Chiropractic Care and Risk For Acute Lumbar Disc Herniation: A Mixed Methods Approach

by

Cesar Alberto Hincapié

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Dalla Lana School of Public Health Division of Epidemiology

University of Toronto

© Copyright by Cesar Alberto Hincapié, 2015

ii

Chiropractic Care and Risk For Acute Lumbar Disc Herniation:

A Mixed Methods Approach

Cesar Alberto Hincapié

Doctor of Philosophy

Dalla Lana School of Public Health Division of Epidemiology

University of Toronto

2015

Abstract

Objective: To advance knowledge of the risk for acute lumbar disc herniation (LDH)

following chiropractic care. The objectives were to: (1) synthesize current evidence on

the incidence and determinants of LDH with radiculopathy in adults; (2) determine

clinicians’ beliefs regarding the risk for acute LDH associated with chiropractic spinal

manipulation treatment (SMT); and, (3) investigate associations between both

chiropractic and primary care physician (PCP) care, and acute LDH with incident early

surgery.

Methods: (1) Systematic review of the literature. (2) Belief elicitation interviews with 47

chiropractors, family physicians and spine surgeons. Probability distributions for the

relative risk (RR) for acute LDH associated with chiropractic SMT were derived. (3) Self-

controlled case series study using Ontario administrative health data. Incidence ratios

for acute LDH with early surgery in exposed periods after chiropractic visits relative to

unexposed periods were estimated within individuals, and compared with incidence

ratios following PCP visits.

iii

Results: (1) The annual incidence of LDH with radiculopathy ranges between 0.1% and

10%, and varies with the case definition. Risk factors included individual, behavioural

and work-related characteristics. (2) In the belief elicitation study, chiropractors

expressed the most optimistic belief (median RR, 0.56; IQR, 0.39-1.03); family

physicians, a neutral belief (median RR, 0.97; IQR, 0.64-1.21); and spine surgeons, a

more pessimistic belief (median RR, 1.07; IQR, 0.95-1.29). (3) The self-controlled case

series study found positive associations between both chiropractic and PCP visits, and

surgically managed acute LDH. There was no more risk associated with visits to

chiropractors than with visits to PCPs.

Conclusions: (1) The best evidence synthesis shows the varying quality and

heterogeneity of the epidemiologic literature on LDH with radiculopathy. There is a need

to develop standardized case definitions that validly classify the clinical spectrum. (2)

Clinicians’ beliefs about the risk for acute LDH associated with chiropractic SMT varied

systematically across professions. The derived probability distributions can serve as

prior probabilities in Bayesian analyses of this exposure-outcome association. (3)

Patients with prodromal back pain related to a developing disc herniation may seek

healthcare from both chiropractors and PCPs before full clinical expression of acute

LDH that is subsequently surgically managed.

iv

Acknowledgments

First, I would like to acknowledge my parents, Juan Manuel and Lilia Hincapié; my

brother, Juan Carlos Hincapié, and sister, Maria Cristina Choy (née Hincapié); and the

rest of my family, for all their love, support and encouragement. Má y Pá, soy quien

soy, en gran parte por tu ejemplo único de amor, sacrificio y esfuerzo. Gracias con toda

alma, vida y corazón!

I would like to express my sincere gratitude and appreciation to my supervisor, J. David

Cassidy, for giving me the opportunity to undertake this PhD under your great

mentorship and for providing many learning and research opportunities throughout the

process. Your dedication, excellence and leadership have burned a bright path for so

many others to follow, and I am truly honoured and blessed to count you as a mentor,

colleague and friend.

A heart-felt thanks to all of my advisory committee members: Pierre Côté, Alejandro

Jadad, George Tomlinson, Raj Rampersaud and Eleanor Boyle, for their enthusiasm,

support and guidance with this thesis. This work and I have been enriched by each of

your brilliant minds and many contributions along the way. A special thanks goes to

George, for your incredible generosity with both your time and sharp analytical mind,

and for your friendship. I will remember with great fondness the many hours of

“R-awesome” coding and analysis that we shared.

I thank my doctoral program director, Nancy Kreiger, and the support staff in the

Division of Epidemiology, at the Dalla Lana School of Public Health, for their helpful

administrative support and guidance throughout the program.

I would like to acknowledge the Canadian Institutes of Health Research and the

Canadian Chiropractic Research Foundation, for their interest and funding of this work.

A special mention goes to Allan Gotlib, for your tireless effort and enthusiasm for the

development of Canadian chiropractic clinician-scientists; thank you, Allan.

Many thanks to my chiropractic and epi research friends: Maja Stupar, Carol

Cancelliere, James Donovan, Heather Shearer, Craig Jacobs, Mana Rezai, Darren

v

Brenner, Paul Arora, Beth Rachlis, Nancy Carnide and Silje Mæland, for the memorable

times shared and many insightful clinical, epidemiologic and personal discussions along

the way. Silje, a special thanks to you and Ketil (and Sunniva, Gustav and Wilhelm), for

your friendship and encouragement, and the great times our families shared.

Thank you to my beautiful, supportive and understanding children, Maia Liliana, Julia

Marie, Daniel Joseph and Michael Juan Hincapié. You are the most precious gift and

blessing in my life, and I give God thanks for allowing me to share in your lives.

Finally, thank you to my wonderful and amazing wife, Barb Hincapié (née Kral). Amor,

there are no words to express my love and gratitude for you and all that you do for our

family. You are a shining beacon of love, dedication and giving. Thank you for your

support, patience, kindness and encouragement. I look forward to continuing on life’s

journey together!

vi

Table of Contents

Acknowledgements iv

Table of Contents vi

List of Tables vii

List of Figures viii

List of Appendices ix

List of Abbreviations x

Thesis Overview xii

Chapter 1 General introduction and outline 1

Chapter 2 Incidence and determinants of lumbar disc herniation with

radiculopathy in adults: a systematic review

In preparation for submission, 2015

22

Chapter 3 Chiropractic spinal manipulation treatment and the risk for acute

lumbar disc herniation: a belief elicitation study

In preparation for submission, 2015

65

Chapter 4 Chiropractic care and risk for acute lumbar disc herniation: a

population-based self-controlled case series study

In preparation for submission, 2015

87

Chapter 5 Summary and synthesis 115

Chapter 6 Appendices 131

vii

List of Tables

Table 1.1. Low risk of bias RCTs on spinal manipulation for lumbar disc herniation

with radiculopathy

Table 2.1. Characteristics of eligible studies examining the incidence of and/or risk

factors for lumbar disc herniation with radiculopathy in adults

Table 2.2. Summary of eligible studies examining the incidence of and/or risk factors

for lumbar disc herniation with radiculopathy in adults

Table 3.1. Characteristics of belief elicitation study participants

Table 3.2. Clinicians’ beliefs about the risk for acute lumbar disc herniation and

surgically managed acute lumbar disc herniation associated with

chiropractic spinal manipulation treatment

Table 3.3. Characteristics of clinicians with optimistic versus pessimistic beliefs about

the risk for acute lumbar disc herniation and surgically managed acute

lumbar disc herniation associated with chiropractic spinal manipulation

treatment

Table 4.1. Characteristics of study participants at the time of the recording of an

acute lumbar disc herniation with incident early surgery

Table 4.2. Distribution of chiropractor and primary care physician visits before the

event index date for all cases and stratified by event fiscal year

Table 4.3. Incidence rate ratios for an acute lumbar disc herniation with first early

surgery event in exposed periods following chiropractic and primary care

physician visits

Table 4.4. Results of sensitivity analyses compared to primary analysis of an acute

lumbar disc herniation with first early surgery event in risk periods 0 to 7

days after chiropractor and primary care physician visits

viii

List of Figures

Figure 1.1. The case-crossover and self-controlled case series designs

Figure 2.1. Flow diagram of information through phases of the systematic review

Figure 3.1. Beliefs regarding the risk for acute lumbar disc herniation associated with

chiropractic spinal manipulation treatment: a. among all clinicians, b.

among clinician groups, c. among optimistic versus pessimistic clinicians

Figure 3.2. Beliefs regarding the risk for acute lumbar disc herniation associated with

chiropractic spinal manipulation treatment: a. by gender, b. by age group,

c. by clinical experience

Figure 4.1. Representation of the self-controlled case series design

Figure 4.2. Case definition flow diagram

Figure 4.3. Bootstrap incidence rate ratios for primary analysis of any visit exposure

with a risk period of 0-7 days following chiropractic versus primary care

physician care

Figure 4.4. Kernel density estimate of the ratio of bootstrap incidence rate ratios for

chiropractic versus primary care physician care

Figure 5.1. Preliminary Bayesian triplots of risk for acute lumbar disc herniation with

early surgery associated with chiropractic care

ix

List of Appendices

Appendix A. Risk of bias assessment summary (Chapter 1)

Appendix B. Comparison of the case-crossover (CCO) and self-controlled case series

(SCCS) designs (Chapter 1)

Appendix C. Systematic review search strategies (Chapter 2)

Appendix D. Detailed evidence tables for systematic review (Chapter 2)

Appendix E. Belief elicitation questionnaire (Chapter 3)

Appendix F. Belief elicitation script (Chapter 3)

Appendix G. Belief elicitation computer protocol (Chapter 3)

Appendix H. Case and exposure definition tables for self-controlled case series study

(Chapter 4)

x

List of Abbreviations

CCO Case-crossover

CI Confidence interval

CT Computerized tomography

DAD Discharge Abstract Database

DC Doctor of chiropractic

ED Emergency department

HEA Home exercise and advice

HNP Herniated nucleus pulposus

HRR Hazard rate ratio

ICD International classification of diseases

IQR Interquartile range

IRR Incidence rate ratio

LBP Low back pain

LDH Lumbar disc herniation

MRI Magnetic resonance imaging

MSK Musculoskeletal

N Number

NACRS National Ambulatory Care Reporting System

ND No data

NR Not reported

NRS Numerical rating scale

NSAID Nonsteroidal anti-inflammatory drug

OHIP Ontario Health Insurance Plan

OR Odds ratio

PCP Primary care physician

PRISMA Preferred Reporting Items for Systematic Reviews and Meta-Analyses

PROSPERO International prospective register of systematic reviews

QTFSD Quebec Task Force on Spinal Disorders

R R project for statistical computing

RCT Randomized controlled trial

xi

REB Research ethics board

ROB Risk of bias

RR Relative risk or risk ratio

SCCS Self-controlled case series

SF-36 Short form-36 health survey

SHELF Sheffield elicitation framework

SIGN Scottish Intercollegiate Guidelines Network

SMR Standardized morbidity ratio

SMT Spinal manipulation treatment

SNP Single nucleotide polymorphism

US United States of America

UK United Kingdom

VAS Visual analog scale

VNTR Variable number of tandem repeats

xii

Thesis Overview

The objective of this thesis is to advance knowledge of the risk for acute lumbar disc

herniation (LDH) following chiropractic care using a mixed methods approach.

Chapter 1 includes a brief introduction to the public health and epidemiologic

challenges of low back pain and LDH with radiculopathy; a description of the

epidemiology of LDH with radiculopathy, chiropractic care and the clinical management

of back pain; and a summary of the uncertainty surrounding the link between spinal

manipulation and LDH. The challenges in studying rare serious adverse events

following spinal manipulation are described. Finally, the rationale for the study designs

of this thesis is provided and the thesis aim and objectives are stated.

Chapter 2 presents a systematic review of the incidence and determinants of LDH with

radiculopathy in adults. This helps clarify the epidemiology of this important condition

and advances understanding of the risk factors for LDH with radiculopathy.

Chapter 3 describes and quantifies clinicians’ beliefs about the risk for acute LDH

associated with chiropractic spinal manipulation treatment. This involves a Bayesian-

inspired approach that can determine the magnitude of a potential risk expected by

clinician experts and describe the presence of uncertainty regarding this exposure-

outcome association.

Chapter 4 investigates the association between chiropractic care and acute LDH with

early surgery, and compares this to the association between primary care physician

care and acute LDH with early surgery. We provide a valid epidemiologic assessment

of the link between both chiropractic and primary medical care and acute LDH with early

surgical intervention.

Chapter 5 is the concluding chapter. The main findings of our studies are summarized,

synthesized and placed in broader clinical, research and public health context. We

elaborate on the implications and limitations of our work and provide guidance for future

research.

1

Chapter 1

General Introduction and Outline 1

Chiropractic care is popular for low back pain, but little is known about its potential link

to acute lumbar disc herniation (LDH) with radiculopathy. Several systematic reviews

suggest that chiropractic treatment or spinal manipulation is a viable therapeutic option

for low back pain, but the summarized studies are of varying quality and too small to

examine rare serious adverse events.1-4 Moreover, randomized clinical trials have

shown benefit of chiropractic spinal manipulation for the management of LDH with

radiculopathy,5-9 yet little is known about serious adverse events possibly related to this

treatment. This void of knowledge represents a significant problem because an

effective treatment is one that cost-effectively improves health outcomes while first

doing no harm.

Today, no valid estimate of the risk for acute LDH with radiculopathy following

chiropractic treatment is available in the scientific literature. The current literature

contains case reports and small case series linking lumbar spine manipulation to acute

LDH with radiculopathy and cauda equina syndrome.10-12 However, case reports and

case series provide the lowest level of evidence with respect to the assessment of risk

and should not be used to make inferences about the lack of safety of a treatment.

They have, however, raised the hypothesis of potential harm.

This chapter includes a brief introduction to the public health and epidemiologic

challenges of low back pain and LDH with radiculopathy; a description of the

epidemiology of LDH with radiculopathy, chiropractic care and the clinical management

of back pain; followed by a summary of the uncertainty surrounding the link between

spinal manipulation and LDH. The challenges in studying rare serious adverse events

following spinal manipulation are described. Finally, the rationale for the study designs

of this thesis is provided and the thesis aim and objectives are stated.

2

1.1 Low Back Pain and Lumbar Disc Herniation With Radiculopathy – Public Health and Epidemiologic Challenges

Around the world, low back pain is recognized as an important public health concern,

mainly because it is a leading cause of disability and work absence, and associated with

a huge economic burden on individuals, industry and society.13 It affects about 70% of

all people in their lifetime, and between 15% to 30% on any given day with varying

types of clinical presentations.14-16 The Global Burden of Disease 2010 project provides

the clearest evidence to date of the huge and increasing burden on global health from

musculoskeletal conditions—it showed that low back pain is the leading cause of

disability worldwide.17

LDH with radiculopathy, defined as the localized displacement of disc material beyond

the normal margins of the intervertebral disc space with pain, weakness or numbness in

a myotomal or dermatomal pattern,18 can be one of the most recognizable presentations

of low back pain. The diagnosis is typically based on a combination of clinical features

suggesting lumbar spinal nerve root compression or irritation, such as: lumbosacral

radiculopathy (i.e., radicular leg pain or sciatica), nerve root tension signs, neurologic

deficits (i.e., muscle weakness and reflex changes), and advanced imaging (i.e., MRI or

CT) findings that correlate with the clinical syndrome.19-21 However, many patients

present with a less definitive clinical picture, involving low back pain in the early

(prodromal) phase that then progresses to radicular leg pain with or without neurologic

signs.22,23 In addition, diagnostic imaging may only be indicated in the prodromal phase

when there is suspicion of underlying pathology, such as infection or malignancy.24

Finally, the predictive values of common physical examination tests for LDH with

radiculopathy are low.20,25,26

All these factors contribute in making the diagnosis during the early phases of the

syndrome especially difficult. Moreover, this multifaceted clinical presentation, resultant

variable case definitions,19 and difficulties with clinical diagnosis and classification,20,25,26

are notable reasons why the descriptive epidemiology of this condition is not precisely

known.

3

1.2 Descriptive Epidemiology of Lumbar Disc Herniation With Radiculopathy

Exact data on the prevalence and incidence of LDH with radiculopathy are lacking.

General population studies involving clinical assessment have estimated the point

prevalence of lumbar disc syndrome at about 5%,27,28 varying by sex and age. It is

most common among persons 30-50 years of age, with men more frequently affected

than women.29 Among acute low back pain patients seeking primary health care, the

prevalence of lumbar radiculopathy suggesting disc herniation is, not surprisingly,

substantially higher. Prospective clinical cohort studies have found as many as 20% to

25% of acute low back pain patients reporting prevalent radiating leg pain below the

knee at the time of initial presentation.30,31 One recent systematic review of the

prevalence of sciatica found considerable variation in estimates, ranging from 1.6% in

the general population to 43% in a population of nursing aides.32

Few epidemiologic studies have estimated the incidence of LDH with radiculopathy.

One study calculated the incidence of hospitalized lumbar disc herniation or sciatica

among the general population at 1.5 per 1,000 person-years,33 while another estimated

7.8 cases of hospitalized lumbar disc disorder with radiculopathy per 1,000 person-

years among young Finnish male military conscripts.34

With respect to risk factors, the literature has suggested several focusing on activities

involving sudden or sustained rotation and flexion of the low back, such as heavy lifting,

twisting, bending, and driving.33,35-41 However, other factors such as smoking,

pregnancy, diabetes, body mass index, hypertension, high cholesterol, and family

history have also been identified as possible determinants.42-45 Overall, little is known

about the incidence of symptomatic LDH, and consequently, risk factors for this

condition are not well understood.

1.3 Chiropractic Care and the Clinical Management of Back Pain

In Canada, medical doctors, chiropractors and physical therapists are the main health

care providers that manage spinal pain.46 Approximately 12% of American and

Canadian adults seek chiropractic care annually, and over 9 out of 10 visits involve

treatment with spinal manipulation.47,48 Canadian chiropractic patients, compared to

4

those seeking medical care for back pain, tend to be younger and have higher

socioeconomic status and fewer health problems.46,49 Patients have reported very good

or excellent satisfaction or perceived helpfulness with chiropractic care that has

included spinal manipulation.50

No widely adopted definition of serious adverse event exists in the back pain literature.

For the purpose of this thesis, we define a serious adverse event as an untoward

occurrence that results in death, is life threatening, requires hospitalization, or results in

significant or permanent disability.12,51 Examples of serious adverse events resulting

from lumbosacral spinal manipulation include disc herniation and cauda equina

syndrome.

The medical and chiropractic management of back pain rests on the delivery of different

therapeutic modalities, with varying risk-benefit profiles. Medical doctors mainly

prescribe medications, such as nonsteroidal anti-inflammatory drugs (NSAIDs), while

chiropractors primarily use spinal manipulation. Both types of interventions can benefit

low back pain,4,52 and are potentially related to adverse events. Compared with nonuse,

the use of NSAIDs has been associated with up to a 3-fold increased risk for

dyspepsia,53 about a 5-fold greater risk for serious upper gastrointestinal bleeding,54,55

and up to a 2-fold increase in the risk of death56. Even the newer and less gastrotoxic

cyclooxegenase-2 selective inhibitors are associated with an increased risk of serious

vascular events, such as stroke and myocardial infarction.57,58

With respect to spinal manipulation, prospective studies of harm have mainly reported

minor and transient adverse events, such as local pain and muscle soreness, but have

not observed the occurrence of serious adverse events. LeBoeuf-Yde et al48 described

a Swedish prospective case series including 625 chiropractic patients who had 1858

treatment visits, of which 99% involved spinal manipulation. Adverse events were

reported to be common, benign and self-limiting (usually within 1 day). The most

commonly reported adverse event was local discomfort in the area of treatment,

accounting for roughly two thirds of all adverse events. Similarly, in their prospective

case series of 4712 treatment visits of 1058 new patients by 102 Norwegian

chiropractors, Senstad et al59 found that 55% of patients reported at least one adverse

event, and 25% of all treatments (75% of which involved the lumbar spine) resulted in at

5

least one event. The onset of most events was on the day or the day after the

treatment visit. Adverse events consisted primarily of short-term local discomfort,

headache, tiredness or radiating discomfort (defined as discomfort in an area away from

the one treated), and were characterized as mild or moderate by 85% of patients.

Neither study, however, reported a serious adverse event associated with spinal

manipulation.

Nonetheless, the reporting of harms is generally poor in the primary literature,4 and

case reports of serious adverse events, such as disc herniation and cauda equina

syndrome, have been described, but their incidence have yet to be validly estimated.60

Hebert et al12 undertook a systematic review of cases of serious adverse events

following lumbopelvic spinal manipulation. They found that over two thirds of all

reported serious adverse events consisted of cauda equina syndrome (29 cases, 38%

of total) or LDH (23 cases, 30% of total). However, important case details were often

not captured in the case reporting, including pre-manipulation presentation of the

patient, specific details of the serious adverse event, time from spinal manipulation to

onset of the event, and clinical outcome. Improved knowledge of the risk for serious

adverse events associated with spinal manipulation could inform clinical decision-

making and is very much needed.

1.4 Spinal Manipulation and Lumbar Disc Herniation – Potential Benefit Or Potential Harm?

Spinal manipulation treatment is generally considered to be safe, although some

concern has been raised about its potential link with LDH.61,62 Several systematic

reviews have found that spinal manipulation can benefit low back pain with little

evidence of serious harm.4,63 Randomized clinical trial evidence also supports the use

of spinal manipulation for the treatment of LDH with radiculopathy (Table 1.1, see

Appendix A for risk of bias assessment summary).7-9

6

Table 1.1. Low risk of bias RCTs on spinal manipulation for lumbar disc herniation with radiculopathy

Authors, year

Subjects, setting, case

definition and study size

Description of Interventions

(n patients)

Outcomes measured and

follow-up Key findings

Santilli et al, 20067 Patients (18-65yo) seen in 2

rehabilitation centers in

Rome, Italy

Case definition: acute LBP

(<10d and pain-free

during prior 3mo) of

moderate to severe

intensity (VAS, ≥5/10),

moderate to severe

radiating pain to one leg

(VAS, ≥5/10) and +ve MRI

n=102

SMT by DC: up to 20

treatments over 30d (53)

Sham SMT by DC: up to 20

treatments over 30d (49)

Validated: VAS1 (local low

back pain), VAS2

(radiating pain), SF-36

(quality of life)

Non-validated: Disc

morphology, Kellner score

(psychological profile)

Follow-up: 15, 30, 45, 90

and 180d

A significantly greater

number of patients treated

with spinal manipulation

had no back, buttock or

leg pain at 180 days. VAS

1: 28% vs. 6 %. VAS 2:

55% vs. 20%.

No significant difference in

the SF-36, psychological

testing and disc

morphology between the

groups.

No adverse events.

McMorland et al, 20108 Patients (>18yo) screened

by 3 spine neurosurgeons

for symptoms of unilateral

lumbar radiculopathy

secondary to LDH at L3-4,

L4-5, L5-S1 disc levels, in

Calgary, Canada

Case definition: leg-

dominant symptoms with

SMT by DC: 2-3 visits/wk

for the first 4 wk reducing

to 1-2 visits/wk for the

next 3-4 wk; at the 8-wk

mark, follow-up visits

were scheduled based on

the patient's symptoms;

average of 21 treatments

(20)

Validated: McGill Pain

Questionnaire, Aberdeen

Back Pain Scale, Roland-

Morris Disability Index,

SF-36 (quality of life)

Follow-up: 3, 6, 12, 24 and

52wk

Significant improvement in

both treatment groups

compared to baseline

scores over time was

observed in all outcome

measures.

Of patients with lumbar

radiculopathy due to LDH

that failed 3mo of medical

7

Authors, year

Subjects, setting, case

definition and study size

Description of Interventions

(n patients)

Outcomes measured and

follow-up Key findings

objective signs of nerve

root tethering ± neurologic

deficit correlated with +ve

MRI

Notable inclusion criterion:

patients had failed ≥3mo

of nonoperative care,

including analgesics,

lifestyle modification,

physiotherapy, massage

therapy and/or

acupuncture

n=40

Surgical microdiscectomy

by neurosurgeon (20)

Crossover to the alternate

treatment was allowed

after 3mo, if patient felt

they were not responding

to primary intervention

management, 60%

benefited from SMT to the

same degree as if they

underwent surgical

intervention.

For the 40% that did not

respond to SMT, surgery

provided an excellent

outcome.

No new neurologic deficits

from either of the

interventions and no

significant adverse

events.

Bronfort et al, 20149 Patients (≥21yo) with

subacute or chronic back-

related leg pain at 2

research centers in

Minnesota and Iowa, US

Case definition: back-

related leg pain based on

QTFSD classifications 2,

3, 4 or 6 (radiating pain

into the proximal or distal

SMT plus HEA: SMT by DC

up to 20 visits and HEA

by DC, exercise therapist

and personal trainer 4

sessions (96)

HEA: HEA by DC, exercise

therapist and personal

trainer 4 sessions (96)

12wk course of care for

both groups

Validated: 11-point NRS,

Roland-Morris Diability

Questionniare, SF-36

(quality of life)

Non-validated: self-reported

LBP

Follow-up: 3, 12, 26 and

52wk

For leg pain, SMT plus HEA

had a clinically important

advantage over HEA

(difference, 10 percentage

points [95%CI, 2 to 19]; P

= 0.008) at 12wk but not

at 52wk (difference, 7

percentage points [CI, -2

to 15]; P = 0.146).

Nearly all secondary

8

Authors, year

Subjects, setting, case

definition and study size

Description of Interventions

(n patients)

Outcomes measured and

follow-up Key findings

part of the lower

extremity, with or without

neurologic signs); severity

of ≥3 on 0-10 scale;

current episode ≥4 weeks;

and stable prescription

medication plan in past

mo

Notable exclusion criterion:

QTFSD classifications 1,

5, 7, 8, 9, 10 and 11 (pain

without radiation into the

lower extremities

N=192

outcomes improved more

with SMT plus HEA at

12wk, but only global

improvement, satisfaction,

and medication use had

sustained improvements

at 52wk.

No serious treatment-

related adverse events or

deaths occurred.

Abbreviations: +ve, positive; CI, confidence interval; d, days; DC, doctor of chiropractic; HEA, home exercise and advice; LBP, low back pain; LDH,

lumbar disc herniation; mo, months; MRI, magnetic resonance imaging; NRS, numerical rating scale; QTFSD, Quebec Task Force on Spinal

Disorders; SF-36, short form-36 health survey; SMT, spinal manipulation treatment; VAS, visual analog scale; wk, weeks; yo, years old

9

On the other hand, some believe, on the basis of case reports and small case series,

that spinal manipulation ought to be contraindicated for the treatment of disc herniation

and have raised the hypothesis of causal harm.10,64-69 The variability in belief was

captured in a survey of the opinions and practices of 453 UK osteopaths, in which over

half of the respondents reported that they use manipulation for disc herniation, while

most of the others described the practice as “dangerous.”70 In sum, there seems to be

a disconnect between the scientific evidence and beliefs in this area, in that the best

available evidence indicates a beneficial effect of chiropractic spinal manipulation on

LDH, yet beliefs and opinion vary and may overly focus on potential harm.

1.5 Challenges In Studying Rare Serious Adverse Events Following Spinal Manipulation Treatment

Attempts to investigate serious adverse events related to chiropractic spinal

manipulation face several challenges. First, the valid study of uncommon adverse

events is limited simply by the rarity of the events. Clinically important but unexpected

adverse effects of spinal manipulation are too rare to be detected in randomized trials.

In addition, well-designed prospective cohort studies are infeasible due to the inability to

recruit and follow a large enough sample of patients that will have a sufficient number of

events.

Another challenge is the issue of confounding. Confounding refers to a situation in

which a noncausal association between a given exposure and an outcome is observed

as a result of the influence of some third variable (or group of variables), usually

designated as a confounder.71 The phenomenon can be described as follows: the

confounding variable is (1) a determinant of the outcome and (2) associated with the

exposure, but (3) is not an intermediate variable in the causal pathway between

exposure and outcome. The essential notion of confounding is that the association

between an exposure and a given outcome can be induced, strengthened, weakened,

or eliminated by a third variable or group of variables. Persons who seek health care

may differ from those who do not in ways that can be difficult to measure and control

for. Some of these differences may also be associated with the future risk for LDH,

10

which makes a conventional observational (case-control or cohort) design a less valid

source of information on this association.

Finally, this is a sensitive topic open to controversy and litigation,72,73 with an extremely

limited evidence base, and thus, susceptible to opinion and anecdotal information.

1.6 The Bayesian Paradigm – A Way Forward and Rationale

A Bayesian approach to health services research provides an opportunity to advance

knowledge in spite of the above challenges.74,75 Bayesian methods start with existing

“prior” beliefs, formally quantified as probability distributions, and update these using

new data to arrive at “posterior” beliefs, which may be used as the basis for inferential

decisions.76 Quantifying currently held beliefs can determine the magnitude of a

potential risk expected by experts and describe the presence of uncertainty or clinical

equipoise about an exposure-outcome association of interest. Experts in a field can

have knowledge of the risk of using a treatment through years of clinical experience.

When quantified, the knowledge gained from their clinical experience can be included in

models estimating risk, which may help to bridge the gap between evidence and beliefs.

A notable strength of this approach is that probability distributions obtained through the

elicitation of beliefs can be used to augment scarce data,74,75,77,78 and be formally

incorporated as Bayesian priors into analyses of association. Although infrequently

used in epidemiologic research, a Bayesian approach may have important utility in the

study of treatment-related rare serious adverse events, particularly in the absence of

definitive scientific evidence.74,75,79,80

1.7 A Primer On Case-Only Designs and Rationale for the Self-Controlled Case Series Design

Epidemiologic research studies, especially those that use secondary data analysis, may

test the hypothesis of interest through the selection of one of several potential study

designs. Conventional designs, including cohort and nested case-control studies, may

yield estimates that are biased by residual confounding because obtaining complete

information on all relevant confounders may be difficult or infeasible.81,82 Recently,

increasing attention has been paid to case-only study designs, including self-controlled

11

case series and case-crossover designs. Case-only designs are considered to offer

improved control over confounding arising from variables that are constant within a

person by focusing on comparisons between different periods of time within each

person’s follow-up.83 The self-controlled case series and case-crossover designs use

restricted samples, only including data for participants who experienced the outcome of

interest.

The self-controlled case series design adopts a prospective cohort perspective by

comparing the rate of the outcome of interest between exposed and unexposed time

periods for each participant (Figure 1.1).84,85 It starts with identifying the exposure of

interest (e.g., a chiropractic visit) and defining the “effect period of exposure”, that is, the

window of time when there is a hypothesized susceptibility to an adverse effect due to

the exposure. The likelihood of the outcome of interest occurring in a time period

following exposure (i.e., “exposed” period) is compared with the likelihood of the

outcome occurring in “unexposed” periods, with the comparison being in the same

individual. Control for risk factors that are constant within an individual is achieved

through fitting Poisson regression models, conditional on the individual, to estimate the

incidence rate ratio (IRR),85 as in a matched cohort study.

The case-crossover design adopts a case-control perspective by comparing exposure

status between a time period just before occurrence of the outcome (i.e., case period)

and one or more referent time periods (i.e., control periods) prior to the case period

when the outcome did not occur (Figure 1.1).83,86 The case-crossover design starts

with identifying the outcome of interest and assessing exposure to the risk factor of

interest in a chosen time period preceding the outcome (i.e., case period). One or more

control periods are then selected, and exposure to the risk factor of interest in the case

period is compared with exposure during these control periods in the same individual.

Control for risk factors that that are constant within an individual is achieved through

using analysis methods that condition on matched time periods,86 analogous to a

matched case-control study.

12

Start ofstudyperiod

Exposurevisit

End ofstudyperiod

Non-sampledperiod

Controlperiod 7d

Outcome index date

Caseperiod 7d

0Days before event index date

Exposurevisit

Start ofstudyperiod

First exposure

visit

End ofstudyperiod

Effect period ofexposure 0-2d

Effect period ofexposure 3-7d

Effect period ofexposure 8-14d

Unexposedperiod

Calendar days after start of study period

Outcome index date

Representation of the case-crossover design. Times are relative to the event index date and the data

are viewed retrospectively from the standpoint of the event index date.

Representation of the self-controlled case series design. Times are expressed in terms of calendar

days from the start of the study period and the data are viewed as in a prospective cohort. Incidence rate

ratios compare the within-person rate of events during exposed periods with the rate of events during

unexposed periods.

Figure 1.1. The case-crossover and self-controlled case series designs

The case-crossover and self-controlled case series designs were originally introduced

for the study of the acute effects of short-term exposures, such as the occurrence of

febrile convulsions after vaccination87 or triggering of myocardial infarction by recent

coffee consumption86. Both designs are appropriate when a brief exposure (e.g., a

chiropractic visit) causes a transient change in risk (i.e., effect period of exposure) of a

rare-onset disease (e.g., acute LDH).

13

For unbiased estimation, the standard self-controlled case series design, which includes

person-time both before and after the outcome event (Figure 1.1), requires that the

exposure distribution be unrelated to event times.88 Bias may be introduced in a

standard self-controlled case series analysis, if the outcome event influences the

likelihood of post-event exposures, as when the prescription of a given drug is

contraindicated by an outcome or disease diagnosis. Recently however, a modification

of the method that ignores post-event exposures,88 has been shown to validly estimate

incidence rate ratios in situations where occurrence of the outcome event affects post-

event exposures. The modified method applies a recursive approach to the problem in

that it starts with the last observed pre-event exposure and works back through the pre-

event exposures, ignoring post-event exposures.88 Given the possibility of the

occurrence of acute LDH influencing the likelihood of post-event visits to a chiropractor

or primary care physician, it is this modified self-controlled case series design that we

use in Chapter 4 of this thesis.

The fundamental advantage of the chosen study design is that the analysis is

conditioned on the individual, yielding only within-person comparisons (i.e., self-

controlled). This is a major strength for assessing potential treatment-induced harm on

a population level because of the concern about potential confounding by factors not

recorded in health care databases. By structuring the analysis so that each person

serves as their own control, we eliminate confounding and selection bias by constant

(time-invariant) characteristics, such as, sex, location, genetics, underlying state of

health, tendency to seek professional health care, average physical activity, long-term

diet, habitual health behaviours, long-past health events such as illnesses and injuries,

occupation, social support, ethnicity, smoking history, and body mass history.

The rationale for using the self-controlled case series design over the case-crossover

design is based on three reasons. First, the case-crossover design may be susceptible

to biased estimates due to time trends in the exposure distribution (i.e., exposure-trend

bias). The case-crossover design is vulnerable to exposure-trend bias because the

control periods always precede the case period; that is, bias may arise because of time

trends in exposure and due to the fixed ordering in time of case and control

periods.83,89,90 Since the self-controlled case series method adopts a prospective cohort

perspective based on using the actual exposure histories of each study participant, it is

14

less susceptible to exposure-trend bias than the case-crossover design as the time

trend in exposure probability can be included in the model.83,85 Second, unlike the

case-crossover design, which typically requires the selection of one or more referent

time periods to serve as a control, the self-controlled case series design makes use of

all available temporal information without the need for selection. This need to choose

comparator time periods in the case-crossover design may make it susceptible to

overlap bias (i.e., a version of the bias that arises from choosing non-disjoint strata to

partition the population in a matched case-control study).91 Unlike the case-crossover

design, the self-controlled case series design is not vulnerable to overlap bias. Third,

the self-controlled case series method is particularly suited to situations when the

investigator can accurately classify each person’s exposure timing.83 This aligns well

with our data given that the timing of exposures (based on billing claims for chiropractic

and primary care physician visits) was likely well captured by the health professionals

involved when claiming for OHIP reimbursement. Appendix B compares and contrasts

the case-crossover and self-controlled case series designs and highlights their

strengths and limitations.

15

1.8 Thesis Aim and Objectives

Overall, three consistent findings emerged from the review of the literature. First, the

epidemiology of LDH with radiculopathy is not well understood. Second, with respect to

serious adverse events following spinal manipulation and perceived risk, beliefs seem to

matter and warrant a better understanding. Third, no valid estimate of the association

between chiropractic treatment and the risk for acute LDH is currently available in the

scientific literature.

The general aim of this thesis is to advance knowledge of the risk for acute LDH

following chiropractic care. To meet this aim, the thesis uses a mixed methods

approach focusing on the following three specific objectives, with the goal of

addressing the current knowledge gaps described previously:

1. To assemble and synthesize the best available evidence on the incidence and

determinants of LDH with radiculopathy in adults. This will help clarify the

epidemiology of this important condition and advance understanding of the risk

factors for LDH with radiculopathy.

2. To describe and quantify clinicians' beliefs (expressed as probability distributions)

about chiropractic spinal manipulation as a risk factor for: (1) acute LDH, and (2)

acute severe LDH that is surgically managed. This involves a Bayesian approach

that can determine the magnitude of a potential risk expected by clinicians, describe

beliefs, and examine the presence of uncertainty regarding this exposure-outcome

association.

3. To investigate the association between chiropractic care and acute LDH with early

surgical intervention (≤8 weeks), and compare this to the association between

primary care physician care and acute LDH with early surgery. This will provide a

valid epidemiologic assessment of the association between chiropractic care and

acute LDH that is managed with early surgical intervention.

16

References 1. Ferreira ML, Ferreira PH, Latimer J, Herbert R, Maher CG. Efficacy of spinal

manipulative therapy for low back pain of less than three months' duration. J Manipulative Physiol Ther. 2003;26:593-601.

2. Bronfort G, Haas M, Evans RL, Bouter LM. Efficacy of spinal manipulation and mobilization for low back pain and neck pain: a systematic review and best evidence synthesis. Spine J. 2004;4:335-56.

3. Chou R, Huffman LH, American Pain Society, American College of Physicians. Nonpharmacologic therapies for acute and chronic low back pain: a review of the evidence for an American Pain Society/American College of Physicians clinical practice guideline. Ann Intern Med. 2007;147:492-504.

4. Furlan AD, Yazdi F, Tsertsvadze A, et al. A systematic review and meta-analysis of efficacy, cost-effectiveness, and safety of selected complementary and alternative medicine for neck and low-back pain. Evid Based Complement Alternat Med. 2012;2012:953139.

5. Burton AK, Tillotson KM, Cleary J. Single-blind randomised controlled trial of chemonucleolysis and manipulation in the treatment of symptomatic lumbar disc herniation. Eur Spine J. 2000;9:202-7.

6. Liu J, Zhang S. Treatment of protrusion of lumbar intervertebral disc by pulling and turning manipulations. J Tradit Chin Med. 2000;20:195-7.

7. Santilli V, Beghi E, Finucci S. Chiropractic manipulation in the treatment of acute back pain and sciatica with disc protrusion: a randomized double-blind clinical trial of active and simulated spinal manipulations. Spine J. 2006;6:131-7.

8. McMorland G, Suter E, Casha S, du Plessis SJ, Hurlbert RJ. Manipulation or microdiskectomy for sciatica? A prospective randomized clinical study. J Manipulative Physiol Ther. 2010;33:576-84.

9. Bronfort G, Hondras MA, Schulz CA, Evans RL, Long CR, Grimm R. Spinal manipulation and home exercise with advice for subacute and chronic back-related leg pain: a trial with adaptive allocation. Ann Intern Med. 2014;161:381-91.

10. Assendelft WJ, Bouter LM, Knipschild PG. Complications of spinal manipulation: a comprehensive review of the literature. J Fam Pract. 1996;42:475-80.

11. Oliphant D. Safety of spinal manipulation in the treatment of lumbar disk herniations: a systematic review and risk assessment. J Manipulative Physiol Ther. 2004;27:197-210.

12. Hebert JJ, Stomski NJ, French SD, Rubinstein SM. Serious adverse events and spinal manipulative therapy of the low back region: a systematic review of cases. J Manipulative Physiol Ther. 2013.

13. Hoy D, Brooks P, Blyth F, Buchbinder R. The epidemiology of low back pain. Best Pract Res Clin Rheumatol. 2010;24:769-81.

14. Kent PM, Keating JL. The epidemiology of low back pain in primary care. Chiropr Osteopat. 2005;13:13.

15. Walker BF. The prevalence of low back pain: a systematic review of the literature from 1966 to 1998. J Spinal Disord. 2000;13:205-17.

16. Andersson GB. Epidemiology of low back pain. Acta Orthop Scand Suppl. 1998;281:28-31.

17

17. Vos T, Flaxman AD, Naghavi M, et al. Years lived with disability (YLDs) for 1160 sequelae of 289 diseases and injuries 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. The Lancet. 2012;380:2163-96.

18. Kreiner DS, Hwang SW, Easa JE, et al. An evidence-based clinical guideline for the diagnosis and treatment of lumbar disc herniation with radiculopathy. Spine J. 2014;14:180-91.

19. Genevay S, Atlas SJ, Katz JN. Variation in eligibility criteria from studies of radiculopathy due to a herniated disc and of neurogenic claudication due to lumbar spinal stenosis: a structured literature review. Spine. 2010;35:803-11.

20. van der Windt DAWM, Simons E, Riphagen II, et al. Physical examination for lumbar radiculopathy due to disc herniation in patients with low-back pain. Cochrane Database Syst Rev. 2010:CD007431.

21. Vroomen PC, de Krom MC, Wilmink JT, Kester AD, Knottnerus JA. Diagnostic value of history and physical examination in patients suspected of lumbosacral nerve root compression. J Neurol Neurosurg Psychiatry. 2002;72:630-4.

22. Deyo RA, Loeser JD, Bigos SJ. Herniated lumbar intervertebral disk. Ann Intern Med. 1990;112:598-603.

23. Weber H. The natural history of disc herniation and the influence of intervention. Spine. 1994;19:2234-8; discussion 3.

24. Koes BW, van Tulder MW, Peul WC. Diagnosis and treatment of sciatica. BMJ. 2007;334:1313-7.

25. Al Nezari NH, Schneiders AG, Hendrick PA. Neurological examination of the peripheral nervous system to diagnose lumbar spinal disc herniation with suspected radiculopathy: a systematic review and meta-analysis. Spine J. 2013;13:657-74.

26. Iversen T, Solberg TK, Romner B, et al. Accuracy of physical examination for chronic lumbar radiculopathy. BMC Musculoskelet Disord. 2013;14:206.

27. Heliovaara M, Impivaara O, Sievers K, et al. Lumbar disc syndrome in Finland. J Epidemiol Community Health. 1987;41:251-8.

28. Kaila-Kangas L, Leino-Arjas P, Karppinen J, Viikari-Juntura E, Nykyri E, Heliovaara M. History of physical work exposures and clinically diagnosed sciatica among working and nonworking Finns aged 30 to 64. Spine. 2009;34:964-9.

29. Heliovaara M. Epidemiology of sciatica and herniated lumbar intervertebral disc [thesis]. Helsinki, Finland, The Social Insurance Institution; 1988.

30. Carey TS, Garrett J, Jackman A, McLaughlin C, Fryer J, Smucker DR. The outcomes and costs of care for acute low back pain among patients seen by primary care practitioners, chiropractors, and orthopedic surgeons. The North Carolina Back Pain Project. N Engl J Med. 1995;333:913-7.

31. Haas M, Goldberg B, Aickin M, Ganger B, Attwood M. A practice-based study of patients with acute and chronic low back pain attending primary care and chiropractic physicians: two-week to 48-month follow-up. J Manipulative Physiol Ther. 2004;27:160-9.

32. Konstantinou K, Dunn KM. Sciatica: review of epidemiological studies and prevalence estimates. Spine. 2008;33:2464-72.

33. Heliovaara M, Knekt P, Aromaa A. Incidence and risk factors of herniated lumbar intervertebral disc or sciatica leading to hospitalization. J Chronic Dis. 1987;40:251-8.

18

34. Mattila VM, Sillanpää P, Visuri T, Pihlajamäki H. Incidence and trends of low back pain hospitalisation during military service--an analysis of 387,070 Finnish young males. BMC Musculoskelet Disord. 2009;10:10.

35. Heliovaara M. Occupation and risk of herniated lumbar intervertebral disc or sciatica leading to hospitalization. J Chronic Dis. 1987;40:259-64.

36. Kelsey JL, Githens PB, Walter SD, et al. An epidemiological study of acute prolapsed cervical intervertebral disc. J Bone Joint Surg Am. 1984;66:907-14.

37. Kelsey JL, Githens PB, White AA, III, et al. An epidemiologic study of lifting and twisting on the job and risk for acute prolapsed lumbar intervertebral disc. J Orthop Res. 1984;2:61-6.

38. Kelsey JL, Hardy RJ. Driving of motor vehicles as a risk factor for acute herniated lumbar intervertebral disc. Am J Epidemiol. 1975;102:63-73.

39. Leino-Arjas P, Kaila-Kangas L, Kauppinen T, Notkola V, Keskimaki I, Mutanen P. Occupational exposures and inpatient hospital care for lumbar intervertebral disc disorders among Finns. Am J Ind Med. 2004;46:513-20.

40. Saftic R, Grgic M, Ebling B, Splavski B. Case-control study of risk factors for lumbar intervertebral disc herniation in Croatian island populations. Croat Med J. 2006;47:593-600.

41. Seidler A, Bolm-Audorff U, Siol T, et al. Occupational risk factors for symptomatic lumbar disc herniation; a case-control study. Occup Environ Med. 2003;60:821-30.

42. Heliovaara M. Body height, obesity, and risk of herniated lumbar intervertebral disc. Spine. 1987;12:469-72.

43. Kelsey JL, Githens PB, O'Conner T, et al. Acute prolapsed lumbar intervertebral disc. An epidemiologic study with special reference to driving automobiles and cigarette smoking. Spine. 1984;9:608-13.

44. Kelsey JL, Greenberg RA, Hardy RJ, Johnson MF. Pregnancy and the syndrome of herniated lumbar intervertebral disc; an epidemiological study. Yale J Biol Med. 1975;48:361-8.

45. Jhawar BS, Fuchs CS, Colditz GA, Stampfer MJ. Cardiovascular risk factors for physician-diagnosed lumbar disc herniation. Spine J. 2006;6:684-91.

46. Côté P, Cassidy JD, Carroll L. The treatment of neck and low back pain: who seeks care? who goes where? Med Care. 2001;39:956-67.

47. Senstad O, Leboeuf-Yde C, Borchgrevink CF. Side-effects of chiropractic spinal manipulation: types frequency, discomfort and course. Scand J Prim Health Care. 1996;14:50-3.

48. Leboeuf-Yde C, Hennius B, Rudberg E, Leufvenmark P, Thunman M. Side effects of chiropractic treatment: a prospective study. J Manipulative Physiol Ther. 1997;20:511-5.

49. Hurwitz EL, Chiang LM. A comparative analysis of chiropractic and general practitioner patients in North America: findings from the joint Canada/United States Survey of Health, 2002-03. BMC Health Serv Res. 2006;6:49.

50. Hurwitz EL. Epidemiology: spinal manipulation utilization. Journal of electromyography and kinesiology. 2012;22:648-54.

51. Edwards IR, Aronson JK. Adverse drug reactions: definitions, diagnosis, and management. The Lancet. 2000;356:1255-9.

52. Roelofs PD, Deyo RA, Koes BW, Scholten RJ, van Tulder MW. Nonsteroidal anti-inflammatory drugs for low back pain: an updated Cochrane review. Spine. 2008;33:1766-74.

19

53. Ofman JJ, Maclean CH, Straus WL, et al. Meta-analysis of dyspepsia and nonsteroidal antiinflammatory drugs. Arthritis Care & Research. 2003;49:508-18.

54. Hernandez-Diaz S, Rodriguez LA. Association between nonsteroidal anti-inflammatory drugs and upper gastrointestinal tract bleeding/perforation: an overview of epidemiologic studies published in the 1990s. Arch Intern Med. 2000;160:2093-9.

55. Ofman JJ, MacLean CH, Straus WL, et al. A metaanalysis of severe upper gastrointestinal complications of nonsteroidal antiinflammatory drugs. J Rheumatol. 2002;29:804-12.

56. Fosbol EL, Gislason GH, Jacobsen S, et al. Risk of myocardial infarction and death associated with the use of nonsteroidal anti-inflammatory drugs (NSAIDs) among healthy individuals: a nationwide cohort study. Clin Pharmacol Ther. 2009;85:190-7.

57. Graham DJ, Campen D, Hui R, et al. Risk of acute myocardial infarction and sudden cardiac death in patients treated with cyclo-oxygenase 2 selective and non-selective non-steroidal anti-inflammatory drugs: nested case-control study. Lancet. 2005;365:475-81.

