. European Journal of Public

Health, 27(2):325–330.

Sundstrom, K., Wahlstrom, H., Ivarsson, S., and Sternberg Lewerin, S. (2014). Economic

effects of introducing alternative Salmonella control strategies in Sweden. PLoS One,

9(5):e96446.

Sundstrom, K. (2018). Cost of illness for five major foodborne illnesses and sequelae in

Sweden. Applied Health Economics and Health Policy, 16(2):243–257.

Taylor, A.J., Lappi, V., Wolfgang, W.J., Lapierre, P., Palumbo, M.J., Medus, C. and Boxrud D.

(2015). Characterization of foodborne outbreaks of Salmonella enterica serovar

Enteritidis with whole-genome sequencing single nucleotide polymorphism-based

analysis for surveillance and outbreak detection. Journal of Clinical Microbiology,

53(10):3334–3340.

The OzFoodNet Working Group. (2006). Burden and causes of foodborne disease in Australia:

Annual report of the OzFoodNet network, 2005. Communicable Diseases Intelligence

Quarterly Report, 30(3):278–300.

104

The OzFoodNet Working Group. (2007). Monitoring the incidence and causes of diseases

potentially transmitted by food in Australia: Annual report of the OzFoodNet network,

2006. Communicable Diseases Intelligence Quarterly Report, 31(4):345–366.



2011. Communicable Diseases Intelligence, 39(2):E236–264.



2012. Communicable Diseases Intelligence, 42:S2209-6051(18)00014-3.

Thompson, C.K., Wang, Q., Bag, S.K., Franklin, N., Shadbolt, C.T., Howard, P., Fearnley, E.J.,

Quinn, H.E., Sintchenko, V. and Hope K.G. Epidemiology and whole genome

sequencing of an ongoing point-source Salmonella Agona outbreak associated with

sushi consumption in western Sydney, Australia 2015. Epidemiology and Infection,

145(10):2062–2071.

Vally, H., Glass, K., Ford, L., Hall, G., Kirk, M.D., Shadbolt, C., Veitch, M.G.K, Fullerton, K.E.,

Musto, J. and Becker, N. (2014). Proportion of illness acquired by foodborne

transmission for nine enteric pathogens in Australia: an expert elicitation. Foodborne

Pathogens and Disease, 11(9):727–733.

Waldram, A., Dolan, G., Ashton, P.M., Jenkins, C. and Dallman, T.J. (2018). Epidemiological

analysis of Salmonella clusters identified by whole genome sequencing, England and

Wales 2014. Food Microbiology, 71:39–45.

WHO (World Health Organization) and FAO (Food and Agriculture Organization of the United

Nations). (2009, November 18). International Food Safety Authorities Network

(INFOSAN): PulseNet International. INFOSAN Information Note No. 4/2009. Retrieved

August 9, 2019, from

https://www.who.int/foodsafety/fs_management/No_04_PulseNet_Nov09_en.pdf.

WHO. (2018, February 20). Salmonella (non-typhoidal). Retrieved January 23, 2017, from

http://www.who.int/mediacentre/factsheets/fs139/en/.

WHO (2018a). Whole genome sequencing for foodborne disease surveillance: landscape

paper. Geneva: World Health Organization. Licence: CC BY-NC-SA 3.0 IGO.

WHO (2019). Antimicrobial resistance surveillance. Retrieved March 27, 2019, from

https://www.who.int/medicines/areas/rational_use/AMR_Surveillance/en/.

Williamson, D.A., Lane, C.R., Easton, M., Valcanis, M., Strachan, J., Veitch, M.G., Kirk, M.D.

and Howden, B.P. (2018). Increasing antimicrobial resistance to nontyphoidal

Salmonella isolates in Australia from 1970 to 2015. Antimicrobial Agents and

Chemotherapy, 62(2):e02012-17.

105

Appendix 1. Supplementary materials for chapter 3 The following supplementary materials were part of the manuscript submission and published

online as a supplement to the paper

Ford L, Glass K, Veitch M, Wardell R, Polkinghorne B, Dobbins T, Lal A, Kirk MD. Increasing

incidence of Salmonella in Australia, 2000-2013. PLoS One. 2016;11(10): e0163989, doi:

10.1371/journal.pone.0163989.

S1 Table. Number and proportion of Salmonella notifications without serovar data by state and

territory, Australia 2000-2013

State Number of notifications

without serovar data

Total number of

notifications

Proportion of notifications

without serovar data (%)

ACT 23 2,028 1.1

NSW 1,355 34,100 4.0

NT 189 5,850 3.2

Qld 819 35,597 2.3

SA 57 9,465 0.6

Tas. 49 2,772 1.8

Vic. 209 24,122 0.9

WA 259 13,261 2.0

Total 2,960 127,195 2.3

S2 Table. Salmonella spp. cases each year, the proportion of cases excluded due to missing

serovar, age or sex data, crude notification rate after exclusions, S. Typhimurium notification

rate, and Non-Typhimurium notification rate (per 100,000 persons, Australia 2000–2013).

Year Total cases

(n)

Proportion

excluded

(%)

Crude rate

(per 100,000

persons)

S. Typhimurium

rate (per 100,000

persons)

Non-Typhimurium

rate (per 100,000

persons)

2000 6,154 5.2 30.6 12.4 18.3

2001 6,995 3.8 34.9 13.7 21.3

2002 7,824 2.9 39.0 15.7 23.2

2003 6,948 3.4 34.1 14.4 19.7

2004 7,752 2.3 38.0 14.8 23.2

2005 8,346 2.0 40.5 16.9 23.6

106

2006 8,168 1.9 39.2 14.0 25.2

2007 9,398 1.7 44.3 19.8 24.5

2008 8,234 2.4 37.8 16.7 21.1

2009 9,443 2.8 42.3 18.3 24.0

2010 11,828 2.7 52.3 24.0 28.2

2011 12,209 1.7 53.7 27.1 26.7

2012 11,178 2.5 48.0 23.2 24.8

2013 12,718 3.6 53.0 25.8 27.2

S3 Table. Proportion (%) of S. Typhimurium and the 20 most notified non-Typhimurium

Salmonella serovars of total notifications included in this study for each state and territory,

Australia, 2000-2013.

ACT NSW NT Qld SA Tas. Vic. WA

Typhimurium 64.1 55.7 10.7 27.0 58.6 35.8 60.3 32.0

Enteritidis 4.7 4.4 2.0 4.1 5.0 3.6 5.7 15.5

Virchow 3.0 3.5 5.9 11.6 2.2 1.7 3.0 1.4

Saintpaul 1.8 1.9 9.3 8.4 1.9 1.2 1.7 4.6

Birkenhead 0.5 3.4 <0.1 4.9 0.1 <0.1 0.4 0.1

Infantis 2.0 2.7 2.1 0.7 3.9 1.1 2.6 2.0

Paratyphi B bv Java 1.9 2.1 2.4 1.4 1.4 1.0 1.9 3.4

Chester 0.6 1.2 3.2 2.7 2.2 0.6 0.7 2.9

Muenchen 0.8 0.9 2.8 2.2 1.4 0.3 0.5 2.9

Bovismorbificans 1.6 2.1 0.5 0.9 2.3 0.9 1.5 1.1

Aberdeen 0.1 0.3 1.3 4.1 0.1 0.2 0.3 0.1

Hvittingfoss 0.6 0.5 1.1 3.3 0.2 0.1 0.6 0.4

Stanley 1.7 1.3 0.3 0.7 1.0 1.2 1.8 1.7

Mississippi 0.3 0.2 0.1 0.2 0.2 41.9 0.5 0.1

Waycross 0.1 1.0 0.1 2.8 0.1 <0.1 0.1 <0.1

Weltevreden 0.7 0.6 3.3 1.2 0.5 0.3 0.7 1.1

Anatum 0.2 0.4 2.7 1.3 1.0 0.2 0.4 1.6

Agona 0.8 1.0 0.6 0.9 1.0 0.6 0.8 0.8

Newport 0.8 0.7 0.3 0.4 1.1 1.1 1.5 1.1

Singapore 0.5 1.3 0.2 0.4 0.8 0.2 0.6 1.2

Potsdam 0.6 0.6 0.4 1.1 0.6 0.7 0.3 0.6

S1

Fig

. Sta

te a

nd

ter

rito

ry c

rud

e (d

ots

) an

d p

red

icte

d (

lines

wit

h 9

5% C

I) n

oti

fica

tio

n r

ates

per

100

,000

per

son

s, A

ust

ralia

200

0-2

013

107

S2 Fig. Salmonella Typhimurium and non-Typhimurium predicted notification rates per

100,000 (with 95% CI) by age group for each State and Territory, 2000-2013

0

20

40

60

80

100

120

140

160

No

tifi

cati

on

ra

te p

er 1

00

,00

0

ACT

Typhimurium

Non-Typhimurium

0

200

400

600

800

1000

1200

1400

No

tifi

cati

on

ra

te p

er 1

00

,00

0

NT

Typhimurium

Non-Typhimurium

0

20

40

60

80

100

120

140

160

No

tifi

cati

on

ra

te p

er 1

00

,00

0

NSW

Typhimurium

Non-Typhimurium

111

0

50

100

150

200

250

300

350

400

450

500N

oti

fica

tio

n r

ate

per

10

0,0

00

QLD

Typhimurium

Non-Typhimurium

0

20

40

60

80

100

120

140

160

180

200

No

tifi

cati

on

ra

te p

er 1

00

,00

0

SA

Typhimurium

Non-Typhimurium

0

50

100

150

200

250

300

No

tifi

cati

on

ra

te p

er 1

00

,00

0

TAS

Typhimurium

Non-Typhimurium

112

0

50

100

150

200

250

300

No

tifi

cati

on

ra

te p

er 1

00

,00

0

WA

Typhimurium

Non-Typhimurium

0

20

40

60

80

100

120

140

No

tifi

cati

on

ra

te p

er 1

00

,00

0

VIC

Typhimurium

Non-Typhimirium

113

114



Ford L, Moffatt C, Fearnley E, Sloan-Gardner T, Miller M, Polkinghorne B, Franklin N,

Williamson DA, Glass K, Kirk MD. The epidemiology of Salmonella outbreaks in Australia,

2001-2016. Frontiers in Sustainable Food Systems. 2018;issue:page, doi:

10.3389/fsufs.2018.00086.

Supplementary Information 1

S1 Table 1: Outbreak setting definitions

Setting Definition

Food premises Food prepared in restaurants, takeaway stores, bakeries,

commercial caterers, national franchised fast food stores, grocery

stores, delicatessens, airlines, or cruises

Home kitchens Food prepared in private residences

Institutions Food prepared in aged care facility, hospital or medical centre,

school, child care centre, military institution, camps, or other

institution not otherwise specified

Market, fair/festival,

or mobile service

Food prepared in markets, fairs, festivals, food trucks, or other

temporary or mobile services

Commercially

manufactured food

Food that is manufactured and distributed commercially

Imported food Imported ready-to-eat food product

Primary produce Food grown and harvested that is typically consumed with no or

minimal cooking or further processing

Other Picnic or other setting not specified

Unknown Setting food prepared in is not known

S1 Table 2: Food categories for salmonellosis outbreaks in Australia

Egg sauces Mayonnaise

Raw egg mayonnaise

Dill mayonnaise

Raw egg mayonnaise in chicken sandwiches

Chicken salad with raw egg mayonnaise

Pork/chicken and salad rolls (with raw egg mayonnaise)

Potato salad with raw egg mayonnaise

Hamburgers with homemade mayonnaise

Raw egg mayonnaise and raw egg milkshakes

Raw egg aioli

Citrus aioli

Garlic aioli

Aioli and mayonnaise-based sauces

Coleslaw with raw egg aioli

Tartare sauce

Tartare sauce, prepared with raw egg

Mayonnaise/tartare/aioli

Raw-egg mayonnaise; raw-egg tartare sauce

Chicken Caesar roll containing raw egg mayonnaise

Caesar salad dressing

Caesar salad dressing with raw egg

Aioli and Caesar salad

Raw egg mayonnaise/Caesar salad dressing

Raw egg salad dressing

Raw egg dressing

Béarnaise sauce

115

Eggs Benedict

Eggs benedict with potato rosti

Eggs benedict; hollandaise sauce

Egg and/or hollandaise sauce

Hollandaise sauce

Raw egg hollandaise sauce

Egg and/or hollandaise sauce

Raw egg sauces

Egg-based sauce

Cold emulsion (raw egg white containing)

Raw egg chipotle cream

Bánh mì (Vietnamese

style sandwiches)

Vietnamese rolls with raw egg butter

Raw egg butter on Vietnamese roll

Vietnamese rolls

Vietnamese-style rolls

Vietnamese pork roll

Pork Rolls (raw egg butter)

Pork rolls with raw egg mayonnaise

Vietnamese pork/chicken/salad rolls containing raw egg

butter

Desserts containing raw

or lightly cooked eggs

Chocolate Mousse

White chocolate mousse

Chocolate mousse with raw eggs

Chocolate mousse cake

Raw egg chocolate mousse cake

Pizza containing egg and chocolate mousse containing raw

egg

Tiramisu containing raw eggs

Tiramisu

Fried ice cream

Deep fried Ice cream containing minimally cooked raw eggs

Fried ice cream with raw egg

Deep fried ice-cream balls

Egg coated fried ice cream

suspect pork in plum sauce, fried ice cream

Homemade ice cream

Ice cream containing raw eggs

Ice cream cake with raw eggs

Raw egg milkshake

Raw egg smoothies

Raw egg semifreddo

Custard

Dessert containing raw egg custard

Custard fruit tart

Bread and butter pudding

Frozen ice cream sponge mixture

Pancake batter

Pancake batter containing raw eggs

Cream and custard profiterole cakes

Chocolate cake with raw egg meringue

116

Chocolate fondant

Rum and raisin bread cake with custard

Raw egg cake mix

French Toast

Various bakery products (custard buns)

Bomb Alaska

Trifle/jelly; Bomb Alaska

Crepes

Custard Berliner bun

Custard eclairs and custard cannoli

Raw brownie batter

Hedgehog - eggs cookie dough raw

Hazelnut gateau with raw egg mousse filling

Dessert containing raw eggs suspected

Dessert containing lightly cooked eggs

Uncooked muffin batter

Pastry custard tart with strawberries & jelly glaze

Eggs used to make cake filling

Desserts, other Chocolate milk (& other foods)

commercial ice cream with homemade chocolate sauce

Cheesecake

Cold set cheesecake

Chocolate eclairs; cream puffs; various cakes; various

bakery products

Engagement cake – cream

Cream filled cakes

Cheese or cream cake

Layered chocolate cake, prepared with cream and ganache

icing (no raw eggs used)

Desserts

Profiteroles

Trifle

Banana Milkshake

Mango pudding dessert

Apple turnover

Bakery products (various)

Egg based bakery products

Bakery cakes and buns

Vietnamese bakery goods

Bakery products, no specific item identified

Various bakery products

Numerous bakery goods

Desserts (containing eggs)

117

Eggs, other Egg nog

Egg wash

Raw eggs

Suspected raw egg

Eggs

Egg shell rinse and contents of cracked eggs

Suspected eggs

Free-range eggs

Raw or runny eggs

Boiled eggs

Scrambled eggs

Eggs (fried soft)

Poached eggs

Egg containing food

Eggs used in uncooked and lightly cooked foods

Menu items containing undercooked egg

Egg dishes

Raw egg dish

Egg dish with cous cous

Asparagus egg surprise dish

Frittata

Breadcrumbs, breakfast eggs served a number of ways

Suspect cross contamination from egg

Sprouts Mung bean sprouts

Snow pea sprouts

Alfalfa sprouts

Poultry Chicken liver pâté

Chicken breast cooked in the home

Chicken

Possibly chicken pieces from franchised restaurant

Roast chicken pieces served cold

Suspected fluid thickener contaminated by raw chicken

mince

Chicken burger

Warm chicken salad

Marinated chicken roll

Suspected BBQ chicken

Chicken (Home BBQ)

Grilled chicken

Chicken suspected

Chicken and gravy

Roasted chicken

Suspect chicken spring rolls

Chicken and pasta salad

Baked chicken

Sticky rice balls with chicken

Chicken meat

Unknown, possible chicken based dish

Chicken meal

Duck prosciutto

118

Raw pigeon meat

Duck parfait

Duck pancakes

Duck liver pâté

Roast duck

Pork Pork suspected

Suspected roast pork

Roast pork

Barbecued pork

Pork spit roast

Sliced Virginian ham

Pork Salami

Spit roasted Pork

Capocollo (cured pork)

Salami

Slow cooked pork hock

Leg of Ham

Homemade unfermented sausage

Fish Ling fish fillets

Fish cakes

Chinese style minced fish balls

Suspected salmon & couscous dish

Barramundi spring rolls

Lamb Lambs liver

Offal (lamb intestine)

Offal stew (lamb intestine, tripe, liver and kidney)

Lambs fry with bacon, onions & gravy

Lamb shanks or salad

Beef Roast meat

Beef burger

Silverside

Beef wraps

Suspect beef products

Frozen raw meat burgers, raw beef steak

Nuts Peanuts

Suspected peanut/cashew mixture

Raw Almonds

Fruit Rockmelon

Pawpaw

Salad, other Assorted salads

Thai salad

Rice salad

Suspect salad

Caesar salad

Chicken Caesar Salad

Potato Salad

Pasta salad

Sweet potato and feta cheese salad

Multiple salads

Crustacean/Mollusc Crab

119

Prawn Soup

Calamari

Prawn salad rolls

Suspect oysters

Vegetables Fresh chillies used to prepare chilli sauce

Taro

Bagged salad products

Suspected cucumber

Sushi Sushi Rolls

Korean Sushi

Sushi (unspecified)

Kimbap style sushi

Tuna and salmon sushi rolls

Tuna mix for sushi

Chicken hand rolls

Sandwich, other Sandwiches

Premade sandwiches

Pork roll, salad roll, sandwiches

Pork/Chicken rolls from bakery

Bread rolls with different fillings

Sandwich containing egg and ham

Egg and bacon roll

Bacon and egg burger

Suspect beef or chicken hamburger with salad, cheese, bacon

Sandwich containing egg and ham

Suspected gourmet rolls including red onion

Chicken and lamb souvlakis

Chicken salad pita bread wrap (using iceberg lettuce)

Pork rolls

Suspected pork rolls

Egg and lettuce sandwich

Kebabs

Pasta Uncooked pasta dough containing eggs

Ravioli

Pasta with lightly cooked egg

Suspected under cooked eggs in pasta dish

Pasta Carbonara

Pasta with raw egg

Tahini/helva Tahini

Imported tahini

Imported helva

Dips Dips

Pesto made with basil, garlic, olive oil

Suspect dip – salad dip and baba ganoush dip

Mixed foods Mayonnaise, chicken

Chicken and fried ice cream

Passionfruit cheesecake; meat pies

Salty fish, pork and egg dish

Baked beans and/or chilli con carne

Hummus and tabouli

120

Fried rice

Chinese food

Variety of Chinese foods

Takeaway chicken, rice, coleslaw, potatoes

Plain hamburger; egg

Chicken; eggs

Chicken, hummus, tabouli

Suspected eggs or chicken

Cooked pork mince and leftover food (mix of tofu, rice,

duck)

Probably prawn dumplings prepared with minced prawn,

Coriander and egg to bind

Salmon/egg/onion/rice patties

Lamb tartare with raw egg

Chicken long soup (with egg and chicken)

Egg-battered chicken and veal schnitzels

Potato bake, lemon meringue, chicken patty

Rice paper rolls

Beef appetiser or frittata

Broken Rice

Noodle dish with chicken and egg

Chicken teriyaki or scrambled eggs

Meals containing chicken pieces and pizza of any kind

Prawn/noodle dish (lightly cooked eggs)

Nasi-Lemak

Suspect steak; suspect fried rice

Combination Chinese omelette

Egg nog or undercooked chicken

Mixed spit roasted meats

Eye fillet meal with onions, potato, salsa verde and red wine

jus

Banana smoothie; Berry smoothie; Any fish

Multiple, likely initial contamination from eggs

Meat based potato pie, rice pudding (both containing raw

eggs)

Multiple foods - hummus and aioli positive

Multiple foods, including sliced deli meats

Undetermined Mixed dishes

Multiple foods

Cordial?

Prawn paste? Tofu eggplant dish

Probably pureed food

Suspected to be ready to eat food such as hand cut fruit and

Pre-prepared meals

Contaminated premade product

Buffet breakfast

Vitamised and soft foods

Various salad rolls from bakery; meals prepared at home

containing egg

Vitamised food

121

Gravy

Unknown pureed food

Contaminated premade products

Unknown, probably assorted salads, wraps and burgers

Unknown, possibly minced or pureed diet

Unknown Unknown

Not identified

Not detected

No vehicle identified

999

None implicated

N/A

Unknown, possible a food source or person-to-person

transmission and then further spread by person.

