Instituto Nacional de Salud Publica Cuernavaca Mexico

Evaluation of the Frontiers Prevention Program Stefano M. Bertozzi IADB Rio de Janeiro November 2003 INSP Paul Gertler Juan Pablo Gutierrez Ken Morrison Int. HIV/AIDS Alliance Sam Mc. Pherson Fiona Samuels ASCI Lalit Dandona Rakhi Dandona IJCG Diana Molina

Why so concerned about evaluation of HIV/AIDS prevention programs? • Epidemic is catastrophic – 40 million people already infected • Prevention efforts have not reversed growth • Large $$ are starting to flow for prevention • Despite success in many small prevention trials, the rare community prevention trials have had very disappointing results • We badly need to know what works and what doesn’t in large-scale prevention

What is the FPP? • HIV/AIDS prevention program in countries at the frontier of the epidemic – implemented by the International HIV/AIDS Alliance – Gates Foundation funding – Evaluated by INSP with ASCI (India) and IJCG (Ecuador) • Delivery of a SOTA comprehensive package of interventions within specific geographic sites – – Ecuador India (Andhra Pradesh) Cambodia Madagascar

How does the Alliance work? • Increase the capacity of local NGOs and CBOs in geographic sites • Key population focused – given hypothesis that in nascent/concentrated epidemics the most effective way to reverse growth is to focus on KPs • Participatory mapping and assessment of KPs and their HIV prevention needs • $$ support for NGOs and CBOs to deliver programs and services with and for all of the epidemiologically-relevant KPs in each site

FPP framework INTERMEDIATE OUTCOMES Enabling Environment Service and Commodity provision for KPs ACTIVITIES (White arrows) PURPOSE GOAL Decrease in KP risky behavior Decrease in KP STI prevalence Decrease in HIV incidence amongst KPs Decrease In HIV incidence in site

FPP Interventions • • Strengthening (NGO & others) capacity to work with KPs KP collective mobilization for advocacy Implementing KP mutual support, cultural and solidarity building activities Implementing anti stigma and HIV prevention with general public --------------------------------------------------------- • • Implementing peer outreach activities Developing & disseminating IEC for KPs Implementing KP risk reduction skills building activities Strengthening clinical capacity (STI, VCT) and quality of services

Evaluation Specific Aims 1. To measure Δ attributable to FPP in sexual risk behavior among SWs and their clients and among MSM – numbers of partners, partner characteristics, sex acts performed, and quality and consistency of condom use. 2. To measure Δ in individual, group and community factors hypothesized to be conducive or a hindrance to increasing safer sexual behavior – knowledge, skills, discrimination, access to services and commodities, social capital, social norms etc.

Specific Aims (2) 3. To measure Δ in the prevalence of STIs – relative importance of sexual risk behavior vs. STI treatment in explaining Δ in prevalence of treatable STIs. 4. To describe – baseline characteristics of individual SWs and MSM – baseline characteristics of the broader community predictive of positive and negative changes in sexual risk behavior over time (w/ and w/o FPP)

Comparison design with randomized intervention & comparison sites INDIA ECUADOR SW MSM PLHA MADAGASCAR

Sites that must be included Site selection & randomization

Sites that can’t be included

Feasible sites for controlled evaluation

Evaluation sample

Comparison design with randomized intervention & comparison sites INDIA ECUADOR MADAGASCAR

New evaluation design? INDIA Andhra Pradesh ECUADOR

New evaluation design INDIA Karnataka IA I FP P- FP P+ Andhra Pradesh ECUADOR

New evaluation design INDIA Andhra Pradesh ECUADOR Karnataka

Evaluation design • Ecuador – still randomized, but underpowered alone with 5 intervention and 3 comparison sites (n=250 per KP per site) • India – Andhra Pradesh: dose-response with both intentional and unintentional variation in dose (quantity and quality of interventions) (37 sites, n=250 per KP per site) – Karnataka: randomized intervention and comparison sites similar to FPP- but site definition is somewhat different (? >24 sites, n≈5000 total)

Dose-response? • Disaggregate the overall FPP intervention package and develop a quantity/quality score for each of the subcomponents • Examine the differential impact of the various components of the intervention package • Seek to explain changes in outcome variables in a site by: – quantity/quality of the different components of the FPP that have been implemented in the site – controlling for differences in individual KP, KP community and site baseline characteristics

OUTCOME (STI prevalence) Dose-response approach FPP Interventions A FPP Interventions B SITE A FPP Interventions C SITE B SITE C

OUTCOME (STI prevalence) Dose-response approach FPP Interventions A A FPP Interventions B FPP Interventions C B SITE A t 1 SITE C SITE B t 2 t 1 C t 2

D-R: intentional & non-intentional • Intentional – FPP+ sites vs. FPP- sites – Non-randomized, but no reason a priori to expect that they would not be well matched • Non-intentional – large heterogeneity expected – how to control for factors that could affect both the “dose” and the “response”? • • Site openness/receptivity (by proxy) Pre-existing NGO work Baseline levels of risk behavior Baseline STI prevalence

Indicators of Openness/receptivity • Family planning programs • Fertility rate • Qualitative assessment (KPs & gatekeepers) • Community regulations/enforcement • Political affiliation • Religious/ethnic mix • Openness of sex work/MSM

Indicators of NGO/CBO presence • Coverage at baseline of HIV/AIDS services by NGO/CBOs • Proportion of related services, esp. FP, delivered by NGOs • Proportion of KPs who feel they have access to or are supported by NGO/CBO

Current STI prevalence • Survey currently in the field – Chlamydia & gonorrhea – HSV 2 – Syphilis

Risk Behavior • Multiple observations per respondent; 3 last sexual encounters: • Type of sex (vaginal, anal, oral, non-penetrative…) • Condom use • Partner characteristics • Perceptions of partner STI/HIV risk/status

Results from Ecuador and India pilots

Internal and External Evaluations SUB-GROUPS OF KPs USED FOR EXTERNAL EVALUATION IN GSs SW IDU PLHA MSM I N T E R V E N T I O NS CLIENT KPs