58. Kearney PM, Baigent C, Godwin J, Halls H, Emberson JR, Patrono C. Do selective cyclo-oxygenase-2 inhibitors and traditional non-steroidal anti-inflammatory drugs increase the risk of atherothrombosis? Meta-analysis of randomised trials. BMJ. 2006;332:1302-8.

59. Senstad O, Leboeuf-Yde C, Borchgrevink C. Frequency and characteristics of side effects of spinal manipulative therapy. Spine. 1997;22:435-40; discussion 40-1.

60. Rubinstein SM. Adverse events following chiropractic care for subjects with neck or low-back pain: do the benefits outweigh the risks? J Manipulative Physiol Ther. 2008;31:461-4.

61. Gouveia LO, Castanho P, Ferreira JJ. Safety of chiropractic interventions: a systematic review. Spine. 2009;34:E405-13.

62. Snelling NJ. Spinal manipulation in patients with disc herniation: A critical review of risk and benefit. Int J Osteopath Med. 2006;9:77-84.

63. Cherkin DC, Sherman KJ, Deyo RA, Shekelle PG. A review of the evidence for the effectiveness, safety, and cost of acupuncture, massage therapy, and spinal manipulation for back pain. Ann Intern Med. 2003;138:898-906.

64. Huang SL, Liu YX, Yuan GL, Zhang J, Yan HW. Characteristics of lumbar disc herniation with exacerbation of presentation due to spinal manipulative therapy. Medicine. 2015;94:e661.

65. Powell FC, Hanigan WC, Olivero WC. A risk/benefit analysis of spinal manipulation therapy for relief of lumbar or cervical pain. Neurosurgery. 1993;33:73-8; discussion 8-9.

66. Schmidley JW, Koch T. The noncerebrovascular complications of chiropractic manipulation. Neurology. 1984;34:684-5.

67. Morandi X, Riffaud L, Houedakor J, Amlashi SF, Brassier G, Gallien P. Caudal spinal cord ischemia after lumbar vertebral manipulation. Joint Bone Spine. 2004;71:334-7.

68. Tamburrelli FC, Genitiempo M, Logroscino CA. Cauda equina syndrome and spine manipulation: case report and review of the literature. Eur Spine J. 2011;20 Suppl 1:S128-31.

20

69. Oppenheim JS, Spitzer DE, Segal DH. Nonvascular complications following spinal manipulation. Spine J. 2005;5:660-6; discussion 6-7.

70. Rebain R, Baxter GD, McDonough S. The passive straight leg raising test in the diagnosis and treatment of lumbar disc herniation: a survey of United kingdom osteopathic opinion and clinical practice. Spine. 2003;28:1717-24.

71. Szklo M, Nieto J. Epidemiology: beyond the basics. 2nd ed. Sudbury, MA: Jones and Bartlett Publishers; 2006.

72. Boucher P, Robidoux S. Lumbar disc herniation and cauda equina syndrome following spinal manipulative therapy: a review of six court decisions in Canada. Journal of forensic and legal medicine. 2014;22:159-69.

73. Jagbandhansingh MP. Most common causes of chiropractic malpractice lawsuits. J Manipulative Physiol Ther. 1997;20:60-4.

74. Spiegelhalter DJ. Incorporating Bayesian ideas into health-care evaluation. Statistical Science. 2004;19:156-74.

75. Spiegelhalter DJ, Myles JP, Jones DR, Abrams KR. Methods in health service research. An introduction to bayesian methods in health technology assessment. BMJ. 1999;319:508-12.

76. Spiegelhalter D, Rice K. Bayesian statistics. Scholarpedia. 2009;4(8):5230. http://www.scholarpedia.org/article/Bayesian_statistics. Accessed 2015-May-15.

77. O'Hagan A, Buck CE, Daneshkhah A, et al. Uncertain judgements: eliciting experts' probabilities. West Sussex: John Wiley & Sons Ltd; 2006.

78. Lilford RJ, Thornton JG, Braunholtz D. Clinical trials and rare diseases: a way out of a conundrum. BMJ. 1995;311:1621-5.

79. Dunson DB. Commentary: practical advantages of Bayesian analysis of epidemiologic data. Am J Epidemiol. 2001;153:1222-6.

80. Hampson LV, Whitehead J, Eleftheriou D, Brogan P. Bayesian methods for the design and interpretation of clinical trials in very rare diseases. Stat Med. 2014;33:4186-201.

81. Hughes MD, Williams PL. Challenges in using observational studies to evaluate adverse effects of treatment. N Engl J Med. 2007;356:1705-7.

82. Smeeth L, Douglas I, Hubbard R. Commentary: we still need observational studies of drugs--they just need to be better. Int J Epidemiol. 2006;35:1310-1.

83. Maclure M, Fireman B, Nelson JC, et al. When should case-only designs be used for safety monitoring of medical products? Pharmacoepidemiol Drug Saf. 2012;21 Suppl 1:50-61.

84. Farrington CP, Nash J, Miller E. Case series analysis of adverse reactions to vaccines: a comparative evaluation. Am J Epidemiol. 1996;143:1165-73.

85. Whitaker HJ, Hocine MN, Farrington CP. The methodology of self-controlled case series studies. Stat Methods Med Res. 2009;18:7-26.

86. Maclure M. The case-crossover design: a method for studying transient effects on the risk of acute events. Am J Epidemiol. 1991;133:144-53.

87. Farrington P, Pugh S, Colville A, et al. A new method for active surveillance of adverse events from diphtheria/tetanus/pertussis and measles/mumps/rubella vaccines. Lancet. 1995;345:567-9.

88. Farrington CP, Whitaker HJ, Hocine MN. Case series analysis for censored, perturbed, or curtailed post-event exposures. Biostatistics. 2009;10:3-16.

89. Greenland S. Confounding and exposure trends in case-crossover and case-time-control designs. Epidemiology. 1996;7:231-9.

21

90. Vines SK, Farrington CP. Within-subject exposure dependency in case-crossover studies. Stat Med. 2001;20:3039-49.

91. Janes H, Sheppard L, Lumley T. Overlap bias in the case-crossover design, with application to air pollution exposures. Stat Med. 2005;24:285-300.

92. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. In: Higgins JPT, Green S, eds: The Cochrane Collaboration; 2011: http://www.cochrane-handbook.org. Accessed 2015-May-15.

22

Chapter 2

Incidence and Determinants Of Lumbar Disc Herniation With 2Radiculopathy In Adults: A Systematic Review

Manuscript 1

Hincapié CA, Cassidy JD, Côté P, Rampersaud YR, Boyle E, Tomlinson GA, Jadad AR.

Incidence and determinants of lumbar disc herniation with radiculopathy in adults: a

systematic review. In preparation for submission, 2015

Incidence and determinants of lumbar disc herniation with radiculopathy in

adults: a systematic review

Authors: Cesar A. Hincapié, DC, MHSc,1,2 J. David Cassidy, PhD, DrMedSc,1,2,3 Pierre

Côté, DC, PhD,4 Y. Raja Rampersaud, MD, MSc,5 Eleanor Boyle, PhD,1,3 George A.

Tomlinson, PhD,1,6 Alejandro R. Jadad, MD, DPhil1,6

Affiliations: 1 Dalla Lana School of Public Health, University of Toronto; 2 Toronto

Western Research Institute, University Health Network; 3 Department of Sports Science

and Clinical Biomechanics, University of Southern Denmark; 4 Faculty of Health

Sciences, University of Ontario Institute of Technology; 5 Division of Orthopaedic

Surgery, Toronto Western Hospital; 6 Toronto General Research Institute, University

Health Network

Corresponding Author: Cesar A. Hincapié, DC, MHSc, University Health Network,

LuCliff Place – 700 Bay Street, Suite 602, Toronto, Ontario M5G 1Z6, Canada. E-mail:

cesar.hincapie@utoronto.ca

Keywords: intervertebral disc displacement; radiculopathy; sciatica; low back pain;

epidemiology; incidence; risk factors; systematic review

Word Count: 6,146

23

Abstract

Background: Lumbar disc herniation (LDH) is the most common cause of lumbosacral

radiculopathy. Compared to nonspecific low back pain, LDH with radiculopathy is

associated with greater disability, healthcare use and intervention. However, little is

known about its epidemiology.

Purpose: To assemble and synthesize the best available evidence on the incidence

and determinants of LDH with radiculopathy in adults.

Study design: Systematic review.

Methods: Medline, Embase and 3 Cochrane Library databases were systematically

searched from 1970 to December 2010. Identified abstracts were screened for

relevance by two independent reviewers using pre-specified inclusion and exclusion

criteria. Eligible articles were critically appraised for quality and risk of bias by teams of

two independent reviewers using the SIGN criteria. Data from eligible studies were

abstracted into evidence tables.

Results: After 7,430 abstracts were screened, 35 studies were critically reviewed, and

26 (74%) were included as scientifically admissible. The annual incidence of

hospitalized or surgically managed LDH with radiculopathy in the general population

ranged from 0.2 to 1.3 per 1,000 persons. However, much LDH with radiculopathy is

not treated in hospitals. The annual incidence of clinically defined LDH with

radiculopathy varied from 6.2 per 1,000 persons (0.6%) among US female nurses, to 93

per 1,000 persons (9.3%) among forest industry workers in Finland. Nonmodifiable risk

factors included age and sex. Modifiable risk factors included lower education, higher

BMI and cardiovascular risk factors in women (diabetes, high cholesterol, hypertension,

and a family history of coronary heart disease), smoking, high cumulative occupational

lumbar load by forward bending (work postures with ≥20-degrees of forward bending or

extreme forward bending of >90-degrees) and manual materials handling, higher levels

of perceived risk of work injury, lower job control (independence and ability to influence

work methods and pace), regular or irregular three-shift work or regular night work in

women, and increased time pressure at work. We found preliminary evidence that

24

manual occupation, genetics, and previous back pain may be associated with LDH with

radiculopathy.

Conclusions: Although the literature is of varying quality and heterogeneous, there is

evidence that LDH with radiculopathy is an important source of pain and disability in

society. The annual incidence ranges roughly between 0.1% and 10%, depending on

case definition, and risk factors are varied. Our findings support the need to develop

standardized case and surveillance definitions that validly classify the clinical spectrum

of this condition. Future research should focus on prospective designs examining

potentially modifiable risk factors and prevention strategies.

2.1 Introduction

Low back pain is a leading cause of disability and associated with a large economic

burden on individuals, industry and society.1-5 Radiculopathy due to lumbar disc

herniation (LDH) is one of the most recognizable presentations of low back pain. The

diagnosis is typically based on a combination of symptoms and signs suggesting lumbar

spinal nerve root compression or irritation, such as nerve root tension signs, neurologic

deficits, and advanced imaging findings that correlate with the clinical syndrome.6-8

LDH, defined as the localized displacement of disc material beyond the margins of the

intervertebral disc space,9 is considered the most common cause of lumbosacral

radiculopathy.10-12 Compared with nonspecific low back pain without radiating leg pain,

LDH with radiculopathy is associated with greater pain, disability, healthcare use and

intervention.13-16 However, little is known about the epidemiology of this important

condition.

Previous studies of symptomatic LDH have reported a point prevalence of about 5%

among the general adult population,17,18 varying by sex and age. In people aged 25–55

years, 95% of symptomatic herniated discs occur at the lower lumbar spine (L4-L5 and

L5-S1 levels).19 Nonetheless, very little is known about the incidence of LDH with

radiculopathy, and consequently, risk factors are not well understood.

The objective of our systematic review was to synthesize the best available evidence on

the incidence and determinants of LDH with radiculopathy in adults. Our aim was to

25

create a baseline of the best scientific evidence to inform clinicians, researchers and

policymakers about the epidemiology of LDH with radiculopathy.

2.2 Methods

Our review was conducted in accordance with the Preferred Reporting Items for

Systematic Reviews and Meta-Analyses (PRISMA)20,21 statement. Details of our

protocol were registered on the Prospective Register of Systematic Reviews

(PROSPERO) (CRD42011001197).22

Data Sources and Searches: We systematically searched Medline, Embase,

Cochrane Central Trials Registry, Cochrane Database of Systematic Reviews, and

Database of Abstracts Related to Effects, from 1970 to December 2010. Search

strategies (Appendix C) combined terms from three key concepts: intervertebral disc

herniation, lumbar spine, and etiology, and were developed in consultation with an

experienced Information Specialist. The reference lists of all articles eligible were

examined for additional relevant studies.

Study Selection: A three-phase screening process was used to identify eligible articles

for our review on the basis of pre-specified inclusion and exclusion criteria. Eligible

were published reports in English, French or Spanish that used cohort, case-control, or

randomized controlled trial designs that examined the incidence and/or determinants

(i.e., risk factors) of LDH with radiculopathy in adults. We excluded studies of

asymptomatic LDH, or LDH due to neoplasm, infection, fracture, dislocation or spinal

cord injury. Narrative reviews, case reports or series, cross-sectional, cadaveric or

animal studies were ineligible due to their inherent limitations for estimating incidence

and inferring etiology. After consulting clinical experts, LDH with radiculopathy was

defined as lumbosacral radiculopathy, sciatica and/or clinically relevant neurologic

deficit, with or without advanced imaging (i.e., MRI or CT) confirmation of disc

herniation.

In the first-level screen, one reviewer (CAH) rated all the citations from the search as

relevant, unsure, or irrelevant. In the second-level screen, three reviewers (CAH, JDC,

PC) independently identified articles meeting the review inclusion and exclusion criteria

among those citations rated as relevant or unsure from the first-level screen (n=282). In

26

the third-level screen, each article considered possibly relevant to the review was

retrieved and underwent full-text analysis for eligibility by two independent reviewers

(CAH/JDC or CAH/PC). When necessary, discrepancies were resolved by consensus

and arbitration by a third reviewer.

Risk of Bias Assessment and Data Extraction: Eligible articles were assessed for

risk of bias by teams of 2 independent reviewers (CAH reviewed all articles; other

reviewers were JDC, PC, ARJ, YRR, and EB) and using checklists based on criteria

recommended by the Scottish Intercollegiate Guidelines Network (SIGN).23 The SIGN

checklists include key questions/items that focus on aspects of study design that have

been shown to influence the validity of the results reported (e.g., selection of subjects,

assessment of exposures and outcomes, confounding assessment), and have been

evaluated and adapted to meet a balance between methodological rigour and

practicality of use.23 An overall assessment of risk of bias was added and rated as low,

moderate or high. When further information was required from original authors, three

attempts were made to contact authors by email. We used consensus and arbitration

by a third reviewer, when necessary, to resolve discrepancies on risk of bias

assessment.

One reviewer (CAH) extracted data from eligible studies into evidence tables relating to

incidence and risk factors of LDH with radiculopathy in adults. We extracted data on

study and population characteristics, case definitions and study outcomes, risk factors

considered, and estimates of incidence and risk.

Data Synthesis and Analysis: We synthesized the literature according to the

principles of best evidence synthesis.24 In best evidence synthesis, scientifically

admissible studies are qualitatively synthesized, with more weight given to evidence

from studies judged to be least vulnerable to bias.25,26 The evidence tables summarize

our findings and form the basis of our recommendations. We defined low and moderate

risk of bias studies as scientifically admissible, but also tabled high risk of bias studies in

order to examine their findings and how they differ from admissible studies.

To better delineate the strength of the evidence, risk factor evidence was categorized as

phase I, II or III.25,27,28 This categorization distinguishes between 3 hierarchic phases of

research aimed at determining causal relationships between a risk factor and a health

27

outcome. Phase I studies describe crude associations between potential determinants

and the outcome of interest. Phase II studies are exploratory analyses that focus on

particular sets of risk factors or attempt to discover which risk factors predict the

development of a health outcome without an explicit attempt to control for confounding.

Phase III investigations are confirmatory studies of explicitly pre-stated hypotheses that

allow for the quantification of the strength, direction, and independence of the proposed

relationship between a risk factor and the development of LDH with radiculopathy. Both

Phase I and II studies provide preliminary evidence of a potential risk factor’s

association with the outcome, whereas Phase III studies provide confirmatory evidence.

Our evidence tables report annual incidence and risk factor effect estimates with 95%

confidence intervals (CIs) as reported in original studies, or calculated from raw data

presented in original studies. Risk factor effect estimates are reported as odds ratios

(ORs), risk ratios (RRs), or hazard rate ratios (HRRs). When necessary, we computed

these statistics using R, version 2.15.1.29 Confidence intervals were calculated using

standard methods.30

Role of the Funding Source: The sponsors of the study had no role in study design,

data collection, data analysis, data interpretation, the writing of the report, or in the

decision to submit the report for publication.

2.3 Results

Search, Selection and Critical Appraisal: Our literature search yielded 8,617

citations, of which 95 articles were deemed potentially eligible after removing duplicates

and screening for relevance; these underwent full-text analysis for eligibility (Figure

2.1). Of these, 46 articles (representing 35 studies) met our eligibility criteria and were

reviewed. After risk of bias assessment and critical appraisal, we rated 9 studies (26%)

as inadmissible (high risk of bias) and 26 studies (74%) as admissible (low or moderate

risk of bias). These 26 admissible studies (represented by 37 articles) are the basis for

our best evidence synthesis.

28

Figure 2.1. Flow diagram of information through phases of the systematic review.

Abbreviations: LDH, lumbar disc herniation; ROB, risk of bias.

Records identified through database searching

(n = 8,559) S

cree

ning

In

clud

ed

Elig

ibili

ty

Iden

tific

atio

n Additional records identified through other sources

(n = 58)

Records after duplicates removed (n = 7,430)

Records screened (n = 7,430)

Records excluded (n = 7,335)

Full-text articles assessed for eligibility

(n = 95)

Full-text articles excluded (n = 49)

No symptomatic LDH (n=20)

Ineligible design (n=20) Not on incidence or risk (n=5) Wrong publication type (n=2)

Ineligible population (n=1) No estimates (n=1)

Articles eligible for review (n = 46) (representing 35

studies)

Low ROB (n=10) Moderate ROB (n=27)

High ROB (n=9)

Admissible studies (n=26) (represented by 37 articles)

Low ROB studies (n=8) Moderate ROB studies (n=18)

Incidence studies (n=15) Risk factor studies (n=22)

Inadmissible (high ROB) studies (n=9)

Incidence studies (n=2) Risk factor studies (n=9)

29

Study Characteristics: Table 2.1 presents a summary of characteristics of the 35

eligible studies, and Table 2.2 provides summary details of the eligible studies. Overall,

8 studies (23%) were rated as having a low risk of bias; 18 studies (51%), a moderate

risk of bias; and, 9 studies (26%), a high risk of bias. The most common sources of bias

included poor or inadequate description of sampling methods, exposure measurement,

and consideration of confounding. Appendix D, Table D.1 provides full descriptions of

the 17 eligible studies (15 admissible, 2 inadmissible) that reported on incidence.31-48

Appendix D, Table D.2 details the 31 eligible studies (22 admissible, 9 inadmissible)

that reported on risk factors.31,33,34,36,37,39,40,42-76

Table 2.1. Characteristics of eligible studies examining the incidence of and/or risk factors for lumbar

disc herniation with radiculopathy in adults

By admissibility*

Admissible studies by

focus†

Characteristic

All Studies

(n=35)

Admissible

(n=26)

Inadmissible

(n=9)

Incidence

(n=15) Risk (n=22)

Risk of bias – n (%)

Low 8 (23) 8 (31) na 8 (53) 5 (23)

Moderate 18 (51) 18 (69) na 7 (47) 17 (77)

High 9 (26) na 9 (100) na na

Focus – n (%)

Incidence only 4 (11) 4 (15) 0 4 (27) na

Risk only 18 (51) 11 (42) 7 (78) na 11 (50)

Both incidence and risk 13 (37) 11 (42) 2 (22) 11 (73) 11 (50)

Study design – n (%)

Cohort 18 (51) 15 (58) 3 (33) 15 (100) 11 (50)

Case-control 17 (49) 11 (42) 6 (67) 0 11 (50)

Case definition type – n (%)

Surgical or hospital 16 (46) 10 (38) 6 (67) 9 (60) 7 (32)

Clinical 15 (43) 13 (50) 2 (22) 6 (40) 12 (55)

Mixed surgical/hospital/clinical 4 (11) 3 (12) 1 (11) 0 3 (14)

30

By admissibility*

Admissible studies by

focus†

Characteristic

All Studies

(n=35)

Admissible

(n=26)

Inadmissible

(n=9)

Incidence

(n=15) Risk (n=22)

Population source – n (%)

General 6 (17) 6 (23) 0 6 (40) 4 (18)

Occupational 12 (34) 8 (31) 4 (44) 8 (53) 7 (32)

Health care 17 (49) 12 (46) 5 (56) 1 (7) 11 (50)

Sex – n (%)

Men only 9 (26) 6 (23) 3 (33) 4 (27) 5 (23)

Women only 2 (6) 2 (8) 0 2 (13) 2 (9)

Both men and women 24 (69) 18 (69) 6 (67) 9 (60) 15 (68)

Incidence estimate type‡ – n (%) (n=17) (n=15) (n=2) (n=15)

Cumulative incidence only 11 (65) 11 (73) 0 11 (73) na

Incidence rate only 5 (29) 4 (27) 1 (50) 4 (27) na

Both estimate types 1 (6) 0 1 (50) 0 na

Risk factors considered§ – n (%) (n=31) (n=22) (n=9) (n=22)

Sociodemographic 24 (77) 19 (86) 5 (56) na 19 (86)

General health, pain, comorb 17 (55) 11 (50) 6 (67) na 11 (50)

Health behaviours 15 (48) 12 (55) 3 (33) na 12 (55)

Personal – Physical 9 (29) 6 (27) 3 (33) na 6 (27)

Personal – Psychosocial 7 (23) 7 (32) 0 na 7 (32)

Workplace – Physical 15 (48) 11 (50) 4 (44) na 11 (50)

Workplace – Psychosocial 6 (19) 5 (23) 1 (11) na 5 (23)

Phase of evidence|| – n (%) (n=31) (n=22) (n=9) (n=22)

I 12 (39) 7 (32) 5 (56) na 7 (32)

II 15 (48) 11 (50) 4 (44) na 11 (50)

III 2 (6) 2 (9) 0 na 2 (9)

Mixed (Phase II and III) 2 (6) 2 (9) 0 na 2 (9)

31

Abbreviation: comorb, comorbidities; na, not applicable

* Admissible studies were those rated has having a low or moderate risk of bias; inadmissible studies

were those rated as having a high risk of bias

† Categories are not mutually exclusive

‡ Proportions among studies on incidence; denominators indicated in header row

§ Proportions among studies on risk factors and not mutually exclusive; denominators indicated in

header row

|| Hierarchical categories of risk factor evidence: phase I and II is preliminary evidence for an

association, phase III is confirmatory evidence from better quality studies

The 26 scientifically admissible studies consisted of 15 cohort studies,31-46,49,50,60 and 11

case-control studies.51-59,61-69 Most of the admissible studies were conducted in Finland

(46%) and the US (23%). Ten studies (38%) used case definitions based either on

surgery or hospital treatment, 13 (50%) used definitions based on clinical signs and

symptoms, and 3 (12%) used mixed case definitions combining clinical, hospital and

surgical definitions.

32

Table 2.2. Summary of eligible studies examining the incidence of and/or risk factors for lumbar disc herniation with radiculopathy in adults

First author,

Year

published

Focus Country

Study

design

Source

population

Participants and setting

Follow-up

Age

range

(mean)

N (% female);

Participation

% Study outcome

Case

definition

type ROB Phase

Heliövaara,

1987 31,49,50

Both*

Finland Cohort

with

case-

control

risk

analyses

General Nationwide cohort linked to

the National Hospital

Discharge Register

Follow-up: 11y

≥15y

(NR)

57,000 (48%);

83%

Hospitalized LDH

or sciatica

Hospital Low II

Burske-

Hohlfeld,

1990 32

Incidence

US Cohort General All LDH surgeries of

Olmsted County residents

Follow-up: 30y

15-78y

(42)

1,028 (NR);

88% incident

surgery

LDH surgery Surgical Low NA

Zitting, 1998 33

Both

Finland Cohort General Birth cohort from 2

provinces linked to the

National Hospital

Discharge Register

Follow-up: 28y

15-28y

among

cases

(NR)

12,058 (NR);

92%

Hospitalized LDH Hospital Low II

Miranda, 2002 34

Both

Finland Cohort Occupational Forest industry workers

responding to a

questionnaire

Follow-up: 1y

NR (45) 2,077 (26%);

77%

Sciatica Clinical Low II

33

First author,

Year

published

Focus Country

Study

design

Source

population

Participants and setting

Follow-up

Age

range

(mean)

N (% female);

Participation

% Study outcome

Case

definition

type ROB Phase

Jarvik, 2005 35

Incidence

US Cohort Health care Veterans Affairs outpatients

at the Puget Sound Health

Care System, Seattle

Division

Follow-up: 3y

35-70y

(median

, 53)

148 (13%);

89%

Clinical LDH Clinical Low NA

Jhawar, 2006 36

Both

US Cohort Occupational Female nurses from the

Nurses’ Health Study

responding to a

questionnaire

Follow-up: 16y

30-55y in

1976

(NR)

98,407

(100%);

≥85%

Clinical LDH Clinical Low III

Mattila, 2008 37

Both

Finland Cohort General Nationwide adolescent

cohort linked to the

National Hospital

Discharge Register

Follow-up: 651,027 person-

years

15-41y

among

cases

(27 at

surgery

)

57,408 (54%);

79%

LDH surgery Surgical Low II

34

First author,

Year

published

Focus Country

Study

design

Source

population

Participants and setting

Follow-up

Age

range

(mean)

N (% female);

Participation

% Study outcome

Case

definition

type ROB Phase

Mattila, 2009 38

Incidence

Finland Cohort Occupational All male military conscripts

linked to the National

Hospital Discharge

Register

Follow-up: 267,700 person-

years

18-29y

(20)

387,070 (0%);

100%

Hospitalized LDH Hospital Low NA

Kelsey, 1975 51-55

Risk

US Case-

control

Health care Cases: 223 radiology and

hospital patients with

LDH, 1971-1973

Controls: 217 controls

matched on age, sex, and

medical setting; 494

unmatched controls

Follow-up: NA

20-64y

(39

among

cases)

934 (41%);

78%

Combined

clinical,

hospitalized

and surgical

LDH

Clinical,

hospital

and

surgical

Mod II

Hurme, 1983 39

Both

Finland Cohort General AII LDH surgeries from

surgical registers of the

Turku University Central

Hospital area

Follow-up: 5y

15-80y

among

cases

(42 at

surgery

)

1,011

surgeries

(44%);

79% incident

surgery

LDH surgery Surgical Mod I

35

First author,

Year

published

Focus Country

Study

design

Source

population

Participants and setting

Follow-up

Age

range

(mean)

N (% female);

Participation

% Study outcome

Case

definition

type ROB Phase

Kelsey, 1984 56,57

Risk

US Case-

control

Health care Cases: 325 orthopaedic,

neurosurgical, and

hospital patients with

LDH, 1979-1981

Controls: 241 controls

matched on age, sex, and

medical setting.

Follow-up: NA

20-64y

(NR)

566 (45%);

72-79%

Combined

clinical,

hospitalized

and surgical

LDH

Clinical,

hospital

and

surgical

Mod II

Riihimäki,

1989 40

Both

Finland Cohort Occupational Male concrete

reinforcement workers

and house painters

responding to a

questionnaire

Follow-up: 5y

25-54y at

baselin

e (NR)

178 (0%);

77-80%

Sciatica Clinical Mod II

Heikkilä, 1989 41

Incidence

Finland Cohort General Finnish adult twin cohort of

the same sex linked to the

National Hospital

Discharge Register

Follow-up: 14y

24-60y+

(NR)

18,730 (55%);

100%

Hospitalized

sciatica

Hospital Mod NA

36

First author,

Year

published

Focus Country

Study

design

Source

population

Participants and setting

Follow-up

Age

range

(mean)

N (% female);

Participation

% Study outcome

Case

definition

type ROB Phase

Mundt, 1993 58,59

Risk

US Case-

control

Health care Cases: 297 orthopaedic,

neurosurgical, and

hospital patients with

LDH, 1986-1988

Controls: 287 clinical and

hospital controls

Follow-up: NA

20-64y

(NR)

585 (41%);

76-79%

Combined

clinical,

hospitalized

and surgical

LDH

Clinical,

hospital

and

surgical

Mod II

Jørgensen,

1994 42

Both

Denmark Cohort Mixed Assistant nurses and all

females linked to the

Danish National Registry

of Hospitalized Patients

Follow-up: 1y

20-69y

(NR)

1,681,152

(100%);

100%

LDH surgery Surgical Mod II

Riihimäki,

1994 43;

Pietri-Taleb,

1995 60

Both

Finland Cohort Occupational Male machine operators,

carpenters and office

workers responding to a

questionnaire

Follow-up: 3y

25-49y

(37)

1,149 (0%);

83%

Sciatica Clinical Mod II

37

First author,

Year

published

Focus Country

Study

design

Source

population

Participants and setting

Follow-up

Age

range

(mean)

N (% female);

Participation

% Study outcome

Case

definition

type ROB Phase

a Leino-Arjas,

2002 44 b Leino-Arjas,

2004 45

Both

Finland Cohort Occupational Nationwide workforce linked

to the National Hospital

Discharge Register

Follow-up: 1y

a 20-64y

(NR) b 25-64y

(NR)

a 2,409,319

(NR);

100% b 1,783,616

(51%);

100%

Hospitalized LDH

and LDH

surgery

Hospital

and

surgical

Mod a II b III

Leclerc, 2003 46

Both

France Cohort Occupational Male workers in the national

electricity and gas

company responding to a

questionnaire

Follow-up: 2y

40-50y at

baselin

e (NR)

841 (0%);

65%

Sciatica Clinical Mod II

Siedler, 2003 61

Risk

Germany Case-

control

Health care Cases: 225 male patients

with acute LDH from

Frankfurt/Main

Controls: 107 population

and 90 hospital controls

Follow-up: NA

25-65y

(42)

437 (0%);

66-93%

Clinical LDH Clinical Mod III

38

First author,

Year

published

Focus Country

Study

design

Source

population

Participants and setting

Follow-up

Age

range

(mean)

N (% female);

Participation

% Study outcome

Case

definition

type ROB Phase

Noponen-

Hietela,

2005 62

Risk

Finland Case-

control

Health care Cases: 155 unrelated

Finnish patients with

sciatica from the Oulu

University Hospital area,

1997-1998

Controls: 179 unrelated

University of Oulu

employees and students

(all Finnish)

Cases:

19-78y

(44)

Controls:

20-69y

(39)

334 (39%

among

cases, 69%

among

controls);

NR

Sciatica Clinical Mod I

Mio, 2007 63

Risk

Japan Case-

control

Health care Cases: 823 hospital patients

of Japanese origin with

LDH

Controls: 841 hospital

controls of Japanese

origin

Follow-up: NA

Cases:

11-83y

(36)

Controls:

13-87y

(61)

1,664 (41%

among

cases, 63%

among

controls;

NR

Clinical LDH Clinical Mod I

39

First author,

Year

published

Focus Country

Study

design

Source

population

Participants and setting

Follow-up

Age

range

(mean)

N (% female);

Participation

% Study outcome

Case

definition

type ROB Phase

Virtanen, 2007 64

Risk

Finland Case-

control

Health care Cases: 243 unrelated

Finnish patients with

sciatica from the Oulu

University Hospital area

Controls: 259 unrelated

Finnish persons from the

same catchment area

Follow-up: NA

NR (NR) 502 (45%);

NR

Sciatica Clinical Mod I

Hirose, 2008 65

Risk

Japan Case-

control

Health care Cases: 847 hospital patients

of Japanese origin with

LDH, 2001-2007

Controls: 896 Japanese

persons from the same

catchment area

Follow-up: NA

Cases:

NR (39)

Controls:

NR (62)

1,743 (38%);

NR

Clinical LDH Clinical Mod I

40

First author,

Year

published

Focus Country

Study

design

Source

population

Participants and setting

Follow-up

Age

range

(mean)

N (% female);

Participation

% Study outcome

Case

definition

type ROB Phase

Karasugi,

2009 66

Risk

Japan

and

Finland

Case-

control

Health care Japan

Cases: 862 hospital patients

of Japanese origin with

LDH, 2001-2007

Controls: 896 Japanese

persons from the same

catchment area

Finland

Cases: 257 unrelated

Finnish patients with

sciatica from the Oulu

University Hospital area

Controls: 249 unrelated

Finnish persons from the

same area

Follow-up: NA

Japan

Cases:

NR (39)

Controls:

NR (62)

Finland

NR (NR)

Japan

1,758 (38%);

NR

Finland

506 (NR);

NR

Clinical LDH and

sciatica

Clinical Mod I

41

First author,

Year

published

Focus Country

Study

design

Source

population

Participants and setting

Follow-up

Age

range

(mean)

N (% female);

Participation

% Study outcome

Case

definition

type ROB Phase

a Siedler, 2009 67

b Schumann,

2010 68

Risk

Germany Case-

control

Health care Cases: 564 patients with

hospital treatment for LDH

pain in 4 regions of

Germany

Controls: 901 population

controls

Follow-up: NA

25-70y

(48)

1,816 (50%);

53-66%

LDH hospital

treatment

Hospital Mod a III b II

Cong, 2010 69

Risk

China Case-

control

Health care Cases: 70 male hospital

patients of Chinese Han

origin with LDH

Controls: 14 male spinal

trauma controls and 113

male healthy blood donor

controls

Follow-up: NA

Cases:

14-41y

(33)

Controls:

20-49y

(38)

197 (0%);

NR

Clinical LDH Clinical Mod I

42

First author,

Year

published

Focus Country

Study

design

Source

population

Participants and setting

Follow-up

Age

range

(mean)

N (% female);

Participation

% Study outcome

Case

definition

type ROB Phase

Hrubec, 1975 70

Risk

US Case-

control

Occupational Cases: 1,132 first admission

records to Army hospitals

for LDH, 1944-1945

Controls: 1,095 records of

Army National Service Life

Insurance policyholders

matched on age and

period of military service

Follow-up: NA

18-56y

(NR)

1,095 case-

control pairs

(0%);

97%

Hospitalized LDH Hospital High II

Bongers, 1988 47

Both

Holland Cohort Occupational Disability pension due to

LDH in male crane

workers and floor workers

at a steel company

Follow-up: 10y

<25-60y+

(NR)

1,405 (0%);

71%

Disability

pension due to

LDH

Clinical High II

Netterstrøm,

1989 48

Both

Denmark Cohort Occupational All full-time male bus drivers

employed by 3 urban bus

companies linked to the

Danish National Patient

Register

Follow-up: 7y

20-69y

(NR)

2,465 (0%);

100%

Hospitalized LDH Hospital High I

43

First author,

Year

published

Focus Country

Study

design

Source

population

Participants and setting

Follow-up

Age

range

(mean)

N (% female);

Participation

% Study outcome

Case

definition

type ROB Phase

An, 1994 71

Risk

US Case-

control

Health care Cases: 163 consecutive

surgical patients with

LDH, 1987-1988

Controls: 205 inpatient

controls matched on sex

and age

Follow-up: NA

16-78y

(45)

368 (39%);

NR

LDH surgery Surgical High I

Chibnall, 2006 72

Risk

US Cohort Occupational African American and non-

Hispanic white workers’

compensation claimants

who filed low back injury

claims

Follow-up: NR

18-55y

(NR)

2,934 (38%);

50%

Clinical and

surgical LDH

Clinical

and

surgical

High II

44

First author,

Year

published

Focus Country

Study

design

Source

population

Participants and setting

Follow-up

Age

range

(mean)

N (% female);

Participation

% Study outcome

Case

definition

type ROB Phase

Saftic, 2006 73

Risk

Croatia Case-

control

Health care Cases: 67 adults from 9

villages on the Croatian

islands with a history of

LDH surgery

Controls: 268 adults

matched on age, gender,

and village of

residence/immigrant

status

Follow-up: NA

≥18y

(NR)

365 (NR); NR LDH surgery Surgical High I

Lee, 2006 74

Risk

Korea Case-

control

Health care Cases: 119 LDH levels in

111 adult patients who

underwent LDH surgery,

2000-2002

Controls: 82 normal disc

levels adjacent to the

herniated levels in the

same patients

Follow-up: NA

40-49y

(NR)

201 adult disc

levels (37%);

NR

LDH surgery Surgical High I

45

First author,

Year

published

Focus Country

Study

design

Source

population

Participants and setting

Follow-up

Age

range

(mean)

N (% female);

Participation

% Study outcome

Case

definition

type ROB Phase

Kunakornsawa

t, 2007 75

Risk

Thailand Case-

control

Health care Cases: 34 LDH levels in 34

adult patients who

underwent LDH surgery,

2001-2003

Controls: 34 normal disc

levels adjacent to the

herniated levels in the

same patients

Follow-up: NA

23-45y

(34)

68 disc levels

(35%);

NR

LDH surgery Surgical High I

Zhang, 2009 76

Risk

China Case-

control

Health care Cases: 2,010 orthopaedic

hospital patients with

LDH, 2005-2007

Controls: 2,170 randomly

selected hospital controls

matched on race, gender,

age, and living area

Follow-up: NA

<30-50y+

(46)

4,180 (40%);

NR

Clinical LDH Clinical High II

Abbreviations: LDH, lumbar disc herniation; Mod, Moderate; N, study size; NA, not applicable; NR, not reported; ROB, risk of bias; y, years

* Study provides data on both incidence and risk factors

46

Incidence: We accepted 15 cohort studies (represented by 16 articles) on the incidence

of LDH with radiculopathy in adults (summarized in Tables 2.1 and 2.2, and fully

detailed in Appendix D, Table D.1).

Ten of the 15 studies (67%) were conducted in Finland,31,33,34,37-41,43-45 3 (20%) in the

United States,32,35,36 and one each in Denmark42 and France.46

Source populations varied. Six studies estimated the incidence in general

populations,31-33,37,39,41 eight studies examined occupational populations,34,36,38,40,42-46

and one study investigated incidence in a health care population.35 The majority of

studies included both genders and a broad age range.

The incidence of LDH with radiculopathy varied by case definition. Among six general

population studies using a surgical or hospital-based case definition, the annual

incidence ranged from 0.2 per 1,000 persons to 1.3 per 1,000 persons (Appendix D,

Table D.1).31-33,37,39,41 In two studies investigating occupational populations and using a

case definition based on surgery or hospitalization, the annual incidence of LDH with

radiculopathy ranged between 1.0 per 1,000 persons and 2.5 per 1,000 persons.42,44,45

One study from Finland found a higher incidence rate of 7.8 per 1,000 person-years of

hospitalized LDH in male military conscripts performing compulsory service between

1990 and 2002.38

Studies using clinical signs and symptoms as the basis for case definition of LDH with

radiculopathy found higher estimates of incidence, suggesting that much symptomatic

LDH is not treated in hospitals. Among female nurses from the US-based Nurses’

Health Study, the annual incidence of clinical LDH with radiculopathy was estimated at

6.2 per 1,000 person-years.36 Five other cohort studies described the incidence of

clinical LDH with radiculopathy in electrical and gas workers,46 Veterans Affairs

outpatients,35 house painters and concrete workers,40 forest industry workers,34 and

office workers, machine operators, and carpenters.43 These found annual incidence

estimates ranging from a low of 28 per 1,000 persons (2.8%) among electrical and gas

workers in France,46 to a high of 93 per 1,000 persons (9.3%) among forest industry

workers in Finland34 (Appendix D, Table D.1).

47

Given the heterogeneity in inception periods, diverse source populations, varied case

definitions for LDH with radiculopathy, and the different methods used to report

incidence in the admissible studies, the variability in incidence estimates presented in

Appendix D, Table D.1 is not surprising, and an overall summary incidence estimate

would not be very informative.

Risk Factors: Our best evidence synthesis included 11 cohort studies (represented by

15 articles) and 11 case-control studies (represented by 18 publications) that

investigated the risk factors for LDH with radiculopathy in adults (summarized in Tables

2.1 and 2.2, and fully detailed in Appendix D, Table D.2).

Ten of the 22 admissible studies (45%) were conducted in Finland,31,33,34,37,39,40,43-

45,49,50,60,62,64 4 (18%) in the US,36,51-59 3 (14%) in Japan,63,65,66 2 (9%) in Germany,61,67,68

and 1 each in China,69 Denmark,42 and France.46

Source populations varied, with 4 studies investigating risk factors in general

populations;31,33,37,39,49,50 7 studies examining occupational populations,34,36,40,42-46,60 and

11 studies focusing on health care populations.51-59,61-69

We present our findings prioritized by the hierarchical phases of evidence as described

in the Methods section, with Phase III studies presenting the strongest evidence of

association, and preliminary evidence provided by Phase II and I studies, in descending

order of strength.

Sociodemographic Characteristics:

Age. Evidence from one Phase III,45 and four Phase II studies31,34,42,44 indicates that

the incidence of LDH with radiculopathy increases with age, peaking about the fourth

and fifth decades of life, and then decreases in later life. Two Phase II studies reported

no association between age and the incidence of clinical LDH (sciatica) among male

worker cohorts.40,46

Gender. We found evidence of LDH with radiculopathy occurring more commonly in

men than in women. One Phase III45 study and 6 Phase II studies31,33,37,39,44,51,52

reported an increased risk for men. However, one Phase II study34 found no

48

association between gender and the development of clinical LDH with radiculopathy

(sciatica) in Finnish forestry workers.

Education. Evidence from one Phase III45 and one Phase II44 study suggests that the

incidence of LDH with radiculopathy decreases with years of education. Two Phase II

studies43,57 found no association between education and the risk of LDH.

Income. The evidence linking income and the risk of hospitalized LDH varies in one

Phase III and one Phase II study. In their Phase III analysis, Leino-Arjas and

colleagues found an increased incidence of hospital care due to LDH across lower

quintiles of personal net income compared to the highest quintile.45 However, in their

Phase II analysis, they reported a greater risk of hospitalized LDH among those with the

highest quartile of personal net income compared to those with the lowest quartile.44

Socioeconomic status. The evidence varies in three Phase II studies investigating

socioeconomic status and the incidence of LDH with radiculopathy. One study found a

higher risk among middle class men.31 Another study reported some indication of an

association between higher social class and LDH in women, but no association in

men.51,52 A third study found no link between socioeconomic background and the

incidence of LDH surgery in men or women.37

Race. Preliminary evidence from two Phase II studies suggests that race is not

associated with the incidence of LDH with radiculopathy.52,53,57

Occupation. We found preliminary evidence that manual occupations are associated

with the risk of LDH with radiculopathy. The studies reported in the Incidence section

above and in Appendix D, Table D.1 show that the incidence varies across

occupations. In addition, Phase II studies reported a greater risk for LDH with

radiculopathy in male concrete workers compared with male house painters;40 in female

assistant nurses compared with all Danish females;42 in machine operators and

carpenters compared with office workers;43 and in manual occupations compared with

upper white-collar occupations.44 However, a French Phase II study by Leclerc et al.

found no significant difference in the risk of developing clinical LDH with radiculopathy

among male managers, supervising employees and operating employees in a national

electricity and gas company.46

49

Body Mass Index, Anthropometrics and Genetics:

Body mass index and obesity. The association between obesity and the risk of LDH

with radiculopathy seems to differ in women and men. In women, evidence from two

Phase III studies36,45 and two Phase II studies37,68 indicates that the incidence of LDH

with radiculopathy increases with obesity. Another Phase II study found no association

in women.49 In men, evidence from one Phase III study45 and two Phase II studies37,46

suggests no association between body mass index and the risk of LDH with

radiculopathy. However, three Phase II studies found a positive association in

men.40,49,68 In addition, three Phase II studies that did not report sex-specific results

found no difference in the incidence of LDH with radiculopathy on the basis of obesity or

body mass index.34,52,57

Height. The evidence linking height and the risk of LDH with radiculopathy varies in 6

Phase II studies. Two studies reported a positive association. Heliövaara and

colleagues found that height (≥180 cm in men and ≥170 cm in women) was associated

with an increased risk for LDH with radiculopathy in Finnish men and women,49 while

Leclerc et al. reported a greater risk for LDH in French male electricity and gas

company workers >180 cm in height.46 However, three Phase II studies reported no

association between height and the incidence of LDH.34,40,57 Finally, one Phase II US

study of radiology and hospitalized LDH patients reported some indication of a positive

association in women, but a negative association in men.52

Genetics. We found preliminary evidence of a genetic predisposition for LDH with

radiculopathy. Five Phase I studies reported an increased risk of LDH for

polymorphisms and sequence variations (mutations) involving inflammatory mediator

genes;62 cartilage collagen genes;63 intervertebral disc extracellular matrix protein

genes;65 the human sickle tail gene;66 and, the cartilage intermediate layer protein

gene.64 One Phase I study in Chinese Han men found an increased risk of LDH for two

allele polymorphisms (A21 and A25 alleles) of the aggrecan gene, and a decreased risk

for a third allele polymorphism (A29 allele) of the aggrecan gene.69

50

General Health, Prior Pain and Comorbidities:

Self-rated health status. The evidence varies in two Phase II studies examining self-

rated health status and the risk of LDH with radiculopathy. In their study of French male

electricity and gas workers, Leclerc et al. found that those who reported poor general

health were more likely to develop clinical LDH with radiculopathy (sciatica).46 In

Finland, Mattila et al. found no association between perceived health status and the

incidence of LDH surgery.37

Muscular strength. Preliminary evidence from one Phase II study suggests that back

and abdominal musculature strength is not associated with the incidence of LDH with

radiculopathy.40

History of musculoskeletal symptoms and injuries. We found preliminary evidence

that a history of back pain increases the risk of LDH with radiculopathy. Three Phase II

studies reported that those with a history of back symptoms or low back pain were at

greater risk of developing clinical LDH (sciatica) than those without a history of back

pain.40,43,46 Three Phase II studies reported no association between a history of back

accidents or low back injuries and clinical LDH with radiculopathy (sciatica).34,40,43 In

addition, in their study of French male electricity and gas workers, Leclerc et al. found

no association between the presence of neck pain and the incidence of sciatica.46

Cardiovascular risk factors. There is evidence from one Phase III study that

cardiovascular risk factors, including diabetes, high cholesterol, hypertension, and a

family history of coronary heart disease, are associated with an increased risk of LDH

with radiculopathy in women.36 We found no study that examined these associations in

men.

Other comorbidities. We found preliminary evidence from two Phase II studies that

lumbar spine degeneration40 or chronic disease or disability37 are not associated with

the incidence of LDH with radiculopathy. However, the evidence linking chronic cough

or respiratory symptoms and the risk of LDH varies in three Phase II studies. One study

found a positive association in women;53 two other studies reported no association in

men or women.31,57

51

Health Behaviours:

Smoking. Evidence from one Phase III study,36 and 6 Phase II studies34,37,43,57,58,68

suggests that the incidence of LDH with radiculopathy increases with smoking.

However, 4 Phase II studies,31,40,46,52 and one crude analysis in a Phase III study45

(using an ecological measure of smoking) found no association between smoking and

the risk of LDH with radiculopathy.

Physical or sports activity. The evidence linking physical or sports activity to the risk

of LDH with radiculopathy varies in 9 Phase II studies. One study reported an

increased incidence of clinical LDH (sciatica) associated with walking, but a decreased

incidence associated with jogging.34 Another study found a positive association

between the frequency of participation in sports clubs (4-5 times per week) and risk of

LDH surgery in women, but not in men.37 In their study of male machine operators,

carpenters and office workers, Riihimäki and colleagues reported an indication of a

positive association between the frequency of physical exercise and the incidence of

clinical LDH (sciatic pain).43 Two other studies reported indications of negative

associations between physical activity (measured as time spent doing housework and

gardening52 and cumulative body building and strength training68) and LDH with

radiculopathy. Finally, 4 other studies found no association between physical or sports

activity and the incidence of LDH with radiculopathy.31,46,57,59

Other health behaviour characteristics. We found preliminary evidence from one

Phase II study that some medication use (not specified) may decrease the risk of

hospitalized LDH in women, but not in men.31 The same study also reported that

frequent use of analgesics was associated with an increased risk of hospitalized LDH,

again in women, but not in men. Another Phase II study of a young general population

in Finland, found no association between drinking style (i.e., abstinence, occasional

drinking, recurrent drinking, recurring drunkenness) and the incidence of LDH surgery.37

Personal Physical Factors:

Non-occupational lifting, bending and other activities. Preliminary evidence from

one Phase II study suggests that off-the-job lifting with the knees straight, back bent,

and starting and ending lifts at the waist may be positively associated with an increased

52

incidence of LDH with radiculopathy.58 In this study, Mundt and colleagues also found

trends toward positive associations between LDH with radiculopathy and lifting children

≥25 pounds, as well as with 2 or more days per week of repeated bending while doing

off-the-job activities. They also reported a tendency toward a negative association

between snow shoveling and LDH, and no association for off-the-job stretching and

carrying.

