CROI 2017 Abstract e-Book

Abstract eBook

Poster and Themed Discussion Abstracts

Poster and Themed Discussion Abstracts

857 VALIDATION OF A RISK SCORE FOR HIV ACQUISITION IN YOUNG AFRICAN WOMEN WITH FACTS 001

Emma K. Burgess 1 , Sinead Delany-Moretlwe 2 , Pedro Pisa 2 , Khatija Ahmed 3 , Sydney Sibiya 4 , Cynthia Gama 5 , Busiswe Nkala 6 , Gustavo Doncel 7 , Glenda Gray 6 , Helen Rees 2 1 Global Hlth Fellows Internship at USAID, Arlington, VA, USA, 2 Wits Reproductive Hlth and HIV Inst, Johannesburg, South Africa, 3 Setshaba Rsr Cntr, Pretoria, South Africa, 4 Qhakaza Mbokodo Rsr Clinic, Ladysmith, South Africa, 5 MatCH, Durban, South Africa, 6 Perinatal HIV Rsr Unit, Soweto, South Africa, 7 CONRAD, Arlington, VA, USA Background: Adolescent Girls and Young Women (AGYW) in South Africa are at high risk of HIV, and understanding which factors may be predictive of their level of risk could improve the efficiency of recruitment into HIV research and targeting AGYW for the prevention programs they urgently need. Using baseline data from the FACTS 001 1% tenofovir gel trial, we sought to externally validate a risk scoring tool to predict HIV-1 acquisition in African women that was derived by Balkus et al. with the VOICE cohort. Methods: We compared FACTS data collected fromwomen across 9 South African sites in 2011 to 2014, to the findings in the VOICE cohort. The full score included 7 items with a maximum total of 11: age, living with a partner, partner provides financial support, partner exclusivity, alcohol use, detection of curable STIs, and HSV-2 serostatus. Scores of ≥5 were associated with HIV incidence of >5/100 PY and identified 91% of infections among only 64% of the women. The score was applied to FACTS participants and evaluated using sensitivity, specificity, and area under the curve (AUC). The 7 risk score factors were also assessed in a multivariate Cox proportional hazards model. Results: Among 2,059 enrolled FACTS participants, 1,115 (54%) had complete data for all 7 indicators. Of these, 81 acquired HIV infections over 1876 person years of follow-up (incidence=4.32/100 PY). Cutting off scores at ≥5 had a sensitivity of 84% and specificity of 23%, therefore, 77% of the women were needed to identify 84% of the infections. Scores of ≥5 were loosely associated with incidences of >3/100 PY. The AUC was 0.56, indicating poor discriminative accuracy. In the full model only 3/7 risk factors had a significant association with HIV: not receiving financial support from a partner (aHR=0.50, CI 0.26, 0.99), not living with one’s partner (aHR=2.48, CI 1.13, 5.45), and being HSV-2 positive (aHR=1.56, CI 1.11, 2.25). Conclusion: Our external validation in FACTS found that the score could not differentiate well between women at lower and higher risk of HIV acquisition as incidence was comparable across risk score categories. The null and opposing risk associations could be attributable to FACTS being a younger cohort, hence the score’s poor discriminative ability in FACTS compared to VOICE. This analysis warrants further investigation for framing a risk score for South African AGYW in high incidence settings, in particular that can be used to identify potential users who would benefit from new HIV prevention options. 858 SELF-ASSESSED HIV RISK 3 YEARS AFTER HIV “TEST AND TREAT” IMPLEMENTED IN SEARCH James Ayieko 1 , Dalsone Kwariisima 2 , Catherine A. Koss 3 , Craig Cohen 3 , Elizabeth A. Bukusi 1 , Tamara D. Clark 3 , Edwin D Charlebois 3 , Moses R. Kamya 4 , Maya L. Petersen 5 , Diane V. Havlir 3 1 Kenya Med Rsr Inst, Nairobi, Kenya, 2 Makerere Univ Joint AIDS Prog, Kampala, Uganda, 3 Univ of California San Francisco, San Francisco, CA, 4 Makerere Univ, Kampala, Uganda, 5 Univ of California Berkeley, Berkeley, CA Background: Data are limited regarding self-perception of HIV acquisition risk where HIV test and treat is ongoing or on how self-perception correlates with true risk. We compared self-assessed HIV risk to an empirically-based risk score for HIV seroconversions to inform PrEP implementation in SEARCH. Methods: Seroconversion data from the SEARCH intervention (HIV test and treat) armwere analyzed using machine-learning to build an HIV risk score that maximized observed seroconversion coverage under a minimized number of persons to be provided PrEP. The risk score included age, sex, marital status, polygamy, education, circumcision, occupation, and alcohol use. After 3 years of HIV test and treat, we measured self-assessed HIV risk (SAR) through structured survey among HIV-uninfected adults in the rural Kenyan community of Nyatoto. We evaluated gender-specific predictors of SAR and its relation to our empirically-derived risk score. Results: Baseline community HIV prevalence was 15%. Overall, 15% of HIV-uninfected adults (N=4159) perceived themselves at risk of HIV. Among men and women, predictors of SAR included age 25-35 yrs (OR 2.3,95%CI:1.6,3.1; OR 1.6, 95%CI:1.2,2.2, relative to 15-25 yrs), employment in the high-risk informal (non-government regulated) sector (OR 3.0,95%CI: 1.9,4.9; OR 2.4,95%CI:1.1,5.3, relative to formal sector) or low-risk informal sector (OR 1.4; 95%CI:1.1,1.9; OR 1.4,95%CI:1.1,1.9, relative to formal sector), polygamous marriage (OR 1.9,95%CI 1.2,1.9; OR 1.8,95%CI:1.4,2.5), and greater than primary education. Among women, SAR was higher among those with an HIV+ spouse (OR 8.5,95%CI:2.0,35.7). Among men, alcohol use predicted SAR (OR 2.6,95%CI:1.5,4.5 for >7 days drinking/month, relative to not drinking). Among both men and women, HIV SAR was higher among those identified as high-risk using our empirically-derived score (OR 1.7,95%CI:1.3,2.4; OR 1.7,95%CI:1.3,2.2). Among all (N=858) persons identified as empirically high-risk, 21% perceived themselves at risk. Conclusion: After 3 years of HIV test and treat,15% of HIV-uninfected subjects perceived themselves at risk of HIV acquisition. Although an empirically-derived risk score predicted self-assessed risk, only 21% of individuals empirically targeted for PrEP perceived themselves at high-risk for HIV. These data suggest that optimizing the uptake of empirically targeted PrEP will require new community sensitization approaches to help persons recognize actual HIV risk. 859 HIV RISK FACTORS AND RISK PERCEPTION AMONG ADOLESCENT GIRLS AND YOUNG WOMEN IN MALAWI Joan T. Price 1 , Nora E. Rosenberg 1 , Bertha Maseko 2 , Savvy Brar 2 , Twambilile Phanga 2 , Audrey Pettifor 1 , Linda-Gail Bekker 3 1 Univ of North Carolina at Chapel Hill, Chapel Hill, NC, USA, 2 Univ of North Carolina Proj–Malawi, Lilongwe, Malawi, 3 Univ of Cape Town, Cape Town, South Africa Background: The high HIV incidence among adolescent girls and young women in sub-Saharan Africa has been associated with a range of individual, social, and structural risk factors. Perceived risk of HIV is a key element in the uptake of prevention programming; an understanding of the association between HIV risk factors and perceived risk in a vulnerable group can inform intervention planning and targeting.

CROI 2017 371