Allocative Efficiency Modelling:

Session on TB

Nicole Fraser

Improving Efficiency in Health Conference

4th February 2016, World Bank Washington

DC

Outline of the session

• Optima TB: Context & progress (Nicole Fraser)

• Technical presentation Optima TB (Tan Doan)

• Overview on other TB models (Nick Menzies)

• General discussion

The greatest public health contribution of the early

models in tuberculosis was that they generated a

scientific attitude towards the whole decision-

making process in control programmes by

employing the techniques of systems analysis and

cost/benefit analysis, thereby making the planner

realize that medicosocial, economic, and operational

parameters were just as important to determine as

epidemiological parameters in the formulation of an

optimum tuberculosis programme.

Halfdan Mahler, 1973

value for money

End TB Strategy

Global Plan to Stop TB

results

resource constraints

cost-effectiveness

a world free of TB equity

patient-centered

integration

accountability

political commitment

universal health coverage

innovation

quality

reach

catastrophic costs

investment case

elimination targets

scope

sustainability

value for money

End TB Strategy

Global Plan to End TB

results

resource constraints

cost-effectiveness

a world free of TB equity

patient-centered

integration

accountability

political commitment

universal health coverage

innovation

quality

reach

catastrophic costs

investment case

elimination targets

scope

sustainability

value for money

End TB Strategy

Global Plan to Stop TB

results

resource constraints

cost-effectiveness

a world free of TB equity

patient-centered

integration

accountability

political commitment

universal health coverage

innovation

quality

reach

catastrophic costs

investment case

elimination targets

scope

sustainability

help understand

heterogeneous

epi settings

use TB data

better

know the costs of

implementation

modes forecast

impact of

NTPs

provide

policy

analysis

identify

opportunities

for better

allocations

Drake et al.

Systematic review,

Health economics 2016

TB incidence has not responded to

TB funding increase

Source: Based on Global TB report 2015, WHO

MDGs SDGs

Stop TB Strategy End TB Strategy 2016-35

2016

High Burden

Country concept

Financial burden of TB on patients…

• Patients risk incurring catastrophic costs esp. if NTP poorly

funded and implemented

• Average total costs to TB patients US$ 847, ~20% are medical

costs, ~20% non-medical costs, and 60% to income loss

• Half of costs incurred before initiating TB treatment

• Equals on average 39% of household income

• higher among the poor and those with MDR-TB

• Financial burden of TB illness greater for patients in poorer

countries without UHC

• OOP expenses typically an inefficient way of buying services

• Need efficient NTPs which minimizes direct and indirect

costs to TB patients

Costs per TB patient highly variable - Within regions (typically higher in former Soviet Union)

- Within GDP bands (China, India, SA, Indonesia, Bangladesh, Pakistan

relatively low)

Launch of the Optima TB partnership: 1 Dec 2015

Partnership for Improving Allocative Efficiency in TB

Title of Presentation 17

Steering

group

Strategic need and

direction, review,

dissemination, scale-up,

advisory

Global Fund, USAID,

Stop TB, Burnet

Institute, World Bank

Technical

group

Sharing of expertise,

identification of priorities

on data and model

development, review,

learning

TB modellers (various

levels of engagement),

economists, public

health experts

National TB

programs,

ministries of

health &

finance

Optima-TB Pilot countries

Subsequent phases of

Optima-TB applications

with NTPs

Mix of TB settings,

includes ex-USSR

country, HIV/TB co-

epidemic, SE Asia, LAC

A software tool that works with selected existing TB epidemic

models to provide evidence for better allocative efficiency in NTPs.

1. TB epidemic model(s) to calibrate to data and project TB trends

with different program/funding combinations or policy scenarios

2. relationships costs of delivering TB services - attained coverage

3. strategic epidemiological objectives and constraints

4. a formal mathematical optimization algorithm to assess optimal

allocation of resources to best achieve the objectives.

AuTuMN is the initial epidemiological model within Optima TB

TB modelling groups have been invited to consider inclusion of their

models in the Optima TB infrastructure and thus be an

epidemiological model selection within Optima TB.

Tan Doan, PhD

Improving Efficiency in Health Washington DC, 3-4 February 2016

with the AuTuMN transmission dynamic model

More to be achieved with each dollar

Governments should provide services that are safe, timely, effective, efficient, equitable, and patient-focused

Allocative efficiency

Maximization of health outcomes using the most cost-effective mix

of health interventions

Efficient allocation of health resources for TB

The following types of questions need to be answered: Knowing your TB epidemics

What are my country’s TB trends and how will these change under different funding scenarios?