Car driving and motor vehicle characteristics. The evidence linking personal car

driving and motor vehicle characteristics to the risk of LDH with radiculopathy varies in 5

Phase II studies. In two older studies by Kelsey and colleagues, an increased incidence

of LDH with radiculopathy was found in persons driving non-Japanese and non-Swedish

cars,57 and in those exposed to driving other than on the job.52,54 However, no

association was reported for driving pattern characteristics, such as use of local roads,

highways, bucket seats, regular seats, automatic or manual transmission, and being the

driver or passenger.57 Three, more recent, Phase II studies reported no association

between car driving and LDH with radiculopathy in workers,34,43 and in health care

workers.58

Other personal physical factors. There is preliminary evidence from one Phase II

study reporting some indication of a positive association between non-professional

home construction/do-it-yourself activities and the risk of LDH with radiculopathy.46 In

addition, no association was reported between the incidence of LDH with radiculopathy

and non-occupational vibration (two Phase II studies)57,58 and the frequency of wearing

shoes with high heels (one Phase II study).57

Personal Psychological Factors:

Psychological stress and personality. The evidence varies in 7 Phase II studies

examining mental stress and the risk for LDH with radiculopathy. Three studies from

Finland found an increased incidence of LDH with radiculopathy in those who reported:

mental stress “to some extent” or “rather much or much”;34 hysteria in blue-collar, but

not in white-collar workers;60 and a greater number of psychological distress symptoms

in women, but not in men.31 However, 4 other studies reported no associations

between the risk of LDH with radiculopathy and personal mental stress;37,40

psychological well-being;46 or, number of stressful life events in the previous year.52

53

Workplace Physical Factors:

Forward bending postures. We found evidence from two Phase III studies that

occupational lumbar load from forward bending is positively associated with the risk of

LDH with radiculopathy. In two German studies, Seidler et al. reported an increased

incidence of LDH with radiculopathy in: i) both men and women with high cumulative

lumbar load by intensive-load postures (postures with trunk inclination of ≥20-degrees)

at work;67 and, ii) men with high cumulative work hours of extreme forward bending

(>90-degrees trunk flexion).61 However, two Phase II studies of worker cohorts found

no association between LDH and working with the trunk forward flexed,34 or bending

forward and backward.46

Other occupational postures. We found evidence from one Phase III study

suggesting no association between inconvenient work postures and LDH with

radiculopathy.45 However, the evidence varies in three Phase II studies examining trunk

twisting movements/rotations and the risk of LDH.34,46,56 In addition, no association was

reported between the incidence of LDH with radiculopathy and sedentary/sitting work

(one Phase III study;45 2 Phase II studies34,46); video display terminal work (one Phase

III study45); working in a kneeling or squatting position (2 Phase II studies34,46); twisted

or bent occupational postures (one Phase II study43); working with hands above

shoulders (one Phase II study34); and prolonged standing at work (one Phase II

study46).

Manual materials handling. There is evidence from two Phase III studies linking

manual materials handling with an increased risk of LDH with radiculopathy. In their

case-control studies, Seidler and colleagues reported strong positive associations

between LDH with radiculopathy and cumulative lumbar load by manual materials

handling and forward bending postures in both women and men.61,67 Their analysis of

exposure by manual materials handling alone, revealed that men that reported higher

cumulative lumbar load by manual materials handling (i.e., lifting, carrying, pushing,

pulling, throwing, catching or shoveling of objects weighing ≥5 kilograms), compared

with those that reported lower cumulative lumbar load by manual materials handling,

were more likely to experience LDH; however, they found no such association in

women.67 One Phase II study also reported positive associations for the frequency of

54

occupational lifting, carrying, lifting while twisting, and lifting and twisting with knees not

bent.56,57 However, two Phase III and three Phase II studies reported no association

between manual materials handling and LDH in women34,45,52,53 and men.34,45,46,52,53,61

Physical workload or strenuousness of work. The evidence varies in 5 studies

examining work physical strenuousness and the risk of LDH with radiculopathy. One

Phase III study from Germany, found some indication of a positive association between

working ≥10 years in occupations with high physical workload and LDH.61 No

association between physical workload and LDH with radiculopathy was reported in one

Phase III study of the general Finnish population,45 one Phase II study of Finnish

forestry workers,34 and one Phase I study of the southwest region of Finland.39

However, one Phase II study of the general Finnish population found a positive

association in women, but no association in men.50

Other workplace physical factors. There is evidence from one Phase III study

indicating a tendency of a positive association between exposure to whole body

vibration and the risk of LDH with radiculopathy.61 Although, one Phase II study found

no such association for vibration.43 We also found preliminary evidence in one Phase II

study that workplace exposure to draft or cold is not associated with the incidence of

clinical LDH (sciatica).43 Finally, the evidence varies in three Phase II studies assessing

the link between occupational driving or truck driving and the risk of LDH with

radiculopathy (2 reported a positive association46,52-54 and 1 reported no association34).

Workplace Psychosocial Factors:

Time pressure, job control and other psychosocial factors at work. We found

evidence in one Phase III study indicating a link between time pressure at work and the

risk of LDH with radiculopathy. In Germany, Seidler and colleagues found that men

who reported a greater number of work years with a high degree of time pressure were

more likely to experience LDH with radiculopathy compared to those who reported no

work years with high time pressure.61 This same study found no association for other

psychosocial characteristics of work, such as monotonous, boring, opportunities to use

knowledge and skills, information about future plans, satisfaction with supervisor,

satisfaction with workmates, psychic strain through contact with clients, and too much

responsibility.61

55

Evidence from one Phase III study of a national Finnish workforce cohort suggests that

job control is negatively associated with the risk of hospitalized LDH with radiculopathy

in both women and men.45 Leino-Arjas and colleagues also reported an increased

incidence of hospitalized LDH with radiculopathy in women who reported having a “3-

shift” work schedule (defined by the authors as “regular or irregular three-shift work or

regular night work”) compared with women with regular daytime work; and in both

women and men who reported higher levels of work injury risk compared with those

who reported no work injury risk. They reported no association for challenging work

tasks and social demands of work.45

Three Phase II studies of Finnish worker cohorts found no associations between the

incidence of LDH with radiculopathy and job satisfaction;34,46 work psychosocial

overload;34 high work pace, monotonous work, or problems with workmates or

superiors;43 and perceived risk of work injury.34

2.4 Discussion

Summary of the main findings: The annual incidence of hospitalized or surgically

managed LDH with radiculopathy in the general population ranged from 0.2 to 1.3 per

1,000 persons. However, much LDH with radiculopathy is not treated in hospitals. The

annual incidence of LDH with radiculopathy defined on the basis of clinical signs and

symptoms varied from 6.2 per 1,000 persons (0.6%) among female nurses in the US, to

93 per 1,000 persons (9.3%) among forest industry workers in Finland.

The evidence suggests that the etiology of this condition is multifaceted. While there is

evidence that several occupational factors are important contributors to the

development of LDH with radiculopathy, it is also clear that other individual and

behavioural factors may contribute to the development of the condition. For instance,

although high cumulative lumbar load from occupational forward bending is a strong risk

factor for LDH with radiculopathy, not every person exposed to high cumulative lumbar

workload will develop LDH.34,61,67 Instead, combinations of risk factors are necessary to

cause LDH with radiculopathy, and the specific combination of risk factors leading to an

episode of the condition likely varies between persons.

56

Our systematic review suggests that LDH with radiculopathy results from complex

relationships between individual, behavioural, and work-related variables. We found

evidence (Phase III) that age, sex, education, BMI and cardiovascular risk factors in

women, smoking, occupational lumbar load by forward bending postures and manual

materials handling, perceived risk of work injury, decision latitude at work, regular or

irregular three-shift work or regular night work in women, and time pressure at work are

associated with the development of LDH with radiculopathy. We also found preliminary

evidence suggesting that manual occupation, genetics, and previous back pain may

contribute to the development of LDH with radiculopathy in adults.

Limitations of the published literature: We accepted 37 (80%) of the 46 eligible

articles that we reviewed. This literature has many limitations, and it is difficult to draw

consistent conclusions from it. Some of these problems are inherent in all studies of

LDH with radiculopathy. For example, there is no universally accepted definition of LDH

with radiculopathy, and the studies are so heterogeneous that it is difficult to compare

incidence rates and risk factors.

Two key challenges are that authors use different criteria to ascertain and define cases

of LDH with radiculopathy, and the lack of a standard definition for the lower range of

severity for LDH with radiculopathy. This leads to case definitions that are susceptible

to information bias, and results in the misclassification of cases. There is a need for

workable clinical and surveillance definitions of LDH with radiculopathy and subsequent

studies to validate various methods of ascertaining cases.6,77 Until there is some

consistency of definitions and their appropriate validation, studies of the incidence of

LDH with radiculopathy will remain so heterogeneous that comparing the incidence

estimates and risk factors will continue to prove challenging.

Another obvious problem is that many studies do not identify the population at risk that

should form the denominator in an incidence calculation. Many studies report the

number of cases admitted to hospitals over a specified time period, but do not provide

information on the population at risk for being admitted. In some cases, it might not be

possible to determine a denominator, since it is difficult to define the population that is

exclusively served by these institutions.

57

Our accepted studies also have variable inclusion and exclusion criteria that impact on

the interpretation and comparability of the findings. Many of the studies include hospital

admissions only. This is problematic since hospital admission policy for LDH with

radiculopathy can vary over time and jurisdictions.31 In addition, these studies typically

fail to capture cases of LDH with radiculopathy treated at the emergency department,

but not admitted to hospital. Studies that use hospital discharge data may be limited by

this selection bias. Even studies that capture emergency department cases miss those

that are treated at outpatient clinics, or receive no treatment at all.

Despite the limitations of this literature, there are some important conclusions that we

can make. The incidence of hospital-treated LDH with radiculopathy in the general

population ranged from 0.2 to 1.3 per 1,000 persons; however, little is known about the

general population incidence of LDH with radiculopathy not treated in hospital. We

found no admissible study estimating the incidence of clinically defined LDH with

radiculopathy in the general population. Our findings suggest that the incidence is

higher among worker populations, and is highly dependent on the method of case

definition and ascertainment. For example, in two studies that examined the incidence

among nurses in Denmark42 and the US36, the incidence of surgical LDH was much

lower than the incidence of clinical LDH (1.3 per 1,000 persons42 versus 6.2 per 1,000

persons36, Appendix D, Table D.1). There is evidence that the incidence of LDH with

radiculopathy is underestimated when rates are based only on those presenting to or

treated at hospital.

Of the 22 admissible studies that investigated risk factors, 7 reported crude associations

(Phase I studies) and 11 reported associations from exploratory multivariable analyses

(Phase II studies); only 4 studies reported associations from confirmatory Phase III

analyses. Thus, we have limited information on risk factors for LDH with radiculopathy

in adults. Nonetheless, the strongest evidence (Phase III studies) suggests that age

and sex are nonmodifiable risk factors for LDH and radiculopathy, and modifiable risk

factors include lower education, higher BMI and the presence of cardiovascular risk

factors in women, smoking, greater cumulative occupational lumbar load by forward

bending postures and manual materials handling, higher levels of perceived risk of work

injury, lower job control or decision latitude at work, regular or irregular three-shift work

or regular night work in women, and increased time pressure at work. We found

58

preliminary evidence that manual occupation, genetics, previous back pain, and lifting

technique may contribute in the development of LDH with radiculopathy in adults.

Limitations of our review: Our systematic review has two main limitations. First,

although all studies in our review were judged to be scientifically admissible, the

strength of our findings is limited by the considerable variation in the methods of the

primary literature. In particular, the source populations, case definitions, and

assessment and control of confounding were very heterogeneous. We tried to minimize

the effect of these potential sources of bias by rating the overall risk of bias as high,

moderate and low, and by classifying studies as Phase I, II and III; giving greater weight

to confirmatory Phase III studies, followed by preliminary Phase II and I studies.

Second, when assessing the evidence about certain risk factors, we judged the

evidence as varying between studies. However, it is possible that this observed

variation might have been due to us combining studies that did not necessarily

contradict each other, but rather provided valid evidence of differing effects in different

populations (i.e., population-specific effects).

Future recommendations: Our review highlights the many existing gaps in knowledge

of the epidemiology of LDH with radiculopathy in adults. Further descriptive and

analytic investigations are needed as the incidence and determinants of this disabling

and costly neuromusculoskeletal condition have not been adequately described or

examined. We found no admissible study that investigated the incidence of clinically

defined LDH with radiculopathy in the general population. To date, the effect of

cardiovascular risk factors in the development of LDH with radiculopathy has yet to be

examined in men. We also found no admissible evidence examining the link between

trauma and LDH with radiculopathy, be it from motor vehicle injury or clinical

iatrogenesis such as spinal manipulation treatment. A number of case reports have

suggested this as a possible risk factor for LDH with radiculopathy;78-85 however, our

best evidence synthesis suggests that further study using appropriate study designs

and methods is warranted in these areas.

Several methodologic issues are important for future research. We recommend that

authors explicitly describe their source and target populations and discuss

characteristics and issues related to their sampling frames. Producing valid incidence

59

estimates requires accurate and complete ascertainment of both the cases and the at-

risk population. There is a need for workable clinical and surveillance definitions of LDH

with radiculopathy and subsequent validation studies. In addition, incidence density

rates, a more accurate measure of the population-time at risk, are preferable to

cumulative incidence proportions. Risk factors should be studied in cohort or case-

control study designs using multivariable statistical analysis that allows for the

identification of independent risk factors after adjustment for other important covariates.

We recommend that better Phase III studies target modifiable risk factors to inform the

design and testing of evidence-based prevention and intervention programs. In sum,

better understanding of the incidence and determinants of LDH with radiculopathy

demands more careful attention to issues of bias and research methods.

2.5 Conclusion

Although the literature is of varying quality and heterogeneous, there is evidence that

LDH with radiculopathy is an important source of pain and disability in society. The

annual incidence ranges roughly between 0.1% and 10%, depending on case definition

and source population, and risk factors are varied. Our findings support the need to

develop standardized case and surveillance definitions that validly classify the clinical

spectrum of this important condition. Future research should focus on prospective

designs examining modifiable risk factors and prevention strategies.

Funding Sources

Dr. Cesar Hincapié was supported by a Fellowship Award in the Area of Knowledge

Translation from the Canadian Institutes of Health Research, with additional support

from the Canadian Chiropractic Research Foundation.

Acknowledgements

We thank Marina Englesakis, Information Specialist at the University Health Network,

for her support and contribution with the review search strategy.

60

References

1. Hoy D, Brooks P, Blyth F, Buchbinder R. The epidemiology of low back pain. Best Pract Res Clin Rheumatol. 2010;24:769-81.

2. Hoy D, March L, Brooks P, et al. The global burden of low back pain: estimates from the Global Burden of Disease 2010 study. Ann Rheum Dis. 2014;73:968-74.

3. Cassidy JD, Côté P, Carroll LJ, Kristman V. Incidence and course of low back pain episodes in the general population. Spine. 2005;30:2817-23.

4. Hincapié CA, Cassidy JD, Côté P. Is a history of work-related low back injury associated with prevalent low back pain and depression in the general population? BMC Musculoskelet Disord. 2008;9:22.

5. Vos T, Flaxman AD, Naghavi M, et al. Years lived with disability (YLDs) for 1160 sequelae of 289 diseases and injuries 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. The Lancet. 2012;380:2163-96.

6. Genevay S, Atlas SJ, Katz JN. Variation in eligibility criteria from studies of radiculopathy due to a herniated disc and of neurogenic claudication due to lumbar spinal stenosis: a structured literature review. Spine. 2010;35:803-11.

7. van der Windt DA, Simons E, Riphagen II, et al. Physical examination for lumbar radiculopathy due to disc herniation in patients with low-back pain. Cochrane Database Syst. Rev. 2010;2:CD007431.

8. Vroomen PC, de Krom MC, Wilmink JT, Kester AD, Knottnerus JA. Diagnostic value of history and physical examination in patients suspected of lumbosacral nerve root compression. J Neurol Neurosurg Psychiatry. 2002;72:630-4.

9. Fardon DF, Milette PC. Nomenclature and classification of lumbar disc pathology. Recommendations of the Combined task Forces of the North American Spine Society, American Society of Spine Radiology, and American Society of Neuroradiology. Spine. 2001;26:E93-E113.

10. Frymoyer JW. Back pain and sciatica. N Engl J Med. 1988;318:291-300. 11. Tarulli AW, Raynor EM. Lumbosacral radiculopathy. Neurol Clin. 2007;25:387-

405. 12. Valat JP, Genevay S, Marty M, Rozenberg S, Koes B. Sciatica. Best Pract Res

Clin Rheumatol. 2010;24:241-52. 13. Selim AJ, Ren XS, Fincke G, et al. The importance of radiating leg pain in

assessing health outcomes among patients with low back pain. Results from the Veterans Health Study. Spine. 1998;23:470-4.

14. BenDebba M, Torgerson WS, Long DM. A validated, practical classification procedure for many persistent low back pain patients. Pain. 2000;87:89-97.

15. Tubach F, Beaute J, Leclerc A. Natural history and prognostic indicators of sciatica. J Clin Epidemiol. 2004;57:174-9.

16. Hill JC, Konstantinou K, Egbewale BE, Dunn KM, Lewis M, van der Windt D. Clinical outcomes among low back pain consulters with referred leg pain in primary care. Spine. 2011;36:2168-75.

17. Heliövaara M, Impivaara O, Sievers K, et al. Lumbar disc syndrome in Finland. J Epidemiol Community Health. 1987;41:251-8.

18. Kaila-Kangas L, Leino-Arjas P, Karppinen J, Viikari-Juntura E, Nykyri E, Heliovaara M. History of physical work exposures and clinically diagnosed sciatica among working and nonworking Finns aged 30 to 64. Spine. 2009;34:964-9.

61

19. Jordan J, Konstantinou K, O'Dowd J. Herniated lumbar disc. Clin Evid (Online). 2011. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3275148/. Accessed 2015-May-15.

20. Moher D, Liberati A, Tetzlaff J, Altman DG, Group P. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med. 2009;6:e1000097.

21. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. PLoS Med. 2009;6:e1000100.

22. Hincapié CA, Cassidy JD, Côté P, et al. The incidence and determinants of symptomatic lumbar disc herniation: a systematic review. PROSPERO. 2011. http://www.crd.york.ac.uk/PROSPERO/full_doc.asp?ID=CRD42011001197.

23. Scottish Intercollegiate Guidelines Network. Section 6: systematic literature review. SIGN 50: a guideline developer's handbook 2008; http://www.sign.ac.uk/guidelines/fulltext/50/section6.html. Accessed 2015-May-15.

24. Slavin RE. Best evidence synthesis: an intelligent alternative to meta-analysis. J Clin Epidemiol. 1995;48:9-18.

25. Cassidy JD, Carroll LJ, Peloso PM, et al. Incidence, risk factors and prevention of mild traumatic brain injury: results of the WHO Collaborating Centre Task Force on Mild Traumatic Brain Injury. J Rehabil Med. 2004:28-60.

26. Côté P, van der Velde G, Cassidy JD, et al. The burden and determinants of neck pain in workers: results of the Bone and Joint Decade 2000-2010 Task Force on Neck Pain and Its Associated Disorders. Spine. 2008;33:S60-S74.

27. Altman DG, Lyman GH. Methodological challenges in the evaluation of prognostic factors in breast cancer. Breast Cancer Res Treat. 1998;52:289-303.

28. Côté P, Cassidy JD, Carroll L, Frank JW, Bombardier C. A systematic review of the prognosis of acute whiplash and a new conceptual framework to synthesize the literature. Spine. 2001;26:E445-58.

29. R: A language and environment for statistical computing [computer program]. Vienna, Austria: R Foundation for Statistical Computing; 2012.

30. Szklo M, Nieto J. Epidemiology: beyond the basics. 2nd ed. Sudbury, MA: Jones and Bartlett Publishers; 2006.

31. Heliövaara M, Knekt P, Aromaa A. Incidence and risk factors of herniated lumbar intervertebral disc or sciatica leading to hospitalization. J Chronic Dis. 1987;40:251-8.

32. Bruske-Hohlfeld I, Merritt JL, Onofrio BM, et al. Incidence of lumbar disc surgery. A population-based study in Olmsted County, Minnesota, 1950-1979. Spine. 1990;15:31-5.

33. Zitting P, Rantakallio P, Vanharanta H. Cumulative incidence of lumbar disc diseases leading to hospitalization up to the age of 28 years. Spine. 1998;23:2337-43; discussion 43-4.

34. Miranda H, Viikari-Juntura E, Martikainen R, Takala EP, Riihimaki H. Individual factors, occupational loading, and physical exercise as predictors of sciatic pain. Spine. 2002;27:1102-9.

35. Jarvik JG, Hollingworth W, Heagerty PJ, Haynor DR, Boyko EJ, Deyo RA. Three-year incidence of low back pain in an initially asymptomatic cohort: clinical and imaging risk factors. Spine. 2005;30:1541-8; discussion 9.

62

36. Jhawar BS, Fuchs CS, Colditz GA, Stampfer MJ. Cardiovascular risk factors for physician-diagnosed lumbar disc herniation. Spine J. 2006;6:684-91.

37. Mattila VM, Saarni L, Parkkari J, et al. Early risk factors for lumbar discectomy: an 11-year follow-up of 57,408 adolescents. Eur Spine J. 2008;17:1317-23.

38. Mattila VM, Sillanpää P, Visuri T, Pihlajamäki H. Incidence and trends of low back pain hospitalisation during military service--an analysis of 387,070 Finnish young males. BMC Musculoskelet Disord. 2009;10:10.

39. Hurme M, Alaranta H, Törmä T, Einola S. Operated lumbar disc herniation: epidemiological aspects. Ann Chir Gynaecol. 1983;72:33-6.

40. Riihimäki H, Wickström G, Hänninen K, Luopajärvi T. Predictors of sciatic pain among concrete reinforcement workers and house painters--a five-year follow-up. Scand J Work Environ Health. 1989;15:415-23.

41. Heikkilä JK, Koskenvuo M, Heliövaara M, et al. Genetic and environmental factors in sciatica. Evidence from a nationwide panel of 9365 adult twin pairs. Ann Med. 1989;21:393-8.

42. Jørgensen S, Hein HO, Gyntelberg F. Heavy lifting at work and risk of genital prolapse and herniated lumbar disc in assistant nurses. Occup Med. 1994;44:47-9.

43. Riihimäki H, Viikari-Juntura E, Moneta G, Kuha J, Videman T, Tola S. Incidence of sciatic pain among men in machine operating, dynamic physical work, and sedentary work. A three-year follow-up. Spine. 1994;19:138-42.

44. Leino-Arjas P, Kaila-Kangas L, Keskimäki I, Notkola V, Mutanen P. Inpatient hospital care for lumbar intervertebral disc disorders in Finland in relation to education, occupational class, income, and employment. Public Health. 2002;116:272-8.

45. Leino-Arjas P, Kaila-Kangas L, Kauppinen T, Notkola V, Keskimäki I, Mutanen P. Occupational exposures and inpatient hospital care for lumbar intervertebral disc disorders among Finns. Am J Ind Med. 2004;46:513-20.

46. Leclerc A, Tubach F, Landre MF, Ozguler A. Personal and occupational predictors of sciatica in the GAZEL cohort. Occup Med. 2003;53:384-91.

47. Bongers PM, Boshuizen HC, Hulshof CT, Koemeester AP. Back disorders in crane operators exposed to whole-body vibration. Int Arch Occup Environ Health. 1988;60:129-37.

48. Netterstrøm B, Juel K. Low back trouble among urban bus drivers in Denmark. Scand J Soc Med. 1989;17:203-6.

49. Heliövaara M. Body height, obesity, and risk of herniated lumbar intervertebral disc. Spine. 1987;12:469-72.

50. Heliövaara M. Occupation and risk of herniated lumbar intervertebral disc or sciatica leading to hospitalization. J Chronic Dis. 1987;40:259-64.

51. Kelsey JL, Ostfeld AM. Demographic characteristics of persons with acute herniated lumbar intervertebral disc. J Chronic Dis. 1975;28:37-50.

52. Kelsey JL. An epidemiological study of acute herniated lumbar intervertebral discs. Rheumatol Rehabil. 1975;14:144-59.

53. Kelsey JL. An epidemiological study of the relationship between occupations and acute herniated lumbar intervertebral discs. Int J Epidemiol. 1975;4:197-205.

54. Kelsey JL, Hardy RJ. Driving of motor vehicles as a risk factor for acute herniated lumbar intervertebral disc. Am J Epidemiol. 1975;102:63-73.

63

55. Kelsey JL, Greenberg RA, Hardy RJ, Johnson MF. Pregnancy and the syndrome of herniated lumbar intervertebral disc; an epidemiological study. Yale J Biol Med. 1975;48:361-8.

56. Kelsey JL, Githens PB, White AA, III, et al. An epidemiologic study of lifting and twisting on the job and risk for acute prolapsed lumbar intervertebral disc. J Orthop Res. 1984;2:61-6.

57. Kelsey JL, Githens PB, O'Conner T, et al. Acute prolapsed lumbar intervertebral disc. An epidemiologic study with special reference to driving automobiles and cigarette smoking. Spine. 1984;9:608-13.

58. Mundt DJ, Kelsey JL, Golden AL, et al. An epidemiologic study of non-occupational lifting as a risk factor for herniated lumbar intervertebral disc. The Northeast Collaborative Group on Low Back Pain. Spine. 1993;18:595-602.

59. Mundt DJ, Kelsey JL, Golden AL, et al. An epidemiologic study of sports and weight lifting as possible risk factors for herniated lumbar and cervical discs. The Northeast Collaborative Group on Low Back Pain. Am J Sports Med. 1993;21:854-60.

60. Pietri-Taleb F, Riihimäki H, Viikari-Juntura E, Lindström K, Moneta GB. The role of psychological distress and personality in the incidence of sciatic pain among working men. Am J Public Health. 1995;85:541-5.

61. Seidler A, Bolm-Audorff U, Siol T, et al. Occupational risk factors for symptomatic lumbar disc herniation; a case-control study. Occup Environ Med. 2003;60:821-30.

62. Noponen-Hietala N, Virtanen I, Karttunen R, et al. Genetic variations in IL6 associate with intervertebral disc disease characterized by sciatica. Pain. 2005;114:186-94.

63. Mio F, Chiba K, Hirose Y, et al. A functional polymorphism in COL11A1, which encodes the alpha 1 chain of type XI collagen, is associated with susceptibility to lumbar disc herniation. Am J Hum Genet. 2007;81:1271-7.

64. Virtanen IM, Song YQ, Cheung KMC, et al. Phenotypic and population differences in the association between CILP and lumbar disc disease. J Med Genet. 2007;44:285-8.

65. Hirose Y, Chiba K, Karasugi T, et al. A functional polymorphism in THBS2 that affects alternative splicing and MMP binding is associated with lumbar-disc herniation. Am J Hum Genet. 2008;82:1122-9.

66. Karasugi T, Semba K, Hirose Y, et al. Association of the tag SNPs in the human SKT gene (KIAA1217) with lumbar disc herniation. J Bone Miner Res. 2009;24:1537-43.

67. Seidler A, Bergmann A, Jäger M, et al. Cumulative occupational lumbar load and lumbar disc disease--results of a German multi-center case-control study (EPILIFT). BMC Musculoskelet Disord. 2009;10:48.

68. Schumann B, Bolm-Audorff U, Bergmann A, et al. Lifestyle factors and lumbar disc disease: results of a German multi-center case-control study (EPILIFT). Arthritis Res Ther. 2010;12:R193.

69. Cong L, Pang H, Xuan D, Tu GJ. Association between the expression of aggrecan and the distribution of aggrecan gene variable number of tandem repeats with symptomatic lumbar disc herniation in Chinese Han of Northern China. Spine. 2010;35:1371-6.

70. Hrubec Z, Nashold BS, Jr. Epidemiology of lumbar disc lesions in the military in World War II. Am J Epidemiol. 1975;102:367-76.

64

71. An HS, Silveri CP, Simpson JM, et al. Comparison of smoking habits between patients with surgically confirmed herniated lumbar and cervical disc disease and controls. J Spinal Disord. 1994;7:369-73.

72. Chibnall JT, Tait RC, Andresen EM, Hadler NM. Race differences in diagnosis and surgery for occupational low back injuries. Spine. 2006;31:1272-5.

73. Saftic R, Grgic M, Ebling B, Splavski B. Case-control study of risk factors for lumbar intervertebral disc herniation in Croatian island populations. Croat Med J. 2006;47:593-600.

74. Lee DY, Ahn Y, Lee SH. The influence of facet tropism on herniation of the lumbar disc in adolescents and adults. J Bone Joint Surg Br. 2006;88:520-3.

75. Kunakornsawat S, Ngamlamaidt K, Tungsiripat R, Prasartritha T. The relationship of facet tropism to lumbar disc herniation. J Med Assoc Thai. 2007;90:1337-41.

76. Zhang Y-g, Sun Z, Zhang Z, Liu J, Guo X. Risk factors for lumbar intervertebral disc herniation in Chinese population: a case-control study. Spine. 2009;34:E918-22.

77. Lin CW, Verwoerd AJ, Maher CG, et al. How is radiating leg pain defined in randomized controlled trials of conservative treatments in primary care? A systematic review. Eur J Pain. 2014;18:455-64.

78. Huang SL, Liu YX, Yuan GL, Zhang J, Yan HW. Characteristics of lumbar disc herniation with exacerbation of presentation due to spinal manipulative therapy. Medicine. 2015;94:e661.

79. Assendelft WJ, Bouter LM, Knipschild PG. Complications of spinal manipulation: a comprehensive review of the literature. J Fam Pract. 1996;42:475-80.

80. Oliphant D. Safety of spinal manipulation in the treatment of lumbar disk herniations: a systematic review and risk assessment. J Manipulative Physiol Ther. 2004;27:197-210.

81. Powell FC, Hanigan WC, Olivero WC. A risk/benefit analysis of spinal manipulation therapy for relief of lumbar or cervical pain. Neurosurgery. 1993;33:73-8; discussion 8-9.

82. Schmidley JW, Koch T. The noncerebrovascular complications of chiropractic manipulation. Neurology. 1984;34:684-5.

83. Morandi X, Riffaud L, Houedakor J, Amlashi SF, Brassier G, Gallien P. Caudal spinal cord ischemia after lumbar vertebral manipulation. Joint Bone Spine. 2004;71:334-7.

84. Rivett DA, Milburn P. Complications arising from spinal manipulative therapy in New Zealand. Physiotherapy. 1997;83:626-32.

85. Tamburrelli FC, Genitiempo M, Logroscino CA. Cauda equina syndrome and spine manipulation: case report and review of the literature. Eur Spine J. 2011;20 Suppl 1:S128-31.

65

Chapter 3

Chiropractic Spinal Manipulation Treatment and the Risk For 3Acute Lumbar Disc Herniation: A Belief Elicitation Study

Manuscript 2

Hincapié CA, Cassidy JD, Rampersaud YR, Côté P, Jadad AR, Tomlinson GA.

Chiropractic spinal manipulation treatment and the risk for acute lumbar disc herniation:

a belief elicitation study. In preparation for submission, 2015

Chiropractic spinal manipulation treatment and

the risk for acute lumbar disc herniation: a belief elicitation study

Authors: Cesar A. Hincapié, DC, MHSc,1,2 J. David Cassidy, PhD, DrMedSc,1,2,3 Y.

Raja Rampersaud, MD, MSc,4 Pierre Côté, DC, PhD,5 Alejandro R. Jadad, MD, DPhil,1,6

George A. Tomlinson, PhD,1,7

Affiliations: 1 Dalla Lana School of Public Health, University of Toronto; 2 Toronto

Western Research Institute, University Health Network; 3 Department of Sports Science

and Clinical Biomechanics, University of Southern Denmark; 4 Division of Orthopaedic

Surgery, Toronto Western Hospital; 5 Faculty of Health Sciences, University of Ontario

Institute of Technology; 6 Centre for Global eHealth Innovation and Techna, University

Health Network; 7 Toronto General Research Institute and Department of Medicine,

University Health Network

Corresponding Author: Cesar A. Hincapié, DC, MHSc, University Health Network,

LuCliff Place – 700 Bay Street, Suite 602, Toronto, Ontario M5G 1Z6, Canada. E-mail:

cesar.hincapie@utoronto.ca

Key Words: chiropractic; spinal manipulation; risk; lumbar disc herniation; beliefs;

Bayesian; priors; belief elicitation

Word Count: 3,996

66

Abstract

Background: Chiropractic spinal manipulation treatment is common for low back pain

and may increase the risk for acute lumbar disc herniation (LDH), but little is known

about this potential link. In the absence of definitive scientific evidence, clinicians can

have knowledge and beliefs about the risk of using spinal manipulation gained through

clinical experience. Our main aim was to elicit those beliefs among clinicians, while

generating new insights to enrich future Bayesian approaches to interpreting outcomes

after spinal manipulation treatment.

Objective: To describe and quantify clinicians' beliefs (expressed as probability

distributions) about chiropractic spinal manipulation as a risk factor for: 1) acute LDH,

and 2) acute severe LDH that is surgically managed.

Study design: Cross-sectional belief elicitation study of chiropractors, family physicians

and spine surgeons in Ontario, Canada.

Methods: Structured interviews were conducted using a validated approach to belief

elicitation with 47 clinicians that treat patients with back pain and disc herniation.

Participants were asked to estimate the 2-month incidence of acute LDH among an

average group of newly diagnosed patients with acute low back pain (without current

LDH) treated with chiropractic spinal manipulation, relative to the incidence in a

counterfactual group not treated with chiropractic spinal manipulation. Probability

distributions for the relative risk (RR) for acute LDH associated with chiropractic spinal

manipulation treatment were derived.

Results: Combining all clinicians, belief of the risk for acute LDH associated with

chiropractic spinal manipulation treatment was neutral (median RR, 0.97; IQR, 0.57-

1.20; 47% probability for a RR>1). Beliefs varied across clinician groups. Chiropractors

expressed the most optimistic belief (median RR, 0.56; IQR, 0.39-1.03; 27% probability

for a RR>1). Family physicians were more neutral in their belief (median RR, 0.97; IQR,

0.64-1.21; 47% probability for a RR>1). Spine surgeons expressed a slightly more

pessimistic belief (median RR, 1.07; IQR, 0.95-1.29; 65% probability for a RR>1).

67

Conclusions: Clinicians’ beliefs about the risk for acute LDH associated with

chiropractic spinal manipulation varied, with important differences among professions.

Our probability distributions quantify and describe these differing beliefs and could be

used as prior probabilities in Bayesian risk analyses of this exposure-outcome

association.

3.1 Introduction

Spinal manipulation treatment is commonly used for back pain and generally considered

to be safe, although some concern has been raised about its potential link with lumbar

disc herniation (LDH).1,2 Several systematic reviews have found that spinal

manipulation can benefit low back pain with little evidence of serious harm.3,4 In

addition, randomized clinical trial evidence supports the use of spinal manipulation for

the treatment of LDH with radiculopathy.5-9 On the other hand, some believe, on the

basis of case reports and small case series, that spinal manipulation should be

contraindicated for the treatment of disc herniation and have raised the hypothesis of

causal harm.10-16 In sum, there seems to be a disconnect between the scientific

evidence and some beliefs in this area.

Today, no valid estimate of the risk for acute disc herniation following chiropractic spinal

manipulation treatment is available in the scientific literature. Attempts to investigate

serious adverse events related to spinal manipulation face several challenges. First,

the valid study of acute LDH with radiculopathy is limited by the rarity of this outcome.

Well-designed prospective studies are impractical due to the inability to recruit and

follow the numbers of patients needed to ensure that sufficient events are observed. In

addition, this is a sensitive topic open to controversy and litigation,17,18 with a very

limited evidence base, and thus, susceptible to opinion and anecdotal information.

A Bayesian approach to health services research and evaluation presents an

opportunity to advance knowledge despite the above challenges.19,20 Bayesian

methods start with existing “prior” beliefs, formally quantified as probability distributions,

and update these using new data to arrive at “posterior” beliefs, which may be used as

the basis for inferential decisions.21 Quantifying currently held beliefs can determine the

magnitude of a potential risk expected by experts and describe the presence of

68

uncertainty or clinical equipoise about an exposure-outcome association. Experts in a

field can have knowledge of the risk of using a treatment through years of clinical

experience. When quantified, the knowledge gained from their clinical experience can

be included in models estimating risk, which may help to bridge the gap between beliefs

and evidence.

A particular strength of this approach is that probability distributions obtained through

the elicitation of beliefs can be used to augment scarce data,19,20,22,23 and be formally

incorporated into Bayesian risk analyses. Although infrequently used in epidemiologic

research, a Bayesian approach may have important utility in the study of treatment-

related rare serious adverse events, particularly in the absence of definitive scientific

evidence.19,20,24,25

No previous study has assessed interprofessional clinician beliefs regarding chiropractic

spinal manipulation and the risk for acute LDH. The objective of our study was to

describe and quantify clinicians' beliefs about the association between chiropractic

spinal manipulation and the development of: (1) acute LDH, and (2) acute severe LDH

that is surgically managed. We elicited beliefs from three groups of clinicians that treat

patients with back pain and disc herniation: chiropractors, family physicians and spine

surgeons. We hypothesized that beliefs about the risk following chiropractic care would

vary by clinician group, with chiropractors believing spinal manipulation to be less risky

than spine surgeons.

3.2 Methods

Study Participants: Since belief elicitation is best conducted as a face-to-face

interaction between participant and investigator,22,26 we considered eligible only those

participants available for an interview within a 2-hour driving radius of the City of

Toronto. The three clinician groups were defined as: (1) chiropractors – sampled from a

list of active registered members with the College of Chiropractors of Ontario within the

Greater Toronto Area (sampling strategy detailed below); 2) family physicians –

selected from a list of referring primary care physicians to a spine surgery service at a

tertiary care hospital in Toronto and active registered members with the College of

Physicians and Surgeons of Ontario; 3) spine surgeons – invited to participate were all

69

eligible active members of the Ontario Division of the Canadian Spine Society. We also

imposed a requirement (self-reported) that the clinician’s practice involve caring for

patients with low back pain and LDH.

There is no consensus on the preferred sampling method and sample size estimate for

a belief elicitation study.22,26,27 A systematic review of belief elicitation methods found a

median sample size in elicitation studies of 11 participants, and suggested that it may

be preferable to include participants chosen non-randomly (e.g., purposive sampling) in

order to capture a range of opinions of the target population.26 We implemented a

pragmatic purposive sampling strategy of the three clinician groups, with the goal of

interviewing 15 members of each group, for a total of 45 respondents. We

hypothesized that beliefs may vary on the basis of clinician group, years of clinical

experience or gender, and attempted to balance these characteristics in our samples.

Case Definitions: Acute LDH was defined as acute onset of lumbar radiculopathy due

to LDH. Acute severe LDH that is surgically managed was defined as acute onset of

severe lumbar radiculopathy (i.e., severe intractable pain and neurologic deficit, but not

cauda equina syndrome) due to LDH and necessitating surgical management.

Study Design: We conducted a belief elicitation study using a computer adaptation of a

validated approach to clinician belief elicitation.27,28 A face-to-face interactive interview

was conducted by a single interviewer (CAH) with each participant, using a

standardized interview questionnaire (Appendix E), script (Appendix F) and elicitation

software. The laptop-based elicitation software (modified SHELF29 application in R30)

allowed us to use real-time graphical representation of a participant’s probability

distributions to provide feedback and ensure that the probability distributions were a true

representation of the participant’s beliefs (Appendix G provides screenshots of the

computer interface and details of the modifications we made to the base SHELF

software). The interview, script and software were pilot tested on a group of 10

clinicians (chiropractors, family physicians and spine surgeons) for face and content

validity, time feasibility (15-minute duration) and clarity. We obtained ethics approval for

our study from the research ethics board of the University Health Network (REB #11-

0240-AE), and administrative approval from the Office of Research Ethics at the

University of Toronto (protocol reference #26588).

70

Belief Elicitation Interview: The interview was developed by applying best practices

for belief elicitation22,26,27 and in consultation with an expert in Bayesian methods (GT).

It aimed at deriving valid probability distributions to describe clinicians' current beliefs

about the effect of chiropractic spinal manipulation on the risk of developing: (1) acute

LDH, and (2) acute severe LDH that is surgically managed.

The belief elicitation interview began with a short introduction, followed by informed

written consent. The participant was then guided through a worked example (Appendix

E, Example Scenario, with Appendix F script) of a belief elicitation to become familiar

with the elicitation process and computer software. The belief elicitation involved two

clinical scenarios, one for each of the outcomes of interest, through which the study

data were collected. For each scenario, a participant was asked to consider the

following at-risk population: a hypothetical average group of 1,000 newly diagnosed

acute low back pain patients without current LDH. The rationale for using 1,000 as the

denominator for this elicitation was based on it being an observable and conceivable

sample of patients that clinicians would be able to imagine. This facilitates the

probabilistic mental manipulations that have to be made during the elicitation interview;

increases the reliability of the elicitation; and reduces some of the common elicitation

biases (e.g., tendency to think in terms of percentages).22,26,27 Our approach meant that

the smallest ‘unit’ of probability was 1/1000 or 0.001. Use of a larger denominator (e.g.,

10,000 or 100,000) would have allowed for a more precise belief to be quantified, but

literature suggests that it is unlikely that people could easily conceive of the difference

between a probability of 0.001 (1/1000) or 0.0001 (1/10,000) or 0.00001 (1/100,000).22

The outcomes of interest were clinicians’ incidence estimates of acute LDH and acute

severe LDH that is surgically managed, among an average group of 1,000 newly

diagnosed acute low back pain patients (without current LDH) that is treated with

chiropractic spinal manipulation, relative to incidence estimates among a counterfactual

group not treated with chiropractic spinal manipulation.

In the final section of the interview, we also collected demographic and additional

characteristics of the participants: gender, age group, clinical background, number of

years in practice treating low back pain patients, number of new low back pain patients

seen per year, history and level of formal statistical training, history of referral for

71

chiropractic care in practice for treatment of low back pain, sources of information used

in formulating beliefs about the effect of chiropractic spinal manipulation on the risk of

developing acute LDH, and history of having seen a patient in clinical practice whose

disc herniation was attributed to chiropractic treatment.

Belief Elicitation Software: We used a modified computer protocol (modified SHELF29

application in R30) that allowed participants to graphically specify a point estimate of

incidence and a plausible range of incidence values that they believed likely for each of

the two clinical scenarios (Appendix G). They indicated their weight of belief for the

incidence values along their specified range by allocating 20 virtual “chips”, each

representing 5% probability, thereby creating a probability distribution or “belief curve”.

This allowed us to provide participants with real-time graphical representation of

participants’ probability distributions. Participants were asked to review the shape and

distribution of their belief curve to ensure that it accurately reflected their true belief prior

to proceeding in the interview.

Statistical Analysis: Medians and proportions were used to summarize characteristics

of the respondents. On each respondent, the probability distribution (i.e., belief) for the

relative risk (RR) for acute LDH was derived by dividing values of the incidence of acute

LDH among patients treated with chiropractic spinal manipulation (the values on the x-

axis in Appendix G) by the point estimate of incidence of acute LDH among patients

not treated with chiropractic spinal manipulation. A similar calculation was carried out

for the second outcome to derive the probability distribution for the RR for surgically

managed acute LDH among patients treated with chiropractic spinal manipulation

compared with those not treated with chiropractic care.

We aggregated all individual belief elicitations to derive an overall distribution of belief

among participants for the RR for each of the two outcomes of interest. We also

aggregated elicitations within each of the clinician groups. Finally, optimistic and

pessimistic probability distributions were estimated by aggregating respondents within

the lowest and highest quartiles of mean RR estimates, respectively.

Prespecified secondary analyses were undertaken to describe and quantify beliefs by

gender, age groups (clinicians ≤50 years of age and clinicians ≥51 years of age), and

72

years of clinical experience (tertiles of years treating patients with back pain: <12, 12-24

and ≥25 years). We used R, version 2.15.130 to carry out our statistical analysis.

Role of the Funding Source: The sponsors of the study had no role in study design,

data collection, data analysis, data interpretation, the writing of the report, or in the

decision to submit the report for publication.

3.3 Results

Study Participants: Of 165 eligible potential participants invited, 87 did not respond

and 31 declined to participate (reason not given). Overall participation was 28%

(47/165) and varied by clinician group: 37% among chiropractors (16/43); 15% among

family physicians (15/99); and, 70% among spine surgeons (16/23).

Table 3.1 describes characteristics of the study participants. All spine surgeons were

men and on average older than the other clinician groups. Most participants (94%)

reported some formal post-secondary statistical training. The median number of years

treating low back pain patients among all participants was 18 (IQR, 10-26 years), with

chiropractors having a lower median of 13 years (IQR, 8-27 years), and spine surgeons

a higher median of 23 years (IQR, 10-26 years).

73

Table 3.1. Characteristics of belief elicitation study participants

Characteristic

All clinicians

(n=47)

Family

physicians

(n=15)

Chiropractors

(n=16)

Spine

surgeons

(n=16)

Men – N (%) 35 (74) 7 (47) 12 (75) 16 (100)

Age group – N (%)

21-40 years 13 (28) 4 (27) 8 (50) 1 (6)

41-60 years 28 (59) 10 (66) 7 (44) 11 (69)

≥61 years 6 (13) 1 (7) 1 (6) 4 (25)

N with post-secondary statistical training –

N (%)

44 (94) 13 (87) 15 (94) 16 (100)

N of years treating LBP patients – Median

(IQR)

18 (10-26) 18 (13-26) 13 (8-27) 23 (10-26)

N of new LBP patients seen per year* –

Median (IQR)

150 (50-500) 150 (55-500) 65 (40-150) 550 (240-

1060)

N that refer for chiropractic care in clinical

practice – N (%)

34 (72) 11 (73) 16 (100) 7 (44)

Elicited incidence of acute LDH among

acute LBP patients not treated with

chiropractic care – Median % (IQR)

5.0 (2.3-10.0) 5.0 (3.8-10.0) 5.0 (3.0-10.0) 3.3 (1.9-10.0)

Elicited incidence of surgically managed

acute LDH among acute LBP patients

not treated with chiropractic care –

Median % (IQR)

1.0 (0.5-2.0) 1.0 (0.3-1.0) 1.5 (0.5-2.6) 1.0 (0.4-1.4)

Abbreviations: IQR, interquartile range; LDH, lumbar disc herniation; LBP, low back pain; N, number

* Self-reported estimate

Clinicians’ Beliefs About the Risk For Acute LDH Associated With Chiropractic

Care: Combining all clinicians, the probability distribution for the risk for acute LDH

associated with chiropractic spinal manipulation was evenly divided between a

decrease in risk and an increase in risk (Figure 3.1.a). The perceived median RR for

74

acute LDH across all clinicians was 0.97 (IQR, 0.57-1.20), with a probability for a RR >1

of 46.7% (Table 3.2).