122

Typhimurium Agona Anatum Bareilly Birkenhead BovismorbificansRaw egg desserts 103 0 0 0 0 0Egg-based sauce 76 0 0 0 0 0Eggs, other 41 0 0 0 0 0Mixed dishes 33 0 0 0 0 0Poultry 28 0 1 0 0 0Sandwiches, other 18 0 1 0 0 1Desserts, other 23 0 0 0 0 0Vietnamese Rolls 17 0 0 0 0 0Pork 9 0 0 0 0 1Salads 8 0 1 0 0 0Sushi 5 1 0 1 0 0Beef 4 0 0 0 0 0Pasta 6 0 0 0 0 0Lamb 3 0 0 0 0 0Fish 4 0 0 0 0 0Sprouts 0 0 0 0 0 0Crustacean/ Mollusc 3 0 0 0 0 0Fruits 0 0 0 0 0 0Vegetables 0 0 1 0 0 0Tahini or helva 1 0 0 0 0 0Dips 3 0 0 0 0 0Nuts 2 0 0 0 0 0Undetermined 23 0 0 0 1 1Total 410 1 4 1 1 3

S2 Table 1: Number of foodborne and suspected foodborne salmonellosis outbreaks with statistical, laboratory or descriptive evidence by serotype, 2001-2016, Australia

123

Raw egg dessertsEgg-based sauceEggs, otherMixed dishesPoultrySandwiches, otherDesserts, otherVietnamese RollsPorkSaladsSushiBeefPastaLambFishSproutsCrustacean/ MolluscFruitsVegetablesTahini or helvaDipsNutsUndeterminedTotal

Enteritidis Chester Hadar Hessarek Hvittingfoss Infantis Johannesburg0 0 0 0 0 0 00 0 0 0 0 0 01 0 1 1 0 0 01 0 0 0 0 1 00 0 0 0 0 1 00 0 0 0 0 1 00 0 0 0 0 0 00 0 0 0 0 0 00 0 0 0 0 0 10 0 0 0 0 0 00 0 0 0 0 0 00 0 0 0 0 0 00 0 0 0 0 0 00 2 0 0 0 0 00 0 0 0 0 0 00 0 0 0 0 0 00 0 0 0 0 1 00 0 0 0 1 0 00 1 0 0 0 0 00 0 0 0 0 0 00 0 0 0 0 0 00 0 0 0 0 0 00 0 0 0 1 1 02 3 1 1 2 5 1

124


Kiambu Litchfield Livingstone London Montevideo Mississippi0 0 0 0 0 01 0 0 0 0 00 0 1 0 0 00 0 0 0 1 00 0 0 0 0 00 0 0 0 0 00 0 0 0 0 00 0 0 0 0 00 0 0 1 0 00 0 0 0 0 10 0 0 0 0 00 0 0 0 0 00 0 0 0 0 00 0 0 0 0 00 0 0 0 0 00 0 0 0 0 00 0 0 0 0 10 1 0 0 0 00 1 0 0 0 00 0 0 0 2 00 0 0 0 0 00 0 0 0 0 00 0 0 0 0 01 2 1 1 3 2

125


Muenchen Newport Oranienburg Oslo Potsdam Saintpaul Singapore0 0 0 0 0 1 10 0 0 0 1 0 00 0 0 0 0 1 00 0 0 0 0 0 00 0 0 1 0 0 10 1 0 0 0 0 00 0 0 0 0 0 00 0 0 0 0 0 00 0 0 1 0 0 00 0 0 0 0 0 00 0 0 0 0 0 10 0 0 0 0 0 10 0 0 0 0 0 00 0 0 0 0 0 00 0 0 0 0 1 01 0 2 0 0 2 00 0 0 0 0 0 00 0 0 0 0 2 00 0 0 0 0 0 00 0 0 0 0 0 00 0 0 0 0 0 00 0 0 0 0 0 00 0 0 0 0 0 01 1 2 2 1 7 4

126


Stanley subsp I Virchow Wangata Zanzibar Zanzibar var 15+ Total0 1 0 0 0 0 1060 0 1 0 0 0 790 0 1 0 0 0 470 0 0 0 0 0 360 1 0 0 1 0 340 0 2 0 0 0 240 0 0 0 0 0 230 0 0 0 0 0 170 1 0 1 0 0 150 0 0 0 0 0 100 0 0 0 0 0 80 0 1 0 0 0 60 0 0 0 0 0 60 0 0 0 0 0 50 0 0 0 0 0 50 0 0 0 0 0 50 0 0 0 0 0 50 0 0 0 0 0 40 0 0 0 0 1 40 0 0 0 0 0 30 0 0 0 0 0 31 0 0 0 0 0 31 0 0 0 0 0 282 3 5 1 1 1 476

127

Supplementary Information 3: The number and proportion of outbreaks due to food vehicles

with significant trends over time, Australia, 2001-2016

S3 Table 1: Number and proportion of foodborne or suspected foodborne salmonellosis

outbreaks due to egg-based sauces and Vietnamese style rolls with statistical, laboratory, or

descriptive evidence, Australia, 2001-2016

Number of outbreaks (%)

Egg-based sauces Vietnamese style rolls

2001 0/18 (0) 0/18 (0)

2002 1/16 (6) 0/16 (0)

2003 0/17 (0) 0/17 (0)

2004 1/20 (5) 0/20 (0)

2005 4/24 (17) 0/24 (0)

2006 1/25 (4) 0/25 (0)

2007 5/26 (19) 0/26 (0)

2008 3/23 (13) 0/23 (0)

2009 4/20 (20) 1/20 (5)

2010 7/31 (23) 0/31 (0)

2011 6/33 (18) 2/33 (6)

2012 5/35 (14) 1/35 (3)

2013 10/31 (32) 1/31 (3)

2014 14/60 (23) 6/60 (10)

2015 12/55 (22) 2/55 (4)

2016 6/42 (14) 4/42 (10)

S3 Table 2: Number and proportion of foodborne or suspected foodborne salmonellosis

outbreaks due to poultry, beef, pork, other sandwiches, and other desserts with statistical,

laboratory, or descriptive evidence, Australia, 2001-2016

Number of outbreaks (%)

Poultry Beef Pork Other sandwiches Other desserts

2001 2/18 (11) 2/18 (11) 1/18 (6) 2/18 (11) 1/18 (6)

2002 3/16 (19) 0/16 (0) 1/16 (6) 2/16 (13) 2/16 (13)

2003 3/17 (18) 1/17 (6) 1/17 (6) 1/17 (6) 1/17 (6)

2004 3/20 (15) 0/20 (0) 1/20 (5) 1/20 (5) 1/20 (5)

2005 2/24 (8) 0/24 (0) 1/24 (4) 2/24 (8) 4/24 (17)

2006 1/25 (4) 1/25 (4) 2/25 (8) 2/25 (8) 0/25 (0)

2007 1/26 (4) 0/26 (0) 1/26 (4) 1/26 (4) 3/26 (12)

2008 3/23 (13) 0/23 (0) 1/23 (4) 1/23 (4) 2/23 (9)

2009 0/20 (0) 0/20 (0) 0/20 (0) 1/20 (5) 1/20 (5)

2010 2/31 (6) 0/31 (0) 1/31 (3) 2/31 (6) 1/31 (3)

2011 2/33 (6) 0/33 (0) 1/33 (3) 0/33 (0) 1/33 (3)

2012 1/35 (3) 0/35 (0) 1/35 (3) 3/35 (9) 1/35 (3)

2013 2/31 (6) 0/31 (0) 0/31 (0) 2/31 (6) 1/31 (3)

2014 2/60 (3) 2/60 (3) 2/60 (3) 1/60 (2) 2/60 (3)

2015 5/55 (9) 0/55 (0) 1/55 (2) 2/55 (4) 1/55 (2)

2016 2/42 (5) 0/42 (0) 0/42 (0) 1/42 (2) 1/42 (2)

128

129

Appendix 3. Supplementary materials for chapter 5 The following supplementary materials were part of the manuscript submission as a

supplement to the paper

Ford L, Haywood P, Kirk MD, Lancsar E, Williamson DA, Glass K. Cost of foodborne

Salmonella infections in Australia, 2015. Journal of Food Protection. 2019;82(9):1607-1614,

doi: 10.4315/0362-028X.JFP-19-105.

130

Su

pp

lem

enta

ry m

ate

rial

1:

Hea

lth

car

e u

sag

e va

riab

les,

dat

a so

urc

es, d

istr

ibu

tio

n, a

nd

ou

tpu

t

Age

gro

up: 0

-4 y

ears

Sal

mo

nel

la

Var

iab

le

Dat

a so

urc

e D

istr

ibu

tio

n

Ou

tpu

t O

utp

ut

un

its

Co

mm

ents

2015

Pop

ulat

ion

AB

S

15

5256

7

Inci

denc

e K

irk e

t al

201

4 (1

0)

Ris

kOut

put 0

.008

, 0.

013,

0.0

23

2092

4 (9

0% C

rI 1

2059

-36

,50)

C

ases

S

tatis

tical

mod

el in

corp

orat

ing

NN

DS

S n

otifi

catio

n nu

mbe

rs fo

r 20

13 (

3003

), 2

014

(370

5), a

nd 2

015

(386

7) w

ith m

ultip

liers

for

unde

rrep

ortin

g an

d do

mes

tic

acqu

isiti

on,

No

med

ical

car

e

Ris

kOut

put 0

.31,

0.4

5,

0.6

9367

(95

% C

rI 4

484-

1810

8)

Cas

es

1 -

(%G

P v

isits

+ %

hosp

italis

ed +

%

ED

vis

its)

GP

vis

its

NG

SII

(11)

A

ltPer

t 0.2

5, 0

.37,

0.5

0 75

86 (

90%

CrI

395

7-14

364)

G

P v

isits

G

P v

isit

dist

ribut

ion

mul

tiplie

d by

in

cide

nce

outp

ut d

istr

ibut

ion

ED

vis

its

NG

SII

(11)

S

ame

met

hod

as

GP

AltP

ert 0

.06,

0.1

24,

0.22

8 25

60 (

90%

CrI

113

0-55

64)

ED

vis

its

ED

vis

it di

strib

utio

n m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n

Hos

pita

lisat

ions

A

IHW

(2,

3)

Ris

kOut

put 0

.02,

0.0

5,

0.09

98

4 (9

0% C

rI 6

61-1

346)

Hos

pita

lisat

ions

Sta

tistic

al m

odel

inco

rpor

atin

g A

IHW

pr

imar

y di

agno

ses

for

2011

/12

(439

), 2

012/

13 (

426)

, 201

3/14

(49

3)

with

mul

tiplie

rs fo

r un

derr

epor

ting

and

dom

estic

acq

uisi

tion

Day

s in

hos

pita

l A

IHW

prim

ary

di

agno

sis

2013

-14

– A

LOS

(3)

2.

85

Day

s

AIH

W p

rovi

des

<1

year

, 1-4

yea

rs,

and

5 ye

ar a

ge g

roup

s af

ter

that

up

to 8

5+. A

wei

ghte

d av

erag

e A

LOS

is

calc

ulat

ed fo

r ou

r ag

e gr

oups

. D

eath

s A

ustr

alia

n B

urea

u of

Sta

tistic

s R

iskO

utpu

t 0.0

0002

, 0.

0000

3, 0

.000

06

0.65

(90

% C

rI 0

.5-0

.82)

D

eath

s S

tatis

tical

mod

el in

corp

orat

ing

AB

S

deat

hs w

ith m

ultip

liers

for

131

unde

rrep

ortin

g an

d do

mes

tic

acqu

isiti

on

Med

icat

ions

N

GS

II -

Adj

uste

d fo

r se

verit

y. 0

-4

age

grou

p on

ly

A

ntid

iarr

hoea

l

Per

t 0.0

025,

0.0

03, 0

.3

860

(90%

CrI

110

-325

0)M

edic

atio

ns

Med

icat

ion

dist

ribut

ions

mul

tiplie

d by

inci

denc

e ou

tput

dis

trib

utio

n

P

aink

iller

AltP

ert 0

.12,

0.3

8, 0

.65

7709

(90

% C

rI 2

738-

1664

5)

Med

icat

ions

A

nti-n

ause

a

Per

t 0.0

025,

0.0

03,

0.30

5 86

0 (9

0% C

rI 1

10-3

250)

Med

icat

ions

A

nti-c

ram

ps

0,

0, 0

0

Med

icat

ions

A

ntib

iotic

s

AltP

ert 0

.003

, 0.0

82,

0.30

5 16

64 (

90%

CrI

133

-63

28)

Med

icat

ions

Sto

ol c

ultu

re a

nd/o

r P

CR

N

ND

SS

S

alm

onel

la

notif

icat

ions

(5)

3003

, 370

5, 3

867

3705

S

tool

cul

ture

s N

ND

SS

Sal

mon

ella

not

ifica

tions

for

age

grou

p fo

r 20

13, 2

014,

and

201

5.

Out

put i

s m

edia

n of

yea

rs.

Day

s lo

st o

f pai

d w

ork

due

to b

eing

ill

NG

SII

- A

djus

ted

for

seve

rity.

0-4

on

ly (

11)

0, 0

, 0

0 D

ays

NG

SII

ques

tion

Q34

B. D

ays

lost

di

strib

utio

n m

ultip

lied

by in

cide

nce

outp

ut d

istr

ibut

ion

Day

s lo

st o

f pai

d w

ork

due

to c

arin

g N

GS

II -

Adj

uste

d fo

r se

verit

y. 0

-4

only

(11

)

AltP

ert 0

.692

, 1.3

, 2.2

2326

896

(90%

CrI

124

09-

5639

3)

Day

s N

GS

II qu

estio

n Q

35B

. D

ays

lost

di

strib

utio

n m

ultip

lied

by in

cide

nce

outp

ut d

istr

ibut

ion

Day

s lo

st o

f act

iviti

es

due

to b

eing

ill

NG

SII

- A

djus

ted

for

seve

rity.

0-4

on

ly (

11)

AltP

ert 1

.7, 2

.6, 3

.81

5398

4 (9

0% C

rI 2

7730

-10

3367

) D

ays

NG

SII

ques

tion

Q34

. Inc

lude

s pa

id

wor

k &

non

-pai

d ac

tiviti

es.

Day

s lo

st d

istr

ibut

ion

mul

tiplie

d by

in

cide

nce

outp

ut d

istr

ibut

ion

Day

s lo

st o

f ac

tiviti

es

due

to c

arin

g N

GS

II -

Adj

uste

d fo

r se

verit

y. 0

-4

only

(11

)

AltP

ert 2

.024

, 3, 4

.283

62

238

(90%

CrI

322

25-

1193

40)

Day

s N

GS

II qu

estio

n Q

35. I

nclu

des

paid

w

ork

& n

on-p

aid

activ

ities

. D

ays

lost

dis

trib

utio

n m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n IB

S f

ollo

win

g S

alm

on

ella

132

Inci

denc

e F

ord

et a

l 201

4 (7

)

1.46

(90

% C

rI 0

.82-

2.57

) C

ases

A

ttrib

utab

le r

isk

of I

BS

app

lied

to t

he

estim

ated

Sal

mon

ella

inci

denc

e fo

r th

e ag

e gr

oup.

G

P v

isits

A

bels

on e

t al

20

06 (

1); F

lik e

t al

2015

(6)

AltP

ert 4

.27,

4.5

, 4.7

3 7

(90%

CrI

4-1

2)

GP

vis

its

GP

vis

it di

strib

utio

n m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n

ED

vis

its

Abe

lson

et a

l 20

06 (

1)

0, 0

, 0

0 E

D v

isits

Hos

pita

lisat

ions

A

IHW

(2,

3)

Ris

kOut

put 0

.07,

0.1

9,

0.41

0.

29 (

90%

CrI

0-0

.53)

H

ospi

talis

atio

nsS

tatis

tical

mod

el in

corp

orat

ing

AIH

W

prim

ary

diag

nose

s fo

r 20

11/1

2 (0

),

2012

/13

(7),

201

3/14

(6)

with

m

ultip

liers

for

unde

rrep

ortin

g,

dom

estic

acq

uisi

tion,

and

pro

port

ion

due

to e

nter

ic in

fect

ion.

The

n 24

.3%

of

IB

S h

ospi

talis

atio

ns a

ttrib

uted

sp

ecifi

cally

to S

alm

onel

la.

Day

s in

hos

pita

l A

IHW

prim

ary

di

agno

sis

2013

-14

– A

LOS

(3)

4

Day

s A

IHW

pro

vide

s <

1 ye

ar, 1

-4 y

ears

, an

d 5

year

age

gro

ups

afte

r th

at u

p to

85+

. A w

eigh

ted

aver

age

ALO

S is

ca

lcul

ated

for

our

age

grou

ps.

Med

icat

ions

or

trea

tmen

ts

Exp

ert o

pini

on

AltP

ert 0

.385

, 0.4

, 0.4

160.

41 (

90%

CrI

0.2

2-0.

76)

Med

icat

ions

M

edic

atio

n di

strib

utio

n m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n

Pat

holo

gy a

nd im

agin

g E

xper

t opi

nion

Tes

t dis

trib

utio

ns m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n

Sto

ol c

ultu

re

P

ert 0

.667

, 1, 1

1.

38 (

90%

CrI

0.7

7-2.

45)

Tes

ts

F

BC

Per

t 0.6

67, 1

, 1

1.38

(90

% C

rI 0

.77-

2.45

) T

ests

E

SR

Per

t 0.6

67, 1

, 1

1.38

(90

% C

rI 0

.77-

2.45

) T

ests

LF

T

P

ert 0

.667

, 1, 1

1.

38 (

90%

CrI

0.7

7-2.

45)

Tes

ts

133

C

RP

Per

t 0.6

67, 1

, 1

1.38

(90

% C

rI 0

.77-

2.45

) T

ests

C

oelia

c di

seas

e

scre

enin

g

Per

t 0.6

67, 1

, 1

1.38

(90

% C

rI 0

.77-

2.45

) T

ests

R

adio

logy

AltP

ert 0

.652

, 0.6

67,

0.68

1 0.

97 (

90%

CrI

0.5

5-1.

71)

X-r

ays

U

ltras

ound

AltP

ert 0

.484

, 0.5

, 0.5

160.

73 (

90%

CrI

0.4

1-1.

28)

Ultr

asou

nds

E

ndos

copy

and

biop

sy

0,

0,

0 0

End

osco

pies

Spe

cial

ist

Exp

ert o

pini

on;

Can

avan

et

al

2014

(4)

AltP

ert 0

.286

, 0.3

, 0.3

150.

44 (

90%

CrI

0.2

5-0.

77)

Spe

cial

ist v

isits

S

peci

alis

t vis

it di

strib

utio

n m

ultip

lied

by in

cide

nce

outp

ut d

istr

ibut

ion

New

ong

oing

illn

ess

Mar

shal

l et a

l 20

07 (

13)

AltP

ert 0

.218

, 0.4

29,

0.66

2.

86 (

90%

CrI

1.3

5-5.

77)

Cas

es

New

ong

oing

illn

ess

dist

ribut

ion

mul

tiplie

d by

GP

vis

it ou

tput

+

hosp

italis

atio

n ou

tput

dis

trib

utio

ns

Day

s of

lost

pai

d w

ork

/ ac

tiviti

es

Abe

lson

et

al

2006

(1)

ass

umed

sa

me

as d

ays

in

hosp

ital +

tim

e vi

sitin

g G

P (

0.5

da

y)

4.

44 (

90%

CrI

2.8

1-7)

D

ays

(Hos

pita

lisat

ions

* A

LOS

) +

(G

P

visi

ts *

0.5

)

Lost

pai

d w

ork/

activ

ities

fr

om o

ngoi

ng il

lnes

s

9

(90%

CrI

4-1

7)

Day

s O

ngoi

ng il

lnes

s di

strib

utio

n m

ultip

lied

by lo

st p

rodu

ctiv

ity

(abo

ve)

ReA

fo

llow

ing

Sal

mo

nel

la

Inci

denc

e F

ord

et a

l 201

4 (7

)

341

(90%

CrI

56-

898)

C

ases

A

ttrib

utab

le r

isk

of R

eA a

pplie

d to

th

e es

timat

ed S

alm

onel

la in

cide

nce

for

the

age

grou

p. R

atio

of

0.09

3 ap

plie

d fo

r <

5s.

G

P v

isits

T

owne

s et

al 2

008

(14)

; A

bels

on e

t al

20

06 (

1)

AltP

ert 0

.66,

0.8

0, 0

.89

286

(90%

CrI

64-

720)

G

P v

isits

G

P v

isit

dist

ribut

ion

mul

tiplie

d by

in

cide

nce

outp

ut d

istr

ibut

ion

134

ED

vis

its

Abe

lson

et a

l 20

06 (

1)

0, 0

, 0

0 E

D v

isits

Hos

pita

lisat

ions

Ris

kOut

put 0

.002

, 0.0

1,

0.03

1.

84 (

90%

CrI

1.0

7-3.