How many TB cases and TB-related deaths have been averted through past TB investments?

Planning and prioritizing TB programs

For a certain level of funding, how should resources be allocated across different programs to maximize health outcomes?

What is the expected epidemic trajectory of TB incidence, prevalence and deaths?

Targeting areas of need

How can targeting national resources to sub-national regions and special population groups improve outcomes?

Predicting outcomes

To achieve predefined TB targets, how much funding will be required, and how should it be allocated?

The approach

Phase 4. Apply optimization algorithm and interpret results

Phase 1. Assess burden of disease

Define diseases, health states and model structure

Compile estimates of prevalence and incidence of disease

Determine baseline projections for each disease state (calibration)

Phase 2. Intervention module

Define clinical interventions to reduce burden of disease

Determine cost functions associated with intervention delivery

Determine effect of increasing coverage

Phase 3. Define objectives and constraints

Set political and strategic objectives

Set time horizons for attaining objectives

Set logistic, ethical, political and other constraints

Institutionalization for sustainability

Transmission dynamic model

TB is a highly complex disease

Transmission dynamic model needs to incorporate the following features:

Three strains of TB: DS-TB, MDR-TB, XDR-TB

Misidentification of strain type due to lack of available drug-resistance testing

Amplification of resistance through default from treatment

Immunity by past treatment and vaccination history

Differential infectiousness by smear status

Latent infection and treatment of latent infection

Sensitivity of diagnostic algorithms for TB

Declining infectiousness with treatment

The ability to distinguish previously treated patients

AuTuMN (Australian Tuberculosis Modelling Network)

TIs

Is

Dssasdfasdf

Lk;asjdfl;as

TIx TIs

Is

Dssadfa

Askldjf;as

TIx On appropriate regimen

Undiagnosed active disease in the community

Correctly diagnosed

On inappropriate regimen

Susceptible compartments

Retreatment cases have same structural links between compartments, but different inter-compartmental flow rates

Defaults and failures pass to re-treatment undiagnosed active disease in the community compartments

Birth

New infection

Reinfection Stabilisation Rapid progression

Late reactivation

Prophylactic treatment

Re-treatment compartments

Extra-pulmonary Smear-negative

Smear-positive

Spontaneous recovery

Treatment commencement by regimen availability

Dss Dss TB diagnosis missed Dssasdfasdf

Lk;asjdfl;as

Dssadfa

Askldjf;as

Strain incorrectly identified

Correct identification as TB and by strain

Correct identification as TB, but strain not identified

Failure of health system to identify patient

Late latent compartments

Early latent compartments

Drug-susceptible TB Multidrug-resistant TB

Extensively drug-resistant TB

Transmission dynamic model

Emma McBryde James Trauer

Transmission dynamic model

Age: children, elderly

HIV status (CD4 count >300, 200-300, <200)

Comorbidity: diabetes mellitus, malnutrition, etc.

Prisoners

Indigenous population

Urban and rural poor

Geographical locations

AuTuMN model is able to stratify population by any demographic and risk profile grouping

AuTuMN successfully implemented in PNG, China, SA, India

Example: Papa New Guinea (PNG)

Transmission dynamic model

Inci

den

ce

% In

cid

ence

at

trib

uta

ble

to

MD

R-T

B

Mo

rtal

ity

DOTs, GeneXpert, pilot PMDT

Scale-up DOTs, GeneXpert, PMDT

in 1 district

Scale-up DOTs, GeneXpert,

province-wide PMDT

AuTuMN successfully implemented in PNG, China, SA, India

Example: India, China, South Africa

Transmission dynamic model

In addition to AuTuMN, other models may be included in Optima TB through re-coding of model into the Optima infrastructure

Models to be provided with optimization algorithm in their existing structures and programming languages

Transmission dynamic model

The approach

Phase 4. Apply optimization algorithm and interpret results

Phase 1. Assess burden of disease

Define diseases, health states and model structure

Compile estimates of prevalence and incidence of disease

Determine baseline projections for each disease state (calibration)

Phase 2. Intervention module

Define clinical interventions to reduce burden of disease

Determine cost functions associated with intervention delivery

Determine effect of increasing coverage

Phase 3. Define objectives and constraints

Set political and strategic objectives

Set time horizons for attaining objectives

Set logistic, ethical, political and other constraints

Institutionalization for sustainability

AuTuMN transmission dynamic model

Epidemic outputs: prevalence, incidence, mortality

Intervention module

Define clinical interventions to reduce TB burden

Measure current levels of activity

Determine the cost-coverage-outcome functions associated with intervention delivery