Beliefs varied by clinician group (Figure 3.1.b). Chiropractors reported the most

optimistic belief of the risk for acute LDH associated with chiropractic care, with a

perceived median RR of 0.56 (IQR, 0.39-1.03) and probability for a RR >1 of 26.7%

(Table 3.2). Family physicians held a more neutral belief, with a perceived median RR

of 0.97 (IQR, 0.64-1.21) and probability for a RR >1 of 47.1%. Spine surgeons reported

a slightly more pessimistic belief, with a perceived median RR of 1.07 (IQR, 0.95-1.29)

and probability for a RR >1 of 64.8%.

Bimodal probability distributions were observed for both chiropractors and family

physicians (Figure 3.1.b), suggesting that within each of these clinician groups there

was a subset of clinicians that believed that there is a protective effect of spinal

manipulation on the risk for acute LDH, and another subset that believed that there is

“no effect” with a chance of harm. The unimodal distribution for the spine surgeons

showed a more homogeneous belief of no benefit on average, with more chance of

harm than benefit.

Optimistic and Pessimistic Beliefs: Optimistic clinicians—those in the lowest quartile

of RR estimates—included 3 family physicians and 9 chiropractors. These clinicians

expressed a belief that chiropractic spinal manipulation reduces the incidence of acute

LDH by about 60% (median RR, 0.42; IQR, 0.29-0.53) (Figure 3.1.c and Table 3.2).

Pessimistic clinicians—those in the highest quartile of RR estimates—included 6 spine

surgeons, 3 family physicians and 3 chiropractors, who believed that chiropractic care

increases the incidence of acute LDH by about 30% (median RR, 1.29; IQR, 1.11-1.59).

Compared to optimists, pessimists were more commonly men, older, spine surgeons,

and reported a greater number of new low back patients per year (Table 3.3). More

pessimists than optimists also reported having seen a patient whose disc herniation was

attributed to chiropractic spinal manipulation.

75

Figure 3.1. Beliefs regarding the risk for acute lumbar disc herniation associated with chiropractic spinal manipulation treatment: a. among all clinicians, b.

among clinician groups, c. among optimistic versus pessimistic clinicians

0.0 0.5 1.0 1.5 2.0 2.5 3.0

0.0

0.5

1.0

1.5

2.0

2.5

a. All participants belief

Relative risk

Wei

ght o

f bel

ief

P(RR<1)=53.3% P(RR>1)=46.7%

0.0 0.5 1.0 1.5 2.0 2.5 3.0

0.0

0.5

1.0

1.5

2.0

2.5

b. Clinician group beliefs

Relative risk

Wei

ght o

f bel

ief

Chiropractors (n=15)Family physicians (n=15)Spine surgeons (n=16)

P(RR>1)=64.8%

P(RR>1)=47.1%

P(RR>1)=26.7%

0.0 0.5 1.0 1.5 2.0 2.5 3.0

0.0

0.5

1.0

1.5

2.0

2.5

c. Optimistic−pessimistic beliefs

Relative risk

Wei

ght o

f bel

ief

Optimists (n=12)Pessimists (n=12)

P(RR>1)=1%

P(RR>1)=85.1%

76

Table 3.2. Clinicians’ beliefs about the risk for acute lumbar disc herniation and surgically managed

acute lumbar disc herniation associated with chiropractic spinal manipulation treatment

Beliefs

Acute LDH Acute LDH-surgery

Median RR*

(IQR) Prob RR>1

Median RR

(IQR) Prob RR>1

All clinicians (n=46) 0.97 (0.57-1.20) 46.7% 1.00 (0.61-1.40) 49.9%

Clinician group

Chiropractors (n=15) 0.56 (0.39-1.03) 26.7% 0.63 (0.35-1.08) 28.3%

Family physicians (n=15) 0.97 (0.64-1.21) 47.1% 1.02 (0.73-1.35) 53.0%

Spine surgeons (n=16) 1.07 (0.95-1.29) 64.8% 1.18 (0.93-1.58) 67.4%

Optimistic-pessimistic beliefs

Optimistic belief (lower quartile of RR)

(n=12)

0.42 (0.29-0.53) 1.0% 0.42 (0.22-0.64) 6.5%

Pessimistic belief (upper quartile of

RR) (n=12)

1.29 (1.11-1.59) 85.1% 1.66 (1.36-2.21) 90.5%

Gender

Women (n=11) 0.77 (0.41-1.10) 34.2% 0.66 (0.30-1.08) 31.0%

Men (n=35) 1.00 (0.68-1.23) 50.6% 1.07 (0.77-1.50) 56.6%

Age group

≤50 years (n=24) 0.96 (0.51-1.24) 46.8% 0.97 (0.57-1.49) 47.4%

≥51 years (n=22) 0.98 (0.66-1.17) 46.7% 1.02 (0.70-1.35) 52.6%

Clinical experience (years treating LBP

patients)

Low tertile – <12 years (n=15) 0.98 (0.54-1.26) 48.8% 0.91 (0.50-1.72) 46.8%

Mid tertile – 12-24 years (n=14) 0.97 (0.52-1.24) 46.9% 0.98 (0.64-1.24) 47.2%

High tertile – ≥25 years (n=17) 0.97 (0.65-1.14) 44.7% 1.04 (0.77-1.36) 54.9%

Abbreviations: IQR, interquartile range; LBP, low back pain; LDH, lumbar disc herniation; n, sample size;

Prob; probability; RR, relative risk

* Perceived estimates of median RRs derived from the elicited incidence probabilities for the outcomes

with and without exposure to chiropractic spinal manipulation treatment

77

Table 3.3. Characteristics of clinicians with optimistic versus pessimistic beliefs about the risk for acute

lumbar disc herniation and surgically managed acute lumbar disc herniation associated with

chiropractic spinal manipulation treatment

Acute LDH Acute LDH-surgery

Characteristic

Optimists

(n=12)

Pessimists

(n=12)

Optimists

(n=12)

Pessimists

(n=12)

Men – N (%) 7 (58) 11 (92) 5 (42) 11 (92)

Age group – N (%)

21-40 years 5 (41) 2 (17) 7 (58) 3 (25)

41-60 years 6 (50) 7 (58) 5 (41) 7 (58)

≥61 years 1 (8) 3 (25) 0 2 (17)

Clinician group – N (%)

Family physicians 3 (25) 3 (25) 3 (25) 3 (25)

Chiropractors 9 (75) 3 (25) 9 (75) 3 (25)

Spine surgeons 0 6 (50) 0 6 (50)

Number with post-secondary statistical

training – N (%)

11 (92) 11 (92) 11 (92) 11 (92)

Number of years treating LBP patients –

Median (IQR)

14 (9-25) 20 (9-26) 12 (7-19) 12 (7-22)

Number of new LBP patients per year –

Median (IQR)

135 (72-187) 250 (50-600) 120 (83-150) 225 (50-763)

Number that refer for chiropractic care in

clinical practice – N (%)

11 (92) 8 (67) 12 (100) 7 (58)

Number that have seen a patient whose

disc herniation was attributed to

chiropractic spinal manipulation – N (%)

1 (8) 7 (58) 0 6 (50)

Abbreviations: IQR, interquartile range; LBP, low back pain; LDH, lumbar disc herniation; N, number

78

Beliefs By Gender, Age Groups and Years of Clinical Experience: Table 3.2 and

Figure 3.2.a show that some differences in belief about the risk for acute LDH

associated with chiropractic care were found between women (n=11) and men (n=35)

clinicians, with women reporting a more optimistic belief (median RR, 0.77; IQR, 0.41-

1.10; 34% probability for a RR >1) and men, a more neutral belief (median RR, 1.00;

IQR, 0.68-1.23; 51% probability for a RR >1).

There were few differences in beliefs according to age group (participants aged ≤50

years, n=24 vs. participants aged ≥51 years, n=22) or tertiles of years of clinical

experience treating patients with low back pain (<12 years, n=15 vs. 12-24 years, n=14

vs. ≥25, n=17) (Table 3.2 and Figures 3.2.b and 3.2.c).

Clinicians’ Beliefs About the Risk For Surgically Managed Acute LDH Associated

With Chiropractic Care: Tables 3.2 and 3.3 show that findings for surgically managed

acute LDH were similar to those detailed above. Compared with RR estimates for the

belief regarding risk for acute LDH associated with chiropractic spinal manipulation, a

small increase in the perceived risk estimates was found among all three clinician

groups. Of note, pessimistic clinicians reported a belief that chiropractic care increases

the incidence of acute LDH that is surgically managed by about 65% (median RR, 1.66;

IQR, 1.36-2.21), with a probability for a RR >1 of over 90%.

79

Figure 3.2. Beliefs regarding the risk for acute lumbar disc herniation associated with chiropractic spinal manipulation treatment: a. by gender, b. by age

group, c. by clinical experience

0.0 0.5 1.0 1.5 2.0 2.5 3.0

0.0

0.5

1.0

1.5

2.0

2.5

a. Beliefs by gender

Relative risk

Wei

ght o

f bel

ief

Women (n=11)Men (n=35)

P(RR>1)=50.6%

P(RR>1)=34.2%

0.0 0.5 1.0 1.5 2.0 2.5 3.0

0.0

0.5

1.0

1.5

2.0

2.5

b. Beliefs by age group

Relative risk

Wei

ght o

f bel

ief

<=50 years of age (n=24)>=51 years of age (n=22)

P(RR>1)=46.8%

P(RR>1)=46.7%

0.0 0.5 1.0 1.5 2.0 2.5 3.0

0.0

0.5

1.0

1.5

2.0

2.5

c. Beliefs by clinical experience

Relative risk

Wei

ght o

f bel

ief

<12 years experience (n=15)12−24 years experience (n=14)>=25 years experience (n=17)

P(RR>1)=48.8%

P(RR>1)=46.9%

P(RR>1)=44.7%

80

3.4 Discussion

We carried out a belief elicitation study to quantify, describe and gain important insights

into beliefs about the risk for acute LDH associated with chiropractic spinal manipulation

held by clinicians experienced with treating patients with back pain and disc herniation.

Clinician beliefs about the incidence of acute LDH and surgically managed acute LDH

with and without chiropractic spinal manipulation treatment varied, with interesting and

important differences observed across professions and specializations.

Our study advances knowledge of the potential link between chiropractic spinal

manipulation and acute LDH in two important respects. First, our study is the first to

examine and provide estimates of clinician beliefs regarding this potential exposure-

outcome association. There are no published estimates of the association between

chiropractic care and the incidence of acute LDH with which to compare our findings. In

addition, given the rarity of acute symptomatic LDH and modest utilization of

chiropractic care in the population, well-designed prospective observational studies may

be challenging to perform. Nonetheless, experts in a field can have knowledge of the

risk of using a treatment through years of clinical experience. By quantifying currently

held clinician beliefs, our results shed light on the magnitude of a potential risk expected

by experts and describe the presence of uncertainty about this exposure-outcome

association.

Among the back pain clinicians in this study, there was a range of beliefs about the risk

associated with chiropractic care. On average, clinicians believed that the incidence of

acute LDH is about the same with or without chiropractic spinal manipulation treatment;

in other words, “no effect” of exposure to chiropractic care on the risk for acute LDH.

However, it is striking to note that optimistic clinicians believed that treatment of acute

low back pain with chiropractic spinal manipulation reduces the incidence of acute LDH

by 58% (IQR, 47-71%), whereas pessimistic clinicians believed that chiropractic care

increases the incidence of acute LDH that necessitates surgical management by 66%

(IQR, 36-121%) (Table 3.2). In addition, the bimodal distributions observed in our

graphical analyses (Figures 3.1 and 3.2) indicate distinct “pools” of belief, with some

clinicians believing in a beneficial effect of chiropractic care and others mainly

81

expressing a belief of no effect with a possibility of harm. Our findings suggest a

divergence of opinion within these back pain clinicians, indicating the presence of

community uncertainty.

Another advance comes as a result of the usefulness of the elicited beliefs under a

Bayesian inferential paradigm. The probability distributions derived from this study may

be relevant to researchers investigating the risk for these rare serious adverse events

following chiropractic spinal manipulation. Using a Bayesian approach, elicited beliefs

in the form of prior probability distributions can be used to augment limited data.22-24,31,32

This could facilitate a Bayesian approach to risk analyses of these exposure-outcome

associations in which informative clinical prior beliefs about the potential risk can be

explicitly and transparently specified as Bayesian priors. This view appears particularly

appropriate in the context of investigating contentious potential treatment-related rare

serious adverse events in the absence of compelling scientific data and evidence.19,20

Our study demonstrates that this belief elicitation method can be used to describe and

quantify clinicians’ beliefs about the safety of an intervention or treatment exposure.

This methodology is not limited to the study of beliefs about the safety of manual

therapy such as spinal manipulation. For example, it has been used to elicit beliefs

regarding: treatment efficacy in pulmonary arterial hypertension28 and pediatric intestinal

failure-associated liver disease;33 and risk factors for falls in community-dwelling older

people.34 This belief elicitation methodology could be generalized to the study of many

uncommon diseases and conditions,22 and could have potential web-based research

applications.35

Our study has limitations. This is a preliminary study involving a relatively small number

of clinicians that may not be representative of the clinician group source populations.

However, our main objective was to describe the beliefs of our study sample and

generate hypotheses, rather than to make strong generalizable inferences about beliefs

in these clinician source populations. The main limitations of our study may relate to

characteristics of the study respondents. Since there is no formal definition for back

pain clinician expert, we defined an expert as a chiropractor, family physician or spine

surgeon with a self-reported active practice treating patients with low back pain. The

large number of years treating patients with low back pain and large number of new

82

patients with low back pain seen per year (Table 3.1) support the notion that the

participants were experienced clinician experts. Our findings are limited to the three

clinician groups interviewed and do not include information on beliefs among physical

therapists. Nonresponse bias may have occurred as a result of clinicians not

responding to our invitation to participate. This may have led to bias if the reasons for

study participation were associated with beliefs regarding chiropractic care safety. We

hypothesize that respondents may have had more extreme beliefs than nonrespondents

(i.e., those without strong opinions might be less likely to take part). If this had been the

case, our estimates may overestimate clinician group beliefs, but on average, would be

similar to our current finding of a neutral belief among all participants. In addition, we

would have preferred a more balanced study sample with respect to gender and age

distribution. However, this proved challenging, as the spine surgery field in Ontario is

predominantly male and older, compared with the more balanced family physician

population and the younger population of chiropractors in the Greater Toronto Area

(Table 3.1). We acknowledge that patient mix and clinical experience likely varied

across clinician groups. We attempted to examine the impact of years of clinical

experience, age and gender by estimating probability distributions on the basis of these

characteristics. Elicited beliefs were similar across age groups and levels of clinical

experience, although differences in beliefs were found between women and men, with

women expressing a more optimistic belief.

Finally, it is also important to note that participants were asked to estimate the incidence

of acute LDH with and without chiropractic spinal manipulation treatment, rather than

provide their belief of the relative risk directly. The probability distribution for the relative

risk was derived from the individual distributions of incidence with and with chiropractic

care. Our rationale for this approach was based on evidence that it is much more

difficult for respondents to directly specify relative risk distributions as these are not

directly observable.22 A clinician can reflect on their knowledge and experience to

estimate a frequency representation of the probability for acute LDH (e.g., “Of 1000

patients with acute low back pain treated with chiropractic spinal manipulation, how

many will go on to develop acute LDH?”), however, it is a much more challenging task

to abstract that information and elicit a valid probability distribution for the ratio of

incidence probabilities (i.e., the relative risk).

83

Our hypothesis-generating study raises some interesting questions warranting further

investigation. What are the reasons for the differences in clinician beliefs that we

found? Do these clinicians interpret the same evidence, as limited as it is, differently?

Do they see different types of patients, so that their beliefs actually reflect their personal

clinical experience? Does spine surgeons’ experience with the more severe end of the

clinical spectrum raise their index of suspicion for viewing spinal manipulation as

potentially harmful? Do they have in-built biases based on their professions and

specializations? Would similar beliefs be observed in health care settings where the

education of these clinician groups is more integrated (e.g., Denmark,36 Switzerland37),

or where family physicians and spine surgeons deliver spinal manipulation? In

Germany, for example, general practitioners and ambulatory orthopaedic surgeons are

often specially trained in spinal manipulation and offer it to their patients.38 This

highlights the need for future well-designed studies to examine determinants of beliefs,

confirm clinicians’ beliefs about the safety of chiropractic spinal manipulation in other

settings, and investigate the association between chiropractic care and acute LDH in

the population.

With regard to public health significance, this study represents an important first step in

the process of understanding current perceptions about the safety of chiropractic spinal

manipulation. It may provide opportunities for interdisciplinary educational initiatives

among the various clinician groups, as well as knowledge translation with patients,

policymakers and other relevant stakeholders with an interest in the safety of

chiropractic care.

3.5 Conclusion

Our study is the first to describe and evaluate clinician beliefs about the association

between chiropractic spinal manipulation and the development of acute LDH and acute

LDH that necessitates surgical intervention. We found that beliefs were diverse, with

important interprofessional differences, indicating the presence of community

uncertainty. Our probability distributions quantify and describe these differing beliefs

and could be used as prior probabilities in Bayesian risk analyses of this exposure-

outcome.

84

Funding Sources

Dr. Cesar Hincapié was supported by a Fellowship Award in the Area of Knowledge

Translation from the Canadian Institutes of Health Research, with additional support

from the Canadian Chiropractic Research Foundation.

Acknowledgements

We thank Drs. Sindhu Johnson and Ivan Diamond for their assistance with the

development of the belief elicitation interview and questionnaire. We acknowledge Drs.

Jeremy Oakley and Anthony O’Hagan for their development and open source sharing of

the SHELF elicitation framework. We thank all the clinician experts who participated in

the belief elicitation interviews.

References

1. Gouveia LO, Castanho P, Ferreira JJ. Safety of chiropractic interventions: a systematic review. Spine. 2009;34:E405-13.

2. Snelling NJ. Spinal manipulation in patients with disc herniation: A critical review of risk and benefit. Int J Osteopath Med. 2006;9:77-84.

3. Cherkin DC, Sherman KJ, Deyo RA, Shekelle PG. A review of the evidence for the effectiveness, safety, and cost of acupuncture, massage therapy, and spinal manipulation for back pain. Ann Intern Med. 2003;138:898-906.

4. Furlan AD, Yazdi F, Tsertsvadze A, et al. A systematic review and meta-analysis of efficacy, cost-effectiveness, and safety of selected complementary and alternative medicine for neck and low-back pain. Evid Based Complement Alternat Med. 2012:953139.

5. Burton AK, Tillotson KM, Cleary J. Single-blind randomised controlled trial of chemonucleolysis and manipulation in the treatment of symptomatic lumbar disc herniation. Eur Spine J. 2000;9:202-7.

6. Liu J, Zhang S. Treatment of protrusion of lumbar intervertebral disc by pulling and turning manipulations. J Tradit Chin Med. 2000;20:195-7.

7. Santilli V, Beghi E, Finucci S. Chiropractic manipulation in the treatment of acute back pain and sciatica with disc protrusion: a randomized double-blind clinical trial of active and simulated spinal manipulations. Spine J. 2006;6:131-7.

8. McMorland G, Suter E, Casha S, du Plessis SJ, Hurlbert RJ. Manipulation or microdiskectomy for sciatica? A prospective randomized clinical study. J Manipulative Physiol Ther. 2010;33:576-84.

9. Bronfort G, Hondras MA, Schulz CA, Evans RL, Long CR, Grimm R. Spinal manipulation and home exercise with advice for subacute and chronic back-related leg pain: a trial with adaptive allocation. Ann Intern Med. 2014;161:381-91.

85

10. Huang SL, Liu YX, Yuan GL, Zhang J, Yan HW. Characteristics of lumbar disc herniation with exacerbation of presentation due to spinal manipulative therapy. Medicine. 2015;94:e661.

11. Assendelft WJ, Bouter LM, Knipschild PG. Complications of spinal manipulation: a comprehensive review of the literature. J Fam Pract. 1996;42:475-80.

12. Powell FC, Hanigan WC, Olivero WC. A risk/benefit analysis of spinal manipulation therapy for relief of lumbar or cervical pain. Neurosurgery. 1993;33:73-8; discussion 8-9.

13. Schmidley JW, Koch T. The noncerebrovascular complications of chiropractic manipulation. Neurology. 1984;34:684-5.

14. Morandi X, Riffaud L, Houedakor J, Amlashi SF, Brassier G, Gallien P. Caudal spinal cord ischemia after lumbar vertebral manipulation. Joint Bone Spine. 2004;71:334-7.

15. Tamburrelli FC, Genitiempo M, Logroscino CA. Cauda equina syndrome and spine manipulation: case report and review of the literature. Eur Spine J. 2011;20 Suppl 1:S128-31.

16. Oppenheim JS, Spitzer DE, Segal DH. Nonvascular complications following spinal manipulation. Spine J. 2005;5:660-6; discussion 6-7.

17. Boucher P, Robidoux S. Lumbar disc herniation and cauda equina syndrome following spinal manipulative therapy: a review of six court decisions in Canada. Journal of forensic and legal medicine. 2014;22:159-69.

18. Jagbandhansingh MP. Most common causes of chiropractic malpractice lawsuits. J Manipulative Physiol Ther. 1997;20:60-4.

19. Spiegelhalter DJ. Incorporating Bayesian ideas into health-care evaluation. Statistical Science. 2004;19:156-74.

20. Spiegelhalter DJ, Myles JP, Jones DR, Abrams KR. Methods in health service research. An introduction to bayesian methods in health technology assessment. BMJ. 1999;319:508-12.

21. Spiegelhalter D, Rice K. Bayesian statistics. Scholarpedia. 2009;4(8):5230. http://www.scholarpedia.org/article/Bayesian_statistics. Accessed 2015-May-15.

22. O'Hagan A, Buck CE, Daneshkhah A, et al. Uncertain judgements: eliciting experts' probabilities. West Sussex: John Wiley & Sons Ltd; 2006.

23. Lilford RJ, Thornton JG, Braunholtz D. Clinical trials and rare diseases: a way out of a conundrum. BMJ. 1995;311:1621-5.

24. Dunson DB. Commentary: practical advantages of Bayesian analysis of epidemiologic data. Am J Epidemiol. 2001;153:1222-6.

25. Hampson LV, Whitehead J, Eleftheriou D, Brogan P. Bayesian methods for the design and interpretation of clinical trials in very rare diseases. Stat Med. 2014;33:4186-201.

26. Johnson SR, Tomlinson GA, Hawker GA, Granton JT, Feldman BM. Methods to elicit beliefs for Bayesian priors: a systematic review. J Clin Epidemiol. 2010;63:355-69.

27. Johnson SR, Tomlinson GA, Hawker GA, Granton JT, Grosbein HA, Feldman BM. A valid and reliable belief elicitation method for Bayesian priors. J Clin Epidemiol. 2010;63:370-83.

28. Johnson SR, Granton JT, Tomlinson GA, Grosbein HA, Hawker GA, Feldman BM. Effect of warfarin on survival in scleroderma-associated pulmonary arterial hypertension (SSc-PAH) and idiopathic PAH. Belief elicitation for Bayesian priors. J Rheumatol. 2011;38:462-9.

86

29. SHELF: the Sheffield Elicitation Framework (version 2.0) [computer program]. Version 2.0. Sheffield, UK: School of Mathematics and Statistics, University of Sheffield; 2010.

30. R: A language and environment for statistical computing [computer program]. Vienna, Austria: R Foundation for Statistical Computing; 2012.

31. Billingham L, Malottki K, Steven N. Small sample sizes in clinical trials: a statistician’s perspective. Clinical Investigation. 2012;2:655-7.

32. Chaloner K, Rhame FS. Quantifying and documenting prior beliefs in clinical trials. Stat Med. 2001;20:581-600.

33. Diamond IR, Grant RC, Feldman BM, et al. Expert beliefs regarding novel lipid-based approaches to pediatric intestinal failure-associated liver disease. JPEN. Journal of parenteral and enteral nutrition. 2014;38:702-10.

34. Deandrea S, Negri E, Ruggeri F. Integrating clinicians’ opinion in the Bayesian meta-analysis of observational studies: the case of risk factors for falls in community-dwelling older people. Epidemiology, Biostatistics and Public Health. 2014;11:e8909.

35. Morris DE, Oakley JE, Crowe JA. A web-based tool for eliciting probability distributions from experts. Environmental Modelling & Software. 2014;52:1-4.

36. Myburgh C, Mouton J. The development of contemporary chiropractic education in Denmark: an exploratory study. J Manipulative Physiol Ther. 2008;31:583-92.

37. Humphreys BK, Peterson CK, Muehlemann D, Haueter P. Are Swiss chiropractors different than other chiropractors? Results of the job analysis survey 2009. J Manipulative Physiol Ther. 2010;33:519-35.

38. Chenot JF, Becker A, Leonhardt C, et al. Use of complementary alternative medicine for low back pain consulting in general practice: a cohort study. BMC Complement Altern Med. 2007;7:42.

87

Chapter 4

Chiropractic Care and Risk For Acute Lumbar Disc 4Herniation: A Population-Based Self-Controlled Case Series

Study

Manuscript 3

Hincapié CA, Tomlinson GA, Côté P, Rampersaud YR, Jadad AR, Cassidy JD.

Chiropractic care and risk for acute lumbar disc herniation: a population-based self-

controlled case series study. In preparation for submission, 2015

Chiropractic care and risk for acute lumbar disc herniation:

a population-based self-controlled case series study

Authors: Cesar A. Hincapié, DC, MHSc,1,2 George A. Tomlinson, PhD,1,3 Pierre Côté,

DC, PhD,4 Y. Raja Rampersaud, MD, MSc,5 Alejandro R. Jadad, MD, DPhil,1,3 J. David

Cassidy, PhD, DrMedSc1,2,6

Affiliations: 1 Dalla Lana School of Public Health, University of Toronto; 2 Toronto

Western Research Institute, University Health Network; 3 Toronto General Research

Institute, University Health Network; 4 Faculty of Health Sciences, University of Ontario

Institute of Technology; 5 Division of Orthopaedic Surgery, Toronto Western Hospital; 6

Department of Sports Science and Clinical Biomechanics, University of Southern

Denmark

Corresponding Author: Cesar A. Hincapié, DC, MHSc, University Health Network,

LuCliff Place – 700 Bay Street, Suite 602, Toronto, Ontario M5G 1Z6, Canada. E-mail:

cesar.hincapie@utoronto.ca

Key Words: chiropractic; spinal manipulation; intervertebral disc displacement;

epidemiologic methods; risk; adverse events; self-controlled case series

Word Count: 4,966

88

Abstract

Background: Chiropractic care is popular for low back pain, but may increase the risk

for acute lumbar disc herniation (LDH). Low back pain is a common early symptom of

LDH, which commonly precedes LDH diagnosis.

Objective: To investigate associations between chiropractic visits and acute LDH with

early surgical intervention (early surgery, ≤8 weeks), and compare this to associations

between primary care physician (PCP) visits and acute LDH with early surgery.

Study design and setting: Self-controlled case series method using 4 population-

based health care databases in Ontario, Canada.

Methods: All patients with a recorded acute LDH with incident early surgery from April

1994 to December 2004 were selected for the study. Health care exposure visits to

chiropractors and PCPs in the year before the acute LDH event index date were

determined from health billing records. Incidence ratios for acute LDH with early

surgery in exposed periods after chiropractic visits relative to unexposed periods were

estimated within individuals, and compared with incidence ratios for acute LDH with

early surgery following PCP visits.

Results: We identified 195 cases of acute LDH with early surgery in a population of

more than 100 million person-years. We found evidence of strong positive associations

between both chiropractic and PCP visits, and acute LDH. Compared with the risk for

acute LDH with early surgery associated with PCP visits, there was no increase in the

risk associated with chiropractic visits. Lumbar spine-related health care visits were

highly associated with subsequent acute LDH with early surgery.

Conclusion: Acute LDH with early surgical intervention is an exceedingly rare event in

Ontario. Both chiropractic and primary medical care may be associated with an

increase in the risk for acute LDH with early surgery. However, these positive

associations are likely due to patients with low back pain from a developing disc

herniation seeking health care before full clinical expression of acute LDH. We found

no evidence of excess risk for acute LDH with early surgery associated with chiropractic

care compared with PCP care.

89

4.1 Introduction

Low back pain is recognized as an important public health concern around the world

because it is associated with considerable disability and burden to individuals, industry

and society.1-5 It affects about 70% of all people in their lifetime, and between 15% to

30% on any given day with varying types of clinical presentations.6-8

Symptomatic lumbar disc herniation (LDH) can be one of the most recognizable

presentations of low back pain. The diagnosis is typically based on a combination of

symptoms and signs suggesting lumbar spinal nerve root compression or irritation, such

as: lumbosacral radiculopathy (i.e., radicular leg pain or sciatica), nerve root tension

signs, neurologic deficits (i.e., muscle weakness and reflex changes), and advanced

imaging (i.e., MRI or CT) findings that correlate with the clinical syndrome.9-11 However,

many patients present with a less clear clinical picture, involving low back pain in the

early (prodromal) phase that then progresses to radicular leg pain with or without

neurologic signs.12,13 In addition, diagnostic imaging may only be indicated in the

prodromal clinical course if there is suspicion of underlying pathology (e.g., infection,

malignancy) other than disc herniation.14 All these factors contribute in making the

diagnosis during the early phase of the syndrome especially difficult.

In North America, primarily medical physicians, chiropractors and physical therapists

clinically manage back pain.15 Approximately 12% of American and Canadian adults

seek chiropractic care annually, and about 95% of chiropractic visits involve spinal

manipulation treatment.16-19 Chiropractic patients, compared to those seeking medical

care for back pain, tend to be younger and have higher socioeconomic status and fewer

health problems.15,16 Several systematic reviews suggest that chiropractic spinal

manipulation can benefit low back pain, but the summarized studies are of varying

quality and too small to evaluate the risk for rare serious adverse events.20-23 In

addition, there is evidence from randomized clinical trials showing some benefit of

spinal manipulation for the management of LDH,24-28 yet little is known about the safety

of this treatment.

Today, no valid estimate of the risk for acute disc herniation following chiropractic

treatment is available in the scientific literature. The current literature presents case

90

reports and small case series linking lumbar spine manipulation to disc herniation and

cauda equina syndrome.29 However, case reports offer the lowest level of scientific

evidence with regard to the determination of risk and cannot be used to make valid

inferences about the lack of safety of a treatment. They have, however, raised the

hypothesis of potential harm.

Individuals in the early phases of a symptomatic LDH often complain of back pain.12,13

As the condition progresses, most develop leg pain or sciatica. At different points in

time along this course they may seek health care for assessment and intervention. If

chiropractic treatment occurs before the disc herniation progresses to the presence of

radiculopathy or neurologic deficits and is thus diagnosed, then the treatment itself can

be erroneously blamed for causing the disc herniation, since the patient may have

progressed regardless of treatment. This systematic error (protopathic bias) is a type of

reverse-causality bias due to processes that occur before a diagnosed or measured

outcome event.30,31 Given that LDH can initially present as low back pain, it is possible

that these patients seek chiropractic care in the prodrome or early-stage of their disc

herniation,32 implying that an observed association between chiropractic care and acute

LDH may not be causal. Since patients also commonly see primary care physicians

(PCPs) for back pain and this health care encounter is unlikely to cause disc herniation,

an observed association between PCP visits and acute LDH could be attributed to care-

seeking for initial symptoms of LDH (protopathic bias).

Rigorous epidemiological studies are needed to assess the possibility of increased risk

for acute LDH following chiropractic care. The purpose of our study was to investigate

the association between chiropractic care and acute LDH with incident early surgery,

and compare it to the association between PCP care and acute LDH with incident early

surgery by using population-based health care data from Ontario, Canada and the self-

controlled case series method. We hypothesized that evidence that chiropractic care

increases the risk for acute LDH would be present if the measured association between

chiropractic visits and acute LDH exceeds the association between PCP visits and

acute LDH.

91

4.2 Methods

Data Sources: We combined data from 4 large population-based health care

databases, totaling over 100 million person-years of observation: 1) the Ontario

Registered Persons Database; 2) the Discharge Abstract Database (DAD); 3) the

National Ambulatory Care Reporting System (NACRS); and 4) the Ontario Health

Insurance Plan Billings Database (OHIP). These health administrative databases

include information on patients’ hospitalizations with diagnosis and surgical intervention

coding, emergency department (ED) visits with diagnosis and intervention coding, and

practitioner (chiropractor and physician) and other health services utilization as

documented by fee-for-service billings with diagnosis coding. Linkage between

datasets was done using encrypted health card numbers as unique identifiers, and all

health information obtained was anonymous. We requested and received these data

from the Health Data Branch of the Ontario Ministry of Health and Long-Term Care.

Research ethics board approval was received at the University Health Network (REB

#09-0668-AE), and administrative approval from the Office of Research Ethics at the

University of Toronto (protocol reference #24781).

Source Population: The source population was all Ontario residents, aged 18 years or

older, who were covered by the provincial universal health care system (Ontario Health

Insurance Plan) between April 1, 1994 and November 30, 2004. We could not study

chiropractic care after November 30, 2004, as this was the last date that chiropractic

care was covered under OHIP and visits after this date could not be identified for the

purpose of our study. Members of the armed forces, federal inmates, the Royal

Canadian Mounted Police, and First Nation individuals living on reserves were ineligible

for the study, as they are not covered by the provincial health care system. Participants

with a history of long term care service in the 2 years prior to their surgery were also

excluded.

Outcome and Case Definition: Eligible cases were all cases of acute LDH with

incident early surgical intervention (≤8 weeks) between April 1, 1994 and November 30,

2004, with at least 2 years of health care coverage prior to their surgery to ensure that

visits to chiropractors and physicians (exposures) could be identified. We used the

92

following 4-step approach to define cases of acute LDH with incident early surgery

based on the coding detailed in Appendix H, Table H.1:

1. Initial cases of incident LDH surgery were identified using at least one disc surgery

intervention code and one LDH diagnosis code from DAD or NACRS data. Hospital

discharge ICD-9 codes for LDH have been shown to have a high positive predictive

value when compared to chart review.33

2. Preliminary cases of acute LDH with incident early surgery were then identified by

including persons (from Step 1 above) who presented to a hospital ED for LDH within 8

weeks prior to their LDH surgery using OHIP and NACRS data.

The 8-week ED visit window (ED window) prior to LDH surgery was chosen in

consultation with a spine surgery expert (YRR) and data suggesting early surgery for

severe or intractable radiculopathy due to disc herniation can occur up to 6 to 12 weeks

following the onset of symptoms.34,35 Given the typically long surgical wait times in the

Ontario health system during our study period and the inherent limitations of

administrative health data,36-40 it was felt that this case scenario (early surgery following

an ED visit for LDH) would most reliably represent a clinically relevant and significant

acute LDH event, and possible rare serious adverse event following chiropractic

treatment. The event index date was defined as this date of service for LDH at the ED

within 8 weeks prior to the LDH surgery date.

3. Using DAD and NACRS data, we excluded persons who had a diagnosis of LDH or

other conditions potentially associated with LDH, 21 months prior to their event index

date. Excluded also were those with other spine surgery interventions on or prior to their

date of incident LDH surgery.

4. Using OHIP data, we excluded persons who had, within 21 months prior to their

event index date, a diagnosis of LDH or other conditions potentially associated with

LDH, specialist visits to neurosurgeons, orthopedic surgeons, neurologists, physiatrists

and rheumatologists, or advanced spine imaging or diagnostic testing related to LDH.

Our outcome of interest was acute LDH with incident early surgical intervention. We did

not include cauda equina syndrome as a target for our case definition because this

93

more severe serious adverse event was investigated in another study41 that formed part

of an overall epidemiologic research program on the safety of chiropractic care.

Exposures: All ambulatory care billing records for visits to chiropractors and PCPs

were identified in the OHIP data (codes in Appendix H, Table H.2) starting from the

one-year period before the event index date (start of the observation period). Multiple

billings on the same date were counted as one visit. Lumbar spine-related chiropractic

visits were defined using diagnosis codes: C07-C09, lumbar, lumbosacral, sacroiliac

and coccyx subluxation; C13-15, multiple site subluxation; C20-24, acquired, postural or

congenital spine curvature; C31-C32, lumbosacral sprain/strain. For PCP visits, we

included community medicine physicians if they submitted ambulatory fee codes to

OHIP. PCP visits for natal care, ophthalmology care, genetic screening and group

counselling were excluded. Lumbar spine related PCP visits were identified using

diagnosis codes: 724, lumbar strain, lumbago, coccydynia, sciatica; and 847, low back,

coccyx sprain/strain. When addressing repeat exposures (repeat chiropractic or PCP

visits), we assumed the risk to remain constant after each exposure, thus not allowing

for a dose effect.

Study Design: The self-controlled case series method is a type of a cohort study in

which the relative risk is based on within-person comparisons rather than between-

person comparisons, with each person contributing to both the exposed and unexposed

observation time.42,43 This means the potential confounding effects of differences

between patients are removed if these factors are fixed (time-invariant) over the

observation period.

We examined the risk for acute LDH after exposure to chiropractic care and PCP care.

Persons who have had health care encounters (chiropractic or medical) may differ from

those who have not in ways that can be difficult to measure and control for. Some of

these differences may also be associated with the future risk for disc herniation, which

makes a conventional observational (case-control or cohort) design a less valid source

of information on this association. Therefore, we used the self-controlled case series

method, which relies on within-person comparisons in a sample of persons, all of which

have had the outcome of interest (cases-only). This design is most appropriate when a

brief, well-defined exposure (chiropractic visit) causes a transient change in risk of a

94

rare-onset disease (acute LDH with early surgery). The main advantage of this design

is that inference is within a person, so that fixed confounders (those that do not vary

with time during the observation period) are controlled for by design. We calculated

incidence rate ratios of events occurring during prespecified risk periods after an

exposure (0-2 days after a visit, 0-7 days after a visit, 0-14 days after a visit, and 0-28

days after a visit), relative to all other observed time (baseline unexposed periods) for

each person. Figure 4.1 illustrates the self-controlled cases series design.

Figure 4.1. Representation of the self-controlled case series design. Times are expressed in terms

of calendar days from the start of the study period and the data are viewed as in a prospective cohort.

Incidence rate ratios compare the rate of events during exposed periods with the rate of events during

unexposed periods. Two possible scenarios (each representing a single patient) for the timing of

chiropractic or primary care physician visits and acute LDH events are shown. A. Patient has three risk

periods (each after a chiropractic or primary care physician visit) and has an acute LDH event at baseline

between the second and third visits. B. Patient has two visits and has an acute LDH during the second

risk period. Each risk period begins the day of a visit and lasts up to 28 days after a visit (risk periods are

not drawn to scale relative to length of baseline periods). Four risk periods were prespecified: 0-2, 0-7, 0-

14, and 0-28 days after a visit

Statistical Analysis: An important assumption underpinning the standard self-

controlled case series method is that the occurrence of an event must not alter the

Start of study period End of study period

Baseline unexposed period

Risk period after a chiropractic or primary care physician visit

Visit Visit Acute LDH event

Acute LDH event

Visit

A

B

95

probability of subsequent exposure. This assumption does not hold for health care

utilization exposures (chiropractic or PCP visits) because the occurrence of a LDH likely

influences the probability of post-event visits to a chiropractor or PCP. When this

assumption fails, a valid and reasonable strategy is to use a recently developed

modification of the method (the pseudo-likelihood approach) that uses counterfactual

modeling and only exposures experienced prior to the event to estimate the relative

incidence.42,44,45 We therefore used this modified self-controlled case series method for

our study.

The observation time in the year prior to the index ED visit was divided into risk periods

and unexposed time. We ran analyses with risk periods of four different lengths: (i) 0-2

days after a visit, (ii) 0-7 days after a visit, (iii) 0-14 days after a visit, and (iv) 0-28 days

after a visit; the baseline period comprised remaining unexposed time. To estimate the

associations between acute LDH and chiropractic care, and acute LDH and PCP care,

we used conditional Poisson regression to calculate incidence rate ratios (IRRs) and

95% confidence intervals for events occurring within each risk period compared to

baseline.42,44-46 Separate analyses were conducted for exposure to lumbar spine related

chiropractic and PCP visits.

We examined the annual incidence of events by fiscal year of health data (1-Apr to 31-

Mar) and observed an unequal distribution of events by fiscal year corresponding to the

implementation of the NACRS database for ED services in Ontario in 2002. To assess

and control for event incidence temporality that may have been related to the initiation

of NACRS on April 1, 2002, we also calculated NACRS-adjusted risk estimates for all

analyses.

Several other sensitivity analyses were undertaken: (1) To assess the impact of our

case conceptualization of early (≤8 weeks) surgical intervention after an ED LDH visit,

all analyses were performed varying the time interval between date of presentation to

the ED for acute LDH and LDH surgery date (in addition to the primary analysis time

interval of up to 8-weeks between an ED LDH visit and LDH surgery date, we examined

time intervals of up to 4- and 12-weeks). (2) To investigate the effect of potential

misclassification of the event diagnosis, we repeated analyses using a broader coding

approach for LDH diagnosis on the event index date (detailed in Appendix H, Table

96

H.1). (3) To examine the effect of event-day visits to chiropractors and PCPs (possible

reverse-causality), we ran analyses with a 1-day lag on exposures in order to estimate

incidence ratios excluding event-day exposure visits. (4) Where possible, we also

performed analyses in subgroups of patients whom only had chiropractic care and those

whom only had PCP care during their observation period. All analyses were performed

using R, version 2.15.147

Finally, we used the nonparametric bootstrap for three additional analyses: (1) To check

the model-based confidence intervals from the conditional Poisson regression for our

primary model, we took 2000 bootstrap resamples from the cases of acute LDH with

early surgery and refitted the models for both DC visits and PCP visits to each sample.

The 2.5th and 97.5th percentiles of the 2000 estimates of the IRRs were used to estimate

the 95% confidence interval for the IRR; (2) For each of the bootstrap samples, we

divided the IRR for a DC visit by the IRR for a PCP visit, thereby estimating the specific

effect on the risk for acute LDH with early surgery of seeing a chiropractor relative to

seeing a primary care physician; (3) We made a scatterplot of the 2000 bootstrapped

pairs of IRRs and a kernel density estimate of the ratio in (2) to better visualize the

relationship between the risks for acute LDH with early surgery associated with each

type of healthcare visit.

Role of the Funding Source: The sponsors of the study had no role in study design,

data collection, data analysis, data interpretation, the writing of the report, or in the

decision to submit the report for publication.

4.3 Results

Identification of Acute LDH Cases: A total of 36,745 persons were identified from the

Ministry of Health data with incident disc surgery during the study period. Of these, we

excluded 423 that were under 18 years of age at the time of their surgery, 180 with a

history of long term care service in the 2 years prior to their surgery, 9,576 with no LDH

diagnosis linked to their disc surgery record, 25,323 with no ED LDH visit within 8

weeks prior to their surgery date, and 1,048 with prevalent LDH or associated

diagnoses and interventions within 21 months prior to the ED event index date. A flow

diagram showing how cases were selected is shown in Figure 4.2. For our primary

97

analysis, 195 incident cases of acute LDH with early surgery met our inclusion/exclusion

criteria over the 11-year study period, and are described in Table 4.1. The mean age of

cases was 43 years at the LDH surgery date, and 60% were male. Of the 195 cases,

72 (37%) had visited a chiropractor during the observation period starting 12 months

prior to their event index date, while 186 (95%) had visited a PCP within that time.

Comorbidities were approximately as common in patients with chiropractic visits as in

patients with PCP visits.

98

Figure 4.2. Case definition flow diagram.

36,745 persons who had incident disc surgery from 1 April 1994 to 30 November 2004

General eligibility exclusions (n=603)

423 under 18 years of age on disc surgery date

180 with long-term care service within 2 years prior to disc surgery date

26,566 initial cases of incident LDH surgery

Excluded persons with no LDH diagnosis linked to disc surgery (n=9,576)

Top 5 diagnoses among 23,830 excluded hospitalization records 722.0 – cervical disc herniation (n=2,175; 9.1%)

724.0 – spinal stenosis other than cervical (n=1,720; 7.2%)

722.5 – degeneration of thoracic or lumbar IVD (n=1,342; 5.6%)

401.9 – unspecified essential hypertension (n=807; 3.4%)

721.1 – cervical spondylosis with myelopathy (n=601; 2.5%)

344.6 & G83.4 – cauda equine syndrome (n=223; 0.9%)

36,142 eligible cases of incident disc surgery

1,243 preliminary cases of acute LDH with incident early surgery

Excluded persons with no ED LDH visit within 8 weeks prior to LDH surgery

date (n=25,323)

195 cases of acute LDH with incident early surgery; 72 with DC visits, 186 with PCP visits

Excluded persons who had other spine surgery interventions on or prior to

LDH surgery date; and, persons with a diagnosis of LDH or other

conditions potentially associated with LDH within 21 months prior to the

event index date, using DAD and NACRS data (n=617)

Excluded persons who had (within 21 months prior to the event index date)

a diagnosis of LDH or other conditions potentially associated with LDH,

specialist visits, or advanced spine imaging or diagnostic testing indicating

LDH, using OHIP data (n=431)

99

Table 4.1. Characteristics of study participants at the time of the recording of an acute lumbar disc

herniation with incident early surgery

Characteristic

All cases

(n=195)*

Cases with

DC visits

(n=72)†

Cases with

PCP visits

(n=186)

Men – n (%) 116 (59.5) 47 (65.3) 110 (59.1)

Age at the time of LDH surgery – mean years (SD) 42.6 (12.9) 42.1 (11.9) 42.5 (12.7)

Comorbidities‡ – n (%)

Diabetes 11 (5.6) 6 (8.3) 11 (5.9)

Hypertension 33 (16.9) 15 (20.8) 32 (17.2)

Coronary heart disease 26 (13.3) 9 (12.5) 25 (13.4)

High cholesterol 17 (8.7) 8 (11.1) 17 (9.1)

Osteoarthritis 28 (14.4) 12 (16.7) 28 (15.1)

Rheumatoid arthritis 0 0 0

Abbreviations: DC, chiropractor; LDH, lumbar disc herniation; PCP, primary care physician

* Persons included in the primary analysis of acute LDH with early surgery

† The majority of cases (n not reportable) had both DC and PCP visits during their observation period

‡ Comorbid conditions identified using OHIP billing diagnosis codes during 2 years prior to the LDH

surgery: diabetes (code 250), hypertension (codes 401 to 403), coronary heart disease (codes 410 to

415 and 426 to 429), high cholesterol (code 272), osteoarthritis (code 715), rheumatoid arthritis (code

714)

Description of Visits: Overall, 22% of cases had a chiropractic visit within 14 days

prior to the event index date, while 59% of cases had a PCP visit within that same time

interval (Table 4.2). A total of 57 cases were identified during the first 8 years of the

study period (April 1, 1994 to March 31, 2002), while 138 cases were identified during

the balance of the study period (April 1, 2002 to November 30, 2004). We found an

uneven distribution of event incidence by fiscal year of Ministry of Health data, which

corresponded with the initiation of the NACRS database for ED services in Ontario, on

April 1, 2002. For cases whose event occurred prior to the 2002 fiscal year (pre-

NACRS), 9 cases (15.8%) had consulted a chiropractor within 7 days of their index

date, and 27 cases (47.4%) had consulted a PCP within the same interval. Among

cases whose event occurred after the start of the NACRS database (event fiscal year

100

≥2002), 14.5% had 4 or more chiropractic visits within the month before the index date,

while 9.4% had 4 or more PCP visits within the same time period.

Table 4.2. Distribution of chiropractor and primary care physician visits before the event index date for

all cases and stratified by event fiscal year*

Exposure

All cases

(n=195)

Event fiscal

year <2002

(n=57)

Event fiscal

year ≥2002

(n=138)

Most recent DC visit before event – n cases (%)

0-2 days 23 (11.8) † 19 (13.8)

0-7 days 38 (19.5) 9 (15.8) 29 (21.0)

0-14 days 42 (21.5) 11 (19.3) 31 (22.5)

Most recent PCP visit before event – n cases (%)

0-2 days 59 (30.3) 19 (33.3) 40 (29.0)

0-7 days 87 (44.6) 27 (47.4) 60 (43.5)

0-14 days 114 (58.5) 34 (59.6) 80 (58.0)

Number of DC visits in month before event – n cases (%)

None 147 (75.4) 44 (77.2) 103 (74.6)

1 to 3 21 (10.8) 6 (10.5) 15 (10.9)

4 or more 27 (13.8) 7 (12.3) 20 (14.5)

Number of PCP visits in month before event – n cases (%)

None 59 (30.3) 17 (29.8) 42 (30.4)

1 to 3 121 (62.1) 40 (70.2)†† 83 (60.1)

4 or more 15 (7.7) † 13 (9.4)

Abbreviations: DC, chiropractor; LDH, lumbar disc herniation; PCP, primary care physician

* We found an unequal distribution of events by fiscal year of incidence corresponding to the

implementation of the NACRS database for ED services in Ontario, Canada in 2002.