52)

Hos

pita

lisat

ions

Sta

tistic

al m

odel

inco

rpor

atin

g A

IHW

pr

imar

y di

agno

ses

for

2011

/12

(10)

, 20

12/1

3 (6

), 2

013/

14 (

4) w

ith

mul

tiplie

rs fo

r un

derr

epor

ting,

do

mes

tic a

cqui

sitio

n an

d pr

opor

tion

due

to e

nter

ic in

fect

ion.

The

n 25

.3%

of

ReA

hos

pita

lisat

ions

att

ribut

ed

spec

ifica

lly to

Sal

mon

ella

. D

ays

in h

ospi

tal

AIH

W p

rima

ry

diag

nosi

s 20

13-1

4 –

ALO

S (

3)

2.

3 D

ays

AIH

W p

rovi

des

<1

year

, 1-4

yea

rs,

and

5 ye

ar a

ge g

roup

s af

ter

that

up

to 8

5+. A

wei

ghte

d av

erag

e A

LOS

is

calc

ulat

ed fo

r ou

r ag

e gr

oups

. M

edic

atio

ns o

r tr

eatm

ents

for

GP

vis

its

Med

icat

ion/

trea

tmen

t dis

trib

utio

ns

mul

tiplie

d by

GP

vis

it ou

tput

di

strib

utio

n

Ant

ibio

tics

Abe

lson

et a

l 20

06 (

1)

AltP

ert 0

.16,

0.2

, 0.2

44

53 (

90%

CrI

13-

146)

M

edic

atio

ns

N

SA

ID

Uot

ila e

t al

201

3 (1

5)

AltP

ert 0

.528

, 0.7

62,

0.91

8 19

9 (9

0% C

rI 4

6-55

1)

Med

icat

ions

E

ye d

rops

A

bels

on e

t al

2006

(1)

A

ltPer

t 0.1

6, 0

.2, 0

.244

53

(90

% C

rI 1

3-14

6)

Med

icat

ions

P

redn

ison

e A

bels

on e

t al

20

06 (

1)

Per

t 0.0

01, 0

.019

, 0.0

996

(90

% C

rI 1

-27)

M

edic

atio

ns

In

ter-

artic

ular

gl

ucoc

ortic

oid

Exp

ert o

pini

on

AltP

ert 0

.16,

0.2

, 0.2

44

53 (

90%

CrI

13-

146)

M

edic

atio

ns

D

MA

RD

U

otila

et

al 2

013

(15)

A

ltPer

t 0.0

12, 0

.095

, 0.

304

22 (

90%

CrI

3-1

15)

Med

icat

ions

Jo

int a

spira

tion

Exp

ert o

pini

on

AltP

ert 0

.16,

0.2

, 0.2

44

53 (

90%

CrI

13-

146)

T

reat

men

t

Pat

holo

gy a

nd im

agin

g fo

r G

P v

isits

T

ests

dis

trib

utio

ns m

ultip

lied

by G

P

visi

t ou

tput

dis

trib

utio

n

135

S

tool

cul

ture

E

xper

t opi

nion

A

ltPer

t 0.0

97, 0

.132

, 0.

174

35 (

90%

CrI

8-9

8)

Sto

ol c

ultu

res

S

erol

ogy

Exp

ert o

pini

on

AltP

ert 0

.097

, 0.1

32,

0.17

4 35

(90

% C

rI 8

-98)

S

erol

ogy

test

s

U

rine

m

icro

/cul

ture

/PC

R

Exp

ert o

pini

on

AltP

ert 0

.097

, 0.1

32,

0.17

4 35

(90

% C

rI 8

-98)

U

rine

cultu

res

C

RP

& U

rate

A

bels

on e

t al

2006

(1)

; Exp

ert

opin

ion

AltP

ert 0

.16,

0.2

, 0.2

44

53 (

90%

CrI

13-

146)

T

ests

F

BC

, ES

R

Abe

lson

et a

l 20

06 (

1); E

xper

t op

inio

n

AltP

ert 0

.16,

0.2

, 0.2

44

53 (

90%

CrI

13-

146)

T

ests

R

heum

atoi

d

fact

or

Abe

lson

et

al

2006

(1)

; Exp

ert

opin

ion

AltP

ert 0

.16,

0.2

, 0.2

44

53 (

90%

CrI

13-

146)

T

ests

R

enal

func

tion

Exp

ert o

pini

on

AltP

ert 0

.16,

0.2

, 0.2

44

53 (

90%

CrI

13-

146)

T

ests

B

lood

HLA

-B27

A

bels

on e

t al

2006

(1)

A

ltPer

t 0.1

6, 0

.2, 0

.244

53

(90

% C

rI 1

3-14

6)

Tes

ts

X

-ra

y E

xper

t opi

nion

A

ltPer

t 0.0

12, 0

.095

, 0.

304

22 (

90%

CrI

3-1

15)

X-r

ays

U

ltras

ound

E

xper

t opi

nion

A

ltPer

t 0.0

17, 0

.034

, 0.

062

9 (9

0% C

rI 2

-29)

U

ltras

ound

s

M

RI

Exp

ert o

pini

on

AltP

ert 0

.002

, 0.0

1, 0

.03

2 (9

0% C

rI 0

-11)

M

RIs

Spe

cial

ist v

isits

A

bels

on e

t al

2006

(1)

assu

med

da

ta a

nd b

ased

on

Han

nu e

t al

2002

(8)

Ref

erra

l to

a sp

ecia

list

20%

who

vis

it a

GP

are

ref

erre

d A

ltPer

t 0.1

6, 0

.2, 0

.244

53

(90

% C

rI 1

3-14

6)

Ref

erra

ls

Ref

erra

l dis

trib

utio

n m

ultip

lied

by G

P

visi

t ou

tput

dis

trib

utio

n

136

Spe

cial

ist v

isits

/yea

r 20

% o

f ref

erre

d ha

ve 2

vis

its p

er

year

AltP

ert 0

.223

, 0.2

4,

0.25

8 64

(90

% C

rI 1

5-17

2)

Spe

cial

ists

vi

sits

S

peci

alis

t dis

trib

utio

n m

ultip

lied

by

GP

vis

it ou

tput

dis

trib

utio

n

New

ong

oing

illn

ess

Leiri

salo

-Rep

o et

al

199

7 (1

2);

Han

nu e

t al 2

005

(9)

AltP

ert 0

.23,

0.5

, 0.7

7 89

(90

% C

rI 2

0-28

5)

Cas

es

New

ong

oing

illn

ess

dist

ribut

ion

mul

tiplie

d by

GP

vis

it ou

tput

+

hosp

italis

atio

n ou

tput

dis

trib

utio

ns

Day

s of

lost

act

iviti

es

Tow

nes

et a

l 200

8 (1

4)

AltP

ert 1

8.49

, 20.

15,

21.9

1 68

73 (

90%

CrI

162

2-18

130)

D

ays

“ReA

sym

ptom

s in

terf

ered

with

us

ual a

ctiv

ities

”. P

rodu

ctiv

ity

dist

ribut

ion

mul

tiplie

d by

inci

denc

e ou

tput

dis

trib

utio

n.

Day

s of

lost

pai

d w

ork

Tow

nes

et a

l 200

8 (1

4)

AltP

ert 2

.59,

3.3

7, 4

.3

1134

(90

% C

rI 2

67-

3117

) D

ays

“Mis

sed

wor

k be

caus

e of

ReA

sy

mpt

om

s”. P

rodu

ctiv

ity d

istr

ibut

ion

mul

tiplie

d by

inci

denc

e ou

tput

di

strib

utio

n.

Lost

pai

d w

ork

from

on

goin

g ill

ness

49

4 (9

5% C

I 195

-966

) D

ays

Ong

oing

illn

ess

dist

ribut

ion

mul

tiplie

d by

lost

pai

d w

ork

dist

ribut

ion

137

Age

gro

up: 5

-19

year

s

Sal

mo

nel

la

Var

iab

le

Dat

a so

urc

e D

istr

ibu

tio

n

Ou

tpu

t O

utp

ut

un

its

Co

mm

ents

Inci

denc

e K

irk e

t al

201

4 (1

0)

14

617

(90%

CrI

840

8-25

384)

C

ases

S

tatis

tical

mod

el in

corp

orat

ing

NN

DS

S

notif

icat

ion

num

bers

for

2013

(20

79),

20

14 (

2659

), a

nd 2

015

(266

4) w

ith

mul

tiplie

rs fo

r un

derr

epor

ting

and

dom

estic

acq

uisi

tion

No

med

ical

car

e

Ris

kOut

put 0

.31,

0.4

6,

0.59

66

44 (

90%

CrI

323

0-12

713)

C

ases

1

- (%

GP

vis

its +

%ho

spita

lised

+ %

ED

vi

sits

)

GP

vis

its

NG

SII

(11)

A

ltPer

t 0.2

5, 0

.37,

0.5

0 52

90 (

90%

CrI

281

3-99

47)

GP

vis

its

GP

vis

it di

strib

utio

n m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n

ED

vis

its

NG

SII

(11)

S

ame

met

hod

as

GP

AltP

ert 0

.06,

0.1

24,

0.22

8 17

89 (

90%

CrI

792

-39

58)

ED

vis

its

ED

vis

it di

strib

utio

n m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n

Hos

pita

lisat

ions

A

IHW

(2,

3)

Ris

kOut

put 0

.02,

0.0

4,

0.09

60

7 (9

0% C

rI 3

91-9

12)

Hos

pita

lisat

ions

Sta

tistic

al m

odel

inco

rpor

atin

g A

IHW

pr

imar

y di

agno

ses

for

2011

/12

(239

),

2012

/13

(271

), 2

013/

14 (

344)

with

m

ultip

liers

for

unde

rrep

ortin

g an

d do

mes

tic a

cqui

sitio

n D

ays

in h

ospi

tal

AIH

W p

rima

ry

diag

nosi

s 20

13-

14 –

ALO

S (

3)

3.

02 d

ays

Day

s

AIH

W p

rovi

des

<1

year

, 1-4

yea

rs, a

nd

5 ye

ar a

ge g

roup

s af

ter

that

up

to 8

5+.

A w

eigh

ted

aver

age

AL

OS

is c

alcu

late

d fo

r ou

r ag

e gr

oups

. D

eath

s A

ustr

alia

n B

urea

u of

Sta

tistic

s R

iskO

utpu

t 0.0

0006

, 0.

0001

2, 0

.000

21

1.7

(90%

CrI

1.3

4-2.

07)

Dea

ths

Sta

tistic

al m

odel

inco

rpor

atin

g A

BS

de

aths

with

mul

tiplie

rs fo

r un

derr

epor

ting

and

dom

estic

ac

quis

ition

M

edic

atio

ns

NG

SII

- A

djus

ted

for

seve

rity.

5-1

9

138

age

grou

p on

ly

(11)

A

ntid

iarr

hoea

l

Per

t 0.0

04, 0

.055

, 0.4

50

1420

(90

% C

rI 2

51-

4340

) M

edic

atio

ns

Med

icat

ion

dist

ribut

ions

mul

tiplie

d by

in

cide

nce

outp

ut d

istr

ibut

ion

P

aink

iller

AltP

ert 0

.042

, 0.1

94,

0.62

9 27

74 (

90%

CrI

656

-94

65)

Med

icat

ions

A

nti-n

ause

a

Per

t 0.0

04, 0

.055

, 0.4

50

1420

(90

% C

rI 2

51-

4340

) M

edic

atio

ns

A

nti-c

ram

ps

P

ert 0

.004

,0.0

45,0

.45

1310

(90

% C

rI 2

20-

4200

) M

edic

atio

ns

A

ntib

iotic

s

Per

t 0.0

04,0

.045

,0.4

5 13

10 (

90%

CrI

220

-42

00)

Med

icat

ions

Sto

ol c

ultu

re a

nd/o

r P

CR

N

ND

SS

S

alm

onel

la

notif

icat

ions

(5)

2079

, 265

9, 2

664

2659

S

tool

cul

ture

s N

ND

SS

Sal

mon

ella

not

ifica

tions

for

age

grou

p fo

r 20

13, 2

014,

and

201

5. O

utpu

t is

med

ian

of y

ears

. D

ays

lost

of p

aid

wor

k du

e to

bei

ng il

l N

GS

II -

Adj

uste

d fo

r se

verit

y. 5

-19

only

(11

)

AltP

ert 0

.04,

0.3

33, 1

.2

4778

(90

% C

rI 7

30-

1752

9)

Day

s N

GS

II qu

estio

n Q

34B

. Day

s lo

st

dist

ribut

ion

mul

tiplie

d by

inci

denc

e ou

tput

dis

trib

utio

n D

ays

lost

of p

aid

wor

k du

e to

car

ing

NG

SII

- A

djus

ted

for

seve

rity.

5-1

9 on

ly (

11)

AltP

ert 0

.103

,0.5

,1.4

61

7238

(90

% C

rI 1

668-

2168

8)

Day

s N

GS

II qu

estio

n Q

35B

. D

ays

lost

di

strib

utio

n m

ultip

lied

by in

cide

nce

outp

ut d

istr

ibut

ion

Day

s lo

st o

f act

iviti

es

due

to b

eing

ill

NG

SII

- A

djus

ted

for

seve

rity.

5-1

9 on

ly (

11)

AltP

ert 5

.18,

7.1

7, 9

.65

1041

21 (

90%

CrI

560

52-

1924

10)

Day

s N

GS

II qu

estio

n Q

34. I

nclu

des

paid

w

ork

& n

on-p

aid

activ

ities

. D

ays

lost

di

strib

utio

n m

ultip

lied

by in

cide

nce

outp

ut d

istr

ibut

ion

Day

s lo

st o

f act

iviti

es

due

to c

arin

g N

GS

II -

Adj

uste

d fo

r se

verit

y. 5

-19

only

(11

)

AltP

ert 0

.686

, 1.5

, 2.8

5 21

608

(90%

CrI

916

3-48

450)

D

ays

NG

SII

ques

tion

Q35

. Inc

lude

s pa

id

wor

k &

non

-pai

d ac

tiviti

es.

Day

s lo

st

dist

ribut

ion

mul

tiplie

d by

inci

denc

e ou

tput

dis

trib

utio

n IB

S f

ollo

win

g S

alm

on

ella

In

cide

nce

For

d et

al 2

014

(7)

12

84 (

90%

CrI

714

-23

08)

Cas

es

Att

ribut

able

ris

k of

IB

S a

pplie

d to

the

es

timat

ed S

alm

onel

la in

cide

nce

for

the

age

grou

p.

139

GP

vis

its

Abe

lson

et a

l 20

06 (

1); F

lik e

t al

201

5 (6

)

AltP

ert 4

.27,

4.5

, 4.7

3 57

66 (

90%

CrI

320

6-10

398)

G

P v

isits

G

P v

isit

dist

ribut

ion

mul

tiplie

d by

in

cide

nce

outp

ut d

istr

ibut

ion

ED

vis

its

Abe

lson

et a

l 20

06 (

1)

0, 0

, 0

0 E

D v

isits

Hos

pita

lisat

ions

A

IHW

(2,

3)

Ris

kOut

put 0

.006

, 0.0

13,

0.02

4 16

(90

% C

rI 9

-25)

H

ospi

talis

atio

nsS

tatis

tical

mod

el in

corp

orat

ing

AIH

W

prim

ary

diag

nose

s fo

r 20

11/1

2 (3

17),

20

12/1

3 (2

95),

201

3/14

(20

6) w

ith

mul

tiplie

rs fo

r un

derr

epor

ting,

dom

estic

ac

quis

ition

, and

pro

port

ion

due

to

ente

ric in

fect

ion.

The

n 24

.3%

of I

BS

ho

spita

lisat

ions

attr

ibut

ed s

peci

fical

ly to

S

alm

onel

la.

Day

s in

hos

pita

l A

IHW

prim

ary

di

agno

sis

2013

-14

– A

LOS

(3

)

1.

66

Day

s A

IHW

pro

vide

s <

1 ye

ar, 1

-4 y

ears

, and

5

year

age

gro

ups

afte

r th

at u

p to

85+

. A

wei

ghte

d av

erag

e A

LO

S is

cal

cula

ted

for

our

age

grou

ps.

Med

icat

ions

or

trea

tmen

ts

Exp

ert o

pini

on

AltP

ert 0

.385

, 0.4

, 0.4

16

514

(90%

CrI

285

-923

) M

edic

atio

ns

Med

icat

ion

dist

ribut

ion

mul

tiplie

d by

in

cide

nce

outp

ut d

istr

ibut

ion

Pat

holo

gy a

nd im

agin

g E

xper

t opi

nion

Tes

t dis

trib

utio

ns m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n

S

tool

cul

ture

Per

t 0.6

67, 1

, 1

1206

(90

% C

rI 6

68-

2190

) T

ests

F

BC

Per

t 0.6

67, 1

, 1

1206

(90

% C

rI 6

68-

2190

) T

ests

E

SR

Per

t 0.6

67, 1

, 1

1206

(90

% C

rI 6

68-

2190

) T

ests

LF

T

P

ert 0

.667

, 1, 1

12

06 (

90%

CrI

668

-21

90)

Tes

ts

C

RP

Per

t 0.6

67, 1

, 1

1206

(90

% C

rI 6

68-

2190

) T

ests

140

C

oelia

c di

seas

e

scre

enin

g

Per

t 0.6

67, 1

, 1

1206

(90

% C

rI 6

68-

2190

) T

ests

R

adio

logy

AltP

ert 0

.652

, 0.6

67,

0.68

1 85

5 (9

0% C

rI 4

75-1

535)

X

-ra

ys

U

ltras

ound

AltP

ert 0

.484

, 0.5

, 0.5

16

642

(90%

CrI

358

-115

3)

Ultr

asou

nds

E

ndos

copy

and

biop

sy

A

ltPer

t 0.0

5, 0

.1, 0

.,15

124

(90%

CrI

57-

257)

E

ndos

copi

es

Spe

cial

ist

Exp

ert

opin

ion;

C

anav

an e

t al

20

14 (

4)

AltP

ert 0

.286

, 0.3

, 0.3

15

385

(90%

CrI

214

-692

) S

peci

alis

t vis

its

Spe

cial

ist v

isit

dist

ribut

ion

mul

tiplie

d by

in

cide

nce

outp

ut d

istr

ibut

ion

New

ong

oing

illn

ess

Mar

shal

l et a

l 20

07 (

13)

AltP

ert 0

.218

, 0.4

29,

0.66

24

10 (

90%

CrI

113

0-50

60)

Cas

es

New

ong

oing

illn

ess

dist

ribut

ion

mul

tiplie

d by

GP

vis

it ou

tput

+

hosp

italis

atio

n ou

tput

dis

trib

utio

ns

Day

s of

lost

pai

d w

ork

/ ac

tiviti

es

Abe

lson

et

al

2006

(1)

as

sum

ed s

ame

as d

ays

in

hosp

ital +

tim

e vi

sitin

g G

P (

0.5

da

y)

(Hos

pita

lisat

ions

*

ALO

S)

+ (

GP

vis

its *

0.5

)29

50 (

90%

CrI

166

7-52

63)

Day

s (H

ospi

talis

atio

ns *

ALO

S)

+ (

GP

vis

its *

0.

5)

Lost

pai

d w

ork/

activ

ities

fr

om o

ngoi

ng il

lnes

s

66

99 (

90%

CrI

377

2-11

191)

D

ays

Ong

oing

illn

ess

dist

ribut

ion

mul

tiplie

d by

lost

pro

duct

ivity

(ab

ove)

ReA

fo

llow

ing

Sal

mo

nel

la

Inci

denc

e F

ord

et a

l 201

4 (7

)

1206

(90

% C

rI 2

90-

3125

) C

ases

A

ttrib

utab

le r

isk

of R

eA a

pplie

d to

the

es

timat

ed S

alm

onel

la in

cide

nce

for

the

age

grou

p.

GP

vis

its

Tow

nes

et a

l 20

08 (

14);

A

bels

on e

t al

20

06 (

1)

AltP

ert 0

.66,

0.8

0, 0

.89

952

(90%

CrI

229

-248

3)

GP

vis

its

GP

vis

it di

strib

utio

n m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n

ED

vis

its

Abe

lson

et a

l 20

06 (

1)

0, 0

, 0

0 E

D v

isits

141

Hos

pita

lisat

ions

A

IHW

(2,

3)

Ris

kOut

put 0

.001

, 0.0

03,

0.01

5 4.

01 (

90%

CrI

2.6

7-6.

32)

Hos

pita

lisat

ions

Sta

tistic

al m

odel

inco

rpor

atin

g A

IHW

pr

imar

y di

agno

ses

for

2011

/12

(18)

, 20

12/1

3 (1

3), 2

013/

14 (

10)

with

m

ultip

liers

for

unde

rrep

ortin

g, d

omes

tic

acqu

isiti

on, a

nd p

ropo

rtio

n du

e to

en

teric

infe

ctio

n. T

hen

25.3

% o

f ReA

ho

spita

lisat

ions

attr

ibut

ed s

peci

fical

ly to

S

alm

onel

la.