Interpret programs and coverage into model parameters

Calibration of the TB transmission dynamic model to the epidemic context under consideration

Intervention module

AuTuMN is complex and flexible enough to simulate all known and future interventions

Increased coverage of DOTs

Improved diagnostic algorithms

Programmatic management of MDR-TB and XDR-TB

Active and intensified case finding

Addressing HIV-TB co-infection

Preventive therapy for latent infection

Short course for MDR-TB

New diagnostics (e.g. GeneXpert)

Laboratory strengthening

Unit cost for well-established programs; total cost for new programs

Countries need to have reasonable cost data accessible for analyses

Cost data can also be obtained from budget planning analysis through existing partnerships between Optima TB and Global Fund, USAID and Stop TB

Costs of TB programs

How should the budget be allocated amongst these ‘n’

programs, modalities, and delivery options, considering their

interactions with synergies and limitations?

Resource allocation

How should the budget be allocated

among these 10 programs?

We have:

Cost-outcome curves that relate spending to model parameters

An transmission dynamic model that translates the parameters into epidemic outcomes

Next question:

Optimization

Aim: for a given amount of money, how should it be allocated to achieve the best outcome?

“Best” could mean:

• Fewest infections

• Fewest deaths

• All of the above

Formally: For resource vector R such that 𝑅 = const. and outcome 𝑂 = 𝑓 𝑅 , find R that minimizes O.

Philippines: already started, to be completed May 2016

Fiji, PNG: early discussion stage, to be completed Dec 2016

4 other countries: planning to be completed Sep 2016

Country implementation

Dec 2015 Feb 2016 Mar 2016 Apr 2016 May 2016

Initial dialogue:

GF, NPT, AuTuMN

Stakeholder engagement

Data collection

and quality check

Preliminary results Final results

Today Philippines

Acknowledgements

AuTuMN Emma McBryde James Trauer Tan Doan Nick Scott Romain Ragonnet Justin Denholm Bosco Ho

David Wilson and team

www.tb-mac.org

Modelling and methods to improve the allocative efficiency of TB programs

Nicolas A Menzies

Harvard T.H. Chan School of Public Health

www.tb-mac.org

• Aim • To improve global TB control by coordinating and promoting mathematical

modelling and other research activities to provide scientific support for policy decisions and implementation

• Objectives • Identify research questions concerning TB control that require input from

mathematical modelling or other quantitative research

• Facilitate sharing of data, information and expertise to achieve consensus on current knowledge and knowledge gaps, methodological standards and current best practice for TB control decision-making

• Fund small analytical/modelling research projects

• Disseminate results and tools to key stakeholders including TB control programmes and donors

www.tb-mac.org

TB MAC Organisation • Consortium

• Open to anyone using mathematical models or other quantitative methods to answer TB control questions

• Committee • Anna Vassall (LSHTM), Katherine Floyd (WHO), David

Dowdy (JHU), Ted Cohen (Yale), Philip Eckhoff (IDM), Michael Kimerling (KNCV), Geoff Garnett (BMGF), Damian Walker (BMGF), Richard White (Chair, LSHTM)

• Secretariat • Rein Houben • Christina Albertsen

• Advisory Panel • Mario Raviglione (WHO/GTB), Lucica Ditiu (STB

Partnership), Jane Carter (Union), Jaap Broekmans (KNCV), Mehran Hosseini (Global Fund), Amy Bloom (USAID), Bruce Levin (Emory), Phil LoBue (CDC), Peter Kim (NIAID/NIH)

www.tb-mac.org

TB MAC activities

• Thematic meetings on modelling issues (TB/HIV, diagnostics, novel drug regimens, UHC & socio-economic determinants)

• Support to foster new entrants to the field (modelling courses, small grants)

• Scientific input to high-level policy meetings or planning processes

• Promote methods development and their application in the service of policy-making

Example:

Modelling the Post-2015 WHO Targets

www.tb-mac.org

• Post-2015 End TB Targets demand TB incidence reduced by 50%, TB mortality reduced by 75% by 2025

Need major gains in high-burden countries

• TB-MAC convened a collaboration of TB modelling groups: can intensified action on TB meet the global TB targets in China, India and South Africa?

Targets modelling collaboration

www.tb-mac.org

• Countries need to understand cost-effectiveness and affordability before committing to efforts

• In planning expanded action on TB, what to do?