† Cell sizes with fewer than six (6) cases were suppressed to ensure confidentiality.

†† Cell was merged with the “4 or more” group to ensure confidentiality.

101

Self-Controlled Cases Series Results: Using the modified self-controlled cases series

method, there was evidence of an increased association between chiropractic visits and

acute LDH with early surgery regardless of the risk period examined (Table 4.3). For

exposed periods 0 to 7 days after a chiropractor visit, there was a cumulative 9 years of

exposed time and 38 patients experienced an incident acute LDH during this exposed

period (adjusted IRR, 12.91; 95% CI, 7.16-23.28).

There was similar evidence of an increased risk of acute LDH for all risk periods

examined up to 28 days after a visit to a PCP. A total of 28 patient-years of exposed

time accumulated for risk periods 0 to 7 days after a PCP visit and 87 persons

experienced an acute LDH within this risk period (adjusted IRR, 14.48; 95% CI, 9.90-

21.18). For patients with PCP visits, 39 had their acute LDH event occur on the same

day as a visit.

When restricting the analyses to visits related to lumbar spine complaints, we observed

sharp increases in associations for PCP visits, but not for associations for chiropractic

visits, which remained about the same (Table 4.3). Adjustment for the initiation of the

NACRS database (April 1, 2002) attenuated the observed associations, however, our

estimates remained consistently high for all risk periods examined for both chiropractic

and PCP visits, with substantial overlap of confidence intervals.

102

Table 4.3. Incidence rate ratios for an acute lumbar disc herniation with first early surgery event in exposed periods following chiropractic and primary

care physician visits

DC Care (n=72) PCP Care (n=186)

Exposure

Patient

years

n

Acute

LDHs*

Crude IRR for Acute

LDH (95% CI)

NACRS-adjusted IRR

for Acute LDH

(95% CI)

Patient

years

n

Acute

LDHs*

Crude IRR for Acute

LDH (95% CI)

NACRS-adjusted IRR

for Acute LDH

(95 % CI)

Any visit

Unexposed Baseline Baseline Baseline Baseline

0-2 days after visit 5.03 23 9.46 (4.87 - 18.41) 8.26 (4.35 - 15.67) 11.42 59 18.00 (11.80 - 27.45) 15.20 (10.17 - 22.73)

0-7 days after visit 9.11 38 15.70 (8.85 - 27.86) 12.91 (7.16 - 23.28) 28.35 87 17.50 (11.72 - 26.15) 14.48 (9.90 - 21.18)

0-14 days after visit 12.74 42 14.06 (7.80 - 25.35) 11.80 (6.42 - 21.67) 47.66 114 19.77 (12.18 - 32.10) 15.72 (9.96 - 24.80)

0-28 days after visit 18.60 48 14.41 (7.66 - 27.11) 11.72 (6.06 - 22.68) 76.82 136 33.57 (17.39 - 64.78) 25.67 (13.56 - 48.60)

Lumbar spine visit

Unexposed Baseline Baseline Baseline Baseline

0-2 days after visit 4.18 20 11.27 (5.86 - 21.68) 9.72 (5.11 - 18.48) 4.48 43 45.94 (26.52 - 79.56) 40.61 (24.15 - 68.31)

0-7 days after visit 7.45 33 16.80 (9.05 - 31.21) 13.98 (7.36 - 26.57) 11.08 63 44.89 (25.07 - 80.36) 37.92 (21.78 - 66.01)

0-14 days after visit 10.47 36 14.91 (7.96 - 27.90) 12.22 (6.32 - 23.63) 18.51 79 54.68 (28.62 - 104.50) 44.12 (23.42 - 83.13)

0-28 days after visit 15.47 41 15.48 (7.91 - 30.29) 12.25 (5.96 - 25.18) 29.38 93 85.29 (41.31 - 176.10) 67.02 (31.37 - 143.18)

Abbreviations: CI, confidence interval; DC, chiropractor; IRR, incidence rate ratio; LDH, lumbar disc herniation; NACRS, national ambulatory care

reporting system; PCP, primary care physician.

* Number of acute LDH with incident early surgery events that occurred during the exposed period.

103

Results of Sensitivity and Secondary Analyses: Qualitatively, the sensitivity

analyses gave similar results to those obtained in the primary analysis (Table 4.4). The

results of the sensitivity analyses showed a consistent trend of strong positive

associations between both chiropractic and PCP visits and acute LDH with incident

early surgical intervention; with increases in associations between acute LDH and PCP

lumbar spine visits.

The bootstrap estimates of the IRRs in the primary analysis of any visit exposure with a

risk period of 0-7 days was in line with our model estimates: adjusted IRR for acute LDH

with early surgery, 9.50 (95%CI, 3.01-35.51) for chiropractic care, compared with 14.45

(95%CI, 7.60-29.28) for PCP care (Figure 4.3).

Finally, the bootstrap analysis of the ratio of IRRs suggested a positive safety profile for

chiropractic care relative to the baseline risk represented by PCP care (Figure 4.4), with

a median ratio of the two IRRs of 0.65 (95%CI, 0.19-2.43). The majority (72%) of the

2000 bootstrapped values had a smaller incidence ratio for a DC visit than for a PCP

visit (Figure 4.3).

104

Table 4.4. Results of sensitivity analyses compared to primary analysis of an acute lumbar disc herniation with first early surgery event in risk periods 0

to 7 days after chiropractor and primary care physician visits*

DC Care PCP Care

Patient

years

n Acute

LDHs†

NACRS-adjusted

IRR for Acute LDH

(95% CI)

Patient

years

n

Acute

LDHs†

NACRS-adjusted

IRR for Acute LDH

(95 % CI)

Primary analysis‡ (n=195)

Any visit 9.11 38 12.91 (7.16 – 23.28) 28.35 87 14.48 (9.90 – 21.18)

Lumbar spine visit 7.45 33 13.98 (7.36 – 26.57) 11.08 63 37.92 (21.78 – 66.01)

Sensitivity analyses by specified criteria

Broad inclusion coding for index ED LDH visit§ (n=724)

8-week ED window, broad coding, any visit 38.39 123 10.79 (7.80 – 14.93) 107.58 333 18.94 (14.67 – 24.44)

8-week ED window, broad coding, lumbar spine visit 29.48 102 12.10 (8.42 – 17.39) 41.54 252 63.49 (43.95 – 91.72)

ED LDH visit within 4-weeks (n=258) and 12-weeks (n=961) prior to LDH surgery date

4-week ED window, narrow coding, any visit 6.11 27 13.35 (6.53 – 27.28) 17.45 59 14.00 (8.77 – 22.36)

4-week ED window, narrow coding, lumbar spine visit 4.82 24 15.72 (6.96 – 35.52) 6.10 46 43.68 (22.28 – 85.63)

12-week ED window, narrow coding, any visit 13.41 50 9.93 (5.82 – 16.95) 40.63 117 14.01 (9.96 – 19.71)

12-week ED window, narrow coding, lumbar spine visit 10.88 41 9.86 (5.51 – 17.62) 16.52 84 37.10 (22.96 – 59.95)

Exposure visits lagged by 1 day (exclude potential event-day reverse-causality) (n=195)

Lagged exposures, 8-week ED window, narrow coding, any visit 9.11 39 14.15 (7.73 – 25.90) 28.38 74 10.85 (7.43 – 15.85)

Lagged exposures, 8-week ED window, narrow coding, lumbar spine

visit 7.46 34 15.60 (8.04 – 30.27) 11.08 51 31.77 (17.96 – 56.19)

105

DC Care PCP Care

Patient

years

n Acute

LDHs†

NACRS-adjusted

IRR for Acute LDH

(95% CI)

Patient

years

n

Acute

LDHs†

NACRS-adjusted

IRR for Acute LDH

(95 % CI)

Patients with only DC care versus patients with only PCP care

8-week ED window, narrow coding, any visit ND ND ND 6.28 33 19.45 (11.89 – 31.82)

8-week ED window, narrow coding, lumbar spine visit ND ND ND 5.65 30 40.58 (18.99 – 86.74)

12-week ED window, broad coding, any visit 2.10 6 8.94 (2.16 – 37.02) 86.61 234 18.49 (13.97 – 24.45)

12-week ED window, broad coding, lumbar spine visit 1.65 || 10.56 (1.77 – 63.04) 33.26 163 62.09 (43.17 – 89.32)

Abbreviations: CI, confidence interval; DC, chiropractor; ED, emergency department; IRR, incidence rate ratio; LDH, lumbar disc herniation; NACRS,

national ambulatory care reporting system; ND, no data; PCP, primary care physician.

* Exposed periods 0-7 days after a visit compared with baseline. Baseline period is all observation time except for the 8-day risk period after a visit.

† Number of acute LDH with incident early surgery events that occurred during the exposed period.

‡ 8-week ED window, narrow inclusion coding (see Step 2 of Appendix H, Table H.1), IRR for acute LDH within exposed periods 0-7 days after visit.

§ In addition to the ED LDH diagnosis codes used for the primary analysis (narrow coding) (Step 2 of Appendix H, Table H.1), broad coding for ED LDH

diagnosis included codes 724, 847 and M545.

|| Cell sizes with fewer than six (6) cases were suppressed to ensure confidentiality.

106

Figure 4.3. Bootstrap incidence rate ratios (IRRs) for primary analysis of any visit exposure with a risk

period of 0-7 days following chiropractic versus primary care physician care.

Figure 4.4. Kernel density estimate of the ratio of bootstrap incidence rate ratios for chiropractic versus

primary care physician care.

107

4.4 Discussion

We investigated the relationship between chiropractic care and the risk for acute LDH

with early surgical intervention in adults by performing a self-controlled case series

study. Our study showed that among adults, chiropractic care may be positively

associated with acute LDH that is managed with early surgery.

We also found strong positive associations between PCP visits and subsequent acute

LDH that is surgically managed. We acknowledge that protopathic bias cannot be

excluded as a plausible explanation for this finding, by which some patients with low

back pain, a common early symptom of LDH,12,13 may have sought health care due to

this prodromal symptom prior to LDH being diagnosed. Since it is unlikely that PCPs

cause LDH through their care for these patients, we posit that that the observed

association between recent PCP visits and early surgically managed acute LDH

represents the background risk associated with patients seeking care for early

prodromal symptoms of LDH. Because the associations between chiropractic care and

acute LDH were not greater than the associations between PCP visits and acute LDH,

there appears to be no excess risk for acute LDH following chiropractic care greater

than the baseline risk associated with primary medical care seeking.

In studies examining the risk for acute LDH following a health care visit exposure, the

potential for confounding is great because persons who seek health care may differ

from those who do not in ways that may be difficult to measure and control for. A major

strength of our study is that we used the self-controlled case series method, in which

within-person comparisons are done, thereby controlling for both known and unknown

confounding factors that do not change over time and which could affect the risk for

acute LDH. This is important because smoking, obesity, average occupational lumbar

spine load and other important risk factors for symptomatic LDH are not commonly

recorded in health administrative data.

Our findings are similar to a previous study examining the association between

chiropractic care and vertebrobasilar stroke, in which similar associations were found

between both chiropractic care and PCP care, and the outcome of vertebrobasilar

stroke.48 In their study, Cassidy and colleagues reported no evidence of excess risk for

108

vertebrobasilar stroke associated with chiropractic care compared to primary medical

care. They attributed their observed associations to the effect of patients seeking health

care, from both chiropractors and PCPs, for prodromal neck pain and headache before

receiving a diagnosis of vertebrobasilar stroke.48

Strengths and Limitations: Our study population included the entire population

registered in Ontario’s provincial health care system over an 11-year period,

representing over 100 million person-years of observation. This allowed us to identify

all early surgically managed cases of acute LDH, as well as OHIP-insured visits to

chiropractors and PCPs. Yet, we found only 195 early-surgically managed acute LDH

cases for our primary analysis, limiting the precision of our estimates. In particular, our

NACRS-adjusted analyses were based on smaller numbers of exposed cases, and

further stratification by lumbar spine related visit diagnosis codes yielded wider

confidence intervals. Nonetheless, there are few other jurisdictions in the world where it

would be possible to carry out a population study linking incident surgically managed

acute LDH with chiropractic and PCP visits.

A notable strength of our study was the use of the self-controlled case series method

and within-person comparisons. This ensured that confounding by time-invariant

factors was controlled for by design. Confounding could have occurred only if patients

had risk factors for surgically managed acute LDH that changed over time, and if these

factors were also associated with the timing of visits to chiropractors and PCPs, and if

these time-dependent effects existed for a high proportion of study participants.

Another strength was that exposures were measured independently of the outcome.

This eliminated the possibility of exposure recall bias. With respect to chiropractic care,

the exposure measure does not necessarily indicate a treatment of lumbar spine

manipulation, although there is evidence that about 95% of chiropractic visits involve

spinal manipulation treatment.16-19

There was some overlap between chiropractic care and PCP care. In our principal

analysis, fewer than six cases (exact number suppressed to ensure confidentiality) had

seen only a chiropractor; and 115 (59.0%) had seen only a PCP. In a sensitivity

analysis we performed using up to a 12-week time interval between ED LDH visit and

LDH surgery date and a broader coding approach for ED LDH diagnosis, we identified a

109

total of 961 cases of acute LDH with incident early surgery, of which 18 cases (1.9%)

had seen only a chiropractor during their observation period, and 514 cases (53.5%)

had seen only a PCP. Although the small number of cases with only chiropractic care

limited the precision of the estimates, this sensitivity analysis made no material

difference to our findings (Table 4.4).

Key limitations of our study should be considered. Misclassification bias is a major

limitation of using administrative health data. Our case conceptualization is an

algorithm of surgical intervention codes, hospital diagnosis codes, and health services

diagnosis and fee codes, from multiple databases and over an 11-year study period.

The reliability and validity of our case definition is not known; however, because of the

high costs of chart abstraction, an attempt to validate the case definition was not

undertaken. It is possible that our case definition algorithm resulted in an overinclusive

or underinclusive cohort of cases, and our definition does not capture events that were

not managed with early surgery. To investigate the impact of our definition of early

surgery, we carried out sensitivity analyses varying the time interval between the LDH

surgery date and the preceding ED visit for disc herniation; from an ED visit within 4

weeks prior to LDH surgery date, to an ED visit within 12 weeks prior to the surgery

date. We also examined the effect of potential misclassification due to the event index

diagnostic coding by repeating analyses using a broader coding approach for ED LDH

diagnosis on the event index date. These sensitivity analyses yielded results similar to

our primary analysis and did not change our conclusions (Table 4.4).

In our study, the exposed period starts on the same day as a visit to a chiropractor or

PCP; therefore, the day of a health encounter visit contributes to the risk period. This

leads to the challenge of events that occurred on the same day as an exposure visit and

the possibility that the exposure visit may have taken place after the ED LDH visit,

which could have plausibly been the case for both chiropractic and PCP event-day

visits. This potential reverse-causality bias would lead to our estimates overestimating

the associations between acute LDH and both chiropractic and PCP care. To assess

the effect of event-day visits to chiropractors and PCPs, we ran a sensitivity analysis

with a 1-day lag on exposure visit dates in order to estimate incidence ratios excluding

event-day exposure visits. This, in effect, re-specified the risk period intervals as: 1-3,

1-8, 1-15, and 1-29 days after a visit, thereby removing the effect of event-day exposure

110

visits on our estimates. Our analysis showed some attenuation of the PCP estimates,

but made no material difference to our findings (Table 4.4).

Although our study period could only extend up until December 2004 due to the

limitation of OHIP coverage for chiropractic care in Ontario, we believe that our findings

are likely still relevant today as the practice of chiropractic care for acute low back pain

and the indications for early LDH surgery (i.e., severe intractable pain and/or neurologic

deficit) have not changed materially over ensuing time period.

Our study is the first epidemiologic investigation of the association between chiropractic

care and acute LDH with early surgical intervention. Our findings should be interpreted

cautiously and considered within clinical context. We have not excluded spinal

manipulation as a possible cause of some cases of acute LDH. Our results suggest

that there is a positive association between a patient seeking chiropractic care and

subsequent acute LDH that is surgically managed. However, the association was not

greater than the association we found between PCP care and surgically managed acute

LDH. The observed associations are likely explained by patients seeking health care

for prodromal symptoms of LDH (i.e., protopathic bias). Chiropractic care is not likely to

be a major cause of these rare events. Nevertheless, we cannot exclude the possibility

that chiropractic spinal manipulation, or even simple physical examination by any health

care practitioner, could cause an exacerbation of a developing or underlying disc

herniation leading to full clinical expression of the pre-existing condition.

Regrettably, there are no clinical screening tests to identify patients with back pain that

may be at increased risk of developing acute disc herniation,49 and current evidence

indicates poor diagnostic performance of most physical tests used to identify LDH.10,50

Future studies would need to be multi-centered and undertake prospective, unbiased

and detailed ascertainment of patients’ reasons for seeking health care. With respect to

clinical practice in light of our current state of knowledge, decisions on how to treat

patients with low back pain should continue to be guided primarily by clinical

effectiveness and current best practice.51,52

111

4.5 Conclusion

Our population-based self-controlled case series study found positive associations

between health care visits to both chiropractors and PCPs, and acute LDH with early

surgical intervention. This suggests that patients with prodromal back pain related to a

developing disc herniation seek health care from both chiropractors and PCPs before

full clinical expression of acute LDH that is subsequently managed with early surgery.

Funding Sources

Dr. Cesar Hincapié was supported by a Fellowship Award in the Area of Knowledge

Translation from the Canadian Institutes of Health Research, with additional support

from the Canadian Chiropractic Research Foundation. The study received additional

support from the Canadian Chiropractic Protective Association.

Acknowledgements

We thank Drs. Heather Whitaker and Ronny Kuhnert for their advice about the self-

controlled case series design and analysis of this study. We acknowledge the Ontario

Ministry of Health and Long-Term Care for support with data acquisition and linkage.

References

1. Hoy D, Brooks P, Blyth F, Buchbinder R. The epidemiology of low back pain. Best Pract Res Clin Rheumatol. 2010;24:769-81.

2. Hoy D, March L, Brooks P, et al. The global burden of low back pain: estimates from the Global Burden of Disease 2010 study. Ann Rheum Dis. 2014;73:968-74.

3. Cassidy JD, Côté P, Carroll LJ, Kristman V. Incidence and course of low back pain episodes in the general population. Spine. 2005;30:2817-23.

4. Hincapié CA, Cassidy JD, Côté P. Is a history of work-related low back injury associated with prevalent low back pain and depression in the general population? BMC Musculoskelet Disord. 2008;9:22.

5. Vos T, Flaxman AD, Naghavi M, et al. Years lived with disability (YLDs) for 1160 sequelae of 289 diseases and injuries 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. The Lancet. 2012;380:2163-96.

6. Kent PM, Keating JL. The epidemiology of low back pain in primary care. Chiropr Osteopat. 2005;13:13.

7. Walker BF. The prevalence of low back pain: a systematic review of the literature from 1966 to 1998. J Spinal Disord. 2000;13:205-17.

8. Andersson GB. Epidemiology of low back pain. Acta Orthop Scand Suppl. 1998;281:28-31.

112

9. Genevay S, Atlas SJ, Katz JN. Variation in eligibility criteria from studies of radiculopathy due to a herniated disc and of neurogenic claudication due to lumbar spinal stenosis: a structured literature review. Spine. 2010;35:803-11.

10. van der Windt DAWM, Simons E, Riphagen II, et al. Physical examination for lumbar radiculopathy due to disc herniation in patients with low-back pain. Cochrane Database Syst Rev. 2010:CD007431.

11. Vroomen PC, de Krom MC, Wilmink JT, Kester AD, Knottnerus JA. Diagnostic value of history and physical examination in patients suspected of lumbosacral nerve root compression. J Neurol Neurosurg Psychiatry. 2002;72:630-4.

12. Deyo RA, Loeser JD, Bigos SJ. Herniated lumbar intervertebral disk. Ann Intern Med. 1990;112:598-603.

13. Weber H. The natural history of disc herniation and the influence of intervention. Spine. 1994;19:2234-8; discussion 3.

14. Koes BW, van Tulder MW, Peul WC. Diagnosis and treatment of sciatica. BMJ. 2007;334:1313-7.

15. Côté P, Cassidy JD, Carroll L. The treatment of neck and low back pain: who seeks care? who goes where? Med Care. 2001;39:956-67.

16. Hurwitz EL, Chiang LM. A comparative analysis of chiropractic and general practitioner patients in North America: findings from the joint Canada/United States Survey of Health, 2002-03. BMC Health Serv Res. 2006;6:49.

17. Hurwitz EL, Coulter ID, Adams AH, Genovese BJ, Shekelle PG. Use of chiropractic services from 1985 through 1991 in the United States and Canada. Am J Public Health. 1998;88:771-6.

18. Senstad O, Leboeuf-Yde C, Borchgrevink CF. Side-effects of chiropractic spinal manipulation: types frequency, discomfort and course. Scand J Prim Health Care. 1996;14:50-3.

19. Leboeuf-Yde C, Hennius B, Rudberg E, Leufvenmark P, Thunman M. Side effects of chiropractic treatment: a prospective study. J Manipulative Physiol Ther. 1997;20:511-5.

20. Ferreira ML, Ferreira PH, Latimer J, Herbert R, Maher CG. Efficacy of spinal manipulative therapy for low back pain of less than three months' duration. J Manipulative Physiol Ther. 2003;26:593-601.

21. Bronfort G, Haas M, Evans RL, Bouter LM. Efficacy of spinal manipulation and mobilization for low back pain and neck pain: a systematic review and best evidence synthesis. Spine J. 2004;4:335-56.

22. Chou R, Huffman LH, American Pain Society, American College of Physicians. Nonpharmacologic therapies for acute and chronic low back pain: a review of the evidence for an American Pain Society/American College of Physicians clinical practice guideline. Ann Intern Med. 2007;147:492-504.

23. Furlan AD, Yazdi F, Tsertsvadze A, et al. A systematic review and meta-analysis of efficacy, cost-effectiveness, and safety of selected complementary and alternative medicine for neck and low-back pain. Evid Based Complement Alternat Med. 2012:953139.

24. Burton AK, Tillotson KM, Cleary J. Single-blind randomised controlled trial of chemonucleolysis and manipulation in the treatment of symptomatic lumbar disc herniation. Eur Spine J. 2000;9:202-7.

25. Liu J, Zhang S. Treatment of protrusion of lumbar intervertebral disc by pulling and turning manipulations. J Tradit Chin Med. 2000;20:195-7.

113

26. Santilli V, Beghi E, Finucci S. Chiropractic manipulation in the treatment of acute back pain and sciatica with disc protrusion: a randomized double-blind clinical trial of active and simulated spinal manipulations. Spine J. 2006;6:131-7.

27. McMorland G, Suter E, Casha S, du Plessis SJ, Hurlbert RJ. Manipulation or microdiskectomy for sciatica? A prospective randomized clinical study. J Manipulative Physiol Ther. 2010;33:576-84.

28. Bronfort G, Hondras MA, Schulz CA, Evans RL, Long CR, Grimm R. Spinal manipulation and home exercise with advice for subacute and chronic back-related leg pain: a trial with adaptive allocation. Ann Intern Med. 2014;161:381-91.

29. Hebert JJ, Stomski NJ, French SD, Rubinstein SM. Serious adverse events and spinal manipulative therapy of the low back region: a systematic review of cases. J Manipulative Physiol Ther. 2013.

30. Maclure M, Fireman B, Nelson JC, et al. When should case-only designs be used for safety monitoring of medical products? Pharmacoepidemiol Drug Saf. 2012;21 Suppl 1:50-61.

31. Salas M, Hofman A, Stricker BH. Confounding by indication: an example of variation in the use of epidemiologic terminology. Am J Epidemiol. 1999;149:981-3.

32. Crawford CM, Hannan RF. Management of acute lumbar disk herniation initially presenting as mechanical low back pain. J Manipulative Physiol Ther. 1999;22:235-44.

33. Faciszewski T, Broste SK, Fardon D. Quality of data regarding diagnoses of spinal disorders in administrative databases. A multicenter study. J Bone Joint Surg Am. 1997;79:1481-8.

34. Peul WC, van den Hout WB, Brand R, Thomeer RT, Koes BW, Leiden-The Hague Spine Intervention Prognostic Study G. Prolonged conservative care versus early surgery in patients with sciatica caused by lumbar disc herniation: two year results of a randomised controlled trial. BMJ. 2008;336:1355-8.

35. Peul WC, van Houwelingen HC, van der Hout WB, et al. Prolonged conservative treatment or 'early' surgery in sciatica caused by a lumbar disc herniation: rationale and design of a randomized trial [ISRCT 26872154]. BMC Musculoskelet Disord. 2005;6:8.

36. Chan BTB, Schull MJ, Schultz SE. Emergency Department Services in Ontario. Toronto: Institute for Clinical Evaluative Services; 2001.

37. Iron K, Jaakkimainen L, Rothwell DM, Ping L, Laupacis A. Investigation of acute lower back pain in Ontario: Are guidelines being followed? Toronto, Ontario: Institute for Clinical Evaluative Sciences; 2004.

38. Jaakkimainen L, Klein-Geltink JE, Guttmann A, et al. Indicators of primary care based on administrative data. In: Jaakkimainen L, Upshur RE, Klein-Geltink JE, et al., eds. Primary care in Ontario: ICES Atlas. Toronto: Institute for Clinical Evaluative Sciences; 2006.

39. Jaakkimainen L, Schultz SE, Klein-Geltink JE, Thiruchelvam D, Kopp A. Ambulatory Physician Care for Adults. In: Jaakkimainen L, Upshur RE, Klein-Geltink JE, et al., eds. Primary care in Ontario: ICES Atlas. Toronto: Institute for Clinical Evaluative Sciences; 2006.

40. Juurlink D, Preyra C, Croxford R, et al. Canadian Institute for Health Information Discharge Abstract Database: A Validation Study. Toronto: Institute for Clinical Evaluative Sciences; 2006.

114

41. Côté P, Boyle E, Choi S, Rampersaud YR, Hincapié CA, Cassidy JD. Chiropractic care and the risk for cauda equina syndrome: A population-based case-crossover study. In Progress. 2015.

42. Whitaker HJ, Hocine MN, Farrington CP. The methodology of self-controlled case series studies. Stat Methods Med Res. 2009;18:7-26.

43. Whitaker HJ, Farrington CP, Spiessens B, Musonda P. Tutorial in biostatistics: the self-controlled case series method. Stat Med. 2006;25:1768-97.

44. Farrington CP, Whitaker HJ, Hocine MN. Case series analysis for censored, perturbed, or curtailed post-event exposures. Biostatistics. 2009;10:3-16.

45. Hua W, Sun G, Dodd CN, et al. A simulation study to compare three self-controlled case series approaches: correction for violation of assumption and evaluation of bias. Pharmacoepidemiol Drug Saf. 2013;22:819-25.

46. Kunhert R, Whitaker HJ. adSCCS: Case series analysis for censored, perturbed or curtailed post-event exposures [computer program]. R package version 1.5. 2012.

47. R: A language and environment for statistical computing [computer program]. Vienna, Austria: R Foundation for Statistical Computing; 2012.

48. Cassidy JD, Boyle E, Côté P, et al. Risk of vertebrobasilar stroke and chiropractic care: results of a population-based case-control and case-crossover study. Spine. 2008;33:S176-83.

49. Hartvigsen L, Kongsted A, Hestbaek L. Clinical examination findings as prognostic factors in low back pain: a systematic review of the literature. Chiropr Man Therap. 2015;23:13.

50. Al Nezari NH, Schneiders AG, Hendrick PA. Neurological examination of the peripheral nervous system to diagnose lumbar spinal disc herniation with suspected radiculopathy: a systematic review and meta-analysis. Spine J. 2013;13:657-74.

51. Koes BW, van Tulder M, Lin CW, Macedo LG, McAuley J, Maher C. An updated overview of clinical guidelines for the management of non-specific low back pain in primary care. Eur Spine J. 2010;19:2075-94.

52. Dagenais S, Tricco AC, Haldeman S. Synthesis of recommendations for the assessment and management of low back pain from recent clinical practice guidelines. Spine J. 2010;10:514-29.

115

Chapter 5

Summary and Synthesis 5

The research reported in this thesis aimed to advance knowledge of the risk for acute

lumbar disc herniation (LDH) following chiropractic care through a mixed methods

scientific approach. First, we synthesized the current evidence on the incidence and

determinants of LDH with radiculopathy in adults. Second, we described and quantified

clinicians’ beliefs about the risk for acute LDH associated with chiropractic spinal

manipulation treatment. Finally, we investigated associations between chiropractic care

and incident acute LDH with early surgical intervention. In this chapter, the main

findings of our studies are summarized, synthesized and placed in broader clinical,

research and public health context. We elaborate on the implications and limitations of

our work and provide guidance for future research.

5.1 Main Findings

In Chapter 2, the incidence and risk factors of LDH with radiculopathy in adults were

examined through a best evidence synthesis of the literature. This chapter provides

evidence of the varying quality and heterogeneity of the literature on the epidemiology

of LDH. The annual incidence ranges roughly between 0.1% and 10%, and is highly

dependent on case definition and source population. Risk factors varied and our

systematic review points to a multifaceted etiology involving relationships between

individual, behavioural and work-related characteristics. We call for the development of

standardized case and surveillance definitions that validly classify the clinical spectrum,

and the need for further well-designed descriptive and analytic studies investigating the

epidemiology of LDH with radiculopathy.

In Chapter 3, we determined clinicians’ beliefs regarding the effect of chiropractic spinal

manipulation on the risk of developing acute LDH and acute LDH that necessitates early

surgical management using a belief elicitation study. Beliefs were diverse, with

interesting and important interprofessional differences, indicating the presence of

community uncertainty. Optimistic clinicians believed that treatment of acute low back

pain with chiropractic spinal manipulation is associated with a 58% decreased risk for

116

acute LDH, while pessimistic clinicians believed that chiropractic care is associated with

a 66% increased risk for surgically managed acute LDH. Our probability distributions

quantify and describe varying clinician beliefs and could be used as prior probabilities in

Bayesian models estimating this exposure-outcome association.

In Chapter 4, the association between chiropractic care and incident acute LDH was

investigated and compared to the association between primary care physician care and

incident early surgically managed acute LDH. Using Ontario administrative health data

and a self-controlled case series design, we found strong positive associations between

both chiropractic and primary care physician visits, and acute LDH with early surgical

intervention. Compared with the risk associated with primary care physician visits, there

was no increase in the risk associated with chiropractic visits. We believe that this

suggests that patients with prodromal back pain related to a developing disc herniation

seek health care from both chiropractors and primary care physicians before full clinical

expression of acute LDH that is subsequently surgically managed.

5.2 Synthesis and Relevance of Findings

Epidemiology of Lumbar Disc Herniation With Radiculopathy: Our systematic

review highlights the many existing gaps in knowledge of the epidemiology of LDH with

radiculopathy in adults. The incidence of hospital-treated LDH with radiculopathy in the

general population ranged from 0.2 to 1.3 per 1,000 person-years, but little is known

about the incidence of clinically defined LDH with radiculopathy in the general

population. As a point of reference, general population studies involving clinical

assessment have estimated the point prevalence of lumbar disc syndrome at about

5%,1,2 varying by sex and age. Our study shows that the incidence of LDH with

radiculopathy is higher among worker populations, ranging from 6.2 per 1,000 person-

years (0.6%) among female nurses in the US to 93 per 1,000 person-years (9.3%)

among forest industry workers in Finland. One other systematic review has reported

even more marked variation in prevalence estimates of sciatica, ranging from 1.6% in

the general population to 43% in a selected working population.3

This study is the first to systematically gather, appraise and synthesize information on

the incidence and determinants of LDH with radiculopathy. It is difficult to draw

117

conclusions about risk factors. These varied and included individual, behavioural and

work-related variables. The best available evidence points to the following modifiable

targets for possible risk reduction: higher BMI, cardiovascular risk factors (in women),

smoking, greater cumulative occupational lumbar load by forward bending postures and

manual materials handling, higher levels of perceived risk of work injury, lower job

control or decision latitude at work, regular or irregular three-shift work or regular night

work in women, and increased time pressure at work.

There is an important need for consensus on clinical and surveillance definitions and

subsequent validation of methods of ascertaining cases. It is encouraging to note the

developing literature in this area,4-10 although difficulties with clinical diagnosis,

prognosis and classification remain. Until there is some consistency of definitions and

their appropriate validation, studies of LDH with radiculopathy will remain so

heterogeneous that comparing incidence estimates and risk factors will continue to

prove challenging.

Our study helps clarify the epidemiology of this important condition and advance

understanding of the risk factors for LDH with radiculopathy. It is a first step in clarifying

what is currently known about the incidence and determinants, so as to inform future

work and understanding. The review advances knowledge for the entire back pain

community including patients, health care practitioners, and back pain researchers.

Clinician Beliefs Regarding Chiropractic Spinal Manipulation and the Risk For

Acute Lumbar Disc Herniation: We have shown that clinician beliefs about

chiropractic treatment as a risk factor for acute symptomatic LDH vary, indicating

uncertainty within a community of back pain clinician experts. Our study supports the

notion of distinct “pools” of belief among chiropractors, family physicians and spine

surgeons, with some clinicians believing in a beneficial effect of chiropractic care and

others mainly expressing a belief of no effect with a chance of harm.

We hypothesize that differences in beliefs might be related to: (1) interprofessional

differences in the interpretation of a very limited body of evidence; (2) the influence of

varying patient-mix and clinical experience across the health professionals; (3) learned

biases as a result of the professionalization and education of health care professionals.

There is evidence from the clinical guideline development literature that clinician

118

judgments (i.e., beliefs) can diverge from research evidence for reasons related to

clinical experience and a weak or limited body of literature.11 In addition, others have

reported on the not surprising tendency for specialists to express overconfidence in the

effectiveness and appropriateness of their intervention.12,13

Using a Bayesian inferential paradigm, elicited beliefs in the form of prior probability

distributions can be used to augment limited epidemiologic data.14-18 The probability

distributions from our study create an opportunity for a Bayesian approach to models

estimating risk, in which informative clinical prior beliefs about the association can be

explicitly and transparently specified as Bayesian priors. This view has been proposed

as a way forward in the field of health services research.19,20 We believe it is particularly

reasonable in the context of investigating contentious potential treatment-related rare

serious adverse events in the absence of compelling scientific data and evidence.

Our study is the first to describe and evaluate clinician beliefs about the association

between chiropractic spinal manipulation and the development of acute LDH. With

respect to clinical and public health significance, this study represents an important first

step in the process of understanding current perceptions about the safety of chiropractic

care. Our hope is that it facilitates interdisciplinary educational initiatives among the

various health professions, as well as knowledge translation with patients, policymakers

and other relevant stakeholders with an interest in the safety of chiropractic care.

Association Between Chiropractic Care and Risk For Acute Lumbar Disc

Herniation: Our study suggests that the risk for acute LDH with early surgery following

chiropractic care is not greater than the baseline risk for acute LDH with early surgery

associated with primary care physician care. We identified 195 cases of early surgically

managed acute LDH in a population of more than 100 million person-years. Using a

self-controlled case series study design, strong positive associations between both

chiropractic and primary care physician visits, and acute LDH were found.

Our study is the first epidemiologic investigation of the association between chiropractic

care and acute LDH that is managed with early surgery. Our findings should be

interpreted cautiously and considered within clinical context. Persons in the early

phases of a symptomatic LDH often complain of back pain, followed by leg pain or

radiculopathy.21,22 With poorer prognosis and quality of life, they have greater pain

119

severity and incur more work loss, medication use, primary health care use and surgery

than those with uncomplicated low back pain.23-27 At different points in time along the

course of their developing disc herniation they may seek health care. If chiropractic

treatment occurs before the disc herniation progresses to the presence of radiculopathy

and is thus diagnosed, then the treatment itself could be erroneously blamed for

causing the disc herniation, since the patient may have progressed regardless of care.

Given our findings of strong positive associations with no excess risk associated with

chiropractic care, we believe that this suggests that patients with prodromal back pain

related to a developing disc herniation seek health care from both chiropractors and

primary care physicians before full clinical expression of acute LDH that subsequently

undergoes early surgical intervention. This is the current best available epidemiologic

assessment of the association between chiropractic care and acute LDH with early

surgical intervention.

5.3 Implications

Clinical Implications: An accurate evaluation of the value of any health care treatment

requires knowledge of both clinical and cost effectiveness, as well as information on the

treatment’s safety profile including valid estimates of the risk for serious adverse events

and the magnitude of risk compared to other treatment alternatives. Our finding of no

excess risk for acute LDH associated with chiropractic care compared to primary

medical care suggests that spinal manipulation should not be considered

contraindicated for the treatment of low back pain. This conclusion, however, should be

qualified by our inability to exclude spinal manipulation as a possible cause of some,

albeit likely very few cases of acute LDH. Nonetheless, in light of the RCT evidence

showing benefit of spinal manipulation for low back pain and LDH with radiculopathy

and our finding of no excess risk associated with chiropractic care, we believe spinal

manipulation should be considered a viable treatment option for these common

disorders. During the clinical management of patients with low back pain with or without

radiculopathy, attention should be paid to presenting signs and symptoms that may

indicate a developing disc herniation and the treatment plan adjusted with the aim of

maximizing potential benefit while reducing potential harm or exacerbation.

120

Methodological Implications: The belief elicitation methodology that we applied to

describe and quantify clinician beliefs is pragmatic, reproducible and not unique to this

clinical area. This method could be translated to other clinical research questions

relevant to the safety profile of chiropractic care (e.g., beliefs and uncertainty

surrounding chiropractic neck manipulation) and generalized to other disciplines as well.

Our belief elicitation method could be used by other researchers who wish to elicit and

quantify belief about risk or treatment effects to develop informative clinical priors for

inclusion in Bayesian models.

The self-controlled case series method that we used is useful for assessing risk and

safety not only for chiropractic care, but also for all types of different care. Given the

dearth of evidence of chiropractic’s safety profile, this method could be used to assess

associations between chiropractic care and cauda equine syndrome,28 and undertake

replication studies to assess the consistency of associations between chiropractic care

and vertebrobasilar stroke examined previously.29 With respect to risk assessments of

other health care, there is already a rich literature of the usefulness of this method.30-32

Finally, our development work for the application of the self-controlled case series

method on the R platform33 may prove useful to other researchers wanting to carry out a

self-controlled case series analysis using the R platform.

Policy Implications: In our self-controlled case series study, we used administrative

health data from the Ontario Ministry of Health and Long-Term Care to carry out our

epidemiologic assessment of chiropractic care. Regrettably, the chiropractic billing data

are no longer available for current or future health services research programs. With

the delisting of chiropractic services from the Ontario Health Insurance Plan (OHIP) as

of December 1, 2004, data collection capturing chiropractic visits and services across

the province ended. Given the current fiscal challenges faced by the Ontario health

system, it is unlikely that chiropractic care will be relisted for coverage through OHIP.

Nonetheless, it would behoove the provincial, national and scientific leadership of the

chiropractic profession with the support and assistance of the provincial and federal

health ministries to attempt to develop a high-quality database that can begin to gather

important clinical practice data once again to facilitate patient-centered research and

improved patient outcomes. It is encouraging to note recent developments on this front,

with the creation of a Canadian Chiropractic Practice-Based Research Network.34

121

Our belief elicitation study represents an important step in the process of understanding

current perceptions about the safety of chiropractic care. This study uses a

characteristic of the Bayesian paradigm—acknowledgement of subjectivity and

context19—as part of the rationale for describing and quantifying clinicians’ beliefs as

probability distributions. The use of subjective priors has a tradition of controversy.

Some object to their use;35 others see value in their explicit and transparent

specification.36 Bayesian methods allow for the possibility that the conclusions of an

analysis may depend on who is conducting it and their available evidence and opinion;

therefore, understanding of subjectivity and context is essential.19

We believe this approach may be particularly useful in the context of investigating

contentious potential treatment-related rare serious adverse events in the absence of

compelling scientific data and evidence. We hope that our findings facilitate

interdisciplinary educational initiatives among the various health professions, and

knowledge translation with policymakers and other relevant stakeholders with an

interest in the safety of chiropractic care.

Legal Implications: There are notable medicolegal implications related to this thesis.

Disc herniation is the leading cause of malpractice claims against chiropractors in

Canada37 and the US.38 Expert opinion is often used on behalf of both plaintiffs and

defendants to provide testimony on scientific and clinical evidence and on causation. In

their recent narrative review of six LDH and cauda equine syndrome court decisions in

Canada, Boucher and Robidoux37 highlighted that all expert witnesses agreed that “no

large scale epidemiological study has yet been conducted to help understand the

relationship between disc herniation and [spinal manipulation].” Our findings fill in this

gap and will likely inform legal proceedings in this area.

5.4 Limitations

Selection Bias: Selection bias is a systematic error due to differences in characteristics

between those who take part in a study and those who do not. The clinicians that

participated in our belief elicitation study (Chapter 3) are a purposive sample. They

may not be representative of the source population of family physicians, chiropractors

and spine surgeons within 2 hours driving distance of the Greater Toronto Area. This

122

limits the external validity of our results, but does not threaten the internal validity of our

study. We hypothesize that respondents may have had more extreme beliefs than

nonrespondents (i.e., those with strong opinions might be more likely to participate), but

had no recourse to assess this possibility. We can, however, consider the implications

on our findings, if this had been the case. If nonparticipants had less extreme beliefs

than participants, there would be little change in the average belief among all

participants, since we found, on average, a neutral belief of “no effect” of chiropractic

manipulation on the risk for disc herniation. Clinician group findings would revert to less

extreme probability distributions, and this would attenuate the chiropractors’ belief of

benefit, as they held the most extreme belief of the three groups. Our study is coherent

with the Bayesian idea that there is value in explicit and transparent formulation of a

“community of priors,”19, and advances knowledge of different viewpoints on the safety

of chiropractic care.

In Chapter 2 and 4, we believe the possibility of selection bias was reduced through our

chosen study designs and research methods. In our systematic review (Chapter 2), we

systematically identified, appraised and synthesized the literature on the incidence and

determinants of LDH with radiculopathy in adults. We applied a best evidence

synthesis methodology to increase the internal validity of our review, which was a

priority given the heterogeneity and limited quality of this body of literature. In Chapter

4, we used a self-controlled case series design. The main strength of this method is

that it uses within-person comparisons, thereby eliminating selection bias and

confounding by constant (time-invariant) characteristics (e.g., smoking history, body

mass history, healthcare-seeking behaviour, average occupational lumbar load) that are

typically not well documented in administrative health data.

Information Bias: Information bias is a systematic error that results because the

information collected about or from study participants is erroneous. These errors may

result in misclassification of exposure status, outcome status, or both. Misclassification

of participants for either exposure or outcome can be either nondifferential or

differential. When the misclassification of outcome is independent of the exposure

status, and vice versa, it is called nondifferential misclassification, and tends to result in

effect estimates that are diluted or attenuated (i.e., biased toward the null value). When

the misclassification of outcome is related to exposure status, or the misclassification of

123

exposure is different for those with and without the outcome, it is differential

misclassification, and the estimate can either exaggerate or underestimate an effect.

Misclassification bias is a major limitation of using administrative health data. In

Chapter 4, we undertook a population-based data linkage study using four Ontario

health care databases. A key limitation is nondifferential misclassification of case

status. The reliability and validity of our case definition is not known. Due to the high

costs of chart abstraction, an attempt to validate the case definition was not undertaken.

Nonetheless, the quality of our case selection was bolstered by a number of strategies.

First, a prespecified 4-step algorithm was developed with careful consideration of the

target outcome of acute LDH with incident early surgical intervention. We reviewed the

methodologies used by previous Ontario health services reports of acute low back

pain,39 primary care services,40,41 emergency department services,42 and the Discharge

Abstract Database,43 and multiple experts in epidemiology and spine surgery were

consulted. Second, we prespecified sensitivity analyses of the case definition to

examine the effect of clinically reasoned variations in our approach to case selection.

Despite our best efforts, it is possible that our case definition was either overinclusive or

underinclusive. However, as the misclassification is nondifferential, this may have lead

to attenuation of our estimates, but would not change our conclusions. Nondifferential

misclassification of the exposure measures was also likely. The reliability and validity of

the codes to classify lumbar spine visits to chiropractors and primary care physicians is

not known. Again, the net result is that risk estimates for acute LDH with early surgery

may be biased toward the null, but this does not affect the conclusions of the study.

Some chiropractic visits (for persons that exhausted the $150 of annual OHIP coverage)

may not have been captured in the OHIP data. However, the proportion of missing

visits is likely small, since the majority of patients have fewer than 15 chiropractic visits

per year.44

In Chapter 3, our findings on clinician beliefs regarding the association between

chiropractic spinal manipulation and acute LDH are preliminary. The reliability and

validity of our approach is not known. We grounded our approach in current best

practice in elicitation, a belief elicitation method45 that has been validated in

rheumatology clinicians, and consulted experts in belief elicitation and the SHELF46

124

elicitation framework in the development of our elicitation interview questionnaire and

methods.

Reverse-Causality Bias: Protopathic bias is a term used if the first (prodromal) signs

and symptoms of a condition are the reason for initial use of a treatment.47-49 This is a

type of reverse-causality bias due to processes that occur before a diagnosed or

measured outcome event.31,49 In Chapter 4, we compared estimates of our focal

association of interest (chiropractic care and acute LDH with early surgery) with

estimates of association between primary care physician care and acute LDH with early

surgery to assess the potential role of protopathic bias. Since patients commonly see

primary care physicians for back pain and these visits are unlikely to cause disc

herniation, we hypothesized that an observed association between primary care

physician visits and acute LDH could be attributed to care-seeking for initial symptoms

of LDH. The consistency of our findings in multiple sensitivity analyses of these

associations gave support to our overall conclusion that the positive associations are

likely due to patients with prodromal back pain from a developing disc herniation

seeking health care before full clinical expression of acute LDH (i.e., protopathic bias).

Chapter 4 also presented us with the additional challenge of assessing reverse-

causality due to event-day exposure visits for both chiropractors and primary care

physicians, and the inability to firmly establish the temporal sequence of a health care

visit (to either chiropractors or primary care physicians) occurring before the ED LDH

visit on the event index date. We applied a statistical solution of specifying a one-day

lag on the exposure visits; thereby removing event-day exposure visits from our

estimates. This sensitivity analysis resulted in attenuation of our estimates, but did not

change our conclusions.

Confounding: Confounding refers to the mixing of effects when an exposure effect is

mixed together with the effect of another variable (confounder) leading to systematic

error in the measurement of the exposure-outcome association. Our studies in

Chapters 2 and 3 are descriptive in nature, so that confounding is not an issue. In

Chapter 4, the main strength of the chosen study design was that comparisons were

within-person, thereby eliminating between-person confounding as well as within-person

confounding due to characteristics (within an individual) that were likely not to change

during the observation period of the analysis. This is an excellent design for assessing

125

potential treatment-induced harm on a population level because of the concern about

potential confounding by factors not recorded in administrative health databases. By

structuring the analysis so that each person serves as their own control, we eliminated

confounding by time-invariant characteristics, such as, sex, location, genetics,

underlying state of health, tendency to seek professional health care, average physical

activity, long-term diet, habitual health behaviours, long-past health events such as

illnesses and injuries, occupation, social support, ethnicity, smoking history, and body

mass history. Nonetheless, the possibility of residual confounding by within-person

factors that do vary with time is still a possibility. For example, a high instantaneous

occupational lumbar load (confounding factor) leading to prodromal back pain and

health-care seeking, which is eventually diagnosed as LDH.

