CROI 2017 Abstract e-Book

Abstract eBook

Poster and Themed Discussion Abstracts

eligible for ART under the CD4/WHO criteria, newly enrolled HIV-positive women and HIV sero-converters became eligible for immediate treatment. Factors associated with uptake of test and treat August 2014- July 2016 were analyzed using logistic regression. Results: We enrolled 502 ART naïve women, 56% of whomwere enrolled after implementation of test and treat. Their mean age was 30.2 (± 6.3) years and 57% had CD4 >500 cells/ml at baseline. Forty-four percent self-identified as FSWs; the rest reported `other’ or `no job’ of whom 70.2% also engaged in transactional sex. Partner violence was reported by 27.3% of women with 52.3% perpetrated by casual partners. During the study period, 73.5% (n=369) of women initiated ART, of whom 27.4% initiated same day treatment. The median time to ART initiation was 3 months. Women who: reported partner violence (OR 0.57; 95% CI 0.33-1.00); had CD4 >500 cells/ml (OR 0.58; 95% CI 0.35-0.98), has shorter follow up duration at the clinic (OR 0.32; 95% CI 0.17-0.61) and self-identified as FSWs (OR 0.43; 95% CI 0.19-0.99) were less likely to take up test and treat. Women who worked in the hospitality industry (OR 2.63; 95% CI 1.01-6.85) and those reporting other/no job (OR 3.31; 95% CI 1.36-8.07) were more likely to take up test and treat compared to those who self-identified as FSWs. Conclusion: Most FSWs initiate ART within two years but relatively few initiate same day treatment. Women who: self-identify as FSWs, report partner violence, have high CD4 counts and have shorter follow up need targeted interventions to increase ART uptake. 916 VALIDATING SELF-REPORTED USE OF ANTIRETROVIRAL THERAPY IN AGINCOURT, SOUTH AFRICA Jennifer Manne-Goehler 1 , Julia K Rohr 2 , Livia Montana 2 , Guy Harling 2 , Xavier Gomez-Olive 3 , Pascal Geldsetzer 2 , Ryan Wagner 4 , Stephen Tollman 5 , Kathleen Kahn 5 , Till Baernighausen 6 1 Beth Israel Deaconess Med Cntr, Boston, MA, USA, 2 Harvard Univ, Cambridge, MA, USA, 3 Tintswalo Hosp, Acornhoek, South Africa, 4 Univ of the Witwatersrand, Acornhoek, South Africa, 5 Univ of the Witwatersrand, Johannesburg, South Africa, 6 Heidelberg Univ, Heidelberg, Germany Background: Knowledge of antiretroviral therapy (ART) use is essential to healthcare providers and those planning policy. Population-level testing for biological exposure to ART (BE-ART) is often not feasible and self-reported ART use (SR-ART) may be inaccurate or biased. We therefore conducted a validation study of SR-ART based on measured BE-ART in a cohort of older HIV-positive adults in South Africa, in order to understand reporting patterns and correlates of accurate self-reporting. Methods: The Health and Aging in Africa: Longitudinal Studies of an INDEPTH community in South Africa (HAALSI) Study is a cohort of 5,059 adults aged 40+ in rural Mpumalanga, South Africa. In the baseline survey conducted in 2014-2015, HAALSI asked about socio-demographic characteristics, self-reported HIV status and ART use. HIV-positive participants also underwent dried blood spot (DBS) testing for emtricitabine (FTC) and lamivudine (3TC). Either FTC or 3TC or both have been included in any of the first- and second-line regimens ever used in the country. We calculate sensitivity and specificity of self-reported ART use for the total population and stratified by sex and age. We plot the positive predictive value (PPV) and negative predictive value (NPV) by ART prevalence. Finally, we use multivariable logistic regression to assess the association between accurate SR-ART use and socio-demographic characteristics. Results: Baseline HIV prevalence in the HAALSI cohort was 23% (n=1,048 of 4,560 with valid DBS results). Of those who were HIV-infected and tested for ART exposure (n=1,035), 662 (64%) were positive for at least one ART drug: 573 (87%) for FTC, 84 (13%) for 3TC and 5 (1%) for both. Among those who were ART exposed only 450 (68%) reported ever accessing an ART program. The sensitivity of SR-ART use was 68% (95% CI: 64–72%) and the specificity 87% (95% CI: 83-90%). Assuming true ART use was 64%, PPV was 90% (95% CI: 87-93) and NPV 61% (95% CI: 56-65). The PPV and NPV are displayed over the full range of prevalence of self-reported ART use in Figure 1. Accurate SR-ART use was not significantly associated with participants’ age, gender, educational attainment or wealth. Conclusion: Large proportions of ART patients will disclose their ART status; and almost everyone who states they are on ART is de facto on ART. In this high-HIV prevalence community in South Africa, the PPV of SR-ART use is very high and NPV moderate. Thus, SR-ART is a useful way to measure ART use in the absence of biological exposure data.

Poster and Themed Discussion Abstracts

917 REDUCING THE DEMAND FOR VIRAL-LOAD MONITORING IN THE CONTEXT OF TREATMENT FOR ALL Denise Evans 1 , Mouhamed Abdou SalamMbengue 1 , Dorina Onoya 1 , Charles Chasela 1 , Matthew Fox 2 1 Univ of the Witwatersrand, Johannesburg, South Africa, 2 Boston Univ, Boston, MA, USA

Background: From September 2016 South Africa will offer antiretroviral treatment for all, regardless of CD4 count. Eliminating CD4 count thresholds for ART eligibility has a number of challenges, including providing adequate resources for treatment and monitoring for the estimated 164,000 additional patients who are expected to start treatment each year. We aim to develop a clinical predictor score (CPS) to screen patients at 6 months on ART, and identify those who need a viral load test. Since most patients fail treatment after the first 12 months on ART, targeting those who need a viral load test at 6 months and reducing the number of tests required may be a reasonable strategy to cope with the extra numbers starting ART. Methods: Prospective cohort study among HIV-positive ART-naïve adults initiating standard first-line ART at Themba Lethu Clinic, Johannesburg between 02/2012-05/2014. We developed a CPS to identify patients likely to have a viral load ≥400copies/ml at 6 months on ART, and should be targeted for viral load testing. Baseline demographic and clinical characteristics, changes from baseline to 6 months, number of missed visits and self-reported adherence at 6 months were included in the CPS. We obtained a risk score by summing adjusted relative risks and determined the overall discriminative value of the score using the area under the ROC curve (AUC) and dichotomized the score using several cut-offs. We determined sensitivity (Se), specificity (Sp), positive predictive value (PPV) and negative predictive value (NPV) of the cut-offs compared to viral load (gold standard). Results: 296 patients were included (64%males; median age 37.2 years). Factors associated with viral load ≥400copies/ml included male gender, platelet count <150/uL, visual analog scale <95%, missing at least two ARV visits by ≥7 days and change in mean cell volume <14.5fL. The optimal diagnostic accuracy was obtained using a risk score ≥5 (Se 65%; Sp 46.7%). CPS performed better than self-reported adherence measures (Se <60%). CPS for targeted viral load testing at 6 months correctly identified 65% (47/72) of

CROI 2017 397