Multiple approaches which could be adopted, useful to compare relative costs and health impact of each

• TB places major economic burden on households, how will expanded TB control affect these costs?

Addressing economic questions

www.tb-mac.org

Form of the analytic model (example)

Uninfected,

Susceptible

Core TB Subdivision

TB Treatmnt

Smear-Neg

1st-Line

TB Treatmnt

Smear-Pos

1st-Line

Latent Slow

/ Recovered

TB Treatmnt

Smear-Neg

2nd-Line

Uninfected,

Partially

Immune

Latent

Fast

Smr-Pos

Active TB

Smr-Neg

Active TB

LTBI

Treatment

TB Treatmnt

Smear-Pos

2nd-Line

Model entry for

foreign born

Basic state

transition

(1) Exits due to

mortality not

shown.

(2) Foreign born

enter distributed

across non-

treatment states.

Model entry for

native born

www.tb-mac.org

HIV

Neg

HIV Subdivision Drug Resistance Subdivision

Pan-

Sensitive

RIF

Resistance

INH

Resistance

MDR-TB MDR+ /

XDR-TB

Treatment

Naive

Treatment

Experienced

Treatment History Subdivision

Basic state

transition

(1) Exits due to mortality not shown.

(2) Foreign born enter distributed across all drug resistance, treatment history and age states, not shown.

ART2

HIV CD4 < 350

ART 1

HIV CD4 > 350

Risk Group Subdivision

Low-Risk High-Risk Long-term

Residents

Age Group Subdivision

0-4 Years 95+ Years 85-94 Years

Model entry for

native born

Model entry for

foreign born

Recent

Arrivals

Native-Born Foreign-Born

10 year age bands

Form of the analytic model (example)

www.tb-mac.org

Small details important

• Example: mechanism for accessing care:

No TB

TB

No TB

TB

No TB

TB

No TB

TB

Presentation Clinical

suspicion Diagnostic

tests

TREA

TMEN

T

www.tb-mac.org

Collaborating models Model Model type Age structure Population strata Countries AuTuMN DC <15 and 15+ MDR/non-MDR-TB, care access. For

South Africa: HIV/ART/CD4 status CH, IN, SA

Harvard DC 15+ HIV/ART/CD4 status, drug resistance, tx history, TB care access

CH, IN, SA

Hopkins DC 15+ HIV/ART/CD4 status, MDR/non-MDR-TB SA

ICPHFI DC 15+ MDR/non-MDR-TB, tx history IN

IDM SM By month MDR/non-MDR-TB, provider and tx history

CH

NTU DC 15+ MDR/non-MDR-TB, health care system, tx history

CH

STAMP SM By month Sex, tx history and type, time since infection and activation

IN

TIME DC <15 and 15+ HIV/ART/CD4 status, MDR/non-MDR-TB, tx history

CH, IN, SA

UGA DC <15 and 15+ HIV/ART status, MDR SA

Planning to serve country needs at scale

Microsims: model details very different

www.tb-mac.org

Defining intervention scenarios • Sought input from program experts in each country to

define scenarios for expanded TB control

e.g. Reduce default between diagnosis and treatment from 10% to 5%

• Worked with country experts to define activities required to achieve scenario goals

e.g. Compensation for patient expenses assoc with diagnosis and treatment, follow-up of defaulters in community

www.tb-mac.org

Analytic approach

• Considered outcomes realized over 20 years

• DALYs used as summary measure of health burden

• Costs assessed from multiple perspectives

TB health services (relevant for affordability)

= Diagnosis, 1st line tx, 2nd line tx, other costs, program overheads

Patients and families (relevant for economic burden)

= Productivity costs, patient medical + non-medical costs

www.tb-mac.org

Cost-effectiveness, South Africa

0 5 10 15

0

5000

10000

15000

Incre

men

tal C

osts

(U

S$

, m

il.)

DALYs Averted (mil.)

Health System Perspective

● ● ● ●IPT for HIV−positives Expand access Improve treatment Combination

● 1000

● 1300

● 850

● 1400

0 5 10 15

0

2000

4000

6000

8000

10000

12000

14000

Incre

men

tal C

osts

(U

S$

, m

il.)

DALYs Averted (mil.)

Societal Perspective

● 730

● 1100

● 600

● 1100

www.tb-mac.org

Cost-effectiveness, China

0 1 2 3 4

0

2000

4000

6000

8000

10000

12000

Incre

men

tal C

osts

(U

S$

, m

il.)

DALYs Averted (mil.)