Random Error: Random error is the variability in the data that cannot be explained by

systematic error (bias). This is related to sample size and is reflected in the precision of

the estimates generated by the data. When the sample size of the study is large, the

width of the confidence intervals for estimates from those data is smaller than those

generated from a smaller study. In Chapter 4, the incidence rate ratios (IRR) are based

on small numbers, and the confidence intervals around the estimates are large. This

makes the interpretation of the IRR a little more difficult. Even though many of the

chiropractic care point estimates were smaller than those for primary care physician

care, interpretation and comparison of the associations was informed more by the

relationship of confidence intervals (overlapping vs. nonoverlapping interval estimates)

and thus lead to our conclusion of similar positive associations for both types of health

care visits and acute LDH.

5.5 Future Research

Our work raises some research questions for future studies. Using a Bayesian

approach to health services research, one logical avenue of inquiry would be to

undertake a Bayesian analysis of the association between chiropractic care and acute

LDH with early surgical intervention. Our objective would be to integrate the probability

distributions from Chapter 3 (Bayesian priors) with the findings of Chapter 4 (the

likelihood), to estimate the posterior probability of the association using Bayes’s

theorem. This would allow us to communicate direct probabilities for acute LDH with

126

early surgery given different prior opinions of the clinicians we elicited beliefs from in

Chapter 3 (e.g., optimists vs. pessimists). As an example, Figure 5.1 provides a

preliminary plot of the results of this Bayesian analysis using the mean and standard

deviation of bootstrap log incidence rate ratios for chiropractic versus primary care

physician care from Chapter 4 to form a normal likelihood, and normal approximations

of the optimistic and pessimistic probability distributions from Chapter 3 as prior

distributions on the log ratio of these incidence rate ratios.

Figure 5.1. Preliminary Bayesian triplots of risk for acute lumbar disc herniation with early surgery

associated with chiropractic care.

The knowledge translation value of this approach to interprofessional education and

communication, and open and transparent treatment of prior beliefs would be interesting

to explore and presents a possible area of novel and innovative scientific study.

127

Given the paucity of evidence and available data that can be used to further examine

the association between chiropractic care and the risk for serious adverse events, it

would seem that this line of inquiry may be limited, at least on the epidemiologic front.

Many research questions with respect to perceptions and beliefs regarding the safety of

chiropractic spinal manipulation can be formulated: What are patients’, clinicians’ and

other stakeholders’ beliefs regarding spinal manipulation as a risk factor for disc

herniation or stroke? What are the determinants of beliefs? Can beliefs change and be

“updated” (as Bayes’s theorem proposes) based on limited evidence? Developing web-

based approaches to belief elicitation may allow future elicitation studies to determine

the beliefs of larger samples of participants in valid and reliable ways, but this is a fairly

nascent area of research.50

With respect to lumbar radiculopathy and symptomatic LDH, many epidemiologic and

clinical questions remain to be addressed as our Chapter 2 highlights. What is the

incidence of and risk factors for lumbar radiculopathy in the general population? Are

cardiovascular risk factors (diabetes, high cholesterol, hypertension, and a family history

of coronary heart disease) determinants of LDH in men? What are the prognostic

factors that determine the course of radiculopathy and symptomatic disc herniation?

Furthermore, there is a great need to develop and assess therapeutic approaches that

could benefit those patients with lumbar radiculopathy and symptomatic disc herniation.

These measures might include conventional clinical therapies, but consideration should

also be given to innovative psychosocial and work-place interventions, given our best

evidence synthesis in Chapter 2.

5.6 Conclusion

In this dissertation, we have met our goal of advancing knowledge of the risk for acute

LDH following chiropractic care through an integrated mixed methods approach. Using

the principles and methods of best evidence synthesis, belief elicitation and

epidemiologic investigation, we have successfully integrated methods that use multiple

types of information of an exposure-outcome association with very limited evidence and

great uncertainty, to make more valid estimates of association and perceptions, and

reduce uncertainty. Our work provides evidence of the varying quality and

heterogeneity of the literature on the epidemiology of LDH with radiculopathy; shows

128

that clinician beliefs about the association between chiropractic care and acute LDH are

diverse, with important interprofessional differences, indicating the presence of

community uncertainty; and reports no evidence of excess risk for acute LDH with early

surgical intervention associated with chiropractic care compared with primary medical

care.

References 1. Heliövaara M, Impivaara O, Sievers K, et al. Lumbar disc syndrome in Finland. J

Epidemiol Community Health. 1987;41:251-8. 2. Kaila-Kangas L, Leino-Arjas P, Karppinen J, Viikari-Juntura E, Nykyri E,

Heliovaara M. History of physical work exposures and clinically diagnosed sciatica among working and nonworking Finns aged 30 to 64. Spine. 2009;34:964-9.

3. Konstantinou K, Dunn KM. Sciatica: review of epidemiological studies and prevalence estimates. Spine. 2008;33:2464-72.

4. Lin CW, Verwoerd AJ, Maher CG, et al. How is radiating leg pain defined in randomized controlled trials of conservative treatments in primary care? A systematic review. Eur J Pain. 2014;18:455-64.

5. Al Nezari NH, Schneiders AG, Hendrick PA. Neurological examination of the peripheral nervous system to diagnose lumbar spinal disc herniation with suspected radiculopathy: a systematic review and meta-analysis. Spine J. 2013;13:657-74.

6. Genevay S, Atlas SJ, Katz JN. Variation in eligibility criteria from studies of radiculopathy due to a herniated disc and of neurogenic claudication due to lumbar spinal stenosis: a structured literature review. Spine. 2010;35:803-11.

7. Konstantinou K, Hider SL, Vogel S, Beardmore R, Somerville S. Development of an assessment schedule for patients with low back-associated leg pain in primary care: a Delphi consensus study. Eur Spine J. 2012;21:1241-9.

8. Kreiner DS, Hwang SW, Easa JE, et al. An evidence-based clinical guideline for the diagnosis and treatment of lumbar disc herniation with radiculopathy. Spine J. 2014;14:180-91.

9. van der Windt DAWM, Simons E, Riphagen II, et al. Physical examination for lumbar radiculopathy due to disc herniation in patients with low-back pain. Cochrane Database Syst Rev. 2010:CD007431.

10. Haswell K, Gilmour J, Moore B. Clinical decision rules for identification of low back pain patients with neurologic involvement in primary care. Spine (Phila Pa 1976). 2008;33:68-73.

11. Raine R, Sanderson C, Hutchings A, Carter S, Larkin K, Black N. An experimental study of determinants of group judgments in clinical guideline development. The Lancet. 2004;364:429-37.

12. Fitch K, Lázaro P, Aguilar MD, Martín Y, Bernstein SJ. Physician recommendations for coronary revascularization. The European Journal of Public Health. 1999;9:181-7.

13. Herrin J, Etchason JA, Kahan JP, Brook RH, Ballard DJ. Effect of panel composition on physician ratings of appropriateness of abdominal aortic

129

aneurysm surgery: elucidating differences between multispecialty panel results and specialty society recommendations. Health Policy. 1997;42:67-81.

14. Billingham L, Malottki K, Steven N. Small sample sizes in clinical trials: a statistician’s perspective. Clinical Investigation. 2012;2:655-7.

15. Chaloner K, Rhame FS. Quantifying and documenting prior beliefs in clinical trials. Stat Med. 2001;20:581-600.

16. Dunson DB. Commentary: practical advantages of Bayesian analysis of epidemiologic data. Am J Epidemiol. 2001;153:1222-6.

17. O'Hagan A, Buck CE, Daneshkhah A, et al. Uncertain judgements: eliciting experts' probabilities. West Sussex: John Wiley & Sons Ltd; 2006.

18. Lilford RJ, Thornton JG, Braunholtz D. Clinical trials and rare diseases: a way out of a conundrum. BMJ. 1995;311:1621-5.

19. Spiegelhalter DJ. Incorporating Bayesian ideas into health-care evaluation. Statistical Science. 2004;19:156-74.

20. Spiegelhalter DJ, Myles JP, Jones DR, Abrams KR. Methods in health service research. An introduction to bayesian methods in health technology assessment. BMJ. 1999;319:508-12.

21. Deyo RA, Loeser JD, Bigos SJ. Herniated lumbar intervertebral disk. Ann Intern Med. 1990;112:598-603.

22. Weber H. The natural history of disc herniation and the influence of intervention. Spine. 1994;19:2234-8; discussion 3.

23. Hill JC, Konstantinou K, Egbewale BE, Dunn KM, Lewis M, van der Windt D. Clinical outcomes among low back pain consulters with referred leg pain in primary care. Spine (Phila Pa 1976). 2011;36:2168-75.

24. Kent PM, Keating JL. Can we predict poor recovery from recent-onset nonspecific low back pain? A systematic review. Man Ther. 2008;13:12-28.

25. Kongsted A, Kent P, Albert H, Jensen TS, Manniche C. Patients with low back pain differ from those who also have leg pain or signs of nerve root involvement - a cross-sectional study. BMC Musculoskelet Disord. 2012;13:236.

26. Konstantinou K, Hider SL, Jordan JL, Lewis M, Dunn KM, Hay EM. The impact of low back-related leg pain on outcomes as compared with low back pain alone: a systematic review of the literature. The Clinical journal of pain. 2013;29:644-54.

27. Selim AJ, Ren XS, Fincke G, et al. The importance of radiating leg pain in assessing health outcomes among patients with low back pain. Results from the Veterans Health Study. Spine (Phila Pa 1976). 1998;23:470-4.

28. Côté P, Boyle E, Choi S, Rampersaud YR, Hincapié CA, Cassidy JD. Chiropractic care and the risk for cauda equina syndrome: A population-based case-crossover study. In Progress. 2015.

29. Cassidy JD, Boyle E, Côté P, et al. Risk of vertebrobasilar stroke and chiropractic care: results of a population-based case-control and case-crossover study. Spine. 2008;33:S176-83.

30. Weldeselassie YG, Whitaker HJ, Farrington CP. Use of the self-controlled case-series method in vaccine safety studies: review and recommendations for best practice. Epidemiol Infect. 2011;139:1805-17.

31. Maclure M, Fireman B, Nelson JC, et al. When should case-only designs be used for safety monitoring of medical products? Pharmacoepidemiol Drug Saf. 2012;21 Suppl 1:50-61.

32. Whitaker HJ. The self-controlled case series method website. 2011; http://statistics.open.ac.uk/sccs. Accessed 2015-May-15.

130

33. R: A language and environment for statistical computing [computer program]. Vienna, Austria: R Foundation for Statistical Computing; 2012.

34. Bussières A, Côté P, French S, et al. Creating a Chiropractic Practice-Based Research Network (PBRN): Enhancing the management of musculoskeletal care. The Journal of the Canadian Chiropractic Association. 2014;58:8-15.

35. Gelman A. Objections to Bayesian statistics. 2008:445-9. 36. Diamond GA, Kaul S. Prior convictions: bayesian approaches to the analysis and

interpretation of clinical megatrials. Journal of the American College of Cardiology. 2004;43:1929-39.

37. Boucher P, Robidoux S. Lumbar disc herniation and cauda equina syndrome following spinal manipulative therapy: a review of six court decisions in Canada. Journal of forensic and legal medicine. 2014;22:159-69.

38. Jagbandhansingh MP. Most common causes of chiropractic malpractice lawsuits. J Manipulative Physiol Ther. 1997;20:60-4.

39. Iron K, Jaakkimainen L, Rothwell DM, Ping L, Laupacis A. Investigation of acute lower back pain in Ontario: Are guidelines being followed? Toronto, Ontario: Institute for Clinical Evaluative Sciences; 2004.

40. Jaakkimainen L, Klein-Geltink JE, Guttmann A, et al. Indicators of primary care based on administrative data. In: Jaakkimainen L, Upshur RE, Klein-Geltink JE, et al., eds. Primary care in Ontario: ICES Atlas. Toronto: Institute for Clinical Evaluative Sciences; 2006.

41. Jaakkimainen L, Schultz SE, Klein-Geltink JE, Thiruchelvam D, Kopp A. Ambulatory Physician Care for Adults. In: Jaakkimainen L, Upshur RE, Klein-Geltink JE, et al., eds. Primary care in Ontario: ICES Atlas. Toronto: Institute for Clinical Evaluative Sciences; 2006.

42. Chan BTB, Schull MJ, Schultz SE. Emergency Department Services in Ontario. Toronto: Institute for Clinical Evaluative Services; 2001.

43. Juurlink D, Preyra C, Croxford R, et al. Canadian Institute for Health Information Discharge Abstract Database: A Validation Study. Toronto: Institute for Clinical Evaluative Sciences; 2006.

44. Rothwell DM, Bondy SJ, Williams JI. Chiropractic manipulation and stroke: a population-based case-control study. Stroke. 2001;32:1054-60.

45. Johnson SR, Tomlinson GA, Hawker GA, Granton JT, Grosbein HA, Feldman BM. A valid and reliable belief elicitation method for Bayesian priors. J Clin Epidemiol. 2010;63:370-83.

46. SHELF: the Sheffield Elicitation Framework (version 2.0) [computer program]. Version 2.0. Sheffield, UK: School of Mathematics and Statistics, University of Sheffield; 2010.

47. Jick H, Vessey MP. Case-control studies in the evaluation of drug-induced illness. Am J Epidemiol. 1978;107:1-7.

48. Lanes SF, Delzell E, Dreyer NA, Rothman KJ. Analgesics and kidney disease. Int J Epidemiol. 1986;15:454-5.

49. Salas M, Hofman A, Stricker BH. Confounding by indication: an example of variation in the use of epidemiologic terminology. Am J Epidemiol. 1999;149:981-3.

50. Morris DE, Oakley JE, Crowe JA. A web-based tool for eliciting probability distributions from experts. Environmental Modelling & Software. 2014;52:1-4.

131

Chapter 6

Appendices

Appendix A. Risk of bias assessment summary (Chapter 1) 132

Appendix B. Comparison of the case-crossover (CCO) and self-controlled

case series (SCCS) designs (Chapter 1)

133

Appendix C. Systematic review search strategies (Chapter 2) 134

Appendix D. Detailed evidence tables for systematic review (Chapter 2) 143

Appendix E. Belief elicitation questionnaire (Chapter 3) 168

Appendix F. Belief elicitation script (Chapter 3) 172

Appendix G. Belief elicitation computer protocol (Chapter 3) 178

Appendix H. Case and exposure definition tables for self-controlled case

series study (Chapter 4)

179

132

Appendix A. Risk of bias assessment summary (Chapter 1)

Ade

quat

e se

quen

ce g

ener

atio

n?

Allo

catio

n co

ncea

lmen

t?

Blin

ding

?

Inco

mpl

ete

outc

ome

data

add

ress

ed?

Free

of s

elec

tive

repo

rting

?

Free

of o

ther

bia

s?

Santilli et al7 Y Y Y Y Y U

McMorland et al8 Y Y U Y Y U

Bronfort et al9 Y Y U Y Y U

Abbreviations: Y = yes; N = no; U = unclear

* Low risk of bias assessed using the risk-of-bias assessment recommended

by the Cochrane Collaboration.92 For a study to be considered low risk of

bias none of the 6 domains recommended by Cochrane (randomization,

allocation concealment, blinding, incomplete outcome data, selective

outcome reporting, other) may be rated “no” and no more than 2 may be

rated “unclear.” Because of the inherent complexities involved in blinding

manual treatments, trials that did not blind providers or participants were

rated unclear, provided that study personnel were unlikely to influence

outcome assessments.

Figure. Risk of bias assessment summary

133

Appendix B. Comparison of the case-crossover (CCO) and self-controlled case series (SCCS) designs (Chapter 1)

Characteristic CCO SCCS

“Zero” time Outcome index date Start of observation period prior to the outcome index date

Analytical approach Matched case-control logic Matched cohort logic

Temporal perspective Data are viewed retrospectively from the outcome index date

Data are viewed prospectively from the “zero” time of the study

Statistical model Conditional logistic regression used to estimate the exposure odds ratio

Poisson regression used to estimate the incidence rate ratio

Comparison of interest Exposure status between a time period just before occurrence of the outcome (i.e., hazard period) and one or more referent time periods (i.e., control periods) prior to the hazard period when the outcome did not occur

The likelihood of the outcome occurring in a time period following exposure (i.e., “exposed” period) is compared with the likelihood of the outcome occurring in “unexposed” periods

Handling of temporal data

Time windows (i.e., hazard and control periods) are selected and exposure status is assessed only within the selected time windows

Makes use of all available temporal information without the need for selection of a comparator time window

Strengths Improved control over time-invariant confounders

Suitable for measuring transient effects of accurately recorded brief exposures (e.g., a chiropractic visit) on the immediate risk of a rare-onset disease (e.g., acute lumbar disc herniation)

Improved control over time-invariant confounders

Suitable for measuring transient effects of accurately recorded brief exposures (e.g., a chiropractic visit) on the immediate risk of a rare-onset disease (e.g., acute lumbar disc herniation)

Modified SCCS validly estimates incidence rate ratios in situations where occurrence of the outcome event affects post-event exposures

Reduced susceptibility to exposure-trend bias

Not vulnerable to overlap bias

Weaknesses Susceptible to exposure-trend bias due to the fixed ordering in time of control and case periods (i.e., control periods always precedes the hazard period

Vulnerable to overlap bias (i.e., a version of the bias that arises from choosing non-disjoint strata to partition the population in a matched case-control study)

Bias may be introduced in a standard SCCS analysis if the outcome event influences the likelihood of post-event exposures

134

Appendix C. Systematic review search strategies (Chapter 2)

Ovid MEDLINE(R) 1950 to November Week 3 2010

# Searches Results Search Type

1 Intervertebral Disk Displacement/ 14310 Advanced

2 Intervertebral Disk Displacement?.mp. 14312 Advanced

3 Intervertebral Disc Displacement?.mp. 12 Advanced

4 Inter-vertebral Disk Displacement?.mp. 0 Advanced

5 Inter-vertebral Disc Displacement?.mp. 1 Advanced

6 Sciatica/ 3930 Advanced

7 sciatica.mp. 4846 Advanced

8 ((hernia* or ruptur* or prolaps* or protru* or bulg* or slip* or displac*) adj3 (disc? or disk? or nuclei or nucleus)).mp. 17799 Advanced

9 ((extru* or sequestra* or migrat*) adj2 (disc? or disk? or nuclei or nucleus)).mp. 751 Advanced

10 (discal adj1 hernia*).mp. 162 Advanced

11 (discus adj1 hernia*).mp. 11 Advanced

12 ischialgi*.mp. 151 Advanced

13 lumboischialgi*.mp. 69 Advanced

14 lumbo-ischialgi*.mp. 19 Advanced

15 cauda equina syndrome?.mp. 783 Advanced

16 or/1-15 22004 Advanced

17 Lumbar Vertebrae/ 33587 Advanced

18 Intervertebral Disk/ 9220 Advanced

19 Lumbosacral Region/ 8354 Advanced

20 ((lumbar or lumbosacral or lumbo-sacral) adj1 (spine or region or area or vertebr* or disk? or disc?)).mp. 53048 Advanced

21 discus intervertebra*.mp. 8 Advanced

22 (spinal adj1 (disk? or disc?)).mp. 138 Advanced

23 (low back adj1 (spine or region or area or vertebr*)).mp. 45 Advanced

24 Intervertebral Disk?.mp. 21747 Advanced

25 Intervertebral Disc?.mp. 5565 Advanced

26 Inter-vertebral Disk?.mp. 4 Advanced

27 Inter-vertebral Disc?.mp. 31 Advanced

28 or/17-27 64882 Advanced

29 16 and 28 16624 Advanced

30 incidence/ 142494 Advanced

31 exp risk/ 651849 Advanced

32 exp Cohort Studies/ 803357 Advanced

33 exp Case-Control Studies/ 495517 Advanced

135

34 epidemiologic studies/ 5002 Advanced

35 incidence.mp. 470248 Advanced

36 risk factor?.mp. 549307 Advanced

37 cohort?.mp. 219800 Advanced

38 case control?.mp. 147466 Advanced

39 risk*.mp. 1189985 Advanced

40 between group*.tw. 48950 Advanced

41 relative risk*.tw. 44395 Advanced

42 etiolog*.mp. 149098 Advanced

43 exp Epidemiology/ 18044 Advanced

44 Epidemiologic Methods/ 26634 Advanced

45 epidemiolog*.mp. 250922 Advanced

46 ep.fs. 966847 Advanced

47 or/30-46 2999198 Advanced

48 29 and 47 4835 Advanced

49 limit 48 to yr="1970 -Current" 4770 Advanced

50 remove duplicates from 49 4442 Advanced

51 limit 50 to humans 4299 Advanced Ovid MEDLINE(R) In-Process & Other Non-Indexed Citations December 21, 2010

# Searches Results Search Type

1 Intervertebral Disk Displacement/ 7 Advanced

2 Intervertebral Disk Displacement?.mp. 7 Advanced

3 Intervertebral Disc Displacement?.mp. 0 Advanced

4 Inter-vertebral Disk Displacement?.mp. 0 Advanced

5 Inter-vertebral Disc Displacement?.mp. 0 Advanced

6 Sciatica/ 2 Advanced

7 sciatica.mp. 111 Advanced

8 ((hernia* or ruptur* or prolaps* or protru* or bulg* or slip* or displac*) adj3 (disc? or disk? or nuclei or nucleus)).mp. 447 Advanced

9 ((extru* or sequestra* or migrat*) adj2 (disc? or disk? or nuclei or nucleus)).mp. 32 Advanced

10 (discal adj1 hernia*).mp. 0 Advanced

11 (discus adj1 hernia*).mp. 0 Advanced

12 ischialgi*.mp. 2 Advanced

13 lumboischialgi*.mp. 4 Advanced

14 lumbo-ischialgi*.mp. 0 Advanced

15 cauda equina syndrome?.mp. 47 Advanced

16 or/1-15 581 Advanced

17 Lumbar Vertebrae/ 41 Advanced

18 Intervertebral Disk/ 13 Advanced

136

19 Lumbosacral Region/ 6 Advanced

20 ((lumbar or lumbosacral or lumbo-sacral) adj1 (spine or region or area or vertebr* or disk? or disc?)).mp. 1200 Advanced

21 discus intervertebra*.mp. 1 Advanced

22 (spinal adj1 (disk? or disc?)).mp. 13 Advanced

23 (low back adj1 (spine or region or area or vertebr*)).mp. 7 Advanced

24 Intervertebral Disk?.mp. 63 Advanced

25 Intervertebral Disc?.mp. 294 Advanced

26 Inter-vertebral Disk?.mp. 0 Advanced

27 Inter-vertebral Disc?.mp. 1 Advanced

28 or/17-27 1479 Advanced

29 16 and 28 254 Advanced

30 incidence/ 262 Advanced

31 exp risk/ 1339 Advanced

32 exp Cohort Studies/ 1342 Advanced

33 exp Case-Control Studies/ 1110 Advanced

34 epidemiologic studies/ 10 Advanced

35 incidence.mp. 17608 Advanced

36 risk factor?.mp. 13224 Advanced

37 cohort?.mp. 10007 Advanced

38 case control?.mp. 2774 Advanced

39 risk*.mp. 47523 Advanced

40 between group*.tw. 2707 Advanced

41 relative risk*.tw. 1309 Advanced

42 etiolog*.mp. 5891 Advanced

43 exp Epidemiology/ 26 Advanced

44 Epidemiologic Methods/ 20 Advanced

45 epidemiolog*.mp. 8393 Advanced

46 ep.fs. 1744 Advanced

47 or/30-46 79129 Advanced

48 29 and 47 45 Advanced EMBASE Classic+EMBASE 1947 to 2010 December 21

# Searches Results Search Type

1 exp intervertebral disk hernia/ or lumbar disk hernia/ 17656 Advanced

2 Intervertebral Disk Displacement?.mp. 622 Advanced

3 Intervertebral Disc Displacement?.mp. 11 Advanced

4 Inter-vertebral Disk Displacement?.mp. 1 Advanced

5 Inter-vertebral Disc Displacement?.mp. 1 Advanced

6 ((hernia* or ruptur* or prolaps* or protru* or bulg* or slip* or displac*) adj3 22852 Advanced

137

(disc? or disk? or nuclei or nucleus)).mp.

7 ((extru* or sequestra* or migrat*) adj2 (disc? or disk? or nuclei or nucleus)).mp. 996 Advanced

8 (discal adj1 hernia*).mp. 352 Advanced

9 (discus adj1 hernia*).mp. 24 Advanced

10 ischialgia/ 6839 Advanced

11 ischialgi*.mp. 6956 Advanced

12 lumboischialgi*.mp. 126 Advanced

13 lumbo-ischialgi*.mp. 32 Advanced

14 sciatica.mp. 4581 Advanced

15 cauda equina syndrome/ 1267 Advanced

16 cauda equina syndrome?.mp. 1702 Advanced

17 or/1-16 30028 Advanced

18 lumbar vertebra/ 14625 Advanced

19 intervertebral disk/ 9886 Advanced

20 back/ 7733 Advanced

21 ((lumbar or lumbosacral or lumbo-sacral) adj1 (spine or region or area or vertebr* or disk? or disc?)).mp. 58172 Advanced

22 discus intervertebra*.mp. 15 Advanced

23 (spinal adj1 (disk? or disc?)).mp. 192 Advanced

24 (low back adj1 (spine or region or area or vertebr*)).mp. 71 Advanced

25 Intervertebral Disk?.mp. 27270 Advanced

26 Intervertebral Disc?.mp. 8682 Advanced

27 Inter-vertebral Disk?.mp. 10 Advanced

28 Inter-vertebral Disc?.mp. 65 Advanced

29 lumbar spine/ 23300 Advanced

30 or/18-29 83215 Advanced

31 17 and 30 21422 Advanced

32 risk:.mp. or exp methodology/ or exp epidemiology/ 5090165 Advanced

33 (cohort or relative risk:).tw. 221597 Advanced

34 incidence/ or standardized incidence ratio/ 165939 Advanced

35 exp risk/ or risk factor/ 977563 Advanced

36 cohort analysis/ 91803 Advanced

37 exp case control study/ 53633 Advanced

38 epidemiology/ 164156 Advanced

39 (incidence or risk factor? or cohort? or case control).mp. 1316967 Advanced

40 etiology/ 333051 Advanced

41 epidemiolog*.mp. 410418 Advanced

42 ep.fs. 786591 Advanced

43 or/32-42 5960537 Advanced

44 31 and 43 6311 Advanced

138

45 limit 44 to yr="1970 -Current" 5974 Advanced

46 remove duplicates from 45 5879 Advanced

47 limit 46 to medline 1784 Advanced

48 46 not 47 4095 Advanced EBM Reviews - Cochrane Database of Systematic Reviews 2005 to December 2010

# Searches Results Search Type

1 [Intervertebral Disk Displacement/] 0 Advanced

2 Intervertebral Disk Displacement?.mp. 6 Advanced

3 Intervertebral Disc Displacement?.mp. 0 Advanced

4 Inter-vertebral Disk Displacement?.mp. 0 Advanced

5 Inter-vertebral Disc Displacement?.mp. 0 Advanced

6 [Sciatica/] 0 Advanced

7 sciatica.mp. 50 Advanced

8 ((hernia* or ruptur* or prolaps* or protru* or bulg* or slip* or displac*) adj3 (disc? or disk? or nuclei or nucleus)).mp. 33 Advanced

9 ((extru* or sequestra* or migrat*) adj2 (disc? or disk? or nuclei or nucleus)).mp. 2 Advanced

10 (discal adj1 hernia*).mp. 0 Advanced

11 (discus adj1 hernia*).mp. 0 Advanced

12 ischialgi*.mp. 16 Advanced

13 lumboischialgi*.mp. 1 Advanced

14 lumbo-ischialgi*.mp. 0 Advanced

15 cauda equina syndrome?.mp. 8 Advanced

16 or/1-15 71 Advanced

17 [Lumbar Vertebrae/] 0 Advanced

18 [Intervertebral Disk/] 0 Advanced

19 [Lumbosacral Region/] 0 Advanced

20 ((lumbar or lumbosacral or lumbo-sacral) adj1 (spine or region or area or vertebr* or disk? or disc?)).mp. 119 Advanced

21 discus intervertebra*.mp. 0 Advanced

22 (spinal adj1 (disk? or disc?)).mp. 2 Advanced

23 (low back adj1 (spine or region or area or vertebr*)).mp. 3 Advanced

24 Intervertebral Disk?.mp. 12 Advanced

25 Intervertebral Disc?.mp. 10 Advanced

26 Inter-vertebral Disk?.mp. 0 Advanced

27 Inter-vertebral Disc?.mp. 2 Advanced

28 or/17-27 126 Advanced

29 16 and 28 47 Advanced

30 [incidence/] 0 Advanced

31 [exp risk/] 0 Advanced

139

32 [exp Cohort Studies/] 0 Advanced

33 [exp Case-Control Studies/] 0 Advanced

34 [epidemiologic studies/] 0 Advanced

35 incidence.mp. 3422 Advanced

36 risk factor?.mp. 1230 Advanced

37 cohort?.mp. 1242 Advanced

38 case control?.mp. 521 Advanced

39 risk*.mp. 6338 Advanced

40 between group*.tw. 6559 Advanced

41 relative risk*.tw. 3145 Advanced

42 etiolog*.mp. 751 Advanced

43 [exp Epidemiology/] 0 Advanced

44 [Epidemiologic Methods/] 0 Advanced

45 epidemiolog*.mp. 1480 Advanced

46 [ep.fs.] 0 Advanced

47 or/30-46 6595 Advanced

48 29 and 47 47 Advanced EBM Reviews - Cochrane Central Register of Controlled Trials 4th Quarter 2010

# Searches Results Search Type

1 Intervertebral Disk Displacement/ 466 Advanced

2 Intervertebral Disk Displacement?.mp. 481 Advanced

3 Intervertebral Disc Displacement?.mp. 0 Advanced

4 Inter-vertebral Disk Displacement?.mp. 0 Advanced

5 Inter-vertebral Disc Displacement?.mp. 0 Advanced

6 Sciatica/ 157 Advanced

7 sciatica.mp. 303 Advanced

8 ((hernia* or ruptur* or prolaps* or protru* or bulg* or slip* or displac*) adj3 (disc? or disk? or nuclei or nucleus)).mp. 880 Advanced

9 ((extru* or sequestra* or migrat*) adj2 (disc? or disk? or nuclei or nucleus)).mp. 8 Advanced

10 (discal adj1 hernia*).mp. 3 Advanced

11 (discus adj1 hernia*).mp. 3 Advanced

12 ischialgi*.mp. 55 Advanced

13 lumboischialgi*.mp. 7 Advanced

14 lumbo-ischialgi*.mp. 2 Advanced

15 cauda equina syndrome?.mp. 6 Advanced

16 or/1-15 1110 Advanced

17 Lumbar Vertebrae/ 1394 Advanced

18 Intervertebral Disk/ 171 Advanced

19 Lumbosacral Region/ 226 Advanced

140

20 ((lumbar or lumbosacral or lumbo-sacral) adj1 (spine or region or area or vertebr* or disk? or disc?)).mp. 3215 Advanced

21 discus intervertebra*.mp. 0 Advanced

22 (spinal adj1 (disk? or disc?)).mp. 8 Advanced

23 (low back adj1 (spine or region or area or vertebr*)).mp. 5 Advanced

24 Intervertebral Disk?.mp. 651 Advanced

25 Intervertebral Disc?.mp. 209 Advanced

26 Inter-vertebral Disk?.mp. 0 Advanced

27 Inter-vertebral Disc?.mp. 0 Advanced

28 or/17-27 3536 Advanced

29 16 and 28 775 Advanced

30 incidence/ 5250 Advanced

31 exp risk/ 19132 Advanced

32 exp Cohort Studies/ 83768 Advanced

33 exp Case-Control Studies/ 6223 Advanced

34 epidemiologic studies/ 28 Advanced

35 incidence.mp. 36403 Advanced

36 risk factor?.mp. 19045 Advanced

37 cohort?.mp. 9285 Advanced

38 case control?.mp. 3182 Advanced

39 risk*.mp. 54509 Advanced

40 between group*.tw. 241268 Advanced

41 relative risk*.tw. 4158 Advanced

42 etiolog*.mp. 5723 Advanced

43 exp Epidemiology/ 15 Advanced

44 Epidemiologic Methods/ 594 Advanced

45 epidemiolog*.mp. 5160 Advanced

46 ep.fs. 19657 Advanced

47 or/30-46 300521 Advanced

48 29 and 47 597 Advanced

49 limit 48 to medline records 462 Advanced

50 limit 48 to embase records 89 Advanced

51 49 or 50 551 Advanced

52 48 not 51 46 Advanced EBM Reviews - Database of Abstracts of Reviews of Effects 4th Quarter 2010

# Searches Results Search Type

1 [Intervertebral Disk Displacement/] 0 Advanced

2 Intervertebral Disk Displacement?.mp. 21 Advanced

3 Intervertebral Disc Displacement?.mp. 0 Advanced

141

4 Inter-vertebral Disk Displacement?.mp. 0 Advanced

5 Inter-vertebral Disc Displacement?.mp. 0 Advanced

6 [Sciatica/] 0 Advanced

7 sciatica.mp. 29 Advanced

8 ((hernia* or ruptur* or prolaps* or protru* or bulg* or slip* or displac*) adj3 (disc? or disk? or nuclei or nucleus)).mp. 43 Advanced

9 ((extru* or sequestra* or migrat*) adj2 (disc? or disk? or nuclei or nucleus)).mp. 1 Advanced

10 (discal adj1 hernia*).mp. 0 Advanced

11 (discus adj1 hernia*).mp. 0 Advanced

12 ischialgi*.mp. 0 Advanced

13 lumboischialgi*.mp. 0 Advanced

14 lumbo-ischialgi*.mp. 0 Advanced

15 cauda equina syndrome?.mp. 4 Advanced

16 or/1-15 61 Advanced

17 [Lumbar Vertebrae/] 0 Advanced

18 [Intervertebral Disk/] 0 Advanced

19 [Lumbosacral Region/] 0 Advanced

20 ((lumbar or lumbosacral or lumbo-sacral) adj1 (spine or region or area or vertebr* or disk? or disc?)).mp. 118 Advanced

21 discus intervertebra*.mp. 0 Advanced

22 (spinal adj1 (disk? or disc?)).mp. 1 Advanced

23 (low back adj1 (spine or region or area or vertebr*)).mp. 0 Advanced

24 Intervertebral Disk?.mp. 29 Advanced

25 Intervertebral Disc?.mp. 4 Advanced

26 Inter-vertebral Disk?.mp. 0 Advanced

27 Inter-vertebral Disc?.mp. 0 Advanced

28 or/17-27 129 Advanced

29 16 and 28 31 Advanced

30 [incidence/] 0 Advanced

31 [exp risk/] 0 Advanced

32 [exp Cohort Studies/] 0 Advanced

33 [exp Case-Control Studies/] 0 Advanced

34 [epidemiologic studies/] 0 Advanced

35 incidence.mp. 1669 Advanced

36 risk factor?.mp. 1193 Advanced

37 cohort?.mp. 1426 Advanced

38 case control?.mp. 713 Advanced

39 risk*.mp. 5161 Advanced

40 between group*.tw. 6841 Advanced

41 relative risk*.tw. 1445 Advanced

142

42 etiolog*.mp. 1577 Advanced

43 [exp Epidemiology/] 0 Advanced

44 [Epidemiologic Methods/] 0 Advanced

45 epidemiolog*.mp. 1313 Advanced

46 [ep.fs.] 0 Advanced

47 or/30-46 9322 Advanced

48 29 and 47 27 Advanced

143

Appendix D. Detailed evidence tables for systematic review (Chapter 2)

Table D.1. Characteristics of eligible studies examining the incidence of lumbar disc herniation with radiculopathy in adults

First author, Year published Country Study design

Study population and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Incidence estimates (95% CI)

Burske-Hohlfeld, 1990 32 USA Cohort

All Olmsted County residents followed from 1950-1979 for any form of LDH surgery

Follow-up: 30y N=1,028 (NR) 15-78y (42) 88% incident surgery

LDH surgery Surgical Any form of back surgery for suspected LDH,

HNP, or fragments of disc material identified by review of complete (inpatient and outpatient) medical records

“Proven” LDH was a protruded, extruded or sequestered disc seen during surgery

“Suspected” LDH was a disc described as bulging or degenerated

Low Incidence rate of first surgery for suspected LDH: age-adjusted in men, 0.57 (0.52-0.62) per 1,000 person-years; age-adjusted in women, 0.36 (0.32-0.40) per 1,000 person-years; age- and sex-adjusted, 0.46 (0.43-0.49) per 1,000 person-years

Incidence rate of first surgery for proven LDH: age-adjusted in men, 0.51 (0.47-0.56) per 1,000 person-years; age-adjusted in women, 0.31 (0.27-0.34) per 1,000 person-years; age- and sex-adjusted, 0.41 (0.38-0.44) per 1,000 person-years

Raw data: numerators: 538, 371, 909, 485, 316, 801;

denominators: NR, but census data used to derive denominator person-years

Zitting, 1998 33 Finland Cohort

Birth cohort from 2 provinces of Finland followed from 1966-1994 for hospitalized LDH in the National Hospital Discharge Register

Follow-up: 28y N=12,058 (NR) 15-28y among cases (NR) 92%

Hospitalized LDH Hospital ICD-9 diagnosis codes and corresponding ICD-

8 codes used to identify all possible lumbar disc disease cases: 7221, 7227, 7244, 7225, 7242, 7245, 7561, 7384, 7385, 7213, 7214

“Confirmed” LDH was a reliable description of herniation in a surgical report or reliable evidence of a herniation on MRI, CT or myelogram with appropriate symptoms (i.e., on the same side) described in the medical records

Low Cumulative incidence of hospitalized LDH: in men, 9.1 (6.9-12.0) per 1,000 persons; in women, 4.2 (2.8-6.4) per 1,000 persons

Average annual incidence: in men, 0.33 (0.25-0.43) per 1,000 persons in women, 0.15 (0.10-0.23) per 1,000 persons

Raw data: in men, 50/5,474; in women, 22/5,218

144

First author, Year published Country Study design

Study population and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Incidence estimates (95% CI)

Mattila, 2008 37 Finland Cohort

Nationwide cohort of adolescents aged 14-18y, without LDH and prior hospitalized nonspecific back-related diagnosis followed up to December 31, 2001, in the National Hospital Discharge Register

Follow-up: 651,027 person-years (11y average follow-up)

N=57,408 (54%) 15-41y among cases (27 at surgery) 79%

LDH surgery Surgical Lumbar discectomy defined as ICD code 9211

between 1979 and 1996, and then with new codes ABC07, ABC16 and ABC26 from 1997 onwards

Low Incidence rate of LDH surgery: 0.41 (0.36-0.46) per 1,000 person-years; in men, 0.55 (0.46-0.63) per 1,000 persons-years; in women, 0.25 (0.19-0.30) per 1,000 persons-years

Raw data: numerators: 251, 166, 85; denominators: NR

Heliövaara, 1987 31 Finland Cohort

Nationwide cohort of Finnish adults followed from 1970-1980 for hospitalized LDH or sciatica in the National Hospital Discharge Register

Follow-up: 11y N=57,000 (48%) ≥15y (NR) 83%

Hospitalized LDH or sciatica Hospital ICD-8 codes indicating principal diagnosis for

hospitalization: 725.10 or 725.19 for LDH; 353.99 for sciatica

Low 11-year cumulative incidence of hospitalized LDH or sciatica: 14.5 (13.5-15.5) per 1,000 persons; of hospitalized LDH: 8.0 (7.3-8.7) per 1,000 persons; of hospitalized sciatica: 6.5 (5.9-7.2) per 1,000 persons

Average annual incidence of hospitalized LDH or sciatica: 1.32 (1.23-1.41) per 1,000 persons; of hospitalized LDH: 0.72 (0.66-0.79) per 1,000 persons; of hospitalized sciatica: 0.59 (0.53-0.65) per 1,000 persons

Raw data: 825/57,000, 454/57,000, 371/57,000

Mattila, 2009 38 Finland Cohort

All male military conscripts without severe back diseases and performing compulsory service between 1990-2002, were followed during their service period by record linkage to the National Hospital Discharge Register

Follow-up: 267,700 person-years; 6- to 12-month military service period

N=387,070 (0%) 18-29y (20) 100%

Hospitalized LDH Hospital Hospitalized LDH (lumbar and other

intervertebral disc disorders with radiculopathy) was ICD-10 diagnosis code M51.1 (lumbar and other intervertebral disc disorders with radiculopathy), and ICD-9 diagnosis codes 7227C and 3539X

Low Incidence rate of hospitalized LDH: 7.8 (6.7-8.3) per 1,000 person-years

Incidence rates ranged from a high of 12.3 (9.6-13.3) per 1,000 person-years in 1993, to a low of 4.2 (3.0-5.0) per 1,000 person-years 2001

Raw data: NR

145

First author, Year published Country Study design

Study population and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Incidence estimates (95% CI)

Leino-Arjas, 2002 44; Leino-Arjas, 2004 45

Finland Cohort

Nationwide workforce cohort followed in 1996 for lumbar intervertebral disc disorders by record linkage to the National Hospital Discharge Register

Follow-up: 1y N=2,409,319 (NR) 20-64y (NR) 100% a n=2,409,319 (NR); 20-64y b n=1,783,616 (51% women); 25-64y

Hospitalized LDH and LDH surgery Hospital and Surgical Hospitalized LDH was primary diagnosis ICD-

10 codes M51.1-M51.9 (intervertebral disc disorders other than those of the cervical spine)

LDH surgery was those with information on surgical decompression or spondylodesis of the lumbar spine during any of the admissions in 1996

Mod 1-year cumulative incidence of hospitalized LDH: among entire 20-64y workforce, 1.9 (1.9-2.0) per 1,000 persons; among those gainfully employed throughout prior year, 2.2 (2.1-2.2) per 1,000 persons; among men gainfully employed throughout prior year, 2.5 (2.4-2.7) per 1,000 persons; among women gainfully employed throughout prior year, 1.8 (1.7-1.9) per 1,000 persons

1-year cumulative incidence of LDH surgery: among entire 20-64y workforce, 1.0 (0.9-1.0) per 1,000 persons; among those gainfully employed throughout prior year, 1.1 (1.1-1.2) per 1,000 persons

Raw data: 4,643/2,409,319 entire workforce; 3,863/1,783,616

gainfully employed workforce; 2,211/868,876 in men gainfully employed; 1,652/914,740 in women gainfully employed; 2,368/2,409,319 surgical cases in entire workforce; 2,015/1,783,616 surgical cases in gainfully employed workforce

Jørgensen, 1994 42 Denmark Cohort

Occupationally active assistant nurses followed for LDH surgery in 1988 by record linkage to the Danish National Registry of Hospitalized Patients, and compared to all Danish females

Follow-up: 1y N=1,681,152 (100%) 20-69y (NR) 100%

LDH surgery Surgical LDH surgery was ICD-8 surgical codes: 82073,

ablatio prolapsus disci intervertebralis lumbalis; 82173, evacuatio disci intervertebralis lumbalis

Mod 1-year cumulative incidence of LDH surgery among nurses aged 30-69y: 1.3 (1.0-1.8) per 1,000 persons

1-year cumulative incidence of LDH surgery among Danish women aged 30-69y: 0.78 (0.74-0.84) per 1,000 persons

Raw data: 37/28,008; 969/1,235,038

Hurme, 1983 39 Finland Cohort

AII lumbar disc herniation surgeries performed in South-West Finland from 1975-1979, based on surgical department registers of the Turku University Central Hospital area (mean area population during study period, 455,000)

Follow-up: 5y N=1,011 surgeries (44%) 15-80y among cases (42 at surgery) 79% incident surgery

LDH surgery Surgical No explicit case definition provided Included all operations performed for LDH,

classified as first operations and reoperations

Mod 5-year cumulative incidence of first LDH surgery: 1.7 (1.6-1.8) per 1,000 persons

Average annual incidence: 0.34 (0.32-0.37) per 1,000 persons Raw data: 778/455,000

146

First author, Year published Country Study design

Study population and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Incidence estimates (95% CI)

Heikkilä, 1989 41 Finland Cohort

Finnish twin cohort consisting of 9,365 adult pairs of the same sex followed from 1972-1985 for hospitalized sciatica by record linkage to the National Hospital Discharge Register

Follow-up: 14y N=18,730 (55%) 24-60y+ (NR) 100%

Hospitalized sciatica Hospital Sciatica was ICD-8 discharge diagnosis code:

353

Mod 14-year cumulative incidence of hospitalized sciatica among twins: 16.5 (14.8-18.5) per 1,000 persons

Average annual incidence: 1.2 (1.1-1.3) per 1,000 persons Raw data: 304/18,370

Jhawar, 2006 36 USA Cohort

Female nurses from the Nurses’ Health Study without prior LDH in 1976, were followed up in 1992

Follow-up: 16y N=98,407 (100%) 30-55y in 1976 (NR) ≥85%

Clinical LDH Clinical Self-reported physician-diagnosis of LDH that

was confirmed by MRI or CT, in follow-up mailed questionnaire

Low Incidence rate: 6.2 (6.0-6.5) per 1,000 person-years Raw data: 2,727/438,662

Miranda, 2002 34 Finland Cohort

Forest industry workers who had no sciatica during the past 12 months in 1994

Follow-up: 1y N=2,077 (26%) NR (45) 77%

Sciatica Clinical Sciatica was more than 7 days of low back pain

radiating below the knee during the preceding 12 months, with a manikin used to denote the anatomic area in mailed questionnaire

Low 1-year cumulative incidence of sciatica: 93.4 (81.6-106.7) per 1,000 persons

Raw data: 194/2,077

Jarvik, 2005 35 USA Cohort

Outpatients without LBP or sciatica in the past 4 months from four clinics at the Veterans Affairs Puget Sound Health Care System, Seattle Division

Follow-up: 3y N=148 (13%) 35-70y (median, 53) 89%

Clinical LDH Clinical LDH was MRI-confirmed disc protrusion or

extrusion, with pain frequency for low back or buttock pain rated as more than “some of the time” [and] sciatic leg pain; or numbness or tingling in the leg, foot, or groin; or weakness in leg or foot, rated as more than “none”

Low 3-year cumulative incidence among persons without MRI disc-findings at baseline: 89.4 (50.7-153.1) per 1,000 persons

Average annual incidence: 29.8 (16.9-51.0) per 1,000 persons Raw data: 11/123

147

First author, Year published Country Study design

Study population and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Incidence estimates (95% CI)

Riihimäki, 1989 40 Finland Cohort

Male concrete reinforcement workers and house painters without prior history of sciatic pain in 1977 followed up in 1982

Follow-up: 5y N=178 (0%) 25-54y at baseline 77-80%

Sciatica Clinical Sciatica was defined as back pain radiating to a

leg in follow-up mailed questionnaire

Mod 5-year cumulative incidence of sciatica: among concrete reinforcement workers, 343 (244-457) per 1,000 persons; among house painters, 229 (159-318) per 1,000 persons

Average annual incidence of sciatica: among concrete reinforcement workers, 68.5 (48.8-91.3) per 1,000 persons; among house painters, 45.7 (31.7-63.5) per 1,000 persons

Raw data: 25/73, 24/105

Leclerc, 2003 46 France Cohort

Male workers in the French national electricity and gas company without LBP during the past 12 months in 1992 were followed up in 1994

Follow-up: 2y N=841 (0%) 40-50y at baseline (NR) 65%

Sciatica Clinical Sciatica was pain, discomfort or stiffness in the

low back region at least 1 day in the previous 12 months, with radiating symptoms in the leg, in follow-up mailed questionnaire

Mod 2-year cumulative incidence of sciatica: 55.9 (42.3-73.5) per 1,000 persons; among those with a history of LBP before 1992, 101.9 (63.7-159.1) per 1,000 persons; among those with no history of LBP before 1992, 45.3 (32.1-63.6) per 1,000 persons

