Is treatment-as-prevention the new “game-changer”? Economic evaluation of HIV combination prevention

Till Bärnighausen1,2, David E. Bloom1, Salal Humair1,3

1Department of Global Health and Population, Harvard School of Public Health, Boston, MA, USA2Africa Centre for Health and Population Studies, University of KwaZulu-Natal, Mtubatuba, South Africa3School of Science and Engineering, Lahore University of Management Sciences (LUMS), Lahore, Pakistan

IAEN Pre-conference meeting, 21 July 2012

Background (1)• On the eve of the June 2011 UN General Assembly High

Level Meeting on AIDS, HIV/AIDS researchers, activists and the press described the results of the HPTN 052 randomized controlled trial (Cohen et al. 2011) as a “game-changer” in the fight against AIDS−Lancet Editorial 2011

−Dickinson 2011

−Clark 2011

−BBC 2011

−Economist 2011

Background (2)• Unclear how TasP could be funded, given flat-lining or

declining financial support for global HIV programs (UNAIDS 2011)

• Unclear when TasP should ideally be implemented. Policymakers could choose not to start implementing TasP until other HIV interventions have reached certain scale levels

• We investigate the cost-effectiveness of different combinations of TasP with two other interventions of proven biological efficacy−Medical male circumcision (MMC) (Auvert et al. 2005; Bailey et al.

2007; Gray et al. 2007)

−Antiretroviral treatment under current eligibility criteria (ART)

ART under current eligibility criteria

0.00

0.20

0.40

0.60

0.80

1.00

1.20

<10% 10-20% 20-30% 30-40% >40%

Adju

sted

haz

ard

ratio

Proportion of all HIV-infected people receiving ART

p=0.002

p<0.001p=0.016

p=0.590

Tanser, Bärnighausen, Grapsa, Newell CROI 2012

Intervention characteristics

MMC ART/TasPDelivery Once-off Life-long

Population in need Larger (e.g. 14 million men for SA)

Smaller (e.g. 2.4 million for TasP for SA)

HIV prevention Acquisition (reduced by 60% per unprotected sex act)

Transmission (reduced by 96% per unprotected sex act)

First-order effect Immediately in men Immediately in women and men

Mortality reduction With delay Immediately

Modeling approach: HIV combination prevention

• Most current models of the HIV epidemic have the limitation that they need to be calibrated to past trends− Limitation that calibration parameter is estimated based on historical

observations and is not guaranteed to remain stable in the future under historically untried interventions (TasP)

− Since calibration parameter is usually a black box and numerically rather than analytically derived from underlying biological and behavioral variables, it is difficult to disaggregate the effects of interventions that operate through different biological or behavioral pathways

• We thus built an analytical model, derived from underlying biological and behavioral mechanisms

Model dynamics

Sexual activityNew HIV infections

Male

HIV+ inflow

Receiving ARTmortality

Not receivingART mortality

Needing ART

HIV stage-specific mortality

Female

Generalmortality

HIV- inflow

Scenarios• A range of different scenarios combining MMC, ART, and TasP

with the following coverage levels (the lowest levels are the current coverage estimates for South Africa)− MMC: 45%, 60%, and 80%

− ART: 50%, 60%, and 80%

− TasP: 0%, 20%, 40%, 60%, and 80%

• All combinations of coverage levels• Coupling of ART and TasP, reflects our belief that that in a real-

life scale-up of TasP, ART coverage of those currently eligible to receive treatment will increase as well

Country application: South Africa• Largest number of HIV-infected individuals in the world

(UNAIDS 2011)• Worldwide largest number of people currently on ART and

needing ART under current guidelines (WHO/UNAIDS/UNICEF 2010)

• High HIV incidence• Political commitment to scaling up MMC but currently

relatively slow rates of increase• Good-quality empirical data

Data• Whenever possible South African national data for base case;

otherwise international sources− Total number HIV-infected and -uninfected: Shisana et al. 2009; StatsSA 2009

− Disease stage distribution and ART coverage: Adam and Johnson 2009

− Time from HIV seroconversion to disease stage: Todd et al. 2007; eART-linc 2008; WHO 2009; Minga et al. 2009

− Annual probability of death by HIV status, disease stage and ART status: Badri et al. 2006; Braitstein et al. 2006; WHO 2009; StatsSA 2009