Health System Perspective

● ● ● ●Expand access Introduce Xpert Improve treatment Combination

●

470

● 16000

●

450

● 3700

0 1 2 3 4

−5000

0

5000

10000

Incre

men

tal C

osts

(U

S$

, m

il.)

DALYs Averted (mil.)

Societal Perspective

● Dom.

● 23000

● Dom.

● 1100

www.tb-mac.org

Cost-effectiveness, India

0 20 40 60

−2000

0

2000

4000

6000

8000

Incre

men

tal C

osts

(U

S$

, m

il.)

DALYs Averted (mil.)

Health System Perspective

● ● ● ● ●Expand access Active case−finding Introduce Xpert Improve treatment Combination

● Dom.

●

3700● 760

● 220

● 94

0 20 40 60

−25000

−20000

−15000

−10000

−5000

0

5000

Incre

men

tal C

osts

(U

S$

, m

il.)

DALYs Averted (mil.)

Societal Perspective

● Dom.

●

2300● 330

● Dom.

● Dom.

www.tb-mac.org

Rankings: impact, cost-effectiveness

Introduce Xpert Expand Access Improve Treatment

Improve Treatment Expand Access IPT for HIV-positives

Introduce Xpert Expand Access Improve Treatment Active Case Finding

DALYs Averted

Introduce Xpert Improve Treatment Expand Access

Expand Access Improve Treatment IPT for HIV-positives

Introduce Xpert Expand Access Improve Treatment Active Case Finding

Cost per DALY Averted

CHINA

INDIA

SOUTH AFRICA

CHINA

INDIA

SOUTH AFRICA

more less

lower higher

= approximately equal

www.tb-mac.org

Conclusions

• Details of expanding TB services likely very different between countries: relevant for structure and process

• Expanding access to care generally impactful and efficient

• Impact of Xpert contingent on DR-TB treatment quality

• In general, could see large reductions in patient econ. burden

• Substantial variation across models used for analysis

What we don’t know matters to decision-making

• Need to ask questions in context of local policy process, more detailed examination of policy space

www.tb-mac.org

Thanks to many Economists

Ines Garcia Baena, Fiammetta Bozzani, Yoko Laurence, Susmita Chatterjee, Sun Qiang, Nicola Foster, Andrew Siroka

Modellers

Rein Houben, Tom Sumner, Grace Huynh, Nimalan Arinaminpathy, Jeremy Goldhaber-Fiebert, Hsien-Ho Lin, Chieh-Yin Wu, Sandip Mandal, Surabhi Pandey, Sze-chuan Suen, Eran Bendavid, Andrew Azman, David Dowdy,Marcus Feldman, Andreas Handel, Christopher Whalen, Stewart Chang, Bradley Wagner, Philip Eckhoff, James Trauer, Justin Denholm, Emma McBryde, Ted Cohen, Joshua Salomon

Country experts, other experts

Carel Pretorius, Marek Lalli, Jeffrey Eaton, Delia Boccia, Mehran Hosseini, Suvanand Sahu, Colleen Daniels, Lucica Ditiu, Daniel Chin, Lixia Wang, Vineet Chadha, Kiran Rade, Puneet Dewan, Piotr Hippner, Salome Charalambous, Alison Grant, Gavin Churchyard, Yogan Pillay, David Mametja, Michael Kimerling, Richard White