Average annual incidence of sciatica: 27.9 (21.1-36.8) per 1,000 persons; among those with a history of LBP before 1992, 51.0 (31.9-79.5) per 1,000 persons; among those with no history of LBP before 1992, 22.6 (16.1-31.8) per 1,000 persons

Raw data: 47/841; 16/157; 31/684

Riihimäki, 1994 43 Finland Cohort

Male machine operators, carpenters and office workers without prior history of sciatica in 1984, followed up in 1987

Follow-up: 3y N=1,149 (0%); 25-49y (37) 83%

Sciatica Clinical Sciatica was LBP radiating to a leg in follow-up

mailed questionnaire

Mod 3-year cumulative incidence of sciatica: among office workers, 141 (111-177) per 1,000 persons; among machine operators, 220 (181-264) per 1,000 persons; among carpenters, 241 (198-290) per 1,000 persons

Average annual incidence of sciatic pain: among carpenters, 80.4 (66.1-96.5) per 1,000 persons; among machine operators, 73.2 (60.4-87.8) per 1,000 persons; among office workers, 46.9 (37.0-59.0) per 1,000 persons

Raw data: 81/336; 85/387; 60/426

148

First author, Year published Country Study design

Study population and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Incidence estimates (95% CI)

Netterstrøm, 1989 48 Denmark Cohort

All full-time male bus drivers employed on April 1, 1978 by 3 urban bus companies followed up from 1978-1984 for hospitalized LDH in the Danish National Patient Register

Follow-up: 7y N=2,465 (0%) 20-69y (NR) 100%

Hospitalized LDH Hospital ICD discharge diagnosis codes for LDH

(725.10-725.11)

High Incidence rate of hospitalized LDH: 3.6 (2.8-4.6) per 1,000 person-years

7-year cumulative incidence of hospitalized LDH: 14.6 (10.6-20.2) per 1,000 persons

Average annual incidence: 2.1 (1.5-2.9) per 1,000 persons Raw data: 62/17,122; 36/2,465

Bongers, 1988 47 Netherlands Cohort

All male crane workers and floor workers employed in the same departments at a steel company on January 1, 1975 and those hired up to the end of 1979, were followed up for incident disability pensions due to LDH up until December 31, 1984

Follow-up: 10y N=1,405 (0%) <25-60y+ (NR) 71%

Disability pension due to LDH Clinical Disability pension due to displacement of

intervertebral disc defined as ICD-9 diagnosis code 722.2, assigned by social insurance physician after evaluation of the social insurance medical records

High Incidence rate of LDH disability pension among crane operators, 2.9 (1.6-4.9) per 1,000 person-years; among floor workers, 1.3 (0.5-2.9) per 1,000 person-years

Raw data: 13/4,454.2; 5/3,805.8

Abbreviations: CT, computerized tomography; HNP, herniated nucleus pulposus; ICD, international classification of diseases; LBP, low back pain; LDH, lumbar disc herniation; Mod, moderate; MRI, magnetic resonance imaging; N, study size; NR, not reported; ROB, risk of bias; y, years

149

Appendix Table D.2. Characteristics of eligible studies examining risk factors of lumbar disc herniation with radiculopathy in adults

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

Zitting, 1998 33 Finland Cohort

Birth cohort from 2 provinces of Finland followed from 1966-1994 for hospitalized LDH in the National Hospital Discharge Register

Follow-up: 28 years N=12,058 (NR) 15-28y among cases (NR) 92%

Hospitalized LDH Hospital ICD-9 diagnosis codes and

corresponding ICD-8 codes used to identify all possible lumbar disc disease cases: 7221, 7227, 7244, 7225, 7242, 7245, 7561, 7384, 7385, 7213, 7214

“Confirmed” LDH was a reliable description of herniation in a surgical report or reliable evidence of a herniation on MRI, CT or myelogram with appropriate symptoms (i.e., on the same side) described in the medical records

Low Sex, age Phase II Sex: women RR 1.0; men RR 2.2 (1.3-3.6) Age: NR; in men, hospitalized LDH first occurred around 15y of

age, and incidence rose more sharply from 20y of age; in women, first cases of hospitalized LDH also appeared around 15y of age

Mattila, 2008 37 Finland Cohort

Nationwide cohort of adolescents aged 14-18y, without LDH and prior hospitalized nonspecific back-related diagnosis followed up to December 31, 2001, in the National Hospital Discharge Register

Follow-up: 651,027 person-years (11y average follow-up)

N=57,408 (54%) 15-41y among cases (27 at

surgery) 79%

LDH surgery Surgical Lumbar discectomy defined as ICD

code 9211 between 1979 and 1996, and then with new codes ABC07, ABC16 and ABC26 from 1997 on

Low Sex, socioeconomic background, perceived health status, chronic disease or disability, number of stress symptoms per week, timing of puberty, overweight, smoking, drinking style, frequency of participation in sports clubs, frequency of other leisure-time physical exercise, school success

Phase II Sex: women RR 1.0; men RR 2.2 (1.7-2.9) In men Smoking: not daily HR 1.0; daily HR 1.5 (1.1-2.2) Timing of puberty: early HR 1.0; normal HR 0.7 (0.5-1.1); late HR

0.6 (0.4-1.0) In women Overweight: no HR 1.0; yes HR 2.1 (1.1-4.1) Frequency of participation in sports clubs: never HR 1.0; 2-3 times

per week or less HR 1.5 (0.9-2.5); 4-5 times per week or more HR 2.7 (1.1-6.3)

Heliövaara, 1987 31,49,50

Finland Cohort with case-

control risk analyses

Nationwide cohort of Finnish adults followed from 1970-1980 for hospitalized LDH or sciatica in the National Hospital Discharge Register

Hospitalized LDH or sciatica Hospital ICD-8 codes indicating principal

diagnosis for hospitalization: 725.10 or 725.19 for LDH; 353.99 for sciatica

Low a Sex, age, geographic region, type of population, social class (men only), number of births (women only), marital status, leisure

Phase II Hospitalized LDH Sex: women RR 1.0; men RR 1.6 (p<0.001) Age at baseline (y): 15-19 RR 1.0; 20-29 RR 3.0 (p<0.001); 30-39

RR 5.5 (p<0.001); 40-49 RR 6.5 (p<0.001); 50-59 RR 3.0 (p<0.001); 60+ RR 1.3

150

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

Follow-up: 11y N=57,000 (48%) ≥15y (NR) 83% a n=2,732; 20-59y b n=1,537; 20-59y c n=2,732; 20-59y

time physical activity, smoking, chronic cough, number of psychological distress symptoms, medication use, frequent use of analgesics

b Height, BMI (as an overall measure of obesity), and triceps skinfold thickness

c Occupation, strenuousness of work

Type of population: urban RR 1.0; rural RR 0.8 (p<0.05); industrial RR 1.1

Hospitalized sciatica Sex: women RR 1.0; men RR 1.3 (p<0.05) Age at baseline (y): 15-19 RR 1.0; 20-29 RR 2.7 (p<0.05); 30-39

RR 8.6 (p<0.001); 40-49 RR 12.1 (p<0.001); 50-59 RR 7.3 (p<0.001); 60+ RR 4.3 (p<0.01)

Type of population: urban RR 1.0; rural RR 0.9; industrial RR 1.5 (p<0.001)

Case-control analyses in men Hospitalized LDH Social class: I OR 1.0; II OR 2.4 (p<0.05); III OR 2.6 (p<0.05); IV

OR 2.0; V OR 1.2 Height (cm): ≤169 OR 1.0; 170-174 OR 1.1 (0.7-1.7); 175-179 OR

1.1 (0.7-1.7); ≥180 OR 2.3 (1.4-3.9) BMI (kg/m2): <21.9 OR 1.0; 22.0-23.9 OR 2.4 (1.3-4.5); 24.0-25.9

OR 3.2 (1.7-6.1); 26.0-27.9 OR 3.1 (1.5-6.3); 28.0-29.9 OR 3.7 (1.7-8.0); ≥30.0 OR 2.3 (0.8-6.2)

Occupation: professional/white collar OR 1.0; intermediate non-manual OR 2.3 (p<0.05); forestry OR 2.9; farmer/agricultural OR 1.4; motor vehicle driver OR 2.9 (p<0.05); metal/machine worker OR 3.0 (p<0.01); construction worker OR 2.4 (p<0.05); chemical processor/paper worker OR 2.4; other industrial OR 2.2 (p<0.05); service and other groups OR 2.3

Hospitalized LDH or sciatica Social class: I OR 1.0; II OR 3.2 (p<0.001); III OR 3.0 (p<0.001); IV

OR 2.6 (p<0.01); V OR 1.5 Occupation: professional/white collar OR 1.0; intermediate non-

manual OR 2.8; forestry OR 3.1 (p<0.05); farmer/agricultural OR 2.5 (p<0.01); motor vehicle driver OR 4.6 (p<0.001); metal/machine worker OR 4.2 (p<0.001); construction worker OR 3.1 (p<0.001); chemical processor/paper worker OR 3.2 (p<0.001); other industrial OR 2.6 (p<0.01); service and other groups OR 3.1 (p<0.01)

Case-control analyses in women Hospitalized LDH Height (cm): ≤159 OR 1.0; 160-164 OR 1.0 (0.6-1.6); 165-169 OR

151

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

1.2 (0.6-2.3); ≥170 OR 3.7 (1.6-6.6) Number of psychological distress symptoms: 0 OR 1.0; 1 OR 2.0

(p<0.05); 2 OR 2.9 (p<0.001); 3-5 OR 1.2 Strenuousness of work: light or very light OR 1.0; normal OR 3.8

(p<0.05); heavy or very heavy OR 2.4 Hospitalized LDH or sciatica Number of psychological distress symptoms: 0 OR 1.0; 1 OR 1.7

(p<0.05); 2 OR 2.0 (p<0.01); 3-5 OR 1.8 (p<0.05) Medication use: None OR 1.0; Some OR 0.7 (p<0.05) Frequent use of analgesics: No OR 1.0; Yes OR 1.9 (p<0.01) Strenuousness of work: light or very light OR 1.0; normal OR 2.0

(p<0.05); heavy or very heavy OR 2.5 (p<0.05) Seidler, 2009 67;

Schumann, 2010 68

Germany Case-control

Cases: 564 patients with inpatient or outpatient hospital treatment for pain associated with LDH in 4 regions of Germany (Frankfurt/Main, Freiburg, Halle/Saale, Regensburg)

Controls: 901 persons randomly selected from a 1% random sample of residents from local population registration offices of the same 4 regions

Follow-up: NA 1,816 (50%) 25-70y (48) 53-66%

LDH hospital treatment Hospital Inpatient or outpatient hospital

treatment because of LDH with radiculopathy, sensory and/or motor deficits (neurological findings), and confirmation of LDH by MRI or CT scan

Mod a Occupational: cumulative lumbar load by manual materials handling (lifting, carrying, pushing, pulling, throwing, catching or shoveling of objects weighing ≥5 kg), cumulative lumbar load by intensive-load postures (postures with trunk inclination of ≥20 degrees)

b Lifestyle factors: BMI,

smoking, and sports activities (endurance sports, ball sports, athletic sports, body building sports)

a Phase III In men Cumulative lumbar load by manual materials handling and/or

intensive-load postures (Nh): 0-<5.0*106 OR 1.0; 5.0-<21.51*106 OR 1.7 (1.1-2.7); ≥21.51*106 OR 3.4 (2.2-5.0)

Cumulative lumbar load by manual materials handling (Nh): 0-<2.34*106 OR 1.0; 2.34-<8.98*106 OR 1.2 (0.7-2.0); ≥8.98*106 OR 2.0 (1.2-3.5)

Cumulative lumbar load by intensive-load postures (Nh): 0 OR 1.0; >0-<4.85*106 OR 1.1 (0.6-2.0); 4.85-14.62*106 OR 1.7 (0.9-3.2); ≥14.62*106 OR 1.9 (1.0-3.5)

In women Cumulative lumbar load by manual materials handling and/or

intensive-load postures (Nh): 0 OR 1.0; 0-<4.04*106 OR 1.6 (1.1-2.7); 4.04-<14.47*106 OR 2.4 (1.6-3.8); ≥14.47*106 OR 2.3 (1.5-3.6)

Cumulative lumbar load by intensive-load postures (Nh): 0 OR 1.0; >0-<2.77*106 OR 1.9 (1.0-3.7); 2.77-8.83*106 OR 2.4 (1.2-4.6); ≥8.83*106 OR 3.2 (1.6-6.3)

b Phase II In men BMI (kg/m2): <21.88 OR 1.0; ≥21.88-<24.30 OR 1.4 (0.8-2.4); ≥24.30-<29.21 OR 2.1 (1.3-3.6); ≥29.21 OR 1.6 (0.7-3.8)

Smoking (pack-years): 0 OR 1.0; >0-<8 OR 1.0 (0.6-1.7); ≥8-<20 OR 1.2 (0.8-1.9); ≥20-<40 OR 1.6 (1.0-2.5); ≥40 OR 0.8 (0.4-

152

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

1.5) Cumulative hours of body building sports (h): 0 OR 1.0; >0-<400

OR 1.0 (0.5-2.4); 400-<1,350 OR 1.0 (0.5-2.2); ≥1,350 OR 0.5 (0.2-1.1)

In women BMI (kg/m2): <21.88 OR 1.0; ≥21.88-<24.30 OR 1.3 (0.9-1.8); ≥24.30-<29.21 OR 1.3 (0.8-1.9); ≥29.21 OR 2.0 (1.1-3.7)

Smoking (pack-years): 0 OR 1.0; >0-<8 OR 1.0 (0.7-1.6); ≥8-<20 OR 1.5 (1.0-2.4); ≥20-<40 OR 1.0 (0.6-1.7); ≥40 OR 1.4 (0.4-4.5)

Leino-Arjas, 2002 44; Leino-Arjas, 2004 45

Finland Cohort

a Nationwide workforce cohort followed in 1996 for lumbar intervertebral disc disorders by record linkage to the National Hospital Discharge Register

Follow-up: 1y N=2,409,319 (NR) 20-64y (NR) 100% b Nationwide occupationally

active workforce cohort followed in 1996 for lumbar intervertebral disc disorders by record linkage to the National Hospital Discharge Register

Follow-up: 1y N=1,783,616 (51%) 25-64y (NR) 100%

a,b Hospitalized LDH Hospital Hospitalized LDH was primary

diagnosis ICD-10 codes M51.1-M51.9 (intervertebral disc disorders other than those of the cervical spine)

Mod a Sex, age, employment status, education, occupational class, income

b Sex, age, education,

income, smoking and BMI

Occupational: physical workload, manual materials handling, accident risk, inconvenient work postures, sedentary work, video display terminal work, challenging tasks, social demands, job control, work time schedule

a Phase II Hospitalized LDH (crude analysis) Age (y): 20-34 RR 1.0; 35-44 RR 1.9 (1.7-2.0); 45-54 RR 1.8 (1.6-

1.9); 55-64 RR 0.9 (0.8-1.1) Employment status at end of year 1995: occupationally active RR

1.0; unemployed RR 0.8 (0.7-0.8); on unemployment pension RR 0.8 (0.5-1.1); on other pension RR 1.7 (1.6-1.9); student, at military service or other RR 0.5 (0.4-0.5)

Hospitalized LDH in those occupationally active throughout 1995 (multivariable analysis)

Sex: men RR 1.0; women RR 0.8 (0.8-0.9) Education: higher RR 1.0; secondary RR 1.3 (1.2-1.5); basic RR 1.5

(1.3-1.8) Occupational class: upper white-collar RR 1.0; lower white-collar

(in supervising position or with independent work tasks) RR 1.0 (0.8-1.1); lower white-collar (in dependent position or with routine work tasks) RR 1.2 (1.0-1.4); specialized manual workers RR 1.4 (1.2-1.6); non-specialized manual workers RR 1.5 (1.3-1.8); farmers RR 1.3 (1.1-1.5); other entrepreneurs RR 1.2 (1.0-1.4)

Income: highest quartile RR 1.0; 2nd RR 1.0 (1.0-1.1); 3rd RR 1.0 (0.9-1.1); lowest RR 0.7 (0.6-0.8)

b Phase III Sex: women RR 1.0; men RR 1.4 (1.3-1.5) In men Age (y): 25-34 RR 1.0; 35-44 RR 1.4 (1.2-1.5); 45-54 RR 1.3 (1.2-

1.5); 55-64 RR 1.1 (0.9-1.3) Education: higher RR 1.0; intermediate RR 1.4 (1.2-1.6); basic RR

153

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

1.7 (1.4-1.9) Personal income: highest quintile RR 1.0; 2nd RR 1.2 (1.0-1.5); 3rd

RR 1.5 (1.3-1.8); two lowest quintiles combined RR 1.4 (1.2-1.6) Accident risk: no RR 1.0; low RR 1.2 (1.0-1.5); high RR 1.5 (1.2-1.8) Job control: low RR 1.0; high RR 0.9 (0.8-1.0) In women Age (y): 25-34 RR 1.0; 35-44 RR 1.8 (1.5-2.0); 45-54 RR 1.7 (1.5-

2.0); 55-64 RR 1.2 (1.0-1.5) Education: higher RR 1.0; intermediate RR 1.6 (1.4-1.8); basic RR

1.8 (1.5-2.1) Personal income: highest quintile RR 1.0; 2nd RR 1.5 (1.3-1.8); 3rd

RR 1.5 (1.2-1.8); two lowest quintiles combined RR 1.5 (1.2-1.8) BMI: lowest tertile RR 1.0; middle tertile RR 1.1 (1.0-1.2); highest

tertile RR 1.3 (1.1-1.5) Accident risk: no RR 1.0; low RR 1.3 (1.1-1.5); high RR 1.4 (1.2-1.8) Job control: low RR 1.0; high RR 0.8 (0.8-0.9) Work time schedule: regular daytime work RR 1.0; two-shift work,

regular evening work, weekend work or other irregular work hours not including night work RR 1.1 (0.9-1.3); regular or irregular three-shift work or regular night work RR 1.3 (1.2-1.6)

Jørgensen, 1994 42

Denmark Cohort

Occupationally active assistant nurses followed for LDH surgery in 1988 by record linkage to the Danish National Registry of Hospitalized Patients, and compared to all Danish females

Follow-up: 1y N=1,681,152 (100%) 20-69y (NR) 100%

LDH surgery Surgical LDH surgery was ICD-8 surgical

codes: 82073, ablatio prolapsus disci intervertebralis lumbalis; 82173, evacuatio disci intervertebralis lumbalis

Mod Age, occupation Phase II Age: NR; incidence of LDH surgery increased with age up to the

45-49y age group in nurses, and the 50-54y age group in all Danish females

Occupation: all Danish females aged 30-69y RR 1.0; assistant nurses aged 30-69y RR 1.6 (1.2-2.2)

Hurme, 1983 39 Finland Cohort

AII lumbar disc herniation surgeries performed in South-West Finland from 1975-1979, based on surgical department registers of the Turku

LDH surgery Surgical No explicit case definition provided Included all operations performed for

LDH, classified as first operations and reoperations

Mod Sex, age, workload, home locality, seasonality

Phase I Sex: NR; 56% of surgeries done on men Age: NR; 90% of patients between 25 and 54 years of age Workload: NR; neither heavy nor light work were more common

among the incident surgery group than among the general population of Finland or South-West Finland

154

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

University Central Hospital area (mean area population during study period, 455,000)

Follow-up: 5y N=1,011 surgeries (44%) 15-80y among cases (42 at

surgery) 79% incident surgery

Jhawar, 2006 36 USA Cohort

Female nurses from the Nurses’ Health Study without prior LDH in 1976, were followed up in 1992

Follow-up: 16 years N=98,407 (100%) 30-55y in 1976 (NR) ≥85%

Clinical LDH Clinical Self-reported physician-diagnosis of

LDH that was confirmed by MRI or CT, in follow-up mailed questionnaire

Low Cardiovascular risk factors: BMI, smoking, diabetes, high cholesterol, hypertension, family history of coronary heart disease

Phase III High cholesterol: no RR 1.0; yes RR 1.3 (1.1-1.4) Diabetes: no RR 1.0; yes RR 1.5 (1.2-2.0) Hypertension: no RR 1.0; yes RR 1.3 (1.1-1.4) Family history: no RR 1.0; yes RR 1.1 (1.0-1.3) BMI (kg/m2): <21.9 RR 1.0; 22.0-24.9 RR 1.0 (0.9-1.1); 25.0-26.9

RR 1.1 (1.0-1.3); 27.0-28.9 RR 1.2 (1.0-1.4); 29+ RR 1.1 (1.0-1.3) Smoking: non-smoker RR 1.0; ex-smoker RR 1.1 (1.0-1.2); current

RR 1.4 (1.3-1.5)

Miranda, 2002 34 Finland Cohort

Forest industry workers who had no sciatica during the past 12 months in 1994

Follow-up: 1y N=2,077 (26%) NR (45) 77%

Sciatica Clinical Sciatica was more than 7 days of low

back pain radiating below the knee during the preceding 12 months, with a manikin used to denote the anatomic area

Low Individual: sex, age, height, BMI, mental stress, smoking, car driving, previous low back injuries

Physical exercise: frequency of physical exercise, sports activity, different types of sports

Work-related: amount of twisting movements of trunk, working with trunk forward flexed, working with a hand above shoulder level, working in sitting position, working in kneeling or squatting position, daily lifting of loads, operating a

Phase II Age (y): <35 OR 1.0; 35-44 OR 2.1 (1.1-4.1); 45-54 OR 2.0 (1.0-

4.0); ≥55 OR 3.3 (1.4-7.6) Smoking: non-smoker OR 1.0; ex-smoker OR 1.3 (0.9-2.0); current

smoker (1-15cig/d, 1-15y) OR 1.3 (0.6-3.1); current smoker (>15cig/d, 1-15y) OR 1.2 (0.3-4.6); current smoker (1-15cig/d, >15y) OR 2.2 (1.2-4.0); current smoker (>15cig/d, >15y) OR 2.3 (1.3-3.9)

Mental stress: not at all OR 1.0; only little OR 1.6 (0.9-2.9); to some extent OR 2.1 (1.2-3.7); rather much or much OR 3.0 (1.5-5.9)

Walking: not at all or only little OR 1.0; moderately OR 1.8 (1.2-2.8); actively OR 1.9 (1.2-3.0)

Jogging: not at all or only little OR 1.0; moderately or actively OR 0.5 (0.3-1.0)

Twisting movements of trunk during the work day: not at all or only little OR 1.0; moderately OR 1.6 (1.1-2.5); much OR 1.9 (1.1-3.2)

155

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

motor vehicle, physical strenuousness of work, job satisfaction, overload at work, risk of accident at work

Seidler, 2003 61 Germany Case-control

Cases: 225 male patients with acute LDH (94 LDH only; 131 LDH and osteochondrosis or spondylosis) from 2 orthopaedic practices and 4 neurosurgical or orthopaedic clinics in the Frankfurt/Main area

Controls: 107 population controls without chronic LBP and 90 patients hospitalized for urolithiasis treatment by lithotripsy without chronic LBP and osteochondrosis/ spondylosis (197 controls in total)

Excluded persons with Bechterew’s disease, spine fractures, malignancies of the spine, and poliomyelitis

Follow-up: NA 437 (0%) 25-65y (42) 66-93%

Clinical LDH Clinical LDH only cases defined as currently

symptomatic, radiographically confirmed (MRI and/or CT scan) LDH or protrusion

Mod Occupational: physical workload, lifting/carrying, extreme forward bending postures, whole body vibration, psychosocial work environment (monotonous, boring, opportunities to use knowledge and skills, information about future plans, satisfaction with supervisor, satisfaction with workmates, psychic strain through contact with clients, time pressure, too much responsibility)

Phase III Physical workload: always working in occupations with low physical

workload OR 1.0; ≥10y in occupations with medium physical workload OR 0.8 (0.4-1.7); ≥10y in occupations with high physical workload OR 2.1 (0.9-4.6)

Cumulative extreme (>90° trunk flexion) forward bending (h): 0 OR 1.0; >0-1,500 OR 1.4 (0.7-2.8); >1,500 OR 2.7 (1.2-6.4)

Cumulative whole body vibration (h): 0 OR 1.0; >0-1,800 OR 2.1 (0.9-4.8); >1,800 OR 1.9 (0.7-4.9)

Psychosocial time pressure (number of working years with high degree of time pressure, classified as 5 or 6): 0 OR 1.0; >0-<10 OR 1.2 (0.6-2.6); ≥10 OR 2.9 (1.3-6.3)

Riihimäki, 1989 40 Finland Cohort

Male concrete reinforcement workers and house painters without prior history of sciatic pain in 1977 followed up in 1982

Follow-up: 5y N=178 (0%)

Sciatica Clinical Sciatica was defined as back pain

radiating to a leg in follow-up mailed questionnaire

Mod Occupation, previous back symptoms, earlier back accidents, lumbar spine degeneration, back muscle strength, abdominal muscle strength, height, BMI,

Phase II Occupation: house painters OR 1.0; concrete reinforcement

workers OR 1.8 (1.2-2.9) Earlier back accidents: no OR 1.0; yes OR 1.6 (1.0-2.7) Previous back symptoms: no symptoms OR 1.0; lumbago or

nonspecific low back pain OR 1.8 (0.9-3.4) BMI (kg/m2): ≤23.9 OR 1.0; 24.0-27.9 OR 1.1 (0.6-2.1); ≥28.0 OR

156

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

25-54y at baseline 77-80%

stress, smoking 1.6 (0.8-3.1)

Leclerc, 2003 46 France Cohort

Male workers in the French national electricity and gas company without LBP during the past 12 months in 1992 were followed up in 1994

Follow-up: 2y N=841 (0%) 40-50y at baseline (NR) 65%

Sciatica Clinical Sciatica was pain, discomfort or

stiffness in the low back region at least 1 day in the previous 12 months, with radiating symptoms in the leg, in follow-up mailed questionnaire

Mod Individual: age, living alone, height, BMI, exercise and sports, gardening, non-professional home construction activities, do-it-yourself, smoking

Health: history of low back pain, presence of neck pain, self-rated general health, score of psychological and psychosomatic well-being

Occupational: occupational category, score of job satisfaction and psychosocial aspects, prolonged sitting, prolonged standing, carrying loads, pulling or pushing heavy loads, bending forward and backward, trunk rotations, kneeling and squatting, driving

Phase II Past history of LBP: no OR 1.0; yes OR 3.7 (1.9-7.4) Height (cm): ≤180 OR 1.0; >180 OR 2.7 (1.2-6.4) Driving for >2h: less than once per week OR 1.0; more than once

per week OR 2.7 (1.2-6.4); daily OR 2.0 (0.9-4.4) Self-rated health: good (score 1 or 2) OR 1.0; medium (score 3 or

4) OR 1.0 (0.5-2.0); bad (score 5-8) OR 2.9 (1.2-7.1) Do-it-yourself activities: no OR 1.0; yes OR 2.0 (0.9-4.8)

Riihimäki, 1994 43; Pietri-Taleb, 1995 60

Finland Cohort

Male machine operators, carpenters and office workers without prior history of sciatica in 1984, followed up in 1987

Follow-up: 3y N=1,149 (0%); 25-49y (37) 83%

Sciatica Clinical Sciatica was LBP radiating to a leg in

follow-up mailed questionnaire

Mod a Age, occupation, seniority in occupation, education level, car driving, physical exercise, smoking, twisted or bent occupational postures, high pace of work, monotonous work, problems with workmates or

Phase II a Occupation: office workers OR 1.0; machine operators OR 1.4

(1.0-1.9); carpenters OR 1.5 (1.1-2.1) Physical exercise: maximum once per week OR 1.0; more than

once per week OR 1.3 (1.0-1.6) Smoking: non-smokers OR 1.0; smokers and ex-smokers OR 1.3

(1.0-1.7) History of other low back pain: no OR 1.0; mild OR 2.7 (1.7-4.2);

severe OR 4.5 (2.7-7.6) b In blue-collar workers

157

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

superiors, draft, cold, vibration, history of back accidents, history of lumbago or other LBP

b Psychological distress

and personality (subscales of the Middlesex Hospital Questionnaire and Maudsley Personality Inventory)

Hysteria (quadratic term): OR per one unit increase, 1.3 (1.1-1.7)

Mio, 2007 63 Japan Case-control

Cases: 823 patients of Japanese origin with LDH from hospitals in the Toyama, Tokyo, and Kyoto areas

Controls: 841 hospital patients of Japanese origin who received medical examinations

Excluded patients with spinal canal stenosis, spondylolisthesis, spondylosis, synovial cysts, spinal tumor, and trauma; and those with occupational and/or habitual risk factors, such as heavy manual laborers, occupational drivers, and heavy smokers

Follow-up: NA 1,664 (41% among cases,

63% among controls Cases:11-83y (36); controls:

13-87y (61) NR

Clinical LDH Clinical LDH was unilateral pain radiating

from the back along the femoral or sciatic nerve to the corresponding dermatome of the nerve root of >3 months duration, with positive MRI findings

Mod Cartilage collagen genes SNPs in 3 type XI

collagen genes: COL11A1, COL11A2, COL2A1

Phase I COL11A1 (rs1676486) SNP, c.4603C→T in exon 62, OR 1.4

(1.2-1.7)

158

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

Hirose, 2008 65 Japan Case-control

Cases: 847 patients of Japanese origin with LDH from 19 hospitals between 2001-2007

Controls: 896 Japanese persons from the same catchment area

Excluded patients with synovial cyst, spinal tumor, spondylosis, spondylolisthesis, trauma, and inflammatory disease

Follow-up: NA 1,743 (38%) Cases: NR (39); controls: NR

(62) NR

Clinical LDH Clinical LDH defined by meeting 3 criteria: i)

a history of unilateral pain radiating from the back along the femoral or sciatic nerve to the corresponding dermatome of the nerve root for >3 months; ii) diagnosis of LDH by MRI; and iii) treatment and monitoring for >1 year by an orthopaedic surgeon

Mod Intervertebral disc extracellular matrix protein genes

SNPs in the 2 thrombospondin genes: THBS1, THBS2

SNPs in 2 matrix metalloproteinases: MMP2, MMP9

Phase I THBS2 SNP (rs9406328), IVS10-8C→T, OR 1.4 (1.2-1.6) MMP9 SNP (rs17576), Q279R, OR 1.3 (1.1-1.5) Genotypes for combined effect of THBS2 (TT/TC/CC) and MMP9

(GG/GA/AA): CC and AA, OR 1.0; TC and GA, OR 1.8 (0.9-3.3); TT and GG, OR 3.0 (1.6-5.8)

Karasugi, 2009 66 Japan & Finland Case-control

Japan Cases: 862 patients of

Japanese origin with LDH from 20 hospitals between 2001-2007

Controls: 896 Japanese persons from the same catchment area

Excluded patients with synovial cyst, spinal tumor, spondylosis, spondylolisthesis, trauma, and inflammatory disease

1,758 (NR) Cases: NR (39); controls: NR

(62) 38% Finland Cases: 257 unrelated Finnish

patients with sciatica from the Oulu University Hospital catchment area

Japan Clinical LDH Clinical LDH was defined by meeting 3

criteria: i) a history of unilateral pain radiating from the back along the femoral or sciatic nerve to the corresponding dermatome of the nerve root for >3 months; ii) diagnosis of LDH by MRI; and iii) treatment and monitoring for >1 year by an orthopaedic surgeon

Finland Sciatica Clinical Disabling unilateral shooting band-like

sciatic pain referring from the back to below the knee (dermatomes L4, L5, and S1) from 3 weeks to 6 months, nonresponsive to nonsteroidal anti-inflammatory agents, with clinical presentation

Mod Human sickle tail (SKT) gene

Japan: 68 tag SNPs of the SKT gene

Finland: SKT SNP (rs16924573) only

Phase I LDH in Japan SKT SNP (rs16924573), OR 1.3 (1.1-1.6) SKT SNP (rs2285592), OR 1.3 (1.1-1.5) SKT SNP (rs17469499), OR 1.2 (1.0-1.5) Sciatica in Finland SKT SNP (rs16924573), OR 2.8 (1.1-7.2)

159

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

Controls: 249 unrelated Finnish persons from the same area

Follow-up: NA 506 (NR) NR (NR) NR

concordant with MRI findings

Cong, 2010 69 China Case-control

Cases: 70 male patients of Chinese Han origin with LDH, from the First Affiliated Hospital of China Medical University

Controls: 14 male spinal trauma patients who underwent surgery at the same hospital and 113 male healthy blood donors without LDH symptoms

Excluded persons with occupational and lifestyle risk factors for LDH, such as heavy manual labour, occupational driving and heavy smoking

Follow-up: NA 197 (0%) Cases: 14-41y (33); controls:

20-49y (38) NR

Clinical LDH Clinical Radicular pain with signs of positive

nerve root tension or neurologic deficit, a confirmatory imaging study (MRI and/or CT scan) indicating LDH corresponding to the symptoms and presence of symptoms for ≥6 weeks

Mod Aggrecan gene Aggrecan gene VNTR

polymorphism Expression of aggrecan

Phase I Aggrecan VNTR polymorphism, A25 allele OR 2.1 (1.1-4.0); A21

allele OR 11.8 (2.1-65.5); A29 allele OR 0.2 (0.1-0.8)

Noponen-Hietela, 2005 62

Finland Case-control

Cases: 155 unrelated Finnish patients with sciatica from Oulu University Hospital area, 1997-1998

Controls: 179 unrelated University of Oulu employees and students (all Finnish), no information on possible MSK disorders

Sciatica Clinical Disabling unilateral shooting band-like

sciatic pain referring from the back to below the knee (dermatomes L4, L5, and S1) from 3 weeks to 6 months, nonresponsive to nonsteroidal anti-inflammatory agents, with clinical presentation

Mod Inflammatory mediator genes

Sequence variations (mutations) in 3 interleukin (IL) genes and 1 tumor necrosis factor (TNF) gene: IL1A, IL1B, IL6, TNFA

16 SNPs in 10 candidate

Phase I Genotypes of the IL6 SNP, T15A in exon 5, (AA and AT v TT)

OR 4.4 (1.2-15.7) Genotypes (haplotype pairs) of IL6, (GGGA/GGGA or

GGGA/other v other/other) OR 5.4 (1.5-19.2) Other findings No mutations identified in the IL1A, IL1B, IL6 or TNFA genes

were associated with sciatica

160

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

Excluded patients with chronic pain syndromes and neurological disorders

Follow-up: NA 334 (39% among cases, 69%

among controls) Cases: 19-78y (44); controls:

20-69y (39) NR

concordant with MRI findings cytokine genes: IL1A, IL1B, IL1 receptor antagonist (IL1RN), IL2, IL4, IL4R, IL6, IL10, TNFA, and interferon γ (IFNG)

Virtanen, 2007 64 Finland & China Case-control

Finland Cases: 243 unrelated Finnish

patients with sciatica from Oulu University Hospital area

Controls: 259 unrelated Finnish persons from the same area

502 (45%) NR (NR) NR China Irrelevant outcome

Sciatica Clinical Disabling unilateral shooting band-like

sciatic pain referring from the back to below the knee (dermatomes L4, L5, and S1) from 3 weeks to 6 months, nonresponsive to nonsteroidal anti-inflammatory agents, with clinical presentation concordant with MRI findings

Mod Cartilage intermediate layer protein (CILP) gene

Functional SNP (rs2073711), +1184T→C, in exon 8 of the CILP gene

Other SNPs in the CILP gene

Phase I CILP SNP (rs2073711), +1184T→C in exon 8, OR 1.4 (1.0-1.9)

Mundt, 1993 58,59 USA Case-control

Cases: 297 patients with confirmed and unconfirmed LDH between 1986-1988, from 38 orthopaedic and neurosurgical practices and 5 hospitals in Massachusetts, New Jersey, and New York

Controls: 287 patients admitted to the same hospital service or practice for a condition not related to the back or neck, matched on age, sex, geographic location, and

Combined clinical, hospitalized and surgical LDH

Clinical, hospital and surgical Confirmed case was herniation,

prolapse, rupture, protrusion, extrusion, extradural defect, or free fragment noted on the surgical report, myelogram, CT scan, or MRI as reported in the medical record

Unconfirmed case was probable or possible herniation based on signs and symptoms consistent with herniation

Probable case was pain, numbness or tingling radiating to the hip,

Mod a In 2y before index date Occupational activities,

use of motor vehicles, whole body and arm vibration, riding in planes, trains, buses, subways and motorcycles, non-occupational lifting (inanimate objects and children), carrying, stretching, bending, shovelling, pregnancy history, smoking

Lifetime history Occupational lifting,

Phase II All cases Smoking (average 10cigs/d): OR per one unit increase, 1.4 (1.2-

1.6) Shovelling: no OR 1.0; yes, 5-15 times OR 0.8 (0.6-1.2); yes, >15

times OR 0.7 (0.4-1.2) Repeated bending while doing off the job activities: no OR 1.0; yes, ≥2 days per week OR 1.4 (0.9-2.2)

Non-occupational lifting ≥11.3kg inanimate objects: no OR 1.0; yes, ≥1/wk for 6mo OR 1.1 (0.8-1.6); yes, knees bent, back straight OR 0.6 (0.4-0.9); yes, knees bent, back bent OR 1.2 (0.6-2.3); yes, knees straight, back bent OR 2.3 (1.1-4.7); yes, starting and ending the lift at the waist OR 2.0 (1.0-4.1)

Used free weights ≥10 times: no OR 1.0; yes OR 0.9 (0.6-1.4); yes, warmed up before workout <½ the time OR 0.5 (0.3-1.1); yes, warmed up before workout ≥½ the time OR 1.1 (0.7-1.7)

161

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

source of medical care Excluded those who had

experienced symptoms for >1 year before index date; those who had previous disc surgery >1 year before index date; those who had other conditions of the back or neck; and those who had experienced activity limitation >1y before index date from back, leg, neck or arm pain of ≥4 weeks duration

Follow-up: NA 585 (41%) 20-64y (NR) 76-79%

buttock, thigh or below the knee in a pattern consistent with nerve root impingement by the disc, with worsening of the symptoms with coughing, stretching the leg, or straining while moving the bowels

Possible case differed from probable LDH in that symptoms did not worsen with cough, stretch or strain; or symptoms referred to the thigh, and worsened with cough, stretch or strain, or positive straight leg raising indicated in the medical record; or symptoms were in the lower leg only, and worsened with cough, stretch or strain, or positive straight leg raising

bending, and twisting b In 2y before index date Participation in specific

sports (baseball or softball, golf, bowling, swimming, diving from a board at a height of at least 3 feet, jogging, aerobics, racquet sports), use of free weights, use of weight lifting equipment, warming up before workout

Used weight lifting equipment ≥10 times: no OR 1.0; yes OR 1.0 (0.7-1.5); yes, warmed up before workout <½ the time OR 0.5 (0.2-1.2); yes, warmed up before workout ≥½ the time OR 1.1 (0.7-1.7)

Confirmed cases Non-occupational lifting ≥11.3kg inanimate objects: no OR 1.0; yes, ≥1/wk for 6mo OR 1.6 (1.0-2.5); yes, knees bent, back straight OR 0.7 (0.4-1.3); yes, knees bent, back bent OR 2.0 (0.8-4.9); yes, knees straight, back bent OR 4.0 (1.6-10.0); yes, starting and ending the lift at the waist OR 2.5 (1.1-6.0); yes, arms extended when lift started ≥½ the time OR 1.9 (1.0-3.5); yes, twisted while lifting ≥½ the time OR 1.9 (0.9-3.9)

Kelsey, 1984 56,57 USA Case-control

Cases: 325 patients with surgical, probable or possible LDH after having lumbar spine x-rays or myelograms taken in 3 area hospitals, 1 neurosurgical practice, or 2 orthopaedic practices between 1979-1981, from the New Haven and Hartford, Connecticut areas

Controls: 241 patients admitted to the same hospital service or practice for a condition not related to the spine, matched on age, sex, and medical setting

Excluded those who had

Combined clinical, hospitalized and surgical LDH

Clinical, hospital and surgical Surgical cases were those in which: i)

the hospital chart indicated that the surgeon saw a herniated disc during surgery (descriptions included ruptured, free fragments, herniated, prolapsed, bulging and extruded, but not disc degeneration without evidence of nerve root involvement); and ii) the patient reported pain distributed along the sciatic nerve; and iii) the patient had a positive straight leg raising test, and/or increased pain in the low back or along the sciatic nerve when stretching or extending the leg from a sitting position, and/or

Mod a Smoking, motor vehicle type, size or model, driving pattern characteristics (local roads, highways, bucket seats, regular seats, driver, passenger, automatic transmission, manual transmission), occupational lifting while twisting body, race, education, marital status, residence, height, weight, number of pregnancies, number of children, respiratory symptoms (chronic phlegm, chronic bronchitis), vibration, frequency of wearing

Phase II All cases combined Smoking: never smoked OR 1.0; former smoker (not in the past

year) OR: 1.0 (0.6-1.7); current smoker (in past year) OR 1.7 (1.0-2.5)

Smoking (average 10cigs/d): OR per one unit increase, 1.2 (1.0-1.4)

Motor vehicle type (driven during the past 5y period): only Japanese or Swedish OR 1.0; other OR 3.0 (1.1-8.3)

Driving non-Japanese or non-Swedish car (average 5h/wk): OR per one unit increase, 1.3 (1.1-1.6)

Married v not married (in women) OR 2.2 (1.0-5.1) Rural residence v urban or suburban residence OR 1.7 (0.9-3.3) Lifting objects >11.3kg, >25 times per day while twisting body:

knees bent (yes v no) OR 1.9 (0.8-4.8); knees not bent (yes v no) OR 7.2 (2.0-25.8);

Lifting objects >11.3kg while twisting body (with and without knees bent): no lifting while twisting body OR 1.0; lifting while twisting body, knees bent OR 2.7 (0.9-7.9); lifting while twisting body, knees almost straight OR 6.1 (1.3-27.9)

162

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

previous prolapsed lumbar or cervical disc, or previous neck, back, leg or arm problems that caused activity restriction for more than 4 consecutive weeks, and all persons who had experienced symptoms for ≥1year prior to study entry

Follow-up: NA 566 (45%) 20-64y (NR) 72-79%

increased pain along the sciatic nerve when coughing, sneezing, or straining at stools

Probable cases were similar to surgical cases except that a herniation need not have been observed at surgery; and included cases in which the sciatic pain was felt in both the thigh and lower leg, and cases in which there was sciatic pain in part of the leg and numbness in another part

Possible cases differed from probable cases in that the sciatic pain was only in the thigh or the lower leg but not in both; or if the leg was numb but straight leg raising increased LBP

shoes with high heels, participation in baseball, golf, bowling, swimming, diving from a board, tennis, bicycling, or jogging

b Occupational: frequency

of lifting, frequency of carrying, frequency of twisting at waist, lifting while twisting body, holding while twisting body, bending knees while lifting and twisting

Frequency of lifting objects >11.3kg: not at all OR 1.0; <5 times/d OR 1.2 (0.7-2.0); 5-25 times/d OR 1.3 (0.7-2.5); >25 times/d OR 3.5 (1.5-8.5)

Frequency of carrying objects >11.3kg: not at all OR 1.0; <5 times/d OR 1.0 (0.6-1.9); 5-25 times/d OR 2.1 (1.0-4.3); >25 times/d OR 2.7 (1.2-5.8)

Lifting objects >11.3kg while twisting body: never or small amount OR 1.0; moderate amount OR 2.5 (0.9-6.8); large amount OR 3.1 (1.3-7.5)

Kelsey, 1975 51-55 USA Case-control

Cases: 223 patients with surgical, probable or possible LDH after having lumbar spine x-rays taken in 3 area hospitals or in the office of 2 private radiologists between 1971-1973, from New Haven, Connecticut

Controls (matched): 217 patients admitted to the same hospital service or radiologist office for a condition not related to the spine, matched on age, sex, and medical setting

Controls (unmatched): 494 patients who had lumbar spine x-rays and were not classified as cases

Excluded those who had

Combined clinical, hospitalized and surgical LDH

Clinical, hospital and surgical Surgical cases were those in which: i)

the hospital chart indicated that the surgeon saw a herniated disc during surgery (descriptions included ruptured, free fragments, herniated, prolapsed, bulging and extruded, but not disc degeneration without evidence of nerve root involvement); and ii) the patient reported pain distributed along the sciatic nerve; and iii) the patient had a positive straight leg raising test, and/or increased pain in the low back or along the sciatic nerve when stretching or extending the leg from a sitting position, and/or increased pain along the sciatic

Mod a,b Sex, age, race, social class, residence, marital status, number of children

b Month of onset of

symptoms, height, weight, body bulk (BMI for men, weight/height for women), weight gain/loss in previous year, respiratory symptoms (chronic phlegm, chronic bronchitis), smoking, physical activity and sports, stressful life events in the previous year

b,c Occupational:

Phase II All cases combined Sex: NR; among surgical cases, male to female ratio 2.1 (p<0.001);

among probable and possible cases, male to female ratio 1.1 (p>0.10)

Age: NR; LDH most common in the 30-49y age groups Residence (matched analysis): urban OR 1.0; suburban OR 1.5

(1.0-2.4) Sedentary jobs, age ≥35y (matched analysis) OR 2.4 (1.3-4.7) Jobs requiring driving (matched analysis) OR 2.8 (1.2-7.1) Truck driving job (matched analysis) OR 4.7 (1.3-25.3) Driving other than on job, drivers v nondrivers: (in men; matched

analysis) OR 2.7 (1.0-8.3); (in women; matched analysis) OR 1.9 (0.9-4.1); (in both men and women; matched analysis) OR 2.2 (1.2-3.9)

Chronic phlegm (in women; matched analysis) OR 2.8 (1.1-8.8) Chronic bronchitis (in women; matched analysis) OR 3.7 (1.0-20.5) Physical activity (matched analysis) OR 0.4 (0.2-1.0) Pregnancy: NR; cases had more pregnancies resulting in live births

than other women of their age Other findings

163

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

previous LDH or other serious back problems, and all persons who had experienced symptoms for ≥1year prior to study entry

Follow-up: NA 934 (41%) 20-64y (39 among cases) 78% a, b, c, d n=934; 41% female e n=407; 100% female

nerve when coughing, sneezing, or straining at stools

Probable cases were similar to surgical cases except that a herniation need not have been observed at surgery; and included cases in which the sciatic pain was felt in both the thigh and lower leg, and cases in which there was sciatic pain in part of the leg and numbness in another part

Possible cases differed from probable cases in that the sciatic pain was only in the thigh or the lower leg but not in both; or if the leg was numb but straight leg raising increased LBP

sedentary jobs (sitting ≥half the time), jobs requiring driving (sitting ≥half the time in a motor vehicle; men only), truck driving (men only), jobs involving any lifting, jobs involving any pushing, jobs involving any pulling, jobs involving any carrying, heavy lifting

b,d Driving: jobs requiring

driving (sitting ≥half the time in a motor vehicle; men only), truck driving (men only), driving other than on job

b,e Pregnancy: average

number of pregnancies, pregnancies resulting in live births, pregnancies not resulting in live births

Social class: NR; some indication of an association between higher social class and LDH in women, but no association in men

Sports: NR; some tendency for cases to play sports rarely and controls more frequently

Lifting on the job: no association found with heavy lifting or with frequent lifting

Hrubec, 1975 70 Risk

Cases: 1,132 first admission records to Army hospitals for LDH in 1944-1945

Controls: 1,095 records of Army National Service Life Insurance policyholders individually matched on age and period of military service

Follow-up: NA 1,095 case-control pairs (0%) 18-56y (NR)

Hospitalized LDH Hospital No explicit case definition provided Records of first admission for LDH

identified using punch cards from the Office of the Surgeon General

High Height, weight, posture, body frame, place of birth, place of residence, education, occupation, marital status, physical defects, military occupation specialty assignments, overseas service, blood type, religion, rank, combat credit (medals or decorations)

