S P. Understanding heterogeneity in HIV infection and efficiency in HIV progam responses

Transcription

1 ASAP Regional Training on Epidemiological and Economic Tools for HIV/AIDS Strategic Planning S P Understanding heterogeneity in HIV infection and efficiency in HIV progam responses

2 80 POPULATION-BASED HIV PREVALENCE IN FRANCISTOWN, BOTSWANA Male Female Sources: BAIS, 2005

3 57 Antigua and Barbuda 58 Trinidad and Tobago 66 St. Kitts and Nevis 71 Mexico 77 Botswana 84 St. Lucia 94 Dominica 95 Belize 97 St. Vincent and the Grenadines 99 Jamaica 116 Suriname 117 Dominican Republic 147 Guyana 176 Haiti

4 GLOBAL SHARE OF HIV INFECTIONS 4

5 HETEROGENEITY OF HIV HIV reached most places at similar time, but spread very differently HIV extraordinarily heterogeneous, yet, sadly, most HIV strategies look similar 5

6 HETEROGENEITY OF HIV IN AFRICA 0-0.1% 1-5% 3-7% 15-35% Sources: UNAIDS 2004 estimates used unless

7 HETEROGENEITY OF HIV: Why is HIV so diverse? In some countries people have more sexual partners than others? 7

8 LIFETIME SEXUAL PARTNERS VARY LITTLE GLOBALLY 8

9 HETEROGENEITY OF HIV: Why is HIV so diverse? In some countries people have more sexual partners than others? -- not very useful Variable infectiousness makes concurrency critical 9

10 HIV TRANSMISSION RISKS Half of all transmission Wawer et al,

11 CONCURRENT PARTNERSHIPS GLOBALLY Female Male Percentage of year olds reporting > 1 regular partner in last year 0 Singapore Sri Lanka Thailand Philippines Kenya Tanzania Zambia Cote Lesotho 11 D'Ivoire Sources: Cassell et al, 2005

12 SMALL DIFFERENCES IN NUMBERS OF PARTNERS CAN CREATE A TRANSMISSION CORE Average Number of Partners Largest components In largest component: 2% 10% 41% 64% Source: Martina Morris, Univ. of Washingtion, used with permission from a presentation given at a meeting on concurrent sexual parnerships and sexually transmitted infections at Princeton University, 6 May 2006.

13 SEXUAL NETWORKING IN LIKOMA,MALAWI Source: Kohler H and Helleringer S. The Structure of Sexual Networks and the Spread of HIV in Sub-Saharan Africa: Evidence from Likoma Island (Malawi). PARC Working Paper Series: WPS Two-thirds linked by single chain of infections over last 3 years

14 HETEROGENEITY OF HIV: MALE CIRCUMCISION Meta-analyses - circumcised men 50-70% less likely to get HIV Randomized trial in South Africa - male circumcision reduced HIV transmission by 60+% 14

15 DIVERSITY OF HIV IN ASIA IN 2005 Sources: UNAIDS, 2004 Bangladesh Pakistan Philippines Indonesia Fiji China Vietnam PNG India Burma Thailand Cambodia (Even lower outside Papua) High male circumcision Low male circumcision

16 HETEROGENEITY OF HIV: THE LETHAL COCKTAIL Concurrent sexual partnerships and limited male circumcision are likely to be two critical factors explaining much heterogeneity in HIV epidemic potential 16

17 TRANSMISSION DYNAMICS (1-2) African evidence underscores importance of understanding transmission dynamics Concentrated - transmission largely among key populations Interventions with those populations would reduce overall infection Generalized - transmission mainly outside key populations Would continue despite effective key population interventions 1% conventionally used as the threshold prevalence to distinguish the two but a crude way to distinguish them 17

18 Successful prevention requires: Knowledge of epidemic profile: Distributions and trends of HIV and STD infections Prevalence and distribution of risk behaviors 18

19 IDU IGNITING A DORMANT EPIDEMIC IN JAKARTA 120,000 Sexual infections originating from IDU , , Cumulative HIV infections 80,000 60,000 40,000 20, , HIV prevalence HIV infections if nothing changes HIV infections in IDUs if nothing changes ,049 HIV infections without IDU epidemic IDU HIV prevalence Sources: Pisani, 2005

20 IMPLICATIONS FOR STRATEGY: SURVEILLANCE EXISTS BUT DOESN T SHAPE STRATEGY (2-3) For years, a major regional project in Southern Africa focused on sex workers, truckers and traders at borders Careful enumeration revealed the following populations at Swaziland s borders Group/Site Lomahsha Namaacha Oshoek Ngwenya Lavumisa Golela Sex Workers Local Visiting Truckers Crossing Sleeping Traders Crossing Sleeping

21 SEXUAL PARTNERSHIPS IN SWAZILAND 60% 50% 40% 30% 20% 10% 0% Guards Drivers Soldiers Police Seasonal Casual sex Commercial sex workers Sources: FHI BSS, 2002

22 Optimizing HIV/AIDS prevention programs: towards multidimensional allocative efficiency México, DF, octubre 2006 Sergio Bautista Paola Gadsden Stefano M Bertozzi

23 State of the Art Until now, the discussion regarding optimal resource allocation for HIV prevention has focused on two dimensions: Allocation among interventions Allocation among subpopulations Both of which we do poorly