− Transmission probability in different disease stages: Boiley et al. 2009

− Number of sex partners: Shisana et al. 2009

− Number of sex acts: Global sex survey 2005

− MMC effect: Auvert et al. 2005; Bailey et al. 2007; Gray et al. 2007

− TasP effect: Cohen et al. 2011

− ART and MMC costs: Schwartländer et al. 2011

Results: incidence, mortality and costs

0.0%

0.1%

0.2%

0.3%

0.4%

0.5%

0.6%

0.7%

0.8%

0.9%

0

5,000,000,000

10,000,000,000

15,000,000,000

20,000,000,000

25,000,000,000

Scenarios

HIV

inci

denc

e

Cum

ulat

ive

cost

s ($

bill

ions

)

0%

1%

2%

3%

4%

5%

6%

7%

8%

9%

10%

0

5,000,000,000

10,000,000,000

15,000,000,000

20,000,000,000

25,000,000,000

Scenarios

Mor

talit

y ra

te in

HIV

-infe

cted

pe

ople

Cum

ulat

ive

Cos

ts ($

bill

ions

)

Results: HIV incidence for all scenarios

2010 2015 2020 2025 2030 2035 20400.0%

0.2%

0.4%

0.6%

0.8%

1.0%

1.2%

A50%:T0%:C45%A80%:T80%:C80%A70%:T0%:C60%

Year

HIV

Inci

denc

e (p

er y

ear)

A = antiretroviral treatment (ART) under current guidelines, T = treatment-as-prevention (TasP), C = medical male circumcision (MMC)

Results: mortality in all scenarios

2010 2015 2020 2025 2030 2035 20403%

4%

5%

6%

7%

8%

9%

10%

A50%:T0%:C80%A80%:T80%:C45%A70%:T0%:C60%

Year

Annu

al M

orta

lity

in H

IV-In

fect

ed

Peop

le

A = antiretroviral treatment (ART) under current guidelines, T = treatment-as-prevention (TasP), C = medical male circumcision (MMC)

Results: costs in all scenarios

2010 2015 2020 2025 2030 2035 20400

500000000

1000000000

1500000000

2000000000

2500000000

A80%:T80%:C45%A50%:T0%:C80%A70%:T0%:C60%

Year

Cost

per

Yea

r ($

billi

ons)

A = antiretroviral treatment (ART) under current guidelines, T = treatment-as-prevention (TasP), C = medical male circumcision (MMC)

Results: incremental cost-effectiveness ratios

ICER = incremental cost-effectiveness ratio, baseline = ART 50%, MMC 45%, TasP 0%

ICER

US$ per infection averted

US$ per death averted

MMC (80%) vs. baseline

1,096 5,198

ART (80%) vs. baseline + MMC(80%)

7,765 5,741

TasP (80%) vs. MMC (80%) and ART(80%)

14,894 16,180

Further results (1)• Combination of high ART coverage under current

guidelines and high MMC coverage provides approximately the same substantial HIV incidence reduction as TasP

• The combination of high ART and high MMC coverage is considerably less expensive than TasP, requiring approximately US$ 5 billion less over the period 2009-2020

Further results (2)• In costs per infection averted, increased MMC coverage

(with costs of about US$1000 per infection averted) outperforms high ART coverage as well as TasP (both with costs close to US$7000 per infection averted) vs. baseline

• The cost-effectiveness of MMC increases over time; and, unlike ART or TasP, MMC becomes cost-saving after 2040

• MMC becomes cost-saving earlier as TasP coverage increases in the counterfactual scenario

Implication• Our results suggest, that we should first expand MMC coverage and ART coverage under current guidelines to near-universal levels – and only then add TasP to the combination prevention package

Ongoing empirical research• New MMC and ART cost data is becoming available

• Data from the TasP trials will inform the assumptions in economic evaluation−Program costs−Productivity effects−Quality-of-life changes

Acknowledgments• Funding

−World Bank−National Institute of Child Health and Human

Development R01 HD058482-01 and National Institute of Mental Health R01 MH083539-01 (Till Bärnighausen)

Results: costs and mortality

0%

1%

2%

3%

4%

5%

6%

7%

8%

9%

10%

0

5,000,000,000

10,000,000,000

15,000,000,000

20,000,000,000

25,000,000,000

Scenarios

Mor

talit

y ra

te in

HIV

-infe

cted

peo

ple

Cum

ulat

ive

Cos

ts ($

bill

ions

)