Day

s in

hos

pita

l A

IHW

prim

ary

di

agno

sis

2013

-14

– A

LOS

(3

)

2.

83

Day

s A

IHW

pro

vide

s <

1 ye

ar, 1

-4 y

ears

, and

5

year

age

gro

ups

afte

r th

at u

p to

85+

. A

wei

ghte

d av

erag

e A

LO

S is

cal

cula

ted

for

our

age

grou

ps.

Med

icat

ions

or

trea

tmen

ts fo

r G

P v

isits

M

edic

atio

n/tr

eatm

ent d

istr

ibut

ions

m

ultip

lied

by G

P v

isit

outp

ut d

istr

ibut

ion

A

ntib

iotic

s A

bels

on e

t al

2006

(1)

A

ltPer

t 0.1

6, 0

.2, 0

.244

18

9 (9

0% C

rI 4

5-50

7)

Med

icat

ions

N

SA

ID

Uot

ila e

t al

201

3 (1

5)

AltP

ert 0

.528

, 0.7

62,

0.91

8 70

3 (9

0% C

rI 1

66-1

900)

M

edic

atio

ns

E

ye d

rops

A

bels

on e

t al

2006

(1)

A

ltPer

t 0.1

6, 0

.2, 0

.244

18

9 (9

0% C

rI 4

5-50

7)

Med

icat

ions

P

redn

ison

e A

bels

on e

t al

20

06 (

1)

Per

t 0.0

01, 0

.019

, 0.0

99

23 (

90%

CrI

4-9

5)

Med

icat

ions

In

ter-

artic

ular

gl

ucoc

ortic

oid

Exp

ert o

pini

on

AltP

ert 0

.16,

0.2

, 0.2

44

189

(90%

CrI

45-

507)

M

edic

atio

ns

D

MA

RD

U

otila

et

al 2

013

(15)

A

ltPer

t 0.0

12, 0

.095

, 0.

304

81 (

90%

CrI

10-

384)

M

edic

atio

ns

Jo

int a

spira

tion

Exp

ert o

pini

on

AltP

ert 0

.16,

0.2

, 0.2

44

189

(90%

CrI

45-

507)

T

reat

men

t

Pat

holo

gy a

nd im

agin

g fo

r G

P v

isits

T

ests

dis

trib

utio

ns m

ultip

lied

by G

P v

isit

outp

ut d

istr

ibut

ion

S

tool

cul

ture

E

xper

t opi

nion

A

ltPer

t 0.0

97, 0

.132

, 0.

174

125

(90%

CrI

29-

336)

S

tool

cul

ture

s

142

S

erol

ogy

Exp

ert o

pini

on

AltP

ert 0

.097

, 0.1

32,

0.17

4 12

5 (9

0% C

rI 2

9-33

6)

Ser

olog

y te

sts

U

rine

m

icro

/cul

ture

/PC

R

Exp

ert o

pini

on

AltP

ert 0

.097

, 0.1

32,

0.17

4 12

5 (9

0% C

rI 2

9-33

6)

Urin

e cu

lture

s

C

RP

& U

rate

A

bels

on e

t al

2006

(1)

; Exp

ert

opin

ion

AltP

ert 0

.16,

0.2

, 0.2

44

189

(90%

CrI

45-

507)

T

ests

F

BC

, ES

R

Abe

lson

et a

l 20

06 (

1); E

xper

t op

inio

n

AltP

ert 0

.16,

0.2

, 0.2

44

189

(90%

CrI

45-

507)

T

ests

R

heum

atoi

d

fact

or

Abe

lson

et

al

2006

(1)

; Exp

ert

opin

ion

AltP

ert 0

.16,

0.2

, 0.2

44

189

(90%

CrI

45-

507)

T

ests

R

enal

func

tion

Exp

ert o

pini

on

AltP

ert 0

.16,

0.2

, 0.2

44

189

(90%

CrI

45-

507)

T

ests

B

lood

HLA

-B27

A

bels

on e

t al

2006

(1)

A

ltPer

t 0.1

6, 0

.2, 0

.244

18

9 (9

0% C

rI 4

5-50

7)

Tes

ts

X

-ra

y E

xper

t opi

nion

A

ltPer

t 0.0

12, 0

.095

, 0.

304

81 (

90%

CrI

10-

384)

X

-ra

ys

U

ltras

ound

E

xper

t opi

nion

A

ltPer

t 0.0

17, 0

.034

, 0.

062

31 (

90%

CrI

7-1

00)

Ultr

asou

nds

M

RI

Exp

ert o

pini

on

AltP

ert 0

.002

, 0.0

1, 0

.03

8 (9

0% C

rI 1

-40)

M

RIs

Spe

cial

ist v

isits

A

bels

on e

t al

2006

(1)

assu

med

da

ta a

nd b

ased

on

Han

nu e

t al

2002

(8)

Ref

erra

l to

a sp

ecia

list

20%

who

vis

it a

GP

are

ref

erre

d A

ltPer

t 0.1

6, 0

.2, 0

.244

18

9 (9

0% C

rI 4

5-50

7)

Ref

erra

ls

Ref

erra

l dis

trib

utio

n m

ultip

lied

by G

P

visi

t ou

tput

dis

trib

utio

n

143

Spe

cial

ist v

isits

/yea

r 20

% o

f ref

erre

d ha

ve 2

vis

its p

er

year

AltP

ert 0

.223

, 0.2

4,

0.25

8 22

8 (9

0% C

rI 5

5-59

5)

Spe

cial

ists

vi

sits

S

peci

alis

t dis

trib

utio

n m

ultip

lied

by G

P

visi

t ou

tput

dis

trib

utio

n

New

ong

oing

illn

ess

Leiri

salo

-Rep

o et

al

199

7 (1

2);

Han

nu e

t al 2

005

(9)

AltP

ert 0

.23,

0.5

, 0.7

7 45

0 (9

0% C

rI 1

03-1

365)

C

ases

N

ew o

ngoi

ng il

lnes

s di

strib

utio

n m

ultip

lied

by G

P v

isit

outp

ut +

ho

spita

lisat

ion

outp

ut d

istr

ibut

ions

Day

s of

lost

act

iviti

es

Tow

nes

et a

l 20

08 (

14)

AltP

ert 1

8.49

, 20.

15,

21.9

1 24

271

(90%

CrI

583

0-62

950)

D

ays

“ReA

sym

ptom

s in

terf

ered

with

usu

al

activ

ities

”. P

rodu

ctiv

ity d

istr

ibut

ion

mul

tiplie

d by

inci

denc

e ou

tput

di

strib

utio

n.

Day

s of

lost

pai

d w

ork

Tow

nes

et a

l 20

08 (

14)

AltP

ert 2

.59,

3.3

7, 4

.3

4029

(90

% C

rI 9

71-

1074

9)

Day

s “M

isse

d w

ork

beca

use

of R

eA

sym

pto

ms”

. Pro

duct

ivity

dis

trib

utio

n m

ultip

lied

by in

cide

nce

outp

ut

dist

ribut

ion.

Lo

st p

aid

wor

k fr

om

ongo

ing

illne

ss

1507

(95

% C

I 342

-473

6)D

ays

Ong

oing

illn

ess

dist

ribut

ion

mul

tiplie

d by

lost

pai

d w

ork

dist

ribut

ion

144

Age

gro

up: 2

0-64

yea

rs

Sal

mo

nel

la

Var

iab

le

Dat

a so

urc

e D

istr

ibu

tio

n

Ou

tpu

t O

utp

ut

un

its

Co

mm

ents

Inci

denc

e K

irk e

t al

201

4 (1

0)

44

525

(90%

CrI

250

12-

7748

9)

Cas

es

Sta

tistic

al m

odel

inco

rpor

atin

g N

ND

SS

not

ifica

tion

num

bers

for

2013

(6

083)

, 201

4 (8

027)

, and

201

5 (8

408)

w

ith m

ultip

liers

for

unde

rrep

ortin

g an

d do

mes

tic a

cqui

sitio

n

No

med

ical

car

e

Ris

kOut

put 0

.47,

0.5

9,

0.70

26

075

(90%

CrI

133

34-

4809

3)

Cas

es

1 -

(%G

P v

isits

+ %

hosp

italis

ed +

%

ED

vis

its)

GP

vis

its

NG

SII

(11)

A

ltPer

t 0.2

5, 0

.37,

0.5

0 16

145

(90%

CrI

823

6-30

686)

G

P v

isits

G

P v

isit

dist

ribut

ion

mul

tiplie

d by

in

cide

nce

outp

ut d

istr

ibut

ion

ED

vis

its

NG

SII

(11)

S

ame

met

hod

as G

P

AltP

ert 0

.06,

0.1

24,

0.22

8 54

56 (

90%

CrI

230

1-12

014)

E

D v

isits

E

D v

isit

dist

ribut

ion

mul

tiplie

d by

in

cide

nce

outp

ut d

istr

ibut

ion

Hos

pita

lisat

ions

A

IHW

(2,

3)

Ris

kOut

put 0

.02,

0.0

4,

0.08

18

14 (

90%

CrI

116

2-27

50)

Hos

pita

lisat

ions

Sta

tistic

al m

odel

inco

rpor

atin

g A

IHW

pr

imar

y di

agno

ses

for

2011

/12

(708

),

2012

/13

(814

), 2

013/

14 (

1041

) w

ith

mul

tiplie

rs fo

r un

derr

epor

ting

and

dom

estic

acq

uisi

tion

Day

s in

hos

pita

l A

IHW

prim

ary

di

agno

sis

2013

-14

– A

LOS

(3

)

3.

80 d

ays

Day

s

AIH

W p

rovi

des

<1

year

, 1-4

yea

rs,

and

5 ye

ar a

ge g

roup

s af

ter

that

up

to

85+

. A w

eigh

ted

aver

age

ALO

S is

ca

lcul

ated

for

our

age

grou

ps.

Dea

ths

Aus

tral

ian

Bur

eau

of

Sta

tistic

s

Ris

kOut

put 0

.000

04,

0.00

008,

0.0

0015

3.

53 (

90%

CrI

2.7

1-4.

38)

Dea

ths

Sta

tistic

al m

odel

inco

rpor

atin

g A

BS

de

aths

with

mul

tiplie

rs fo

r un

derr

epor

ting

and

dom

estic

ac

quis

ition

M

edic

atio

ns

NG

SII

- A

djus

ted

for

seve

rity.

20-

64

145

age

grou

p on

ly

(11)

A

ntid

iarr

hoea

l

AltP

ert 0

.166

, 0.2

86,

0.43

9 12

569

(90%

CrI

605

6-24

515)

M

edic

atio

ns

Med

icat

ion

dist

ribut

ions

mul

tiplie

d by

in

cide

nce

outp

ut d

istr

ibut

ion

P

aink

iller

AltP

ert 0

.127

, 0.2

42,

0.38

5 10

630

(90%

CrI

481

0-21

791)

M

edic

atio

ns

A

nti-n

ause

a

AltP

ert 0

.042

, 0.1

2,

0.23

7 51

84 (

90%

CrI

187

0-12

430)

M

edic

atio

ns

A

nti-c

ram

ps

P

ert 0

.028

, 0.0

77,

0.20

4 37

89 (

90%

CrI

153

1-83

39)

Med

icat

ions

A

ntib

iotic

s

AltP

ert 0

.016

, 0.0

67,

0.16

9 29

34 (

90%

CrI

782

-819

4)

Med

icat

ions

Sto

ol c

ultu

re a

nd/o

r P

CR

N

ND

SS

S

alm

onel

la

notif

icat

ions

(5)

6083

, 802

7, 8

408

8207

S

tool

cul

ture

s N

ND

SS

Sal

mon

ella

not

ifica

tions

for

age

grou

p fo

r 20

13, 2

014,

and

201

5.

Out

put i

s m

edia

n of

yea

rs.

Day

s lo

st o

f pai

d w

ork

due

to b

eing

ill

NG

SII

- A

djus

ted

for

seve

rity.

20-

64

only

(11

)

AltP

ert 0

.929

, 1.2

56,

1.66

1 55

790

(90%

CrI

295

24-

1022

69)

Day

s N

GS

II qu

estio

n Q

34B

. Day

s lo

st

dist

ribut

ion

mul

tiplie

d by

inci

denc

e ou

tput

dis

trib

utio

n

Day

s lo

st o

f pai

d w

ork

due

to c

arin

g N

GS

II -

Adj

uste

d fo

r se

verit

y. 2

0-64

on

ly (

11)

AltP

ert 0

.159

, 0.3

08,

0.53

7 13

366

(90%

CrI

606

7-29

351)

D

ays

NG

SII

ques

tion

Q35

B.

Day

s lo

st

dist

ribut

ion

mul

tiplie

d by

inci

denc

e ou

tput

dis

trib

utio

n

Day

s lo

st o

f act

iviti

es

due

to b

eing

ill

NG

SII

- A

djus

ted

for

seve

rity.

20-

64

only

(11

)

AltP

ert 3

.847

, 4.4

87,

5.20

3 19

9771

(90

% C

rI 1

0970

6-35

3563

) D

ays

NG

SII

ques

tion

Q34

. Inc

lude

s pa

id

wor

k &

non

-pai

d ac

tiviti

es.

Day

s lo

st

dist

ribut

ion

mul

tiplie

d by

inci

denc

e ou

tput

dis

trib

utio

n D

ays

lost

of a

ctiv

ities

du

e to

car

ing

NG

SII

- A

djus

ted

for

seve

rity.

20-

64

only

(11

)

AltP

ert 0

.498

, 0.7

44,

1.06

8

3286

8 (9

0% C

rI 1

6828

-62

858)

D

ays

NG

SII

ques

tion

Q35

. Inc

lude

s pa

id

wor

k &

non

-pai

d ac

tiviti

es.

Day

s lo

st

dist

ribut

ion

mul

tiplie

d by

inci

denc

e ou

tput

dis

trib

utio

n IB

S f

ollo

win

g S

alm

on

ella

146

Inci

denc

e F

ord

et a

l 201

4 (7

)

3906

(90

% C

rI 2

104-

7077

) C

ases

A

ttrib

utab

le r

isk

of IB

S a

pplie

d to

the

estim

ated

Sal

mon

ella

inci

denc

e fo

r th

e ag

e gr

oup.

G

P v

isits

A

bels

on e

t al

20

06 (

1); F

lik e

t al

201

5 (6

)

AltP

ert 4

.27,

4.5

, 4.7

3 17

582

(90%

CrI

947

1-31

969)

G

P v

isits

G

P v

isit

dist

ribut

ion

mul

tiplie

d by

in

cide

nce

outp

ut d

istr

ibut

ion

ED

vis

its

Abe

lson

et a

l 20

06 (

1)

0, 0

, 0

0 E

D v

isits

Hos

pita

lisat

ions

A

IHW

(2,

3)

Ris

kOut

put 0

.04,

0.0

8,

0.18

32

6 (9

0% C

rI 1

86-5

17)

Hos

pita

lisat

ions

Sta

tistic

al m

odel

inco

rpor

atin

g A

IHW

pr

imar

y di

agno

ses

for

2011

/12

(681

0),

2012

/13

(555

8), 2

013/

14 (

5027

) w

ith

mul

tiplie

rs fo

r un

derr

epor

ting,

do

mes

tic a

cqui

sitio

n, a

nd p

ropo

rtio

n du

e to

ent

eric

infe

ctio

n. T

hen

24.3

%

of I

BS

hos

pita

lisat

ions

attr

ibut

ed

spec

ifica

lly to

Sal

mon

ella

. D

ays

in h

ospi

tal

AIH

W p

rima

ry

diag

nosi

s 20

13-

14 –

ALO

S (

3)

1.

22

Day

s A

IHW

pro

vide

s <

1 ye

ar, 1

-4 y

ears

, an

d 5

year

age

gro

ups

afte

r th

at u

p to

85

+. A

wei

ghte

d av

erag

e A

LOS

is

calc

ulat

ed fo

r ou

r ag

e gr

oups

. M

edic

atio

ns o

r tr

eatm

ents

E

xper

t opi

nion

A

ltPer

t 0.3

85, 0

.4,

0.41

6 15

62 (

90%

CrI

840

-283

8)

Med

icat

ions

M

edic

atio

n di

strib

utio

n m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n

Pat

holo

gy o

r im

agin

g E

xper

t opi

nion

Tes

t dis

trib

utio

ns m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n

S

tool

cul

ture

Per

t 0.6

67, 1

, 1

3684

(90

% C

rI 1

984-

6696

) T

ests

F

BC

Per

t 0.6

67, 1

, 1

3684

(90

% C

rI 1

984-

6696

) T

ests

E

SR

Per

t 0.6

67, 1

, 1

3684

(90

% C

rI 1

984-

6696

) T

ests

LF

T

P

ert 0

.667

, 1, 1

36

84 (

90%

CrI

198

4-66

96)

Tes

ts

147

C

RP

Per

t 0.6

67, 1

, 1

3684

(90

% C

rI 1

984-

6696

) T

ests

C

oelia

c di

seas

e

scre

enin

g

Per

t 0.6

67, 1

, 1

3684

(90

% C

rI 1

984-

6696

) T

ests

R

adio

logy

AltP

ert 0

.652

, 0.6

67,

0.68

1 26

06 (

90%

CrI

140

5-47

17)

X-r

ays

U

ltras

ound

AltP

ert 0

.484

, 0.5

, 0.

516

1951

(90

% C

rI 1

051-

3542

) U

ltras

ound

s

E

ndos

copy

and

biop

sy

A

ltPer

t 0.0

5, 0

.1, 0

.,15

379

(90%

CrI

168

-798

) E

ndos

copi

es

Spe

cial

ist

Exp

ert o

pini

on;

Can

avan

et

al

2014

(4)

AltP

ert 0

.286

, 0.3

, 0.

315

1171

(90

% C

rI 6

32-2

120)

S

peci

alis

t vis

its

Spe

cial

ist v

isit

dist

ribut

ion

mul

tiplie

d by

inci

denc

e ou

tput

dis

trib

utio

n

New

ong

oing

illn

ess

Mar

shal

l et a

l 20

07 (

13)

AltP

ert 0

.218

, 0.4

29,

0.66

74

74 (

90%

CrI

337

0-15

720)

C

ases

N

ew o

ngoi

ng il

lnes

s di

strib

utio

n m

ultip

lied

by G

P v

isit

outp

ut +

ho

spita

lisat

ion

outp

ut d

istr

ibut

ions

D

ays

of lo

st p

aid

wor

k /

activ

ities

A

bels

on e

t al

20

06 (

1)

assu

med

sam

e as

day

s in

ho

spita

l + ti

me

visi

ting

GP

(0.

5 d

ay)

(Hos

pita

lisat

ions

*

ALO

S)

+ (

GP

vis

its *

0.

5)

1012

5 (9

0% C

rI 6

002-

1732

1)

Day

s (H

ospi

talis

atio

ns *

ALO

S)

+ (

GP

vis

its

* 0.

5)

Lost

pai

d w

ork/

activ

ities

fr

om o

ngoi

ng il

lnes

s

17

705

(90%

CrI

796

8-34

322)

D

ays

Ong

oing

illn

ess

dist

ribut

ion

mul

tiplie

d by

lost

pro

duct

ivity

(ab

ove)

ReA

fo

llow

ing

Sal

mo

nel

la

Inci

denc

e F

ord

et a

l 201

4 (7

)

2883

(90

% C

rI 6

88-7

718)

C

ases

A

ttrib

utab

le r

isk

of R

eA a

pplie

d to

the

estim

ated

Sal

mon

ella

inci

denc

e fo

r th

e ag

e gr

oup.

G

P v

isits

T

owne

s et

al

2008

(14

);

Abe

lson

et

al

2006

(1)

AltP

ert 0

.66,

0.8

, 0.8

9 28

83 (

90%

CrI

688

-771

8)

GP

vis

its

GP

vis

it di

strib

utio

n m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n

148

ED

vis

its

Abe

lson

et a

l 20

06 (

1)

0, 0

, 0

0 E

D v

isits

Hos

pita

lisat

ions

A

IHW

(2,

3)

Ris

kOut

put 0

.002

, 0.

005,

0.0

23

20 (

90%

CrI

13-

26)

Hos

pita

lisat

ions

Sta

tistic

al m

odel

inco

rpor

atin

g A

IHW

pr

imar

y di

agno

ses

for

2011

/12

(71)

, 20

12/1

3 (7

1), 2

013/

14 (

50)

with

m

ultip

liers

for

unde

rrep

ortin

g,

dom

estic

acq

uisi

tion,

and

pro

port

ion

due

to e

nter

ic in

fect

ion.

The

n 25

.3%

of

ReA

hos

pita

lisat

ions

att

ribut

ed

spec

ifica

lly to

Sal

mon

ella

. D

ays

in h

ospi

tal

AIH

W p

rima

ry

diag

nosi

s 20

13-

14 –

ALO

S (

3)

3.