24 Evidence of CE for prevention interventions 3 Low-level Concentrated Generalized Generalized Intervention epidemic epidemic low-level high-level Blood safety 1 study found 1 study found 4 studies found 2 studies found ART to reduce MTCT 2 studies found 3 studies found 3 studies found Sterile injection 1 study found 2 studies found 1 study found 1 study found VCT 1 study found 2 studies found Peer-based programs 4 studies found 4 studies found STI treatment 3 studies found 1 study found School-based education Harm reduction for IDU ART for prevention and postexposure prophylaxis Condom promotion, distribution and IEC Condom social marketing Surveillance IEC Abstinence education MTCT, feeding substitution Drug substitution for IDU Universal precautions Vaccines Behavior change those HIV+ 1 study found 2 studies found 1 study found 1 study found 1 study found No cost-effectiveness studies found Source: Bertozzi SM, Padian N, Wegbreit J. et al HIV/AIDS prevention and treatment. In: DCP2 Poor data = inefficient allocation

25 Dimension 1: Example 2 Allocation of resources to interventions 100% IEC Condoms VCT STI PMTCT Blood Safety Other 80% 60% 40% 20% 0% Mauritania (0.6%) Madagascar (0.9%) Ghana (4.6%) Burkina (7.2%) Cameroon (7.2%) Uganda (10%) Mozambique (14.5%) Zambia (20%) Source: World Bank Review of National HIV/AIDS Strategies for Countries Participating in the World Bank s Africa Multi-Country AIDS Program (MAP). The Concentrated epidemic of Generalized these countries is Generalized characterized by heterosexual transmission Low High

26 State of the Art Until now, the discussion regarding optimal resource allocation for HIV prevention has focused on two dimensions: Allocation among interventions Allocation among subpopulations Both of which we do poorly

27 PREVENTION SPENDING AND PERCENTAGE OF AIDS CASES AMONG LATIN AMERICAN MSM Argentina Bolivia Brazil Chile Costa Rica DR El Salvador Guatemala Mexico Panama Paraguay Peru Uruguay Preventive Expenditure on MSM AIDS Cases among MSM Source: SIDALAC/ UNAIDS. 2005

28 Dimension 2: Example Coverage of subpopulations by region (2003) IDU Sex Workers MSM Prisoners Children living on the street 100% Primary Students Secondary Students 80% 60% 40% 20% 0% Africa Eastern Mediterranean Eastern Europe Americas South-East Asia Western Pacific All Reporting Countries Source: USAID, et al Coverage of selected services for HIV/AIDS prevention, care and support in low and middle income countries in Less than 20% of coverage: IDU, SW, MSM

29 Dimension 1: Allocation among interventions To interventions that produce the greatest value for money Cost-effectiveness Analysis (CEA)

30 Dimension 1: Allocation among interventions Y Comparison of two different interventions in the same subpopulation Benefit Y2 Y1 Intervention VCT Intervention IEC X1 Investment X

31 Dimension 2: Allocation among subpopulations To subpopulations at greatest risk of acquiring and/or transmitting HIV infection

32 Dimension 2: Allocation among subpopulations Y Comparison of the same intervention in two different subpopulations Benefit Y1 Street-based Sex Workers Y2 Brothel-based Sex Workers X1 Investment X

33 Interventions What? (1) Subpopulations Where? (2) How? Inputs (3)

34 Dimension 3: Allocation among inputs Optimal combination of inputs within a given intervention Cost-effectiveness analysis typically assumes constant (optimal?) efficiency of implementation across sites A cost-effective intervention may turn out to be very cost-ineffective if implemented inefficiently i.e. if you don t purchase the optimal combination of inputs including purchasing too many villas in Switzerland

35 Y Dimension 3: Allocation among inputs Not all implementations of VCT intervention are at its efficiency frontier Y2 Intervention VCT Y1 Benefit Investment X1 X

36 1,000 Scale and Average Unit Cost of VCT programs in 5 countries US$ Average Unit Costs ,000 10, ,000 Annual clients receiving VCT Mexico Uganda Russia India South Africa Source: Preliminary analysis of PANCEA data. Unpublished data. 2006

37 1,000 Scale and Average Unit Cost of VCT programs in 5 countries US$ Average Unit Costs ,000 10, ,000 Mexico Annual clients receiving VCT Source: Preliminary analysis of PANCEA data. Unpublished data. 2006

38 Optimal Allocation Framework 100 Inputs (3) 100 Subpopulations (2) Optimal allocation C 0 Interventions (1) 100 B A

39 Optimal Allocation Framework Optimal allocation Inputs (3) Subpopulations (2) Isoeffectiveness parabola Interventions (1)

40 Allocation Costs -- Subpopulations Cost of Allocation Cultural costs lnformation costs 0 Allocation efficiency 100 % among subpopulations

41 Allocation Costs -- Interventions Cost of Allocation Cultural costs lnformation costs GLOBAL PUBLIC GOOD 0 Allocation efficiency 100 % among interventions

42 Allocation Costs -- Inputs Cost of Allocation Cultural costs lnformation costs VERY LOCAL GOOD e.g. clinic 0 Allocation efficiency 100 % among inputs

43 Summary Three dimensions to achieve allocative efficiency in prevention: Allocation among interventions (WHAT) Allocation among subpopulations (WHO) Allocation among inputs (HOW) Each one of these 3 dimensions is essential to achieving the best use of resources We have been ignoring the 3 rd and it may well be the one where the marginal return to investment in improving efficiency is greatest

44