www.tb-mac.org

Health service costs

0

5000

10000

15000

Au

TuM

N

Harv

ard

IDM

NT

U

TIM

E

Me

an

Au

TuM

N

Harv

ard

IDM

NT

U

TIM

E

Me

an

Au

TuM

N

Harv

ard

IDM

NT

U

TIM

E

Me

an

Au

TuM

N

Harv

ard

IDM

NT

U

TIM

E

Me

an

Expand access Introduce Xpert Improve treatment Combination

Incr

em

en

tal C

ost

s (U

SD

, m

il.) China

−2000

0

2000

4000

6000

8000

10000

Au

TuM

N

Harv

ard

ICP

HF

I

STA

MP

TIM

E

Me

an

Au

TuM

N

Harv

ard

ICP

HF

I

STA

MP

TIM

E

Me

an

Au

TuM

N

Harv

ard

ICP

HF

I

STA

MP

TIM

E

Me

an

Au

TuM

N

Harv

ard

ICP

HF

I

STA

MP

TIM

E

Me

an

Au

TuM

N

Harv

ard

ICP

HF

I

STA

MP

TIM

E

Me

an

Expand access Active case−finding Introduce Xpert Improve treatment Combination

Incr

em

en

tal C

ost

s (U

SD

, m

il.) India

0

5000

10000

15000

AuTu

MN

Harv

ard

Hop

kin

s

TIM

E

UG

A

Me

an

AuTu

MN

Harv

ard

Hop

kin

s

TIM

E

UG

A

Me

an

AuTu

MN

Harv

ard

Hop

kin

s

TIM

E

UG

A

Me

an

AuTu

MN

Harv

ard

Hop

kin

s

TIM

E

UG

A

Me

an

IPT for HIV−positives Expand access Improve treatment Combination

Incr

em

en

tal C

ost

s (U

SD

, m

il.) South Africa

●

Diagnosis

First−line treatment

MDR−TB treatment

Other provider costs

Program overheads

Total TB service costs

www.tb-mac.org

Patient costs

−10000

−5000

0

5000

Au

TuM

N

Harv

ard

IDM

NT

U

TIM

E

Me

an

Au

TuM

N

Harv

ard

IDM

NT

U

TIM

E

Me

an

Au

TuM

N

Harv

ard

IDM

NT

U

TIM

E

Me

an

Au

TuM

N

Harv

ard

IDM

NT

U

TIM

E

Me

an

Expand access Introduce Xpert Improve treatment Combination

Incr

em

en

tal C

ost

s (U

SD

, m

il.) China

−30000

−20000

−10000

0

Au

TuM

N

Ha

rvard

ICP

HF

I

STA

MP

TIM

E

Me

an

Au

TuM

N

Ha

rvard

ICP

HF

I

STA

MP

TIM

E

Me

an

Au

TuM

N

Ha

rvard

ICP

HF

I

STA

MP

TIM

E

Me

an

Au

TuM

N

Ha

rvard

ICP

HF

I

STA

MP

TIM

E

Me

an

Au

TuM

N

Ha

rvard

ICP

HF

I

STA

MP

TIM

E

Me

an

Expand access Active case−finding Introduce Xpert Improve treatment Combination

Incr

em

en

tal C

ost

s (U

SD

, m

il.) India

−5000

−4000

−3000

−2000

−1000

0

AuTu

MN

Harv

ard

Hop

kin

s

TIM

E

UG

A

Me

an

AuTu

MN

Harv

ard

Hop

kin

s

TIM

E

UG

A

Me

an

AuTu

MN

Harv

ard

Hop

kin

s

TIM

E

UG

A

Me

an

AuTu

MN

Harv

ard

Hop

kin

s

TIM

E

UG

A

Me

an

IPT for HIV−positives Expand access Improve treatment Combination

Incr

em

en

tal C

ost

s (U

SD

, m

il.) South Africa

●Patient productivity costs Patient non−medical costs Patient medical costs Total patient−incurred costs)

www.tb-mac.org

Budget impact

2020 2025 2030 2035

0

200

400

600

800

1000

1200

1400

China, Base Case Scenario

Tota

l Co

sts

(US

D,

mil.

)

2020 2025 2030 2035

China, Combination Scenario

2020 2025 2030 2035

0

500

1000

1500

2000

India, Base Case Scenario

Tota

l Co

sts

(US

D,

mil.

)

2020 2025 2030 2035

India, Combination Scenario

2020 2025 2030 2035

0

500

1000

1500

South Africa, Base Case Scenario

Tota

l Cost

s (U

SD

, m

il.)

2020 2025 2030 2035

South Africa, Combination Scenario

DiagnosisFirst−line treatment

MDR−TB treatmentOther provider costs

Program overheads

Title of Presentation 58

Discussion

Steering group: Coordination of AE analyses

Title of Presentation 59

• Coordination across pilot countries and modelling

initiatives

• Maximizing learning and application value, e.g.:

• National vs. provincial analyses

• Complementary objectives and research questions

• Implementation modes/program planning vs. investment

case focus

• Sharing of primary country data when appropriate

• Model comparisons

• Coordination of messages

Technical group: Joint work areas

60

1. Intervention and expenditure/budget categories

(nomenclature and what actually used in tools, this includes

non-service areas like training, infrastructure)

2. Costs

• Cost functions – need to collate existing cost work

• Patient-side costs – specific TB meeting focussing on this?

• Outcomes (incl. catastrophic costs)

• Understand costs & cost drivers, incl. broader system/operational costs

3. Intervention effectiveness

• Ongoing work at McGill

• Ongoing TB treatment systematic review by WHO

4. Other evidence for informing methods & parameterisation

61

Thank you