56

Day

s A

IHW

pro

vide

s <

1 ye

ar, 1

-4 y

ears

, an

d 5

year

age

gro

ups

afte

r th

at u

p to

85

+. A

wei

ghte

d av

erag

e A

LOS

is

calc

ulat

ed fo

r ou

r ag

e gr

oups

. M

edic

atio

ns o

r tr

eatm

ents

for

thos

e w

ho

visi

t a G

P

Med

icat

ion/

trea

tmen

t dis

trib

utio

ns

mul

tiplie

d by

GP

vis

it ou

tput

di

strib

utio

n

Ant

ibio

tics

Abe

lson

et a

l 20

06 (

1)

AltP

ert 0

.16,

0.2

, 0.2

44

576

(90%

CrI

133

-155

1)

Med

icat

ions

N

SA

ID

Uot

ila e

t al

201

3 (1

5)

AltP

ert 0

.528

, 0.7

62,

0.91

8 21

31 (

90%

CrI

500

-589

7)

Med

icat

ions

E

ye d

rops

A

bels

on e

t al

2006

(1)

A

ltPer

t 0.1

6, 0

.2, 0

.244

57

6 (9

0% C

rI 1

33-1

551)

M

edic

atio

ns

P

redn

ison

e A

bels

on e

t al

20

06 (

1)

Per

t 0.0

01, 0

.019

, 0.

099

69 (

90%

CrI

10-

294)

M

edic

atio

ns

In

ter-

artic

ular

gl

ucoc

ortic

oid

Exp

ert o

pini

on

AltP

ert 0

.16,

0.2

, 0.2

44

576

(90%

CrI

133

-155

1)

Med

icat

ions

D

MA

RD

U

otila

et

al 2

013

(15)

A

ltPer

t 0.0

12, 0

.095

, 0.

304

240

(90%

CrI

30-

1213

) M

edic

atio

ns

Jo

int a

spira

tion

Exp

ert o

pini

on

AltP

ert 0

.16,

0.2

, 0.2

44

576

(90%

CrI

133

-155

1)

Tre

atm

ent

149

Pat

holo

gy a

nd im

agin

g fo

r th

ose

who

vis

it a

GP

T

ests

dis

trib

utio

ns m

ultip

lied

by G

P

visi

t ou

tput

dis

trib

utio

n

S

tool

cul

ture

E

xper

t opi

nion

A

ltPer

t 0.0

97, 0

.132

, 0.

174

378

(90%

CrI

88-

1044

) S

tool

cul

ture

s

S

erol

ogy

Exp

ert o

pini

on

AltP

ert 0

.097

, 0.1

32,

0.17

4 37

8 (9

0% C

rI 8

8-10

44

Ser

olog

y te

sts

U

rine

m

icro

/cul

ture

/PC

R

Exp

ert o

pini

on

AltP

ert 0

.097

, 0.1

32,

0.17

4 37

8 (9

0% C

rI 8

8-10

44

Urin

e cu

lture

s

C

RP

& U

rate

A

bels

on e

t al

2006

(1)

; Exp

ert

opin

ion

AltP

ert 0

.16,

0.2

, 0.2

44

576

(90%

CrI

133

-155

1)

Tes

ts

F

BC

, ES

R

Abe

lson

et a

l 20

06 (

1); E

xper

t op

inio

n

AltP

ert 0

.16,

0.2

, 0.2

44

576

(90%

CrI

133

-155

1)

Tes

ts

R

heum

atoi

d

fact

or

Abe

lson

et

al

2006

(1)

; Exp

ert

opin

ion

AltP

ert 0

.16,

0.2

, 0.2

44

576

(90%

CrI

133

-155

1)

Tes

ts

R

enal

func

tion

Exp

ert o

pini

on

AltP

ert 0

.16,

0.2

, 0.2

44

576

(90%

CrI

133

-155

1)

Tes

ts

B

lood

HLA

-B27

A

bels

on e

t al

2006

(1)

A

ltPer

t 0.1

6, 0

.2, 0

.244

57

6 (9

0% C

rI 1

33-1

551)

T

ests

X

-ra

y E

xper

t opi

nion

A

ltPer

t 0.0

12, 0

.095

, 0.

304

240

(90%

CrI

30-

1213

) X

-ra

ys

U

ltras

ound

E

xper

t opi

nion

A

ltPer

t 0.0

17, 0

.034

, 0.

062

94 (

90%

CrI

21-

308)

U

ltras

ound

s

M

RI

Exp

ert o

pini

on

AltP

ert 0

.002

, 0.0

1,

0.03

26

(90

% C

rI 4

-121

) M

RIs

Spe

cial

ist v

isits

A

bels

on e

t al

2006

(1

)ass

umed

da

ta a

nd b

ased

150

on H

annu

et a

l 20

02 (

8)

Ref

erra

l to

rheu

mat

olog

ist

20%

who

vis

it a

GP

are

ref

erre

d A

ltPer

t 0.1

6, 0

.2, 0

.244

57

6 (9

0% C

rI 1

33-1

551)

R

efer

rals

R

efer

ral d

istr

ibut

ion

mul

tiplie

d by

GP

vi

sit

outp

ut d

istr

ibut

ion

Spe

cial

ist v

isits

/yea

r 20

% o

f ref

erre

d ha

ve 2

vis

its p

er

year

AltP

ert 0

.223

, 0.2

4,

0.25

8 69

2 (9

0% C

rI 1

65-1

847)

S

peci

alis

ts

visi

ts

Spe

cial

ist d

istr

ibut

ion

mul

tiplie

d by

GP

vi

sit

outp

ut d

istr

ibut

ion

New

ong

oing

illn

ess

Leiri

salo

-Rep

o et

al 1

997

(12)

; H

annu

et a

l 20

05 (

9)

AltP

ert 0

.23,

0.5

, 0.7

7 13

73 (

90%

CrI

309

-418

5)

Cas

es

New

ong

oing

illn

ess

dist

ribut

ion

mul

tiplie

d by

GP

vis

it ou

tput

+

hosp

italis

atio

n ou

tput

dis

trib

utio

ns

Day

s of

lost

pai

d ac

tiviti

es

Tow

nes

et a

l 20

08 (

14)

AltP

ert 1

8.49

, 20.

15,

21.9

1 73

538

(90%

CrI

173

00-

1952

52)

Day

s “R

eA s

ympt

oms

inte

rfer

ed w

ith u

sual

ac

tiviti

es”.

Pro

duct

ivity

dis

trib

utio

n m

ultip

lied

by in

cide

nce

outp

ut

dist

ribut

ion.

D

ays

of lo

st p

aid

wor

k T

owne

s et

al

2008

(14

) A

ltPer

t 2.5

9, 3

.37,

4.3

12

231

(90%

CrI

286

8-33

333)

D

ays

“Mis

sed

wor

k be

caus

e of

ReA

sy

mpt

om

s”. P

rodu

ctiv

ity d

istr

ibut

ion

mul

tiplie

d by

inci

denc

e ou

tput

di

strib

utio

n.

Lost

pai

d w

ork

from

on

goin

g ill

ness

53

13 (

95%

CI 2

100-

1039

0)

Day

s O

ngoi

ng il

lnes

s di

strib

utio

n m

ultip

lied

by lo

st p

aid

wor

k di

strib

utio

n

151

Age

gro

up: 6

5+ y

ears

Var

iab

le

Dat

a so

urc

e D

istr

ibu

tio

n

Ou

tpu

t O

utp

ut

un

its

Co

mm

ents

Inci

denc

e K

irk e

t al

20

14 (

10)

10

767

(90%

CrI

610

4-19

042)

C

ases

S

tatis

tical

mod

el in

corp

orat

ing

NN

DS

S

notif

icat

ion

num

bers

for

2013

(15

24),

20

14 (

1867

), a

nd 2

015

(207

7) w

ith

mul

tiplie

rs fo

r un

derr

epor

ting

and

dom

estic

acq

uisi

tion

N

o m

edic

al c

are

R

iskO

utpu

t 0.4

0, 0

.54,

0.

66

5829

(90

% C

rI 2

754-

1132

6)

1

- (%

GP

vis

its +

%ho

spita

lised

+ %

ED

vi

sits

)

GP

vis

its

NG

SII

(11)

A

ltPer

t 0.2

5, 0

.37,

0.5

0 39

14 (

90%

CI 2

041-

7438

) G

P v

isits

G

P v

isit

dist

ribut

ion

mul

tiplie

d by

in

cide

nce

outp

ut d

istr

ibut

ion

ED

vis

its

NG

SII

(11)

S

ame

met

hod

as

GP

AltP

ert 0

.06,

0.1

24,

0.22

8 13

20 (

90%

CrI

568

-29

05)

ED

vis

its

ED

vis

it di

strib

utio

n m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n

Hos

pita

lisat

ions

A

IHW

(2,

3)

Ris

kOut

put 0

.04,

0.0

8,

0.19

90

7 (9

0% C

I 570

-138

2)

Hos

pita

lisat

ions

Sta

tistic

al m

odel

inco

rpor

atin

g A

IHW

pr

imar

y di

agno

ses

for

2011

/12

(341

),

2012

/13

(419

), 2

013/

14 (

519)

with

m

ultip

liers

for

unde

rrep

ortin

g an

d do

mes

tic a

cqui

sitio

n D

ays

in h

ospi

tal

AIH

W

prim

ary

diag

nosi

s 20

13-1

4 –

ALO

S (

3)

6.

80 d

ays

Day

s

AIH

W p

rovi

des

<1

year

, 1-4

yea

rs, a

nd 5

ye

ar a

ge g

roup

s af

ter

that

up

to 8

5+. A

w

eigh

ted

aver

age

ALO

S is

cal

cula

ted

for

our

age

grou

ps.

Dea

ths

Aus

tral

ian

Bur

eau

of

Sta

tistic

s

Ris

kOut

put 0

.000

6,

0.00

12, 0

.002

1 13

(90

% C

I 10-

15)

Dea

ths

Sta

tistic

al m

odel

inco

rpor

atin

g A

BS

de

aths

with

mul

tiplie

rs fo

r un

derr

epor

ting

and

dom

estic

acq

uisi

tion

Med

icat

ions

N

GS

II -

Adj

uste

d fo

r se

verit

y. 6

5+

152

age

grou

p on

ly (

11)

A

ntid

iarr

hoea

l

AltP

ert 0

.094

, 0.6

53,

0.90

6

6573

(90

% C

rI 1

799-

1335

7)

Med

icat

ions

M

edic

atio

n di

strib

utio

ns m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n

P

aink

iller

AltP

ert 0

.008

, 0.3

07,

0.70

8 31

62 (

90%

CrI

381

-85

59)

Med

icat

ions

A

nti-n

ause

a

AltP

ert 0

.008

, 0.3

07,

0.70

8 31

62 (

90%

CrI

381

-85

59)

Med

icat

ions

A

nti-c

ram

ps

A

ltPer

t 0.0

08, 0

.277

, 0.

708

2874

(90

% C

rI 3

18-

8416

) M

edic

atio

ns

A

ntib

iotic

s

0, 0

.051

, 0.1

04

531

(90%

CrI

59-

1309

) M

edic

atio

ns

Sto

ol c

ultu

re a

nd/o

r P

CR

N

ND

SS

S

alm

onel

la

notif

icat

ions

(5

)

1524

, 186

7, 2

077

1867

S

tool

cul

ture

s N

ND

SS

Sal

mon

ella

not

ifica

tions

for

age

grou

p fo

r 20

13, 2

014,

and

201

5. O

utpu

t is

med

ian

of y

ears

.

Day

s lo

st o

f pai

d w

ork

due

to b

eing

ill

NG

SII

- A

djus

ted

for

seve

rity.

65+

on

ly (

11)

AltP

ert 0

.081

, 0.6

67,

2.40

8 70

79 (

90%

CrI

109

0-25

866)

D

ays

NG

SII

ques

tion

Q34

B. D

ays

lost

di

strib

utio

n m

ultip

lied

by in

cide

nce

outp

ut

dist

ribut

ion

Day

s lo

st o

f pai

d w

ork

due

to b

eing

car

ed f

or

NG

SII

- A

djus

ted

for

seve

rity.

65+

on

ly (

11)

Per

t 0.0

08, 0

.33,

1.8

57

5084

(90

% C

rI 1

006-

1463

2)

Day

s N

GS

II qu

estio

n Q

35B

. D

ays

lost

di

strib

utio

n m

ultip

lied

by in

cide

nce

outp

ut

dist

ribut

ion

Day

s lo

st o

f act

iviti

es

due

to b

eing

ill

NG

SII

- A

djus

ted

for

seve

rity.

65+

on

ly (

11)

AltP

ert 4

.86,

7.6

67,

11.5

04

8215

8 (9

0% C

rI 4

1194

-16

1843

) D

ays

NG

SII

ques

tion

Q34

. Inc

lude

s pa

id w

ork

&

non-

paid

act

iviti

es.

Day

s lo

st d

istr

ibut

ion

mul

tiplie

d by

inci

denc

e ou

tput

dis

trib

utio

n

Day

s lo

st o

f act

iviti

es

due

to c

arin

g N

GS

II -

Adj

uste

d fo

r se

verit

y. 6

5+

only

(11

)

Per

t 0.0

08, 0

.33,

1.8

57

5071

(90

% C

rI 9

98-

1458

7)

Day

s N

GS

II qu

estio

n Q

35. I

nclu

des

paid

wor

k &

no

n-pa

id a

ctiv

ities

. D

ays

lost

dis

trib

utio

n m

ultip

lied

by in

cide

nce

outp

ut d

istr

ibut

ion

IBS

fo

llow

ing

Sal

mo

nel

la

153

Inci

denc

e F

ord

et a

l 20

14 (

7)

94

2 (9

0% C

rI 5

16-1

703)

C

ases

A

ttrib

utab

le r

isk

of IB

S a

pplie

d to

the

estim

ated

Sal

mon

ella

inci

denc

e fo

r th

e ag

e gr

oup.

G

P v

isits

A

bels

on e

t al

20

06 (

1); F

lik

et a

l 201

5 (6

)

AltP

ert 4

.27,

4.5

, 4.7

3 42

29 (

90%

CrI

231

6-76

55)

GP

vis

its

GP

vis

it di

strib

utio

n m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n

ED

vis

its

Abe

lson

et a

l 20

06 (

1)

0, 0

, 0

0 E

D v

isits

Hos

pita

lisat

ions

A

IHW

(2,

3)

Ris

kOut

put 0

.03,

0.0

7,

0.16

67

(90

% C

rI 3

9-10

4)

Hos

pita

lisat

ions

Sta

tistic

al m

odel

inco

rpor

atin

g A

IHW

pr

imar

y di

agno

ses

for

2011

/12

(136

8),

2012

/13

(113

1), 2

013/

14 (

1086

) w

ith

mul

tiplie

rs fo

r un

derr

epor

ting,

dom

estic

ac

quis

ition

, and

pro

port

ion

due

to e

nter

ic

infe

ctio

n. T

hen

24.3

% o

f IB

S

hosp

italis

atio

ns a

ttrib

uted

spe

cific

ally

to

Sal

mon

ella

. D

ays

in h

ospi

tal

AIH

W

prim

ary

diag

nosi

s 20

13-1

4 –

ALO

S (

3)

1.

2 D

ays

AIH

W p

rovi

des

<1

year

, 1-4

yea

rs, a

nd 5

ye

ar a

ge g

roup

s af

ter

that

up

to 8

5+. A

w

eigh

ted

aver

age

ALO

S is

cal

cula

ted

for

our

age

grou

ps.

Med

icat

ions

or

trea

tmen

ts

Exp

ert

opin

ion

AltP

ert 0

.385

, 0.4

, 0.4

16

377

(90%

CrI

207

-684

) M

edic

atio

ns

Med

icat

ion

dist

ribut

ion

mul

tiplie

d by

in

cide

nce

outp

ut d

istr

ibut

ion

Pat

holo

gy o

r im

agin

g E

xper

t op

inio

n

Tes

t dis

trib

utio

ns m

ultip

lied

by in

cide

nce

outp

ut d

istr

ibut

ion

S

tool

cul

ture

Per

t 0.6

67, 1

, 1

889

(90%

CrI

485

-161

0)

Tes

ts

F

BC

Per

t 0.6

67, 1

, 1

889

(90%

CrI

485

-161

0)

Tes

ts

E

SR

Per

t 0.6

67, 1

, 1

889

(90%

CrI

485

-161

0)

Tes

ts

LF

T

P

ert 0

.667

, 1, 1

88

9 (9

0% C

rI 4

85-1

610)

T

ests

154

C

RP

Per

t 0.6

67, 1

, 1

889

(90%

CrI

485

-161

0)

Tes

ts

C

oelia

c di

seas

e

scre

enin

g

Per

t 0.6

67, 1

, 1

889

(90%

CrI

485

-161

0)

Tes

ts

R

adio

logy

AltP

ert 0

.652

, 0.6

67,

0.68

1 62

9 (9

0% C

rI 3

44-1

134)

X

-ra

ys

U

ltras

ound

AltP

ert 0

.484

, 0.5

, 0.5

16

472

(90%

CrI

258

-849

) U

ltras

ound

s

E

ndos

copy

and

biop

sy

A

ltPer

t 0.1

5, 0

.2, 0

.25

186

(90%

CrI

99-

353)

E

ndos

copi

es

Spe

cial

ist

Exp

ert

opin

ion;

C

anav

an e

t al

201

4 (4

)

AltP

ert 0

.286

, 0.3

, 0.3

15

282

(90%

CrI

155

-512

) S

peci

alis

t vis

its

Spe

cial

ist v

isit

dist

ribut

ion

mul

tiplie

d by

in

cide

nce

outp

ut d

istr

ibut

ion

New

ong

oing

illn

ess

Mar

shal

l et a

l 20

07 (

13)

AltP

ert 0

.218

, 0.4

29,

0.66

17

92 (

90%

CrI

816

-37

56)

Cas

es

New

ong

oing

illn

ess

dist

ribut

ion

mul

tiplie

d by

GP

vis

it ou

tput

+ h

ospi

talis

atio

n ou

tput

di

strib

utio

ns

Day

s of

lost

pai

d w

ork

/ ac

tiviti

es

Abe

lson

et

al

2006

(1)

as

sum

ed

sam

e as

da

ys in

ho

spita

l +

time

visi

ting

GP

(0.

5 da

y)

(Hos

pita

lisat

ions

*

ALO

S)

+ (

GP

vis

its *

0.5

)23

94 (

90%

CrI

142

5-41

15)

Day

s (H

ospi

talis

atio

ns *

ALO

S)

+ (

GP

vis

its *

0.

5)

Lost

pai

d w

ork/

activ

ities

fr

om o

ngoi

ng il

lnes

s

42

03 (

90%

CrI

190

2-81

08)

Day

s O

ngoi

ng il

lnes

s di

strib

utio

n m

ultip

lied

by

lost

pro

duct

ivity

(ab

ove

) R

eA f

ollo

win

g S

alm

on

ella

In

cide

nce

For

d et

al

2014

(7)

881

(90%

CrI

209

-231

0)

Cas

es

Att

ribut

able

ris

k of

ReA

app

lied

to th

e es

timat

ed S

alm

onel

la in

cide

nce

for

the

age

grou

p.

GP

vis

its

Tow

nes

et a

l 20

08 (

14);

A

ltPer

t 0.6

6, 0

.80,

0.8

9 69

9 (9

0% C

I 165

-184

3)

GP

vis

its

GP

vis

it di

strib

utio

n m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n

155

Abe

lson

et

al

2006

(1)

E

D v

isits

A

bels

on e

t al

2006

(1)

0,

0, 0

0

ED

vis

its

Hos

pita

lisat

ions

A

IHW

(2,

3)

Ris

kOut

pu 0

.001

, 0.0

03,

0.01

3 2.

54 (

90%

CI 1

.87-

3.88

) H

ospi

talis

atio

nsS

tatis

tical

mod

el in

corp

orat

ing

AIH

W

prim

ary

diag

nose

s fo

r 20

11/1

2 (8

),

2012

/13

(7),

201

3/14

(11

) w

ith m

ultip

liers

fo

r un

derr

epor

ting,

dom

estic

acq

uisi

tion,

an

d pr

opor

tion

due

to e

nter

ic in

fect

ion.

T

hen

25.3

% o

f R

eA h

ospi

talis

atio

ns

attr

ibut

ed s

peci

fical

ly t

o S

alm

onel

la.

Day

s in

hos

pita

l A

IHW

pr

imar

y di

agno

sis

2013

-14

– A

LOS

(3)

4.

2 D

ays

AIH

W p

rovi

des

<1

year

, 1-4

yea

rs, a

nd 5

ye

ar a

ge g

roup

s af

ter

that

up

to 8

5+. A

w

eigh

ted

aver

age

ALO

S is

cal

cula

ted

for

our

age

grou

ps.

