1. Community-Level HIV Incidence Outcomes of NIMH Project Accept (HPTN 043) Glenda Gray for the Project Accept Study Team IAS 2013 2 July 2013 Kuala Lumpur, Malaysia

2. Majority of persons unaware of HIV status –Low testing motivation Limitations to individual clinic-based VCT: –Passive, inaccessible to certain groups HIV silent and hidden ART slowly rolled out nationally and globally Context matters… in 2002/3

3. The first community- randomized trial designed to: –test a combination of social, behavioral, and structural approaches for HIV prevention –assess the impact of an integrated strategy for HIV prevention on HIV incidence –assess the impact of an integrated strategy for HIV prevention on behavioral and social outcomes at the community level. NIMH Project Accept (HPTN 043)

4. Community-level approach chosen because earlier VCT research in Africa found that while it lead to increased information and risk reduction, many avoided testing as it was not normative, because of stigma and no available support services or effective treatment for those testing positive. Rationale for NIMH Project Accept (HPTN 043)

5. To determine whether communities that received at least 36 months of intervention would have lower HIV incidence, increased rate of HIV testing, lower rates of sexual risk behavior and lower stigma compared to control communities Objective

6. 48 communities in 5 study sites Vulindlela, South Africa Chiang Mai, Thailand Kisarawe, Tanzania Soweto, South Africa Mutoko, Zimbabwe

7. Phase III cluster community (pair) randomized trial of a community-level behavioral intervention to reduce HIV incidence: –8 in rural Zimbabwe, 10 in rural Tanzania, 8 in Soweto and 8 in rural KwaZulu Natal, South Africa, and 14 in rural northern Thailand –Thailand data not included due to low prevalence (<1%) and negligible incidence Trial Design

8. Communities randomized to 2 VCT approaches Community-based VCT (CBVCT N = 24 communities) 1.Community preparation, outreach, mobilization 2.Mobile VCT 3.Post-test support services a.Stigma-reduction skills training b.Coping effectiveness training c.Ongoing counseling 4.Ongoing data feedback and field adjustments Standard VCT (SVCT N = 24 communities) 1.Clinic-based VCT 2.Standard VCT services normally provided in that community

9. The COMPLETE INTERVENTION PACKAGE for community based VCT (CBVCT) Community Mobilization Mobile VCT brought to where people are Testing Support Services TSS club guests receive stigma and HIV/AIDS info: Mobilized for testing Participants receive risk reduction information and mobilize partners for testing Community members mobilized: Social networks, door-to-door, mob talks, community events Social networks are identified and secured for information sessions Update from community members around caravan Participants tested, move on to TSS for support and referrals DATA

10. Study Design: Timeline Pilot studies in Zimbabwe and Thailand Community Selection, Recruitment, Funding Baseline Survey 2001 2004 2003 2005 200620072008 2002 2009 2010 2011 INTERVENTION Community Random- ization Post- Intervention Assessment Qualitative Cohort Probability sample of 18-32 year olds Survey only (N=14,567) Total N = 48 communities 24 intervention / 24 control Assessment of a random sample of 18-32 year olds in each intervention and control community Behavioral survey (N=56,683). Biologic assays to estimate HIV incidence

11. Goal was to impact entire community, not just a study cohort Intervention: provided to anyone in the community could participate Outcomes: evaluated among probability sample of 54,326 community residents 18 to 32 years of age (89% response rate) Incident infections: used a multi-assay algorithm (MAA) developed by HPTN Core Lab at Hopkins and the Core Statistical Unit at SCHARP and Charles University (Prague) community Primary outcome = HIV incidence, evaluated at community level

12. HIV incidence estimated using a cross-sectional laboratory-based measure that was extensively validated by the HPTN Central Laboratory No HIV testing done at baseline, since HIV testing was the mechanism by which we anticipated a reduction in HIV incidence (i.e., we could not “contaminate” the communities) HIV was not evaluated based on participation in the intervention – rather, it was measured on a random sample (at the community level) who may or may not have participated in any intervention activities community Primary outcome = HIV incidence, evaluated at community level

13. Prevalence and Estimated Incidence CountryPrevalenceIncidence Population Size South Africa--Soweto14.11.2 152,000 (8 communities) South Africa--Vulindlela30.83.9 67,200 (8 communities) Zimbabwe12.90.9 93,300 (8 communities) Tanzania 5.90.8 54,900 (10 communities) Thailand 1.0<0.1 103,200 (14 communities)

14. Incidence Differences: Intervention vs. Control Communities Subgroup (N of Incident Infections) Effect a 95% CIp-value All participants (464)0.860.73 – 1.020.0822 Women (316) Men (148) 0.88 0.81 0.73 – 1.06 0.57 – 1.15 0.1691 0.1934 Age 18-24 years (271) Age 25-32 years (193) 0.98 0.75 0.80 – 1.22 0.54 – 1.04 0.8554 0.0777 Women, age 18-24 years (201) Women, age 25-32 years (115) 1.00 0.70 0.78 – 1.28 0.54 – 0.90 0.9833 0.0085 Men, age 18-24 years (69) Men, age 25-32 years (79) 0.95 0.78 0.64 – 1.40 0.41 – 1.47 0.6934 0.3914 a Relative risk of infection (CBVCT vs. SVCT); weighted incidence ratio

15. Our findings among older women suggest that their risk may have been reduced due to the risk reduction reported by men, especially those who were found to be HIV-negative Conclusions

16. Our modest reductions in HIV incidence at a population level: –Provides a benchmark –The addition of other components — linkage and retention in care, early ART treatment, male circumcision, pre-exposure prophylaxis — might be successful in achieving greater reductions in HIV incidence in entire communities Conclusions

17. Important to understand what happens in entire communities and not just in study cohorts participating in experiments Bridge from clinical trials proving the concept to intervention studies demonstrating effectiveness Major challenges in prevention science

18. Principal Investigators –Soweto, South Africa: Thomas Coates / Glenda Gray –Tanzania: Michael Sweat / Jessie Mbwambo –Thailand: David Celentano / Suwat Chariyalertsak –Vulindlela, South Africa: Thomas Coates / Linda Richter / Heidi van Rooyen –Zimbabwe: Steve Morin / Alfred Chingono NIMH Cooperative Agreement Project Officer: Chris Gordon Core Lab: Susan Eshleman/Estelle Piwowar-Manning Statistical Core: Michal Kulich, Deborah Donnell Collaborators: NIMH Project Accept (HPTN 043)

19. We thank the communities that partnered with us in conducting this research, and all study participants for their contributions. We also thank study staff and volunteers at all participating institutions for their work and dedication. Acknowledgements