Phase II All statistically significant associations: Height (cm): 152-164 OR 0.4; 165-181 OR 0.9; 182-203 OR 2.6 Weight (kg): 45-58 OR 0.5; 59-81 OR 1.0; 82-136 OR 1.6 Craftsman, foreman or kindred occupation: OR 1.5 Clerical or kindred occupation: OR 0.6 Married or re-married at induction: OR 1.4 Place of birth <2500 population or rural: OR 1.6 Rural free delivery address: OR 1.6 Good posture at induction: OR 1.3 Defects at induction relevant to HNP affecting back or legs: OR

3.5

164

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

97% Heavy frame at induction: OR 1.5 Military occupation specialty, ground combat: OR 1.6 Combat credit, 2+ battle stars: OR 0.7 Rank, staff sergeant or sergeant: OR 1.3

Netterstrøm, 1989 48

Denmark Cohort

All full-time male bus drivers employed on April 1, 1978 by 3 urban bus companies followed up from 1978-1984 for hospitalized LDH in the Danish National Patient Register

Follow-up: 7y N=2,465 (0%) 20-69y (NR) 100%

Hospitalized LDH Hospital ICD discharge diagnosis codes for

LDH (725.10-725.11)

High Occupation Phase I Occupation: hospitalized LDH in male urban bus drivers compared

to all Danish men: SMR 137 (105-176)

An, 1994 71 Risk

Cases: 163 consecutive surgical patients with LDH at Pennsylvania Hospital between 1987-1988

Controls: 205 inpatients without lumbar disc disease from medical and surgical services at the same hospital, matched on sex and age

Follow-up: NA 368 (39%) 16-78y (45) NR

LDH surgery Surgical Confirmed LDH surgery case defined

by meeting 3 criteria: i) prolapsed lumbar intervertebral disc was the primary diagnosis; ii) all were symptomatic and admitted for surgical management (100% surgically confirmed); iii) all had radiographic evidence (myelogram, CT scan or MRI) of intervertebral herniation

High Smoking Phase I Smoking (current and ex-smokers v non-smokers) OR 2.2 Smoking (current smokers v non-smokers) OR 3.0 Other findings Incorrect analysis for a matched CC study; compares proportion of

current and ex-smokers among cases v controls instead of discordant pairs, results not sex- and age-adjusted

Saftic, 2006 73 Risk

Cases: 67 adults from 9 villages on the Croatian islands of Rab, Vis, Lastovo, and Mljet, who had a positive history of LDH surgery

Controls: 268 adults matched on age, gender, and village of residence/immigrant

LDH surgery Surgical LDH surgery was a positive history of

surgery due to lumbar intervertebral disc herniation at the level of L4/L5 or L5/S1 based on examination of medical histories and medical records

High BMI, occupation type, intensity of physical labor at work, intensity of physical labour at home, smoking index, claudication index, self-assessed limitation in physical activity, level of education,

Phase I BMI (kg/m2): ≥25.7 v <25.7 OR 2.8 (1.1-4.5) Occupation: involving hard physical labour (agriculture workers,

soldiers, construction workers, mechanics or fishermen) v involving sitting or standing (clerks, lawyers, economists, tailors, waiters, cooks, salespersons, teachers, policemen, electricians, and housewives) OR 1.9 (1.1-3.8)

Intensity of physical labour at work: hard v sitting, easy or moderate) OR 2.9 (1.1-4.8)

165

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

status on a 4:1 basis Excluded persons who

underwent surgery of the lower spine due to other causes, such as degenerative changes or stenosis

Follow-up: NA 365 (NR) ≥18y (NR) NR

socioeconomic status, and family history of lower spine surgery

Positive family history of spine surgery (having a parent who underwent spine surgery): yes v no OR 4.0 (1.9-6.1)

Lee, 2006 74 Risk

Cases: 119 herniated lumbar disc levels in 111 adult patients who underwent LDH surgery between 2000-2002

Controls: 82 normal discs levels adjacent to the herniated levels in the same adult patients

Excluded patients who had not had pre-operative CT scans

Follow-up: NA 201 adult disc levels (37%) 40-49y (NR) NR

LDH surgery Surgical Herniated disc level noted during

open discectomy or percutaneous endoscopic discectomy for radiculopathy

High Facet joint tropism Phase I NR; degree of facet tropism at the L4-L5 level was significantly

greater in herniated discs than in normal discs (6.9±5.5° v 3.6±3.0°, p<0.001)

No significant difference found between herniated and normal discs at the L3-L4 and L5-S1 levels

Kunakornsawat, 2007 75

Risk

Cases: 34 herniated lumbar disc levels in 34 adult patients who underwent discectomy at Lerdsin Hospital between 2001-2003

Controls: 34 normal discs levels adjacent to the herniated levels in the same patients

Excluded MRIs of persons

LDH surgery Surgical Patient aged <45y who underwent

conventional discectomy for LDH (single protrusion or extrusion disc at L3-L4, L4-L5 or L5-S1), without previous surgery, and with symmetrical MRI scan of each disc level

High Facet joint tropism Phase I Facet tropism at L4-L5 OR 1.8 (0.3-10) Facet tropism at L5-S1 OR 1.7 (0.3-10) Other findings No association found between facet tropism and LDH in all levels

166

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

who had associated spinal abnormality such as spina bifida, unilateral sacralization, enlarged transverse processes, or spondylolisthesis

Follow-up: NA 68 disc levels (35%) 23-45y (34) NR

Bongers, 1988 47 Netherlands Cohort

All male crane workers and floor workers employed in the same departments at a steel company on January 1, 1975 and those hired up to 1979, followed up for incident disability pensions due to LDH until December 1984

Follow-up: 10y N=1,405 (0%) <25-60y+ (NR) 71%

Disability pension due to LDH Clinical Disability pension due to

displacement of intervertebral disc defined as ICD-9 diagnosis code 722.2, assigned by social insurance physician after evaluation of the social insurance medical records

High Whole-body vibration Phase II Disability pension due to LDH in crane operators exposed to

whole-body vibration v floor workers, IDR 2.5 (90% CI, 1.2-12.5)

Zhang, 2009 76 Risk

Cases: 2,010 patients with LDH in orthopaedic departments at 3 hospitals between 2005-2007

Controls: 2,170 randomly selected patients (in-patients or medical exam) without current back/sciatic pain, or back/sciatic pain for >1 month ever, matched on race, gender, age, and living area

Excluded patients with lumbar spinal stenosis,

Clinical LDH Clinical No explicit case definition provided LDH diagnosis evaluated by ≥2

orthopaedic experts in terms of patient’s symptoms, signs and imaging examination (MRI and/or CT scan) among patients presenting with back-leg pain and typical sciatica

High Sex, age, height, weight, smoking, drinking, family history of lumbar disc herniation, bed characteristics (hard, soft), educational background (primary school, secondary school, high school, university), physical exercise, occupational lumbar load (quite light, light, medium, heavy), occupational character (non-manual, half non-

Phase II Among all persons Family history OR 3.6 (1.9-6.5) Lumbar load OR 2.1 (1.7-2.6) Hard-working OR 1.8 (1.1-2.5) Educational background OR 0.8 (0.6-1.0) Physical exercise OR 0.4 (0.2-0.8) Bed characteristics (hard v soft) OR 0.4 (0.2-0.6) Other findings Model omitted sex, age, BMI, occupational character, drinking,

smoking, and vocational activities

167

First author, Year of publication

Country Study design

Participants and setting Follow-up N (% female) Age range (mean) Participation %

Study outcome Case definition type Case definition ROB Risk factors considered

Phase of evidence Risk estimates (95% CI)

spinal congenital dysplasia, intraspinal tumor, spinal instability from trauma, scoliosis, and spondylolisthesis

Follow-up: NA 4,180 (40%) <30-50y+ (46) NR

manual/half manual, manual), psychosocial factors at work (monotonous, boring, time urgency, too much responsibility, life pressure, hard-working)

Chibnall, 2006 72 Risk

African American and non-Hispanic white workers’ compensation claimants who filed low back injury claims in St. Louis City or Kansas City, Missouri, and whose claims were settled between January 1, 2001, and June 1, 2002

Follow-up: NA 2,934 (38%) 18-55y (NR) 50%

Clinical and surgical LDH Clinical and surgical No explicit case definition provided From a series of medical diagnosis

questions, participants were categorized into 2 diagnosis groups: LDH v low back sprain/strain/pain (regional backache)

For claimants with a LDH diagnosis, surgery was classified as “no” v “yes”

High Sex, age, race, socioeconomic status, presence of lower extremity pain, lumbar degeneration, legal representation, legal representation due to dissatisfaction with medical treatment, current work status

Phase II LDH diagnosis (v regional backache diagnosis) Sex: women OR 1.0; men OR 1.7 (1.4-2.2) Age (y): OR per one unit increase, 1.2 (1.1-1.3) Race: white OR 1.0; African American OR 0.6 (0.4-0.7) Lower extremity pain: no OR 1.0; yes OR 3.3 (2.3-4.9) Lumbar degeneration: no OR 1.0; yes OR 2.1 (1.6-2.6) Legal representation: no OR 1.0; yes OR 1.0 (0.7-1.4) Legal representation due to dissatisfaction with medical treatment:

no OR 1.0; yes OR 0.6 (0.5-0.8) Surgery among those with LDH diagnosis Race: white OR 1.0; African American OR 0.3 (0.2-0.5) Socioeconomic status: OR per one unit increase, 1.1 (1.0-1.4) Lower extremity pain: no OR 1.0; yes OR 2.9 (1.2-6.4) Legal representation: no OR 1.0; yes OR 1.4 (0.9-2.3) Legal representation due to dissatisfaction with medical treatment:

no OR 1.0; yes OR 0.6 (0.4-0.8) Abbreviations: BMI, body mass index; cig, cigarettes; cm, centimeter; CT, computerized tomography; d, day; h, hour; ICD, international classification of diseases; kg, kilogram; LBP, low back pain;

LDH, lumbar disc herniation; m, meter; mo, month; MRI, magnetic resonance imaging; MSK, musculoskeletal; N, study size; NA, not applicable; Nh, Newton-hours; NR, not reported; OR, odds ratio; ROB, risk of bias; RR, relative risk; SMR, standardized morbidity ratio; SNP, single nucleotide polymorphism; v, versus; VNTR, variable number of tandem repeats; wk, week; y, year

168

Appendix E. Belief elicitation questionnaire (Chapter 3)

!"#$%&'(&& &

Version date: 2011-03-25 Belief Elicitation Questionnaire – Final 1&

Belief Elicitation – Example Scenario

Treatment: No health care Outcome: Tension-type headache E.1. Among an average group of 1,000 newly diagnosed acute low back pain patients

receiving no health care treatment, how many do you believe will go on to

develop tension-type headache within 2 months of diagnosis?

Estimate: 20 patients

E.2. There may be some uncertainty around your estimate. What do you believe is a

plausible range for the number of patients that will develop tension-type headache

within 2 months of diagnosis?

Lower bound (L): 10 patients Upper bound (U): 25 patients

E.3. You have been given 20 virtual chips. Each chip represents 5% probability, for a

total of 100% probability. Placing the chips in the bins, indicate the weight of belief

for your estimates of the number of patients developing tension-type headache

within 2 months of diagnosis.

Do you have any questions? Please review the shape and distribution of your answer. Does this reflect

what you truly believe? If not, please feel free to revise your placement of chips.

Do you have any questions?

169

!"#$%&'(&& &

Version date: 2011-03-25 Belief Elicitation Questionnaire – Final 2&

Belief Elicitation Questionnaire

Scenario 1 Treatment: Chiropractic care Outcome: Acute lumbar disc herniation 1. Among an average group of 1,000 newly diagnosed acute low back pain patients

not treated with chiropractic care, how many do you believe will go on to develop acute lumbar disc herniation within 2 months of diagnosis? Estimate:

2. Among an average group of 1,000 newly diagnosed acute low back pain patients

treated with chiropractic care, how many do you believe will go on to develop acute lumbar disc herniation within 2 months of diagnosis? Estimate:

3. There may be some uncertainty around your estimate. What do you believe is a

plausible range for the number of patients treated with chiropractic care that will develop acute lumbar spine disc herniation within 2 months of diagnosis? Lower bound (L): Upper bound (U):

4. You have been given 20 virtual chips. Each chip represents 5% probability, for a

total of 100% probability. Placing the chips in the bins, indicate the weight of belief for your estimates of the number of patients treated with chiropractic care that will develop acute lumbar disc herniation within 2 months of diagnosis.

Please review the shape and distribution of your answer. Does this reflect what you truly believe? If not, please feel free to revise your placement of chips. 5. Compared to primary medical care, what overall effect do you believe

chiropractic care has on the risk of developing acute lumbar disc herniation?

! Decreases the risk ! No effect on the risk ! Increases the risk

170

!"#$%&'(&& &

Version date: 2011-03-25 Belief Elicitation Questionnaire – Final 3&

Scenario 2 Treatment: Chiropractic care Outcome: Acute severe lumbar disc herniation that is surgically managed 6. Among an average group of 1,000 newly diagnosed acute low back pain patients

not treated with chiropractic care, how many do you believe will go on, within 2 months, to develop acute severe lumbar disc herniation that is surgically managed? Estimate:

7. Among an average group of 1,000 newly diagnosed acute low back pain patients

treated with chiropractic care, how many do you believe will go on, within 2 months, to develop acute severe lumbar disc herniation that is surgically managed? Estimate:

8. There may be some uncertainty around your estimate. What do you believe is a

plausible range for the number of patients treated with chiropractic care that will go on, within 2 months, to develop acute severe lumbar disc herniation that is surgically managed? Lower bound (L): Upper bound (U):

9. You have been given 20 virtual chips. Each chip represents 5% probability, for a

total of 100% probability. Placing the chips in the bins, indicate the weight of belief for your estimates of the number of patients treated with chiropractic care that will go on, within 2 months, to develop acute severe lumbar disc herniation that is surgically managed.

Please review the shape and distribution of your answer. Does this reflect what you truly believe? If not, please feel free to revise your placement of chips. 10. Compared to primary medical care, what overall effect do you believe

chiropractic care has on the risk of developing acute severe lumbar disc herniation that is surgically managed?

! Decreases the risk ! No effect on the risk ! Increases the risk

171

!"#$%&'(&& &

Version date: 2011-03-25 Belief Elicitation Questionnaire – Final 4&

Demographic information 11. Sex: ! Male ! Female 12. Age group (years): ! 21-30 ! 31-40 ! 41-50 ! 51-60 ! !61 13. Clinical background: ! Primary care medicine

! Chiropractic ! Spine surgery or Neurosurgery

14. How many years have you been treating low back pain patients? 15. How many new low back pain patients do you see per year? 16. Have you ever had formal statistical training? ! Yes ! No

Highest level of statistical training: ! None ! Graduate ! Undergraduate ! Postgraduate 17. Do you refer for chiropractic care in your practice to treat low back pain patients? ! Yes ! No Rationale: 18. Which source(s) of information do you use in formulating your beliefs about the

effect of chiropractic care on the risk of developing acute lumbar spine disc herniation? Check all that apply.

! Research evidence ! Clinical experience ! News media ! Other:

Comment:

19. In your clinical practice, have you ever seen a patient whose disc herniation was

due to chiropractic spinal manipulation? ! Yes ! No 20. Please feel free to provide any additional comments:

172

Appendix F. Belief elicitation script (Chapter 3)

Version date: 2011-03-25 Elicitation Script – Final 1!

Standardized Belief Elicitation Interview Script

Thank you for agreeing to meet with me. The purpose of this study is to document clinicians’ beliefs about the effect of chiropractic care on the risk of developing two outcomes: 1) acute lumbar disc herniation (aLDH); and 2) acute severe lumbar disc herniation that is surgically managed (aLDH-surgery). I am a PhD candidate in epidemiology at the University of Toronto. This study will form part of my PhD thesis. This interactive questionnaire will take approximately 15 minutes to complete. If you have any questions, I am here to assist you. Participation in this study is voluntary. Your responses will be kept anonymous, and will only be used for research purposes. You have been assigned a unique identifier code that will be written at the top of your questionnaire. The code will be kept confidential. You will not be named or identified in any reports, publications, or presentations that may come from this study. If you do not treat a low back pain patient population, or do not wish to participate in this study, please feel free to decline. Are you agreeable to participation in this study? If no—thank the participant for their time. If yes—proceed to the example. Before we begin the questionnaire, I have a worked example of what will be asked of you. For each scenario, you will be asked to consider an average group of 1,000 newly diagnosed acute low back pain patients without current lumbar disc herniation. Acute lumbar disc herniation is defined as acute onset of lumbar radiculopathy due to disc herniation. The health care treatment options under consideration include:

No health care No additional health care after LBP diagnosis Primary medical care Course of NSAIDs/painkillers prescribed by family physician Chiropractic care Course of spinal manipulation prescribed by chiropractor This sheet lists these patient and treatment characteristics so that they will be easy to recall. Provide sheet. In the case of the example, a hypothetical participant is asked about his beliefs about the development of tension-type headache (TTH) in acute low back pain patients that receive no health care treatment. The questions and process in this worked example are similar to the questionnaire you will be asked to complete. Do you have any questions? If yes—answer any questions. Once all questions have been addressed, give the participant the example questionnaire. If no—give the participant the example questionnaire and read each question aloud.

173

Version date: 2011-03-25 Elicitation Script – Final 2!

Let’s start with the example: E.1. Among an average group of 1,000 newly diagnosed acute low back pain patients receiving no

health care treatment, how many do you believe will go on to develop TTH within 2 months of diagnosis? Estimate: 20 patients In this example, the participant thought that the number of patients that would go on to develop TTH within 2 months was 20; so he gave an estimate of 20 patients.

E.2. There may be some uncertainty around your estimate. What do you believe is a plausible range for the number of patients that will develop TTH within 2 months of diagnosis? Lower bound (L): 10 patients Upper bound (U): 25 patients Here, the participant thought that a plausible range for the number of patients that would go on to develop TTH within 2 months could be as low as 10 and as high as 25; so he gave a lower bound of 10 and an upper bound of 25.

E.3. In Question 2, the participant indicated his range of plausible values for the number of patients developing TTH within 2 months. However, he may believe that some estimates along that range would be more likely than others. Using virtual chips, each representing 5% probability, he placed the chips in the bins to indicate his weight of belief for the estimates along his range. In this example, he had more belief in an estimate of 20 and less belief in estimates of 10 and 25. Do you have any questions? If yes—answer any questions. Once all questions have been addressed, continue example. If no—continue example. Once all the chips have been placed, you can see that it creates a shape and distribution; we’ll call this a “belief curve”. At this point, the participant is asked to take a moment and check if the shape and distribution of the curve is a fair representation of what he truly believes. If he feels that the curve does not reflect what he believes, he is able to rearrange his chips until he is satisfied that his true belief is represented.

Do you have any questions? If yes—answer any questions. Once all questions have been addressed, proceed to the elicitation questionnaire. If no—proceed to the elicitation questionnaire.

174

Version date: 2011-03-25 Elicitation Script – Final 3!

Standardized Interview Script – Belief Elicitation Scenarios You have been shown an example of this belief elicitation exercise using a worked example of no health care treatment and TTH. The structure of the questionnaire I am giving you now is very similar. You will be asked to consider 2 scenarios: 1) chiropractic care and the risk of developing aLDH, and 2) chiropractic care and the risk of developing aLDH-surgery. When answering these questions, remember to keep in mind the patient and treatment profiles listed on the sheet that you have in front of you: Patient Profile: An average group of 1,000 newly diagnosed acute low back pain patients without current lumbar disc herniation. Outcome Profile: Acute onset of lumbar radiculopathy due to lumbar disc herniation Treatment Profiles:

No health care No additional health care after LBP diagnosis Primary medical care Course of NSAIDs/painkillers prescribed by family physician Chiropractic care Course of spinal manipulation prescribed by chiropractor I am here to assist you throughout the elicitation. Do you have any questions? If yes—answer any questions. Once all questions have been addressed, give the participant the elicitation questionnaire and read each question aloud. If no— give the participant the elicitation questionnaire and read each question aloud. Treatment: Chiropractic care Outcome: Acute lumbar disc herniation 1. Among an average group of 1,000 newly diagnosed acute low back pain patients not treated

with chiropractic care, how many do you believe will go on to develop aLDH within 2 months of diagnosis? Estimate: Please provide your estimate of the number of patients developing aLDH within 2 months of diagnosis.

2. Among an average group of 1,000 newly diagnosed acute low back pain patients treated with

chiropractic care, how many do you believe will go on to develop aLDH within 2 months of diagnosis? Estimate: Please provide your estimate of the number of patients developing aLDH within 2 months of diagnosis.

3. There may be some uncertainty around your estimate. What do you believe is a plausible

range for the number of patients treated with chiropractic care that will develop aLDH within 2 months of diagnosis? Lower bound (L): Upper bound (U): Elicit the range of plausible values for the number of patients developing aLDH within 2 months. Identify a

175

Version date: 2011-03-25 Elicitation Script – Final 4!

range such that it is extremely unlikely (but not necessarily impossible) that the point estimate lies outside.

4. You have indicated that the lower bound of the range is L and the upper bound is U. I will divide the range from L to U into equal-width bins. I have placed 1 virtual chip, representing 5% probability, in each of the L and U bins. I will now give you 18 more chips, each representing 5% probability, for a total of 100%. Using these chips, please indicate the weight of belief for your estimates of the number of patients treated with chiropractic care that will develop aLDH within 2 months of diagnosis. Do you have any questions? If yes—answer any questions. Once all questions have been addressed, continue scenario. If no—continue scenario. Once all the chips have been placed and distribution fitted: Please review the shape and distribution of your answer. Does this reflect what you truly believe? If not, please feel free to revise your placement of chips until you are satisfied that the resulting distribution reflects your true belief. Do you have any questions? If yes—answer any questions. Once all questions have been addressed, proceed to question 5. If no— proceed to question 4.

5. Compared to primary medical care, what overall effect do you believe chiropractic care has on the risk of developing aLDH? Decreases the risk No effect on the risk Increases the risk

Do you have any questions? If yes—answer any questions. Once all questions have been addressed, proceed to next scenario. If no— proceed to next scenario.

Let’s move on. The outcome of interest now is aLDH-surgery: Treatment: Chiropractic care Outcome: Acute severe lumbar disc herniation that is surgically managed 6. Among an average group of 1,000 newly diagnosed acute low back pain patients not treated

with chiropractic care, how many do you believe will go on, within 2 months, to develop aLDH-surgery? Estimate: Please provide your estimate of the number of patients developing aLDH-surgery within 2 months.

176

Version date: 2011-03-25 Elicitation Script – Final 5!

7. Among an average group of 1,000 newly diagnosed acute low back pain patients treated with chiropractic care, how many do you believe will go on, within 2 months, to develop aLDH-surgery? Estimate: Please provide your estimate of the number of patients developing aLDH-surgery within 2 months.

8. There may be some uncertainty around your estimate. What do you believe is a plausible

range for the number of patients treated with chiropractic care that will go on, within 2 months, to develop aLDH-surgery? Lower bound (L): Upper bound (U): Elicit range of plausible values for the number of patients developing aLDH-surgery within 2 months. Identify a range such that it is extremely unlikely (but not necessarily impossible) that the point estimate lies outside.

9. You have indicated that the lower bound of the range is L and the upper bound is U. I will divide the range from L to U into equal-width bins. I have placed 1 virtual chip, representing 5% probability, in each of the L and U bins. I will now give you 18 more chips, each representing 5% probability, for a total of 100%. Using these chips, please indicate the weight of belief for your estimates of the number of patients treated with chiropractic care that will go on, within 2 months, to develop aLDH-surgery. Do you have any questions? If yes—answer any questions. Once all questions have been addressed, continue scenario. If no—continue scenario. Once all the chips have been placed and distribution fitted: Please review the shape and distribution of your answer. Does this reflect what you truly believe? If not, please feel free to revise your placement of chips until you are satisfied that the resulting distribution reflects your true belief. Do you have any questions? If yes—answer any questions. Once all questions have been addressed, proceed to question 10. If no— proceed to question 10.

10. Compared to primary medical care, what overall effect do you believe chiropractic care has on the risk of developing aLDH-surgery? Decreases the risk No effect on the risk Increases the risk

Do you have any questions? If yes—answer any questions. Once all questions have been addressed, proceed to next section. If no— proceed to next section.

The demographic section is the final part of this questionnaire. Please complete it now and let me know when you are finished.

177

Version date: 2011-03-25 Elicitation Script – Final 6!

After completion of the questionnaire: Thank you very much for sharing your beliefs with us and participating in this study. Your beliefs will be synthesized with that of your colleagues. If you would like, when the study is completed, we can email you the results of the study for your interest. Would this be of interest to you? If yes—record email address. If no—continue. Thank you for your time.

178

Appendix G. Belief elicitation computer protocol (Chapter 3)

Figure. Eliciting a probability distribution using the modified SHELF computer protocol. In the top screen, a respondent first expresses a point estimate, lower and upper bounds of their uncertainty, and indicates the distribution of the weight of their belief by interactively allocating “chips in bins”. In the bottom screen, a best fitting curve, from among 6 choices, is fitted to the chips in bins and presented back to the respondent, with the median and quartiles to ensure that the “belief curve” reflects their true belief. The software was modified so that: (1) the bins for the choices adapted to the initial range of possible values the respondent chose; (2) there was feedback on the number of chips placed and the number remaining; (3) the smoothed plot at the bottom gave a little more feedback on estimated mean, median IQR for the fitted function; (4) the entire set of data (number of chips in each bin, bin width, fitted functions, etc.) and plots of results could be saved with the respondent ID in the filenames.

179

Appendix H. Case and exposure definition tables for self-controlled case series study (Chapter 4)

Table H.1. Defining incident cases of acute LDH with early surgical intervention – Inclusion and exclusion

criteria and corresponding diagnostic/fee service codes for administrative health databases

Code Description Type Source

Step 1 – Initial cases of LDH surgery Initial cases of LDH surgery identified using at least one disc surgery intervention code and one LDH diagnosis

code from DAD and NACRS data

Disc surgery intervention codes Discectomy 92.31 Excision or destruction of intervertebral disc ICD-9 DAD Chemonucleolysis 1.SE.59 Destruction of intervertebral disc ICD-10 DAD, NACRS Discotomy/Excision 1.SE.87 Intervertebral disc discotomy/partial excision ICD-10 DAD, NACRS Discectomy/Excision/Laminectomy/La

minotomy 1.SE.89 Intervertebral disc discectomy/total excision/

laminectomy/laminotomy ICD-10 DAD, NACRS

LDH diagnosis codes Disc herniation 722.1 Displacement of thoracic or lumbar

intervertebral disc without myelopathy ICD-9 DAD

722.2 Displacement of intervertebral disc site unspecified without myelopathy

ICD-9 DAD

722.7 Intervertebral disc disorder with myelopathy ICD-9 DAD M51.0 Lumbar and other intervertebral disc

disorders with myelopathy ICD-10 DAD, NACRS

M51.1 Lumbar and other intervertebral disc disorders with radiculopathy

ICD-10 DAD, NACRS

M51.2 Other specified intervertebral disc displacement

ICD-10 DAD, NACRS

Sciatica 724.3 Syndrome characterized by pain radiating from the back into the buttock and into the lower extremity along its posterior or lateral aspect, and most commonly caused by protrusion of a lower lumbar intervertebral disc

ICD-9 DAD

Lumbosacral neuritis or radiculitis unspecified

724.4 Radicular syndrome of lower limbs ICD-9 DAD

Other relevant syndromes M54.1 Radiculopathy ICD-10 DAD, NACRS M54.3 Sciatica ICD-10 DAD, NACRS M54.4 Lumbago with sciatica ICD-10 DAD, NACRS

Step 2 – Preliminary cases of surgically managed acute LDH Preliminary cases of acute LDH with early surgery identified by including persons (from Step 1 above) who

presented to a hospital ED for LDH within 8 weeks prior to their LDH surgery using OHIP and NACRS data; this ED LDH visit date is defined as the event index date for the study

Emergency department LDH diagnosis codes (codes used for sensitivity analysis 2 as detailed in the Statistical Analysis section of the Methods; primary analysis does not include codes that may be less disc-specific as indicated below by *)

Diseases of musculoskeletal system 722 Intervertebral disc disorder MOH OHIP 724* Lumbar strain, lumbago, coccydynia, sciatica MOH OHIP Sprains, strains and other trauma 847* Low back, coccyx, neck MOH OHIP Disc herniation M51.0 Lumbar and other intervertebral disc

disorders with myelopathy ICD-10 NACRS

180

Code Description Type Source

M51.1 Lumbar and other intervertebral disc disorders with radiculopathy

ICD-10 NACRS

M51.2 Other specified intervertebral disc displacement

ICD-10 NACRS

Other relevant syndromes M54.1 Radiculopathy ICD-10 NACRS M54.3 Sciatica ICD-10 NACRS M54.4 Lumbago with sciatica ICD-10 NACRS M54.5* Low back pain ICD-10 NACRS

Emergency department fee codes H codes H101 Minor Assess.-G.P.-Emerg.-Dr.On Duty MOH OHIP H102 Comprehensive Assess. & Care MOH OHIP H103 Multiple Systems Assessment-G.P.-

Emergency-Dr. On Duty MOH OHIP

H104 GP-Reassess-Emerg Dept-Physician On Duty M-F Days

MOH OHIP

H121 Emerg.Dept.Phys.On Duty 12mn-8am Minor Assess.

MOH OHIP

H122 12 Midnight To 8:00 A.M. Comprehensive Assess. & Care

MOH OHIP

H123 Emerg.Dept.Phys.On Duty 12mn-8am Mult.Syst.Assess.

MOH OHIP

H124 Emerg.Dept.Phys.On Duty 12mn-8am Re-Assess.

MOH OHIP

H151 Emerg.Dept.Phys.On Duty Sat./Sun./Holiday Minor Assess.

MOH OHIP

H152 Sat/Sun & Holidays Comprehensive Assess. & Care

MOH OHIP

H153 Emerg.Dept.Phys.On Duty Sat./Sun/Holiday Mult.Syst.Assess.

MOH OHIP

H154 Emerg.Dept.Phys.On Duty Sat./Sun./Holiday Reassess.

MOH OHIP

H055 Emergency Department - Physician On Duty MOH OHIP H065 Emerg.Phys.Consult.(Non Spec. .In

Emerg.Med.) MOH OHIP

H100 Description Unknown MOH OHIP H105 Interim Inpatient Admission Orders MOH OHIP H106 Description Unknown MOH OHIP H107 Description Unknown MOH OHIP H112 Emerg.Dept.-Dr.On Duty-Extra To Proc.- 12

Mn To 8 Am. MOH OHIP

H113 Emerg.Dept.-Dr.On Duty-Extra To Proc.- Sat./Sun./Holidays.

MOH OHIP

H131 Physician On Duty ED - Evenings (18:00-24:00)

MOH OHIP

H132 Physician On Duty ED - Evenings (18:00-24:00)

MOH OHIP

H133 Physician On Duty ED - Evenings (18:00-24:00)

MOH OHIP

H134 GP-Reassess-Emerg Dept-Physician On Duty M-F Evenings

MOH OHIP

H980 Spec Vis Emerg Dept Phys., Wk/Daytime MOH OHIP H981 Spec Vis Emerg Dept Phys., Addit'l Pt., MOH OHIP

181

Code Description Type Source Wk/Daytime

H984 Spec Vis Emerg Dept Phys., Mon-Fri., Eve MOH OHIP H985 Spec Vis Emerg Dept Phys., - Addit'l Pt., Eve MOH OHIP H986 Spec Vis Emerg Dept Phys., Nights MOH OHIP H987 Spec Vis Emerg Dept Phys., Addit'l Pt.,

Nights MOH OHIP

H988 Spec Vis Emerg Dept Phys., Sat-Sun-Hols MOH OHIP H989 Spec Vis Emerg Dept Phys., Sat-Sun-Hols,

Addit'l Pt. MOH OHIP

K codes K990 Spec Vis Emerg Dept., Wk/Daytime MOH OHIP K991 Spec Vis Emerg Dept., Wk/Daytime, Addit'l

Pt. MOH OHIP

K992 Spec Vis Emerg Dept.-Sac.Off.Hrs. Wk/Daytime

MOH OHIP

K993 Spec Vis Emerg Dept.-Sac.Off.Hrs. Wk/Daytime Addit'l Pt.

MOH OHIP

K994 Spec Vis - Emerg Dept., Mon-Fri., Eve. MOH OHIP K995 Spec Vis - Emerg Dept., Mon-Fri., Eve.

Addit'l Pt. MOH OHIP

K996 Spec Vis - Emerg Dept., Nights MOH OHIP K997 Spec Vis - Emerg Dept., Nights Addit'l Pt. MOH OHIP K998 Spec Vis - Hosp Emerg Dept., Sat-Sun-Hols MOH OHIP K999 Spec Vis - Hosp Emerg Dept., Additional

Patients MOH OHIP

Step 3 – DAD and NACRS Exclusions Excluded persons who had other spine surgery interventions on or prior to their date of incident LDH surgery;

or, diagnosis of LDH or other conditions potentially associated with LDH within 21 months prior to the event index date using DAD and NACRS data

Other spine surgery codes Exploration and decompression of

spinal canal 16.01 Removal of foreign body from spinal canal ICD-9 DAD

16.02 Reopening of laminectomy site ICD-9 DAD 16.09 Other exploration and decompression ICD-9 DAD Division of intraspinal nerve root 16.1 ICD-9 DAD Chordotomy 16.2 ICD-9 DAD Excision or destruction of lesion 16.3 Excision or destruction of lesion of spinal

cord and spinal meninges ICD-9 DAD

Plastic operations 16.4x Plastic operations on spinal cord and spinal meninges

ICD-9 DAD

Freeing of adhesions 16.5 Freeing of adhesions of spinal cord and nerve roots

ICD-9 DAD

Shunt and drainage 16.6x Shunt and drainage by shunt of spinal theca ICD-9 DAD Injection of destructive agent 16.7 Injection of destructive agent into spinal

canal ICD-9 DAD

Invasive diagnostic procedures 16.8x Invasive diagnostic procedures on spinal cord and spinal canal structures

ICD-9 DAD

Other operations 16.9x Other operations on spinal cord and canal structures

ICD-9 DAD

Decompression/ Release

1.AW.72 Spinal cord decompression/release ICD-10 DAD, NACRS

Destruction/ 1.AW.59 Spinal cord destruction/rhizotomy ICD-10 DAD, NACRS

182

Code Description Type Source Rhizotomy

Discectomy 1.SC.75 Multilevel spine discectomy with interbody fusion

ICD-10 DAD, NACRS

1.SE.53 Discectomy with replacement of intervertebral spacer device

ICD-10 DAD, NACRS

Implantation 1.SE.53 Intervertebral disc device implantation ICD-10 DAD, NACRS Inspection 2.AX.70 Spinal canal inspection ICD-10 DAD, NACRS Laminectomy/

Laminotomy 1.SF.80 Sacral (for decompression of sacral spinal

canal) ICD-10 DAD, NACRS

1.SC.80 Spinal vertebrae (for decompression of spinal cord)

ICD-10 DAD, NACRS

1.SC.74 Spinal vertebrae with instrumentation or with discectomy and instrumentation

ICD-10 DAD, NACRS

1.SC.75 Spinal vertebrae with fusion or with discectomy and multilevel fusion

ICD-10 DAD, NACRS

Neuroectomy 1.AW.87 Spinal nerves [T2-S5] neuroectomy ICD-10 DAD, NACRS

Prior LDH diagnosis codes (exclude prior LDH as defined by Step 1 LDH diagnosis codes)

Disc herniation 722.1 Displacement of thoracic or lumbar intervertebral disc without myelopathy

ICD-9 DAD

722.2 Displacement of intervertebral disc site unspecified without myelopathy

ICD-9 DAD

722.7 Intervertebral disc disorder with myelopathy ICD-9 DAD M51.0 Lumbar and other intervertebral disc

disorders with myelopathy ICD-10 DAD, NACRS

M51.1 Lumbar and other intervertebral disc disorders with radiculopathy

ICD-10 DAD, NACRS

M51.2 Other specified intervertebral disc displacement

ICD-10 DAD, NACRS

Sciatica 724.3 Syndrome characterized by pain radiating from the back into the buttock and into the lower extremity along its posterior or lateral aspect, and most commonly caused by protrusion of a lower lumbar intervertebral disc

ICD-9 DAD

Lumbosacral neuritis or radiculitis unspecified

724.4 Radicular syndrome of lower limbs ICD-9 DAD

Other relevant syndromes M54.1 Radiculopathy ICD-10 DAD, NACRS M54.3 Sciatica ICD-10 DAD, NACRS M54.4 Lumbago with sciatica ICD-10 DAD, NACRS

Prior other diagnosis codes (spine-related disorders of the nervous and MSK systems, congenital anomalies, fractures)

Disorders of the nervous system 324.1 Intraspinal abscess ICD-9 DAD 334.8 Other spinocerebellar diseases ICD-9 DAD 334.9 Spinocerebellar disease unspecified ICD-9 DAD 335.x Anterior horn cell diseases ICD-9 DAD 336.x Other diseases of spinal cord ICD-9 DAD 340 Multiple sclerosis ICD-9 DAD 342.x Hemiplegia and hemiparesis ICD-9 DAD 344.x Other paralytic syndromes ICD-9 DAD 349.0 Reaction to spinal or lumbar puncture ICD-9 DAD

183

Code Description Type Source

349.81 Cerebrospinal fluid rhinorrhea ICD-9 DAD 353.x Nerve root and plexus disorders ICD-9 DAD 355.x Mononeuritis of lower limb and unspecified

site ICD-9 DAD

356.x Hereditary and idiopathic peripheral neuropathy

ICD-9 DAD

357.x Inflammatory and toxic neuropathy ICD-9 DAD 358.x Myoneural disorders ICD-9 DAD 359.x Muscular dystrophies and other myopathies ICD-9 DAD G06.1 Intraspinal abscess and granuloma ICD-10 DAD, NACRS G11.3 Cerebellar ataxia with defective DNA repair ICD-10 DAD, NACRS G11.8 Other hereditary ataxias ICD-10 DAD, NACRS G11.9 Hereditary ataxia, unspecified ICD-10 DAD, NACRS G12.x Spinal muscular atrophy and related

syndromes ICD-10 DAD, NACRS

G95.x Other and unspecified diseases of spinal cord

ICD-10 DAD, NACRS

G32.0 Subacute combined degeneration of spinal cord in diseases classified elsewhere

ICD-10 DAD, NACRS

G99.2 Myelopathy in diseases classified elsewhere ICD-10 DAD, NACRS G35 Multiple sclerosis ICD-10 DAD, NACRS G81.x Hemiplegia and hemiparesis ICD-10 DAD, NACRS G82.x Paraplegia (paraparesis) and quadriplegia

(quadriparesis) ICD-10 DAD, NACRS

G83.x Other paralytic syndromes ICD-10 DAD, NACRS G97.1 Other reaction to spinal and lumbar

puncture ICD-10 DAD, NACRS

G96.0 Cerebrospinal fluid leak ICD-10 DAD, NACRS G54.x Nerve root and plexus disorders ICD-10 DAD, NACRS G57.x Mononeuropathies of lower limb ICD-10 DAD, NACRS G60.x Hereditary and idiopathic neuropathy ICD-10 DAD, NACRS G61.x Inflammatory polyneuropathy ICD-10 DAD, NACRS G62.x Other and unspecified polyneuropathies ICD-10 DAD, NACRS G63.x Polyneuropathy in diseases classified

elsewhere ICD-10 DAD, NACRS

E08.42 Diabetes mellitus due to underlying condition with diabetic polyneuropathy

ICD-10 DAD, NACRS

E09.42 Drug or chemical induced diabetes mellitus with neurological complications with diabetic polyneuropathy

ICD-10 DAD, NACRS

E10.42 Type 1 diabetes mellitus with diabetic polyneuropathy

ICD-10 DAD, NACRS

E11.42 Type 2 diabetes mellitus with diabetic polyneuropathy

ICD-10 DAD, NACRS

E13.42 Other specified diabetes mellitus with diabetic polyneuropathy

ICD-10 DAD, NACRS

G70.x Myasthenia gravis and other myoneural disorders

ICD-10 DAD, NACRS

G71.x Primary disorders of muscles ICD-10 DAD, NACRS G72.x Other and unspecified myopathies ICD-10 DAD, NACRS G73.x Disorders of myoneural junction and muscle ICD-10 DAD, NACRS

184

Code Description Type Source in diseases classified elsewhere

Disorders of the MSK system

722.x Intervertebral disc disorders ICD-9 DAD

738.5 Other acquired deformity of back or spine ICD-9 DAD M51.x Thoracic, thoracolumbar, and lumbosacral

intervertebral disc disorders ICD-10 DAD, NACRS

M96.1 Postlaminectomy syndrome, not elsewhere classified

ICD-10 DAD, NACRS

M99.83 Other biomechanical lesions of lumbar region

ICD-10 DAD, NACRS

M99.84 Other biomechanical lesions of sacral region ICD-10 DAD, NACRS Congenital anomalies 740.x Anencephalus and similar anomalies ICD-9 DAD 741.x Spina bifida ICD-9 DAD 742.x Other congenital anomalies of nervous

system ICD-9 DAD

754.2 Congenital musculoskeletal deformities of spine

ICD-9 DAD

Q00.x Anencephaly and similar malformations ICD-10 DAD, NACRS Q05.x Spina bifida ICD-10 DAD, NACRS Q07.x Other congenital malformations of nervous

system ICD-10 DAD, NACRS

Q01.x Encephalocele ICD-10 DAD, NACRS Q02.x Microcephaly ICD-10 DAD, NACRS Q03.x Congenital hydrocephalus ICD-10 DAD, NACRS Q04.x Other congenital malformations of brain ICD-10 DAD, NACRS Q06.x Other congenital malformations of spinal

cord ICD-10 DAD, NACRS

Q67.5 Congenital deformity of spine ICD-10 DAD, NACRS Q76.x Congenital malformations of spine and bony

thorax ICD-10 DAD, NACRS

Fractures, dislocations, sprains and strains of the spine

805.x Fracture of vertebral column without mention of spinal cord injury

ICD-9 DAD

806.x Fracture of vertebral column with spinal cord injury

ICD-9 DAD

S32.x Fracture of lumbar spine and pelvis ICD-10 DAD, NACRS Injury To Nerves And Spinal Cord

952.x Spinal cord injury without evidence of spinal bone injury

ICD-9 DAD

953.x Injury to nerve roots and spinal plexus ICD-9 DAD S34.x Injury of lumbar and sacral spinal cord and

nerves at abdomen, lower back and pelvis level

ICD-10 DAD, NACRS

Step 4 – OHIP Exclusions Excluded persons who had, within 21 months prior to their event index date, a diagnosis of disc herniation or

other conditions potentially associated with LDH, specialist visits to neurosurgeons, orthopedic surgeons, neurologists, physiatrists or rheumatologists, or advanced spine imaging or diagnostic testing, using OHIP data

Prior disc disorder diagnosis codes Diseases of musculoskeletal system 722 Intervertebral disc disorder MOH OHIP Lumbar spine chiropractic diagnosis C42 Neuritis and neuralgia – Lumbar MOH OHIP C43 Neuritis and neuralgia – Pelvic, ileoiguinal MOH OHIP C45 Sciatic – acute MOH OHIP

185

Code Description Type Source

C46 Sciatic – chronic MOH OHIP C47 Sciatic – leg, other than sciatic MOH OHIP C48 Sciatic – discopathic or discogenic MOH OHIP C52 Radiculitis – lumbar, lumbosacral MOH OHIP C54 Radiculitis – vertebrogenic or pressure MOH OHIP

Prior other diagnosis codes Neoplasms 140 – 239 Malignant neoplasms, benign neoplasms,

carcinoma in situ, neoplasms of uncertain behaviour

MOH OHIP

Nervous system disorders 320 – 330 333 – 344 348 – 349

Central nervous system diseases MOH OHIP

353 – 359 Peripheral nervous system diseases MOH OHIP Congenital anomalies 741 – 759 Congenital anomalies MOH OHIP Fractures 805 Vertebral column - without spinal cord

damage MOH OHIP

806 Vertebral column - with spinal cord damage MOH OHIP 829 Other fractures MOH OHIP

Prior specialty visit codes Specialty of billing physician 04 Neurosurgery MOH OHIP 06 Orthopaedic surgery MOH OHIP 18 Neurology MOH OHIP 31 Physical medicine MOH OHIP 48 Rheumatology MOH OHIP

Prior spine imaging and diagnostic testing fee codes MRI X490 Limited spine (one segment) – multislice MOH OHIP X492 Limited spine (one segment) – repeat MOH OHIP X493 Intermediate spine (2 segments) – multislice MOH OHIP X495 Intermediate spine (2 segments) – repeat MOH OHIP X496 Complex spine – multislice sequence MOH OHIP X498 Complex spine – repeat MOH OHIP CT X415 Spine CT without IV contrast MOH OHIP X416 Spine CT with IV contrast MOH OHIP X417 3-dimensional spine CT acquisition

sequencing MOH OHIP

X128 Spine CT with and without IV contrast MOH OHIP EMG G455 Complete-Nerve Conduct.-Tech.Comp. MOH OHIP G456 EMG And Nerve Conduction Studies-P1 MOH OHIP G457 EMG And/Or Perf/Super/Interp. MOH OHIP G458 Single Fibre Electromyography MOH OHIP

G459 EMG And Nerve Conduction Studies-

Interpret-P2 MOH OHIP

G465 Thera. Proc.-Manipulation Joint-Major MOH OHIP

G466 Nerve Conduct Studies Tech.Comp.To

G469 Sched.B MOH OHIP

G467 Phys Med - Misc Therap Proc (Phys's Own

Patients) MOH OHIP

186

Code Description Type Source

G469 Phys.Med Ltd Nerve Conduct Studies/EMG

Interpr.Only Sched.B MOH OHIP

Other diagnostic testing G368 Chemonucl.If Lumbosacral Disc Incl. MOH OHIP G386 Inj./Infusion Lat. Discography Subs. MOH OHIP

J006 Diag. Radiol. Clinic. Proc.-Discogram-One

Disc MOH OHIP

J011 Diag. Radiol. Clinic. Proc.-Myelogram. MOH OHIP

J020 Diag. Radiol. Clinic. Proc.-Mylogram-

Post.Fossa Views MOH OHIP

J030 Diag. Radiol. Clinic. Proc.-Discogram-Ea.

Add. Disc. MOH OHIP

J038 Diag. Radiol. Clinic. Proc.-Myelogram/Supine

Vw. MOH OHIP

X164 Diag. Radiology-Discogram(S)-One/More

Levels MOH OHIP

X173 Diag. Radiology-Myelogram-Spine +/Or

Posterior Fossa MOH OHIP

Z454 Chemonucleolysis-Lateral Discography-1st

Disk MOH OHIP

187

Table H.2. Chiropractor and primary care physician exposure codes.

Specialist Specialty code Fee code Fee code description

Chiropractor* 59 - Chiropractor V101 Chiropractic subsequent visit office/institution V102 Chiropractic home service V103 Chiropractic initial service office/institution Primary Care Physician† 00 - Family practice and general practice 05 - Community medicine A001 Minor assessment A005 Consultation A003 General assessment A004 General re-assessment A006 Repeat consultation A007 Intermediate assessment A008 Mini assessment A050 Special community medicine consultation A051 Complex medical specific re-assessment A053 Medical specific assessment A054 Medical specific re-assessment A055 Community medicine consultation – office A056 Repeat consultation A058 Partial assessment A400 Comprehensive community medicine consultation A405 Community medicine limited consultation – office A901 General/family practice – house call assessment A905 General/family practice – limited consultation A911 General/family practice – special consultation A912 General/family practice – comprehensive consultation K005 Individual care K006 Hypnotherapy K007 Psychotherapy K013 Individual Care – first three units K017 Annual health examination K022 HIV primary care K028 STD management K030 Diabetic management assessment K033 Individual care – additional units K037 Fibromyalgia/chronic fatigue syndrome care

* Chiropractor visit was defined as a billing record with a combination of relevant specialty code and fee code as listed above

† Primary care physician visit was defined as a billing record with a combination of relevant specialty code and fee code as listed above