Med

icat

ions

or

trea

tmen

ts fo

r th

ose

who

vi

sit a

GP

Med

icat

ion/

trea

tmen

t di

strib

utio

ns

mul

tiplie

d by

GP

vis

it ou

tput

dis

trib

utio

n

A

ntib

iotic

s A

bels

on e

t al

2006

(1)

A

ltPer

t 0.1

6, 0

.2, 0

.244

13

8 (9

0% C

rI 3

2-37

5)

Med

icat

ions

N

SA

ID

Uot

ila e

t al

20

13 (

15)

A

ltPer

t 0.5

28, 0

.762

, 0.

918

520

(90%

CrI

119

-140

3)

Med

icat

ions

E

ye d

rops

A

bels

on e

t al

2006

(1)

A

ltPer

t 0.1

6, 0

.2, 0

.244

13

8 (9

0% C

rI 3

2-37

5)

Med

icat

ions

P

redn

ison

e A

bels

on e

t al

20

06 (

1)

Per

t 0.0

01, 0

.019

, 0.0

99

17 (

90%

CrI

3-6

8)

Med

icat

ions

In

ter-

artic

ular

gl

ucoc

ortic

oid

Exp

ert

opin

ion

A

ltPer

t 0.1

6, 0

.2, 0

.244

13

8 (9

0% C

rI 3

2-37

5)

Med

icat

ions

D

MA

RD

U

otila

et a

l 20

13 (

15)

AltP

ert 0

.012

, 0.0

95,

0.30

4 59

(90

% C

rI 7

-281

) M

edic

atio

ns

Jo

int a

spira

tion

Exp

ert

opin

ion

AltP

ert 0

.16,

0.2

, 0.2

44

138

(90%

CrI

32-

375)

T

reat

men

t

156

Pat

holo

gy a

nd im

agin

g fo

r th

ose

who

vis

it a

GP

T

ests

dis

trib

utio

ns m

ultip

lied

by G

P v

isit

outp

ut d

istr

ibut

ion

S

tool

cul

ture

E

xper

t op

inio

n A

ltPer

t 0.0

97, 0

.132

, 0.

174

92 (

90%

CrI

21-

254)

S

tool

cul

ture

s

S

erol

ogy

Exp

ert

opin

ion

AltP

ert 0

.097

, 0.1

32,

0.17

4 92

(90

% C

rI 2

1-25

4)

Ser

olog

y te

sts

U

rine

m

icro

/cul

ture

/PC

R

Exp

ert

opin

ion

AltP

ert 0

.097

, 0.1

32,

0.17

4 92

(90

% C

rI 2

1-25

4)

Urin

e cu

lture

s

C

RP

& U

rate

A

bels

on e

t al

2006

(1)

; E

xper

t op

inio

n

AltP

ert 0

.16,

0.2

, 0.2

44

138

(90%

CrI

32-

375)

T

ests

F

BC

, ES

R

Abe

lson

et a

l 20

06 (

1);

Exp

ert

opin

ion

AltP

ert 0

.16,

0.2

, 0.2

44

138

(90%

CrI

32-

375)

T

ests

R

heum

atoi

d

fact

or

Abe

lson

et

al

2006

(1)

; E

xper

t op

inio

n

AltP

ert 0

.16,

0.2

, 0.2

44

138

(90%

CrI

32-

375)

T

ests

R

enal

func

tion

Exp

ert

opin

ion

AltP

ert 0

.16,

0.2

, 0.2

44

138

(90%

CrI

32-

375)

T

ests

B

lood

HLA

-B27

A

bels

on e

t al

2006

(1)

A

ltPer

t 0.1

6, 0

.2, 0

.244

13

8 (9

0% C

rI 3

2-37

5)

Tes

ts

X

-ra

y E

xper

t op

inio

n A

ltPer

t 0.0

12, 0

.095

, 0.

304

59 (

90%

CrI

7-2

81)

X-r

ays

U

ltras

ound

E

xper

t op

inio

n A

ltPer

t 0.0

17, 0

.034

, 0.

062

23 (

90%

CrI

5-7

2)

Ultr

asou

nds

M

RI

Exp

ert

opin

ion

AltP

ert 0

.002

, 0.0

1, 0

.03

6 (9

0% C

rI 1

-29)

M

RIs

Spe

cial

ist v

isits

A

bels

on e

t al

2006

(1)

157

assu

med

da

ta a

nd

base

d on

H

annu

et a

l 20

02 (

8)

Ref

erra

l to

rheu

mat

olog

ist

20%

who

vi

sit a

GP

are

re

ferr

ed

AltP

ert 0

.16,

0.2

, 0.2

44

138

(90%

CrI

32-

375)

R

efer

rals

R

efer

ral d

istr

ibut

ion

mul

tiplie

d by

GP

vis

it ou

tput

dis

trib

utio

n

Spe

cial

ist v

isits

/yea

r 20

% o

f re

ferr

ed h

ave

2 vi

sits

per

ye

ar

AltP

ert 0

.223

, 0.2

4,

0.25

8 16

7 (9

0% C

rI 3

9-44

5)

Spe

cial

ists

vi

sits

S

peci

alis

t dis

trib

utio

n m

ultip

lied

by G

P

visi

t ou

tput

dis

trib

utio

n

New

ong

oing

illn

ess

Leiri

salo

-R

epo

et a

l 19

97 (

12);

H

annu

et a

l 20

05 (

9)

AltP

ert 0

.23,

0.5

, 0.7

7 32

9 (9

0% C

rI 7

4-99

9)

Cas

es

New

ong

oing

illn

ess

dist

ribut

ion

mul

tiplie

d by

GP

vis

it ou

tput

+ h

ospi

talis

atio

n ou

tput

di

strib

utio

ns

Day

s of

lost

pai

d ac

tiviti

es

Tow

nes

et a

l 20

08 (

14)

AltP

ert 1

8.49

, 20.

15,

21.9

1 17

757

(90%

CrI

418

9-46

649)

D

ays

“ReA

sym

ptom

s in

terf

ered

with

usu

al

activ

ities

”. P

rodu

ctiv

ity d

istr

ibut

ion

mul

tiplie

d by

inci

denc

e ou

tput

dis

trib

utio

n.

Day

s of

lost

pai

d w

ork

Tow

nes

et a

l 20

08 (

14)

AltP

ert 2

.59,

3.3

7, 4

.3

2939

(90

% C

rI 6

96-

7959

) D

ays

“Mis

sed

wor

k be

caus

e of

ReA

sym

ptom

s”.

Pro

duct

ivity

dis

trib

utio

n m

ultip

lied

by

inci

denc

e ou

tput

dis

trib

utio

n.

Lost

pai

d w

ork

from

on

goin

g ill

ness

10

98 (

95%

CI 2

47-3

441)

D

ays

Ong

oing

illn

ess

dist

ribut

ion

mul

tiplie

d by

lo

st p

aid

wor

k di

strib

utio

n

158

References

1. Abelson, P., M. Forbes, and G. Hall. 2006. The annual cost of foodborne illness in Australia. In Australian Government Department of Health and Ageing, Canberra. 2. Australian Institute of Health and Welfare. Date, 2015, Separation statistics by principal diagnosis (ICD-10-AM 7th edition), Australia, 2011-12 to 2012-13. Available at: https://reporting.aihw.gov.au/Reports/openRVUrl.do. Accessed June, 2015. 3. Australian Institute of Health and Welfare. Date, 2017, Separation statistics by principal diagnosis (ICD-10-AM 7th edition), Australia, 2013-14 to 2014-15. Available at: http://www.aihw.gov.au/hospitals-data/principal-diagnosis-data-cubes/. Accessed, 2015. 4. Canavan, C., J. West, and T. Card. 2014. Review article: the economic impact of the irritable bowel syndrome. Aliment Pharmacol Ther. 40:1023-34. 5. Department of Health. Date, 2017, National Notifiable Disease Surveillance System. Notifications of a selected disease by age group, sex and year. Available at: http://www9.health.gov.au/cda/source/rpt_5_sel.cfm. Accessed, 2015. 6. Flik, C. E., W. Laan, A. J. Smout, B. L. Weusten, and N. J. de Wit. 2015. Comparison of medical costs generated by IBS patients in primary and secondary care in the Netherlands. BMC Gastroenterol. 15:168. 7. Ford, L., M. Kirk, K. Glass, and G. Hall. 2014. Sequelae of foodborne illness caused by 5 pathogens, Australia, circa 2010. Emerg Infect Dis. 20:1865-71. 8. Hannu, T., L. Mattila, H. Rautelin, P. Pelkonen, P. Lahdenne, A. Siitonen, and M. Leirisalo-Repo. 2002. Campylobacter-triggered reactive arthritis: a population-based study. Rheumatology (Oxford). 41:312-8. 9. Hannu, T., L. Mattila, A. Siitonen, and M. Leirisalo-Repo. 2005. Reactive arthritis attributable to Shigella infection: a clinical and epidemiological nationwide study. Ann Rheum Dis. 64:594-8. 10. Kirk, M., L. Ford, K. Glass, and G. Hall. 2014. Foodborne illness, Australia, circa 2000 and circa 2010. Emerg Infect Dis. 20:1857-64. 11. Kirk, M., C. McKercher, and G. Hall. 2011. Gastroenteritis in Australia: report of the National Gastroenteritis Survey II 2008. In OzFoodNet and the National Centre for Epidemiology and Population Health, Canberra. 12. Leirisalo-Repo, M., P. Helenius, T. Hannu, A. Lehtinen, J. Kreula, M. Taavitsainen, and S. Koskimies. 1997. Long-term prognosis of reactive salmonella arthritis. Ann Rheum Dis. 56:516-20. 13. Marshall, J. K., M. Thabane, M. R. Borgaonkar, and C. James. 2007. Postinfectious irritable bowel syndrome after a food-borne outbreak of acute gastroenteritis attributed to a viral pathogen. Clin Gastroenterol Hepatol. 5:457-60. 14. Townes, J. M., A. A. Deodhar, E. S. Laine, K. Smith, H. E. Krug, A. Barkhuizen, M. E. Thompson, P. R. Cieslak, and J. Sobel. 2008. Reactive arthritis following culture-confirmed infections with bacterial enteric pathogens in Minnesota and Oregon: a population-based study. Ann Rheum Dis. 67:1689-96. 15. Uotila, T. M., J. A. Antonen, A. S. Paakkala, J. T. Mustonen, and M. M. Korpela. 2013. Outcome of reactive arthritis after an extensive Finnish waterborne gastroenteritis outbreak: a 1-year prospective follow-up study. Clin Rheumatol. 32:1139-45.

159

Supplementary material 2: Sources of data for costing health care services

General practitioner visits

We estimated the probability that an incident case of Salmonella would consult a general

practitioner (GP) using weighted data from the National Gastroenteritis Survey II (NGSII) (7)

of 0.367 (95% Credible Intervals 0.246-0.501).

We adopted the assumptions of the earlier Abelson study for Australia (1) for irritable bowel

syndrome (IBS) and reactive arthritis (ReA) following Salmonella infection. For IBS we

assumed 4.5 visits (95% CrI 4.27-4.73) per patient, which was based on data from Bettering

the Evaluation of Care and Health (BEACH) program (11). For ReA, we assumed 20% visited

a GP for 4 visits, based on Hannu et al (5), resulting in a probability of 0.8 (95% CrI 0.66-0.89).

We used 95% credible intervals about these point estimates to reflect uncertainty in the

assumptions.

We assumed a normal consult visit for Salmonella and that 25% of GP visits for IBS and ReA

would be long consults and 75% would be normal consults. The costs for a normal consult of

$37.29 and a long consults of $72.06, calculated by dividing total benefits by services from

Medicare item reports(4) for Medicare Benefits Schedule (MBS) (3) item number 23 and item

number 36 respectively for the calendar year of 2015, were considered unrepresentative of the

true cost for society. We therefore adopted an average cost of $80 for a normal consult and

$110 for a long consult.

Emergency department visits

As with GP visits, we used weighted data from the NGSII to estimate the probability that an

incident case of foodborne Salmonella visited the Emergency department (ED), but was not

admitted to hospital, a probability of 0.124 (95% CrI 0.06-0.228) for ED visits.

For ED presentation costs we used an estimate from 2009/10 in New South Wales of $396

multiplied by urgency and disposition group triage weights (8). We used ED Only Triage 1-5

cost weights, where people in triage category 1 are categorised as having an immediate life-

threatening condition and people in triage category 5 are categorised as having a less urgent

condition. We weighted 5 the heaviest and 1 the lightest, to calculate a weighted average of

0.87. This equated to a total ED presentation cost of $345.31.

Hospital costs

Hospital cost data were extracted as Australian Refined Diagnosis Related Groups (AR-DRG)

costs and ALOS as published by the Independent Hospital Pricing Authority for 2013/14.(6)

For Salmonella, we used AR-DRG G67B for age groups 0-4, 5-19, and 20-64 and AR-DRG

160

67A for the age group 65+. We believe this is a conservative cost estimate as the ALOS we

extracted from AIHW was higher than the ALOS for these DRG codes. For IBS, we used AR-

DRG G67B costs for age groups 5-19, 20-64, and 65+, and AR-DRG G67A for age group 0-4

due to the longer ALOS in this age group. For ReA, we used AR-DGR I66B for all age groups.

See Table S2.1 for ALOS and Costs.

Table S2.1: AR-DRG codes used, average length of stay (ALOS), and costs, 2013-14

AR-DRG

Code

Description ALOS Cost

G67A Oesophageal, gastroenteritis, with complications 4.46 $7,054

G67B Oesophageal, gastroenteritis, without

complications

1.55 $2,328

I66B Inflammatory Musculoskeletal Disorders without

catastrophic or severe complications

4.04 $7,138

Treatments

To estimate treatments or medications taken for foodborne Salmonella infection, we used data

from the NGSII weighted for severity of diarrhoea by age group as with GP and ED visits (Table

2). We used expert clinical opinion to estimate that 40% (CrI 38.5%-41.6%) of IBS cases

would take medication (Table 3). For ReA, we used prior Australian assumptions (1), the

literature (12), and expert clinical opinion to estimate the proportion of cases that would receive

each of the medications or treatments in Table 4.

Costs for medications and treatments extracted from either the Pharmaceutical Benefits

Schedule (PBS) or from Chemist Warehouse (https://www.chemistwarehouse.com.au) for

Salmonella, IBS, and ReA can be found in Tables S2.2-S2.4.

Table S2.2: Salmonella medications and treatments, relevant PBS item numbers and brands,

and average estimated costs, 2015

Treatment PBS item number and any

brand costs taken into

account if applicable

Chemist

Warehouse

costs

used?

Average

estimated

cost

Antibiotics $11.36

Amoxycillin 1884E, 1886G 1887H, 1888J,

1889K, Amoxil brand

No $7.80

161

Trimethaprim-

sulphamethoxazole

2951H, 3103H, Septin Forte

brand

No $9.86

Metronidazole 1621H, 1630T, 1636D, 1642K,

Flagyl brand

No $11.04

Ciprofloxacin 1208N, 1209P, Ciproxin 250

brand

No $17.85

Ceftriaxone 1783W, 1784X, 1785Y, 1788D No $11.63

Azithromycin Yes $9.99

Diarrhoea relief $11.60

Electrolyte Yes $10.99

Loperamide 10889D, 10592L, 1571Q,

Immodium brand

Yes $14.34

Diphenoxylate-atropine 2501P No $9.46

Pain relief $10.02

Paracetamol & codeine 1215Y, 3316M, 10186D, 4170L,

4171M, 4275B, Panadeine Forte

brand

No $10.61

Paracetamol (Dymadon) Yes $9.19

Paracetamol 10582Y, 1770E, 3348F, 3349G,

5196L

Yes $7.12

Ibuprofen (Nurofen) Yes $8.99

Tramadol hydrochloride 8523N, Tramal SR 100 brand No $14.19

Nausea relief $15.46

Metoclopramide 1206L, Maxalon brand No $12.27

Prochlorperazine 2369Q, 2893G, Stemetil brand No $10.44

Promethazine 1948M, 3374N, 3488N No $30.16

Antacid Yes $8.99

Cramp relief $17.23

Hyoscine butylobromide 3473T Yes $17.23

Table S2.3: IBS medications and treatments, relevant PBS item numbers and brands, and

average estimated costs, 2015



account if applicable

Chemist

Warehouse

Average

estimated

cost

162

costs

used?

Average IBS treatment cost $72.65

Psyllium fibre Yes $19.99

Loperamide 10889D, 10592L, 1571Q,

Immodium brand

Yes $14.34

Prochlorperazine 2369Q, 2893G, Stemetil brand No $10.44

Hyoscine butylobromide 3473T Yes $17.23

Mebeverine hydrochloride 4328T, Colofac brand Yes $27.44

Rifaximin 10001J No $481.08

Colestyramine 2967E, 9249T No $55.27

Antidepressants

Sertraline 2236Q, 8836C, 2237R, 8837D No $7.05

Citalopram 8702B, 8703C, 8220P No $6.48

Escitalopram 8701Y, 9433L, 10181W, 8700X,

9432K

No $14.41

Paroxetine 2242B No $11.29

Fluoxetine 1434L, 8270G No $15.15

Fluvoxamine 8512B, 8174F No $10.87

Table S2.4: Reactive arthritis medications and treatments, relevant PBS item numbers and

brands, and average estimated costs, 2015



account

Chemist

Warehouse

costs

used?

Average

estimated

cost

Antibiotics As described in Table 1 $11.36

NSAID (Naproxen) Yes $10.50

Eye drops Yes $8.99

Prednisone 1934T, 1935W, 1936X,

Panafcorte brand

No $7.38

Inter-articular glucocorticoid

(Beetamethosone)

2694T No $20.75

DMARD (Methotrexate)*

11288D, 11283W, 11275K,

1622J, 11268C, 2396D, 4502Y,

No $87.65

163

4512L, 7250N, 7251P, 11295L,

2395C, 2272N

DMARD (Infliximab)* 10057H, 10067W, 10184B,

10196P, 4284L, 5753T, 5754W,

5755X, 5756Y, 5757B, 5758C,

6397Q, 6448J, 6496X, 9612X,

9613Y, 9617E, 9654D, 9674E

Yes $3,228.79

Joint Aspiration No $76.00

*Assumed 80% of those on DMARDs would be prescribed Methotrexate and 20% would be

prescribed Infliximab

Pathology and Imaging

For Salmonella, we extracted a cost of $45.19 for stool microscopy, culture, and sensitivity

(MCS) and $36.58 for polymerase chain reaction (PCR) by dividing total benefits by services

from Medicare item reports (4) for MBS (3) item number 69345 and item number 69496

respectively for the calendar year of 2015. We assumed 45% of tests would be MCS only, 9%

would be PCR only, and 46% would be both MCS and PCR (9).

Assumptions for the proportion of IBS cases undergoing pathology testing and imaging were

based on expert clinical opinion. We assumed that everyone who had an endoscopy would

also have anaesthesia and a biopsy. We extracted costs for these tests by dividing total

benefits by services for each item number from Medicare item reports (4) for the calendar year

of 2015. Item numbers and costs are detailed in Table S2.5.

Table S2.5: IBS Pathology and imaging, Medicare Benefits Schedule item numbers, and

estimated cost, 2015

Pathology test or Imaging MBS item

number(3)

Estimated cost

Stool MCS 69345 $45.19

FBC 65070 $14.13

ESR 65060 $6.56

LFT 66512 $14.75

CRP 66500 $8.17

Coeliac disease screening 71163 $21.12

Abdominal X-ray 58900 $31.60

Abdominal Ultrasound 55036 $103.39

Endoscopy 30473 $96.47

164

Anaesthesia for endoscopy 20902 $111.85

Biopsy 72818 $89.82

We used the prior Australian estimates (1) and expert clinical opinion to estimate the proportion

of ReA cases receiving pathology testing and imaging. We extracted costs for these tests by

dividing total benefits by services for each item number from Medicare item reports (4) for the

calendar year of 2015. Item numbers and costs are detailed in Table S2.6.

Table S2.6: ReA Pathology and Imaging, Medicare Benefits Schedule item numbers, and

estimated cost, 2015

Pathology test or Imaging MBS item

number(3)

Estimated cost

Stool MCS 69345 $45.19

IgG, IgM & IgA 71071 $26.17

Urinalysis 69333 $17.39

CRP & Urate 66503 $9.86

FBC 65070 $14.13

ESR 65060 $6.56

Rheumatoid factor 71106 $9.60

Renal Function with imaging

& 2 blood samples

12527 $71.92

Blood HLA-B27 71147 $34.61

Lumbosacral X-ray 58106 $75.98

Lower limb ultrasound 55834 $78.04

MRI 63328 $380.42

We have not incorporated episode coning, an arrangement which places an upper limit on the

number of services an episode for which Medicare benefits are payable, into our cost

estimates. However, we took a conservative approach in estimating the number of tests

ordered, so believe this will balance out any underestimation by not taking episode coning into

account.

Specialist Visits

For IBS, we used the literature (2) and expert clinical opinion to estimate that 30% (95% CrI

21%-49%) of cases would visit a specialist. For ReA, we used past assumptions and the

literature to estimate that 20% of cases would be referred to a specialist and 20% of these

165

would have 2 visits per year, resulting in a distribution of 0.24 (95% CrI 0.223-0.258). For cases

with ongoing illness, we assumed all cases would visit a specialist, of which 20% would be

initial visits and 80% would be repeat visits.

The costs for an initial specialist visit of $128.16 and repeat visit of $63.19, calculated by

dividing total benefits by services from Medicare item reports (4) for Medicare Benefits

Schedule (MBS) (3) item number 110 and item number 116 respectively for the calendar year

of 2015, were considered unrepresentative of the true cost for society. We therefore adopted

an average cost of $315 for an initial visit and $166 for a repeat visit based on the Australian

Medical Association’s recommended fees (10).

References

1. Abelson, P., M. P. Forbes, and G. Hall. 2006. The annual cost of foodborne illness in

Australia. Australian Government Department of Health and Ageing, Canberra.

2. Canavan, C., J. West, and T. Card. 2014. Review article: the economic impact of the

irritable bowel syndrome. Aliment Pharmacol Ther. 40:1023-1034.

3. Department of Health. MBS Online. Available at:

http://www.mbsonline.gov.au/internet/mbsonline/publishing.nsf/Content/Home. . Accessed 7

June 2018.

4. Department of Human Services. Medicare Item Reports. Available at:

http://medicarestatistics.humanservices.gov.au/statistics/mbs_item.jsp. Accessed 7 June

2018.

5. Hannu, T., L. Mattila, H. Rautelin, P. Pelkonen, P. Lahdenne, A. Siitonen, and M.

Leirisalo-Repo. 2002. Campylobacter-triggered reactive arthritis: a population-based study.

Rheumatology (Oxford). 41:312-318.

6. Independent Hospital Pricing Authority. 2016. National Hospital Cost Data Collection,

Australian Public Hospitals Cost Report, Round 18 (Financial year 2013-14).

7. Kirk, M., C. McKercher, and G. Hall. 2011. Gastroenteritis in Australia: report of the

National Gastroenteritis Survey II 2008. OzFoodNet and the National Centre for Epidemiology

and Population Health, Canberra.

8. Reeve, R., and M. Haas. 2013. Estimating the Cost of Emergency Department

Presentations in NSW. In, vol. Working Paper 2014/01. Centre for Health Economics Research

and Evaluation. University of Technology, Sydney, Ultimo NSW.

9. Roper, K., S. Vilkins, K. Glass, and M. Kirk. 2017. Implications of culture independent

diagonstic testing (CIDT) on Salmonella surveillance in Australia. Australian National

University, Canberra.

166

10. Sivey, P. 2016, How much?! Seeing private specialists often costs more than you

bargained for. Available at: https://theconversation.com/how-much-seeing-private-specialists-

often-costs-more-than-you-bargained-for-53445. Accessed 3 July 2018.

11. The University of Sydney. 2017, Better the Evaluation and Care of Health (BEACH).

Available at: http://sydney.edu.au/medicine/fmrc/beach/. Accessed 7 Jun, 2018.

12. Uotila, T. M., J. A. Antonen, A. S. Paakkala, J. T. Mustonen, and M. M. Korpela. 2013.

Outcome of reactive arthritis after an extensive Finnish waterborne gastroenteritis outbreak: a

1-year prospective follow-up study. Clin Rheumatol. 32:1139-1145.

167

Supplementary materials 3: Estimated mean cost of illness, Salmonella, circa 2015,

Australia

TABLE S3.1: Estimated mean annual cost of illness (millions AUD) of health care usage, lost productivity, premature mortality, for acute and ongoing illness, circa 2015, Australia

Mean cost in millions AUD (90% CrI)

NTSa IBSa ReAa Totalb

Health care usage

Acute illness 23.8 (19.3-28.9) 5.7 (4.0-8.0) 2.0 (1.0-3.6) 31.6 (26.5-37.2)

Ongoing illness - 3.5 (1.8-6.9) 0.86 (0.34-1.7) 4.3 (2.7-6.5)

Lost productivity

Acute illness 22.0 (13.7-33.6) 2.1 (1.3-3.3) 3.1 (1.0-6.5) 27.2 (18.5-38.4)

Ongoing illness - 3.8 (1.8-6.9) 1.2 (0.37-2.8) 5.1 (2.7-8.4)

Premature mortality

Acute illness 79.0 (66.0-92.1) - - 79.0 (66.0-92.1)

Total costs

Acute illness 124.7 (107.4-143.1) 7.9 (5.3-11.3) 5.1 (2.0-10.0) 137.7 (119.7-157.0)

Ongoing illness - 7.3 (3.9-12.3) 2.1 (0.73-4.4) 9.4 (5.5-14.8)

Total 124.7 (107.4-143.1) 15.1 (9.5-23.0) 7.2 (2.4-14.2) 147.2 (127.8-167.9)

aNTS: non-typhoidal Salmonella; IBS: irritable bowel syndrome; ReA: reactive arthritis

bNumbers may not sum due to simulation and rounding

168

TABLE S3.2: Estimated mean annual cost of acute and ongoing illness in 1 year (millions AUD) by age group, circa 2015, Australia

Mean cost in millions AUD (90% CrI)

NTSa IBSa ReAa Totalb

Health care usage

0-4 years 4.3 (3.0-5.9) 0.004 (0.002-0.006) 0.16 (0.04-0.37) 4.5 (3.2-6.1)

5-19 years 2.9 (2.0-4.0) 1.7 (0.90-3.0) 0.56 (0.13-1.3) 5.2 (3.7-6.9)

20-64 years 8.8 (5.9-12.4) 6.0 (3.3-9.8) 1.8 (0.46-4.0) 16.6 (11.9-22.0)

65+ years 7.8 (5.1-11.0) 1.5 (0.80-2.4) 0.41 (0.10-0.93) 9.6 (6.8-13.0)

Lost Productivity

0-4 years 5.3 (2.3-10.0) 0.003 (0.001-0.004) 0.38 (0.07-0.78) 5.7 (2.6-10.3)

5-19 years 1.9 (0.46-4.4) 0.37 (0.17-0.65) 0.26 (0.05-0.61) 2.5 (1.0-5.0)

20-64 years 13.4 (6.7-22.8) 5.4 (2.7-9.4) 3.6 (0.73-8.4) 22.5 (13.9-33.6)

65+ years 1.4 (0.33-3.0) 0.20 (0.10-0.35) 0.13 (0.03-0.31) 1.7 (0.64-3.3)

Premature mortality

0-4 years 2.8 (2.1-3.5) - - 2.8 (2.1-3.5)

5-19 years 7.3 (5.7-8.8) - - 7.3 (5.7-8.8)

20-64 years 15.1 (11.6-18.7) - - 15.1 (11.6-18.7)

65+ years 53.9 (41.6-66.3) - - 53.9 (41.6-66.3)

Total costs

0-4 years 12.4 (8.5-18.1) 0.006 (0.004-0.009) 0.50 (0.11-1.1) 12.9 (8.9-18.6)

5-19 years 12.0 (9.4-15.3) 2.1 (1.1-3.6) 0.82 (0.09-1.9) 14.9 (11.9-18.6)

20-64 years 37.3 (27.7-49.8) 11.4 (6.0-19.1) 5.3 (1.2-12.2) 54.1 (41.2-70.0)

65+ years 63.0 (50.3-75.9) 1.6 (0.9-2.7) 0.54 (0.12-1.2) 65.1 (52.5-78.2)


bNumbers may not sum due to simulation and rounding

169

TABLE S3.3: Estimated mean annual cost of acute and ongoing illness ($AUD) per case by age group, circa 2015, Australia

Mean cost AUD

NTSa IBSa ReAa Total

Health care usage only

0-4 years 209.17 905.15 314.42 217.70

5-19 years 199.67 439.31 297.20 366.26

20-64 years 200.67 500.86 312.14 386.87

65+ years 754.01 500.21 293.93 936.82

Total 254.69 480.42 297.22 385.73

Health care usage with lost productivity

0-4 years 451.10 1,457.92 925.03 476.91

5-19 years 320.19 530.18 432.12 532.25

20-64 years 486.25 942.31 922.75 889.34

65+ years 872.56 567.33 387.99 1,088.16

Total 482.66 784.35 722.37 725.81

Health care usage, lost productivity, and premature mortality

0-4 years 592.44 1,457.92 925.03 618.25

5-19 years 846.17 530.18 432.12 1,058.22

20-64 years 848.07 942.31 922.75 1,251.17

65+ years 6,175.07 567.33 387.99 6,390.67

Total* 1,339.11 784.35 722.37 1,582.21


170

The following supplementary information will be part of the manuscript submission as a

supplement to the paper.

Ford L, Glass K, Williamson DA, Sintchenko V, Robson JMB, Stafford R, Kirk D. Cost of whole

genome sequencing for non-typhoidal Salmonella enterica. To be submitted 2019.

National notification numbers

In 2017, 5687/16051 (35%) of non-typhoidal Salmonella notifications were serotyped as

Typhimurium (Department of Health 2018) (Table S1).

Table S1: National number of Salmonella spp. notifications by type, Australia, 2017

Number of notifications

Salmonella Typhimurium 5687

Non-Typhimurium Salmonella 8718

Unspecified 1646

Total 16051

Outbreak scenarios

We obtained dates of illness onset and case numbers for several outbreaks occurring over the

last 20 years in Australia from Queensland Health. We then generated epidemiological curves

and cumulative outbreak costs based on current laboratory subtyping methods (serotyping and

MLVA) compared with whole genome sequencing (WGS) and polymerase chain reaction

(PCR) testing for a point source outbreak and two prolonged outbreaks.

Point source outbreak: Salmonella Typhimurium outbreak in January 2015

In this outbreak, there were 123 cases of Salmonella Typhimurium with illness onsets over 6

days (Figure S1). As this was an outbreak of Salmonella Typhimurium, serotyping and MLVA

would have been used to type the isolates. As in our simulated point source outbreak,

cumulative outbreak costs were slightly less if WGS had been used, costing USD 152,123

(90% CrI 92,528-257,067) for serotyping and MLVA, compared with USD 150,200 (90% CrI

90,894-255,326) for WGS or USD 139,163 (90% 79,922-244,329) for PCR-only (Figure S2).

171

Figure S1: Epidemiological curve of Salmonella Typhimurium outbreak in January 2015,

Queensland Australia

Figure S2: Cumulative outbreak costs using serotyping and MLVA compared to WGS in a

Salmonella Typhimurium outbreak in January 2015, Queensland Australia

Prolonged outbreak: Salmonella Bovismorbificans outbreak in May and June 2001

In this outbreak, there were 29 cases of Salmonella Bovismorbificans with illness onsets over

44 days (Figure S3). As with the prolonged outbreak scenario, cumulative outbreak costs with

WGS (USD 35,465 (90% CrI 21,504-60,616)) were higher than with serotyping (USD 34,237

(90% CrI 20,308-59,395)) or PCR (USD 32,863 (90% CrI 18,927-57,974)) (Figure S4). If WGS

data were able to detect the outbreak earlier, and an intervention was put in place earlier, WGS

would result in a cost savings. However, the difference in costs is not as large at our first

modelled earlier intervention point (2 weeks) as was seen in our first modelled earlier

0

10

20

30

40

50

60

03‐Jan‐15 04‐Jan‐15 05‐Jan‐15 06‐Jan‐15 07‐Jan‐15 08‐Jan‐15

Number of cases

Date of illness onset

$0

$20,000

$40,000

$60,000

$80,000

$100,000

$120,000

$140,000

$160,000

03‐Jan‐15 04‐Jan‐15 05‐Jan‐15 06‐Jan‐15 07‐Jan‐15 08‐Jan‐15

Cumulative outbreak cost USD


Serotyping and MLVA WGS PCR

172

intervention points in our simulated prolonged outbreak. This is due to the length of the

outbreak and the shape of the epidemiological curve.

Figure S3: Epidemiological curve of Salmonella Bovismorbificans outbreak in May and June

2001, Queensland Australia

Figure S4: Cumulative outbreak costs using serotyping compared to WGS, with earlier

modelled interventions, in a Salmonella Bovismorbificans outbreak in May and June 2001,

Queensland Australia

0

1

2

3

4

5

16‐M

ay‐01

18‐M

ay‐01

20‐M

ay‐01

22‐M

ay‐01

24‐M

ay‐01

26‐M

ay‐01

28‐M

ay‐01

30‐M

ay‐01

01‐Jun‐01

03‐Jun‐01

05‐Jun‐01

07‐Jun‐01

09‐Jun‐01

11‐Jun‐01

13‐Jun‐01

15‐Jun‐01

17‐Jun‐01

19‐Jun‐01

21‐Jun‐01

23‐Jun‐01

25‐Jun‐01

27‐Jun‐01

Number of cases


$0

$5,000

$10,000

$15,000

$20,000

$25,000

$30,000

$35,000

$40,000

16‐M

ay‐01

18‐M

ay‐01

20‐M

ay‐01

22‐M

ay‐01

24‐M

ay‐01

26‐M

ay‐01

28‐M

ay‐01

30‐M

ay‐01

01‐Jun‐01

03‐Jun‐01

05‐Jun‐01

07‐Jun‐01

09‐Jun‐01

11‐Jun‐01

13‐Jun‐01

15‐Jun‐01

17‐Jun‐01

19‐Jun‐01

21‐Jun‐01

23‐Jun‐01

25‐Jun‐01

27‐Jun‐01Cumulative outbreak cost USD


Serotyping WGS

WGS, 2 week earlier intervention WGS, 4 week earlier intervention

PCR

173

As this was a Salmonella Bovismorbificans outbreak occurring in 2001, the isolates also were

phage typed. If we added phage typing costs on top of the serotyping costs, the cumulative

outbreak costs for WGS would likely be lower.

Prolonged outbreak: Salmonella Saintpaul outbreak in April 2011

In this outbreak, there were 17 cases of Salmonella Saintpaul cases occurring over 39 days

(Figure S5). As with the previous example and the simulated prolonged outbreak scenario,

cumulative outbreak costs with WGS (USD 20,689 (90% CrI 12,655-35,477) were higher than

with serotyping (USD 20,011 (90% CrI 11,969-34,824)) or PCR (USD 19,159 (90% CrI 11,133-

33,982)) (Figure S6). If an intervention could be put into place and prevent just one case, then

WGS would result in cost savings.

Figure S5: Epidemiological curve of Salmonella Saintpaul outbreak September–November

2005, Queensland Australia

0

1

2

3

4

29‐Sep

‐05

01‐Oct‐05

03‐Oct‐05

05‐Oct‐05

07‐Oct‐05

09‐Oct‐05

11‐Oct‐05

13‐Oct‐05

15‐Oct‐05

17‐Oct‐05

19‐Oct‐05

21‐Oct‐05

23‐Oct‐05

25‐Oct‐05

27‐Oct‐05

29‐Oct‐05

31‐Oct‐05

02‐Nov‐05

04‐Nov‐05

06‐Nov‐05

Number of cases


174

Figure S6: Cumulative outbreak costs using serotyping compared to WGS in a Salmonella

Saintpaul outbreak in 2005, Queensland Australia

References

Department of Health 2018. Salmonella public data set. Retrieved 21 November 2018 from

http://www9.health.gov.au/cda/source/pub_salmo.cfm.

$0

$5,000

$10,000

$15,000

$20,000

$25,000

29‐Sep

‐05

01‐Oct‐05

03‐Oct‐05

05‐Oct‐05

07‐Oct‐05

09‐Oct‐05

11‐Oct‐05

13‐Oct‐05

15‐Oct‐05

17‐Oct‐05

19‐Oct‐05

21‐Oct‐05

23‐Oct‐05

25‐Oct‐05

27‐Oct‐05

29‐Oct‐05

31‐Oct‐05

02‐Nov‐05

04‐Nov‐05

06‐Nov‐05

Cumulative outbreak costs USD


Serotyping WGS PCR

175

Appendix 4. Supplementary materials for chapter 6 The following supplementary information was part of the manuscript submission and published

online as a supplement to the paper.


Valcanis M, Castillo CFS, Sait M, Howden BP, Kirk MD. Incorporating whole-genome

sequencing into public health surveillance: lessons from prospective sequencing of Salmonella

Typhimurium in Australia. Foodborne Pathogens and Disease. 2018;15(3): 161-167, doi:

10.10189/fpd.2017.2352.

Supplementary Data

Isolates linked through multilocus variable-number tandem-repeat analysis (MLVA) and epidemiological data were as-sociated with 11 identified clusters or point-source outbreaks(Supplementary Fig. S1).

Outbreak investigations indicated that 5 of these 11 events,which occurred over a large geographic area between Januaryand March 2016, were related to a common egg grading fa-cility. The details of these outbreak investigations, as well astwo additionally linked investigations, are discussed else-where (Ford et al., 2017). The whole-genome sequencing(WGS) data show that there were an additional nine isolateswith illness onsets between February and May that werewithin eight single-nucleotide polymorphisms (SNPs) ofknown outbreak isolates. While it is possible that the twoisolates with illness onsets around the same time as theAustralian Capital Territory (ACT) outbreaks in Februarywere unknowingly linked to one of the two implicated foodpremises, it is unlikely that the other seven isolates with ill-ness onsets in later months had an unmentioned associationwith the food premises. The genetic relatedness of these casesto outbreak cases suggests that the source of illness may bethe same as the suspected source of illness (eggs) of theoutbreak isolates in this clade.

Fifteen isolates were linked to an outbreak associated witha food premises (Outbreak 6) in May 2016 identified through

case interviews. Cases in this outbreak reported eating at thefood premises over a period of 11 d and those who ate thereon the 11th day had a different MLVA profile (03-10-14-11-496) and were *90 SNPs apart from those who reportedeating at the premises the previous 10 d (MLVA: 03-12-18-14-523). The source of the outbreak is unknown. An addi-tional two isolates were linked by WGS to cases from the first10 d and one isolate was linked by WGS to the cases from the11th day who did not mention eating at the food premisesduring their interview. Illness onsets for these three caseswere within 2 weeks of the outbreak.

Six isolates with no SNP differences (Outbreak 7) werelinked to an outbreak in a private residence in February 2016identified through case interviews. The source of the outbreakwas unknown.

There were two sets of siblings included, which—based onillness onset dates—were likely transmitted from person toperson (Clusters 1 and 2).

There were two isolates from residents of the ACT, whohad travelled interstate during their incubation and werecases in a known outbreak in that state (Outbreak 8).

Finally, there were five isolates from four cases who werepossibly associated with a common food premises, but therewas insufficient epidemiological and environmental evidenceto implicate the food premises (Outbreak 9).

176

SUPPLEMENTARY FIG. S1. Maximum likelihood core genome SNP phylogeny of Salmonella Typhimurium isolates withepidemiological data, ACT, January to June 2016. Clusters identified by WGS are highlighted in the tree. Clusters and outbreaksidentified by epidemiological data are labeled in the tree. Figure created with Interactive Tree of Life (https://itol.embl.de).

177

178

The following supplementary information was part of the manuscript submission and published




Williamson DA, Sintchenko V, Howden BP, Kirk MD. Seven Salmonella Typhimurium

outbreaks in Australia linked by trace-back and whole-genome sequencing. Foodborne

Pathogens and Disease. 2018;15(5):285-292, doi: 10.1089/fpd.2017.2353.

Supplementary Information

Outbreak summaries

Australian Capital Territory (ACT)

In the ACT there were two small outbreaks of S. Typhimurium 03-26-13-08-523 in

February 2016. In the first outbreak there were 4 confirmed cases who were linked to

restaurant A and reported eating an eggs benedict dish. A public health officer inspected

restaurant A, with no food safety compliance issues identified. Three samples of egg,

spinach, and commercially produced hollandaise sauce were collected for testing. No

Salmonella was detected in any of the samples. In the second outbreak, there were 5

confirmed cases who were linked to restaurant B. An egg and lettuce sandwich was the most

frequently reported food among cases. A public health officer inspected restaurant B, with no

food safety compliance issues identified. Three food samples (boiled egg, egg and lettuce

sandwich, and shredded lettuce) were collected for testing. No Salmonella was detected in

any of the samples. Egg supplier details were obtained during both inspections and both

restaurant A and restaurant B used eggs produced by company X.

New South Wales (NSW)

In NSW there was one outbreak of S. Typhimurium 03-26-13-08-523 in October-

November 2015 and two in January 2016. In November 2015 a retrospective cohort study

was conducted into a large number of patrons with gastrointestinal illness related to a

degustation dinner prepared and consumed at Restaurant C. A total of 40 out of 60

individuals who were interviewed or surveyed met the case definition (Supplementary Table

S2). Seven cases from the cohort had clinical samples taken and all were confirmed as

Salmonella Typhimurium 03-26-13-08-523. Coriander mayonnaise, containing raw egg was

179

identified as a probable source of infection through the epidemiological investigation

(relative risk: 3.6; p value: 0.042). The NSW Food Authority conducted a basic inspection of

the premises on 6 November 2015, followed by a comprehensive inspection on 10 November

2015. The Food Authority discovered inadequate food handling practices: use of

unpasteurised eggs in dessert products did not reach adequate temperatures to ensure

pathogen reduction/kill step and separation of eggs was done by hand in the shell. Twelve

food samples (eggs, marinated chevre, double brie cheese, chilli jam, beetroot, micro herbs,

quail breast fillet, crème anglaise, crème patissiere, aioli, 2 x mayonnaise mixtures), and 6

environmental samples (whisk kitchen, kitchen aid x 2, aioli bench and sink, profiterole

bench and sink, boot swabs) were collected by the NSW Food Authority. No Salmonella was

detected in any of the samples. Only 7 of the food specimens, not including the mayonnaise,

were possible leftovers from the dinner. Egg supplier details were obtained and eggs were

produced by company X.

In one outbreak identified in January 2016, there were 203 cases were linked to a

bakery, of which 91 were confirmed with Salmonella and 81 of these were S. Typhimurium

03-26-13-08-523 (Supplementary Table S2). A survey of a sample of 100 suspected cases

identified chicken, pork and salad-filled rolls as the most commonly consumed foods. The

NSW Food Authority inspected the bakery at which time non-compliances with food safety

standards were identified such as lack of appropriate sanitising of utensils or equipment, other

cleanliness issues, and some evidence of rodent droppings. A prohibition order was served

and numerous food and environmental samples were collected for testing. Results of the

bacterial analysis showed the presence of Salmonella Typhimurium 3-26-13-08-523 in 9 food

samples including ready-to-eat salad items that were the basic ingredients of the filled rolls.

Salmonella Typhimurium 3-26-13-08-523 was also detected in 14 environmental samples.

The eggs used by the bakery were produced by company X.

180

A second outbreak was identified in January 2016 involving 2 cases at an aged care

facility. Investigations were unable to identify a potential food or source of infection. The

environmental investigation did not identify any issues, and no food specimens or

environmental samples were collected for testing.

Queensland

Two outbreaks of S. Typhimurium 03-26-13-08-523 infection were reported in

Queensland during February and March 2016 respectively. The first outbreak investigation

identified 12 cases of gastrointestinal illness, including six Queensland residents whom were

laboratory confirmed (Supplementary Table S2), linked to the consumption of French crepes

purchased from one of three market stalls operated by the same vendor. Investigations

identified that an egg batter used to make the crepes was not discarded after each event and

left-over batter was potentially used over a three day period. Food and environmental samples

collected the following week tested negative for bacterial pathogens. Invoice records indicate

the eggs used by the business during the outbreak were from company X. The second

outbreak involved 10 cases of Salmonella Typhimurium 03-26-13-08-524 infection among

persons who had attended an Italian Festival held in a community park over a 3 day period

(Supplementary Table S2). Investigations were unable to identify a potential food vehicle or

source of infection due to poor recall among cases and the time delay for the investigation.

The festival also consisted of market stalls for the sale of foods.

181

Supp

lem

enta

ry T

able

1

Supp

lem

enta

ry T

able

S2.

Cas

e de

finiti

ons u

sed

in o

utbr

eaks

of S. T

yphi

mur

ium

03-

26-1

3-08

-523

in th

e A

ustra

lian

Cap

ital T

errit

ory

(AC

T),

New

Sou

th W

ales

(NSW

), an

d Q

ueen

slan

d (Q

ld) i

nves

tigat

ed b

etw

een

Nov

embe

r 201

5 an

d M

arch

201

6.

Stat

e M

onth

and

ye

ar o

f in

vest

igat

ion

Sett

ing

Cas

e de

finiti

on

No.

of c

ases

N

o. o

f lab

-co

nfir

med

ca

ses

NSW

N

ov 2

015

Res

taur

ant

An

indi

vidu

al w

ho a

te o

r ser

ved

dinn

er a

t the

Res

taur

ant o

n 29

Oct

ober

20

15 a

nd su

bseq

uent

ly d

evel

oped

acu

te g

astro

ente

ritis

. 40

7

NSW

Ja

n 20

16

Bak

ery

Vom

iting

or d

iarr

hoea

follo

win

g co

nsum

ptio

n of

take

aw

ay fo

ods

purc

hase

d fr

om B

aker

y X

from

Wed

nesd

ay 2

2 Ja

nuar

y 20

16.

203

81

NSW

Ja

n 20

16

Age

d ca

re

faci

lity

Pers

ons w

ith d

iarr

hoea

AN

D/O

R v

omiti

ng w

ithin

a re

side

ntia

l age

d ca

re

faci

lity,

occ

urrin

g in

a 2

4 ho

ur p

erio

d, w

hich

can

not b

e at

tribu

ted

to a

kn

own

caus

e - e

.g. a

n ex

istin

g m

edic

al c

ondi

tion

or th

e us

e of

med

icat

ions

su

ch a

s ant

ibio

tics o

r ape

rient

s.

2 2

AC

T Fe

b 20

16

Res

taur

ant

An

indi

vidu

al w

ho a

te a

t the

Res

taur

ant o

n 12

/2/1

6 an

d su

bseq

uent

ly

deve

lope

d ac

ute

gast

roen

terit

is

5 4

Qld

Fe

b 20

16

Mar

ket

A p

erso

n w

ho a

ttend

ed a

ny o

f the

thre

e im

plic

ated

mar

kets

bet

wee

n 19

an

d 21

Feb

ruar

y 20

16 a

nd w

as e

ither

dia

gnos

ed w

ith S. T

yphi

mur

ium

03-

26-1

3-08

-523

infe

ctio

n or

atte

nded

the

mar

kets

with

a c

onfir

med

cas

e an

d de

velo

ped

acut

e ga

stro

ente

ritis

dur

ing

the

follo

win

g w

eek.

12

6

AC

T Fe

b 20

16

Res

taur

ant

A c

ase

of S. T

yphi

mur

ium

03-

26-1

3-08

-523

who

ate

at t

he R

esta

uran

t be

twee

n 24

/2/1

6 an

d 29

/2/1

6 an

d su

bseq

uent

ly d

evel

oped

acu

te

gast

roen

terit

is

5 5

Qld

M

ar 2

016

Fest

ival

A

per

son

who

atte

nded

the

Italia

n Fe

stiv

al b

etw

een

25 a

nd 2

7 M

arch

201

6 an

d w

as d

iagn

osed

with

S. T

yphi

mur

ium

03-

26-1

3-08

-523

infe

ctio

n.

10

10

182

Supp

lem

enta

ry T

able

2

Supp

lem

enta

ry T

able

S1.

Foo

d sa

mpl

es a

nd e

nviro

nmen

tal s

wab

s col

lect

ed in

the S.

Typ

him

uriu

m 0

3-26

-13-

08-5

23 o

utbr

eaks

in th

e A

ustra

lian

Cap

ital T

errit

ory

(AC

T), N

ew S

outh

Wal

es (N

SW),

and

Que

ensl

and

(Qld

) inv

estig

ated

bet

wee

n N

ovem

ber 2

015

and

Mar

ch 2

016.

St

ate

Mon

th a

nd

year

of

inve

stig

atio

n

Sett

ing

Food

sam

ples

col

lect

ed

Env

iron

men

tal s

wab

s co

llect

ed

Sam

ples

with

S. T

yphi

mur

ium

03

-26-

13-0

8-52

3 de

tect

ed

NSW

N

ov 2

015

Res

taur

ant

Mar

inat

ed c

hevr

e*; b

rie*;

be

etro

ot*;

chi

lli ja

m*;

mic

ro

herb

s (4)

*; q

uail

brea

st fi

llet

(2)*

; Who

le e

ggs;

crè

me

angl

aise

; crè

me

patis

sier

e; a

ioli;

m

ayon

nais

e (2

)

Whi

sk; s

tand

mix

ers (

2); a

ioli

benc

h an

d si

nk; p

rofit

erol

e be

nch

and

sink

; boo

t sw

abs

kitc

hen

Non

e

NSW

Ja

n 20

16

Bak

ery

Chi

cken

roll;

coo

ked

pork

mea

t; w

hite

rect

angu

lar c

ut m

eat;

red

rect

angu

lar c

ut m

eat;

chic

ken

shre

dded

(2);

pate

; Cae

sar s

alad

w

ith sl

iced

boi

led

eggs

; co

lesl

aw; t

abou

li; le

ttuce

sh

redd

ed; c

oria

nder

; cuc

umbe

r sl

iced

; tom

ato

slic

ed; c

arro

ts

shre

dded

; oni

ons s

liced

; bee

troot

sl

iced

; boi

led

eggs

(3);

chee

se

slic

ed; m

ayon

nais

e an

d m

arga

rine

mix

(2);

may

onna

ise

sauc

e (2

); fr

ozen

pat

e to

be

re-

cook

ed; c

hick

en st

uffin

g; w

hole

ra

w c

hick

en; w

hole

shel

l egg

s (tr

ay o

f 30)

; fro

zen

cook

ed p

ork;

fr

ozen

slic

ed w

hite

pro

cess

ed

mea

t; fr

ozen

slic

ed re

d

Boo

t sw

abs (

3); w

ash

basi

n;

sink

; coo

l roo

m (2

); be

nch;

fr

eeze

r; di

spla

y ca

bine

ts;

woo

den

chop

ping

boa

rds;

pla

stic

ch

oppi

ng b

oard

s; c

ompo

site

ut

ensi

ls; c

ompo

site

mix

ing

bow

ls; c

ompo

site

mic

row

ave

and

rice

cook

er; c

ompo

site

po

wer

poi

nts a

nd li

ght s

witc

hes;

ch

icke

n co

oker

(2);

chic

ken

skew

ers;

chi

cken

ute

nsils

; raw

ch

icke

n st

orag

e tu

b; ra

w c

hick

en

prep

sink

; chi

cken

seas

onin

g bu

cket

; woo

den

tabl

es; s

tove

; pr

oces

sed

mea

t kni

fe; w

indo

w

sill;

pat

e fr

idge

; gra

vy p

ot a

nd

uten

sils

; whi

sk

14 fo

od sa

mpl

es: c

hick

en ro

ll;

whi

te re

ctan

gula

r cut

mea

t; re

d re

ctan

gula

r cut

mea

t; ch

icke

n sh

redd

ed; C

aesa

r sal

ad w

ith

slic

ed b

oile

d eg

gs; l

ettu

ce

shre

dded

; cor

iand

er; c

ucum

ber

slic

ed; t

omat

o sl

iced

; car

rots

sl

iced

; oni

ons s

liced

; bee

troot

sl

iced

; boi

led

eggs

(3);

chee

se

slic

ed

9 en

viro

nmen

tal s

wab

s: b

oot

swab

s (3)

; sin

k; c

ool r

oom

; co

mpo

site

mix

ing

bow

l; ch

icke

n sk

ewer

s; ra

w c

hick

en p

rep

sink

; pa

te fr

idge

183

proc

esse

d m

eat;

froz

en c

hick

en

schn

itzel

; fro

zen

pate

(2);

mar

garin

e; c

rack

ed b

lack

pe

pper

; red

food

col

our

NSW

Ja

n 20

16

Age

d ca

re

faci

lity

No

sam

ples

col

lect

ed

No

swab

s col

lect

ed

-

AC

T Fe

b 20

16

Res

taur

ant

Who

le e

ggs (

3); s

pina

ch;

com

mer

cial

ly p

rodu

ced

holla

ndai

se sa

uce

No

swab

s col

lect

ed

Non

e

Qld

Fe

b 20

16

Mar

ket

Panc

ake

batte

r; ch

icke

n; h

am;

spin

ach;

che

ese;

chi

cken

cre

pe

Prep

arat

ion

benc

h (k

itche

n);

hand

mix

er; c

repe

spre

ader

; ba

tter b

ottle

; cut

ting

boar

d

Non

e

AC

T Fe

b 20

16

Res

taur

ant

Boi

led

egg;

egg

and

lettu

ce

sand

wic

h; sh

redd

ed le

ttuce

N

o sw

abs c

olle

cted

N

one

Qld

M

ar 2

016

Fest

ival

N

o sa

mpl

es c

olle

cted

N

o sw

abs c

olle

cted

-

*Pos

sibl

e le

ft ov

er fo

ods f

rom

the

day(

s) o

f the

out

brea

k.

184

185



Ford L, Ingle D, Glass K, Veitch M, Williamson DA, Harlock M, Gregory J, Stafford R, French

N, Bloomfield S, Grange Z, Conway ML, Kirk MD. Whole-genome sequencing of Salmonella

Mississippi and Typhimurium Definitive Type 160, Australia and New Zealand. Emerging

Infectious Diseases. 2019;25(9):1690-1697, doi: 10.3201/eid2509.181811.

Article DOI: https://doi.org/10.3201/eid2509.181811

Whole-Genome Sequencing of Salmonella Mississippi and Typhimurium Definitive Type 160, Australia and New Zealand

Appendix

Methods

Salmonella Mississippi Sampling Strategy

We sampled from 529 Salmonella Mississippi human isolates with a Tasmanian postcode

at the Microbiological Diagnostic Unit Public Health Laboratory (MDU PHL) and isolated

between 2011 and 2015. We excluded any isolate with an indication of overseas travel (n = 2),

sorted by isolation date, and then sampled every 1 in 15 isolates (n = 36). When comparing the

characteristics of sampled isolates to the population, there was a higher proportion of males in

the sample (Appendix Table 1).

Therefore, we re-sampled, sorting by isolation date and sampling every 2nd in 15 isolates

(n = 36). As the characteristics of this sample were closer to the population, they were chosen for

sequencing (Appendix Table 2, Appendix Figures 2, 3).

We also sampled from 74 isolates from other states that were isolated between 2011 and

2015 and at MDU PHL. We excluded isolates with an indication of overseas acquisition (n = 12).

We selected the 1 isolate available from the Australian Capital Territory (ACT), 1 from South

Australia (SA), and 1 from Queensland (Qld), all with travel to Tasmania reported. There were 6

isolates with a New South Wales postcode and 2 isolates with a Western Australia postcode. We

randomly selected 1 from each of these states, both with travel to Tasmania reported. There were

51 isolates with a Victorian postcode. We sorted by postcode and selected 1 in 5, sampling 10

isolates from Victoria. The characteristics of cases with selected isolates compared to all isolates

in other states is in Appendix Table 3.

186

In addition, we opportunistically sampled 12 Salmonella Mississippi human isolates that

were at Queensland Health Forensic and Scientific Services, which were isolated between 2004

and 2009. Of these, 42% were from males, and the median age was 33.5. Some isolates were

from cases involved in an unpublished 2008 case–control study. The isolates were sent to MDU

PHL for sequencing.

Additional Genomics Information

As there is no publicly available complete Salmonella Mississippi genome, a local

reference was assembled using one of the MDU PHL Salmonella Mississippi isolates.

Preliminary analysis showed the Salmonella Mississippi genomes had similar QC stats and the

oldest sequenced strain (AUSMDU00020775) was selected (2000). Illumina reads of the isolate

were assembled using Unicycler (1) using a minimum contig length on 200 bp. The resulting

assembly was used as the reference for the analysis of the core genome. AUSMDU00020775

assembly comprised 92 contigs, with total number of reference bases 4672632 and an N50 of

366033.

We looked at 11 publicly available draft assemblies of Salmonella Mississippi (2–4). The

sequence type (ST) of the publicly available draft assemblies were determined using mlst (v2.11-

dev) (https:/github.com/tseemann/mlst) in conjunction with the “senterica” MLST scheme. Of

note, the majority of the public and the MDU PHL Salmonella Mississippi isolates did not have a

known ST. One publicly available isolate (Mississippi_BCW_4007) (4) had no alleles the same

as the other Salmonella Mississippi genomes, indicative that this was a highly divergent strain of

Salmonella Mississippi. This isolate was excluded from further analysis. Nine of the remaining

included publicly available draft assemblies were from soil or Takahe from 2011–2013 in New

Zealand (3) and 1 was from a human in 2010 from the United States of America (2).

Results

Single-Nucleotide Polymorphisms (SNPs) between Australian and New Zealand Isolates

While there were only 9 publicly available Salmonella Mississippi isolates from New

Zealand, there were a large number of SNPs between these isolates and those from Australia

(Appendix Figure 3). There was also considerable heterogeneity within the Australian isolates,

with a maximum of 649 SNPs between 2 isolates.

187

Comparatively, there was much less variability among the Australian and New Zealand

Salmonella Typhimurium DT160 isolates (Appendix Figure 4).

Violin plots were made using Stata SE 14, with the vioplot package.

Selected Risk Factors for Salmonella Mississippi and Salmonella Typhimurium DT160

We compared data on risk factors for Salmonella Mississippi and Salmonella

Typhimurium DT160 cases in the week before the onset of illness using a 2-sample test of

proportions (Appendix Table 4).

Salmonella Typhimurium DT160 Australian Case Animal Contact

Of the 88% (36/41) of Australian Salmonella Typhimurium DT160 cases that reported

direct animal contact, animals included dogs (n = 27), cats (n = 18), chickens (n = 12), birds (n =

6), possum (n = 3), wallabies or kangaroos (n = 3), horses (n = 3), sheep (n = 2), rabbits (n = 2),

and 1 each of cows, goats, pigs, fish, guinea pigs, deer, bandicoot, and worms and snails.

References

1. Wick RR, Judd LM, Gorrie CL, Holt KE. Unicycler: resolving bacterial genome assemblies from short

and long sequencing reads. PLOS Comput Biol. 2017;13:e1005595. PubMed

https://doi.org/10.1371/journal.pcbi.1005595

2. Gupta R, Schmidtke A, Sabol A, Castillo D, Ribot E, Trees E. Salmonella enterica subsp. enterica

serovar Mississippi str. 2010K–1406, whole genome shotgun sequencing project. Acession

ALPQ000000001. GenBank. 2012 [cited 2018 Apr 23].

https://www.ncbi.nlm.nih.gov/nuccore/ALPQ00000000.1

3. Grange ZL, Biggs PJ, Rose SP, Gartrell BD, Nelson NJ, French NP. Genomic epidemiology and

management of Salmonella in island ecosystems used for takahe conservation. Microb Ecol.

2017;74:735–44. PubMed https://doi.org/10.1007/s00248-017-0959-1

4. den Bakker HC, Moreno Switt AI, Govoni G, Cummings CA, Ranieri ML, Degoricija L, et al. Genome

sequencing reveals diversification of virulence factor content and possible host adaptation in

distinct subpopulations of Salmonella enterica. BMC Genomics. 2011;12:425. PubMed

https://doi.org/10.1186/1471-2164-12-425

188

Appendix Table 1. Sex and median age of cases with sampled isolates and all Salmonella Mississippi isolates, Tasmania, 2011–2015, MDU PHL, sample 1Variable All isolates Sampled isolates% M 47 64Median age, y 45 50

Appendix Table 2. Sex and median age of cases with sampled isolates and all Salmonella Mississippi isolates, Tasmania, 2011–2015, MDU PHL, sample 2Variable All isolates Selected isolates% M 47 53Median age, y 45 42

Appendix Table 3. Sex and median age of cases with sampled isolates and all Salmonella Mississippi isolates, mainland Australia2011–2015, MDU PHLVariable All isolates Selected isolates% M 40 38Median age, y 45 54

Appendix Table 4. Salmonella Mississippi and Salmonella Typhimurium DT160 cases who answered yes to risk factor questions ofthose who answered the question, Australia

VariableSalmonella Mississippi,

no. (%)Salmonella Typhimurium DT160,

no. (%) p valueBushwalking 2/25 (8) 3/40 (7.5) 0.94Camping 2/26 (8) 0/40 (0) 0.07Gardening 4/25 (16) 9/41 (22) 0.56Swimming 4/25 (16) 7/39 (18) 0.84Drinking from an untreated raw water source 14/23 (61) 12/30 (40)* 0.1Animal contact 19/28 (68) 36/41 (88) 0.04*2/12 (17%) reported boiling their water before drinking.

Appendix Figure 1. Age distribution of all Salmonella Mississippi cases with isolates, Tasmania,

Australia, 2011–2015, Microbiological Diagnostic Unit Public Health Laboratory, sample 2.

189

Appendix Figure 2. Month of isolation of all Salmonella Mississippi cases with isolates, Tasmania,

Australia, 2011–2015, Microbiological Diagnostic Unit Public Health Laboratory, sample 2.

Appendix Figure 3. Violin plot of single nucleotide polymorphisms within Australian and New Zealand

clades and between them, Salmonella Mississippi, 2008–2015.

190

Appendix Figure 4. Violin plot of single nucleotide polymorphisms within Australian and New Zealand

clades and between them, Salmonella Typhimurium definitive type 160, 1998–2016.

191