CROI 2019 Abstract eBook

Abstract eBook

Poster Abstracts

outcomes. STATA v14.2 was used to calculate frequencies and proportions and to test for differences in characteristics and outcomes. Results: During the 9-month period, of the 119,950 clients screened for HIV testing, 66% (79,021) were patients, and 34% (40,929) were non-patients. Overall, 73% (57,873) of patients and 90% (36,892) of non-patients were eligible for testing; testing uptake was >95% in both groups. Among 92,153 clients tested, the median age was 29 years, 57% (52,215) were women, and 40% (36,728) were non-patients. Although more non-patients were men (45% vs. 42%; p-value=<0.001), a greater proportion of patients were younger than 19 years (16% vs. 9%; p-value=<0.001) or older than 49 years (20% vs. 6%; p-value=<0.001). In total, 1.3% (1,185) of clients were HIV positive. Percent yield was higher among non-patients than among patients (1.5% vs. 1.2%; p-value=<0.001), overall and across age categories (Figure 1). Non-patients accounted for 45% (539) of all PLHIV identified, including 57% (117/205) of HIV- positive women aged 15–24 years, 46% (24/52) of HIV-positive men aged 15–24 years, 45% of HIV-positive men (169/377) and 44% of HIV-positive women (188/427) aged 25–49 years. Conclusion: Nearly half of all HIV-positive individuals identified in the OPD were non-patients. Our findings suggest that in the setting of a generalized HIV epidemic, routine provider-initiated HIV testing and counseling of non-patients is a key strategy for timely diagnosis of PLHIV.

viral load <500 copies/mL. In terms of risk during the most recent year in care, 61%were sexually active, 13% had alcohol abuse, and 12% had active drug use. Annual CT/GC positivity rates were stable across calendar years with estimates ranging from 1.9% to 3.4% (p=0.36). Prevalence was inversely associated with age: 2016 CT/GC positivity was 16%/3.9% in ages 18-24 compared to 1.1%/0.7% in age 50+. In every predictive model, despite including a variety of potential STI predictors (including race, region, recent STI, CD4/VL, sex partner characteristics, and substance use), age was the most important variable in predicting CT/GC positivity. In the full machine learning model with good performance (area under the curve [AUC] >0.85), women age <35 years were more likely to have CT/GC and older age (55+) was protective against STI. (see Fig) Conclusion: In a nationally representative sample of US women living with HIV, younger age (<35 years) was the most important predictor of CT/GC infection in a complex machine-learning model. Age-based STI screening among women engaged in HIV care should be reasonable to adopt and simple to implement, although a precise age threshold is yet to be defined.

Poster Abstracts

926 SELF-TESTS FOR AT-HOME PARTNER TESTING ARE ACCEPTABLE & UTILIZED AMONG PREGNANT WOMEN Jillian Pintye 1 , John Kinuthia 2 , Felix Abuna 3 , Nancy M. Ngumbau 2 , Salphine A. Wattoyi 2 , Mary M. Marwa 2 , Anjuli D. Wagner 1 , Julia C. Dettinger 1 , Laurén Gómez 1 , Ben O. Odhiambo 2 , Jared Baeten 1 , Grace John-Stewart 1 , for the PrEP Implementation for Mothers in Antenatal Care (PrIMA) Study 1 University of Washington, Seattle, WA, USA, 2 Kenyatta National Hospital, Nairobi, Kenya, 3 University of Washington in Kenya, Nairobi, Kenya Background: Increasing male partner and couples HIV testing among pregnant women in HIV high-burden settings remains a challenge. Secondary distribution of HIV self-tests within routine antenatal care (ANC) could provide an opportunity to close the gap on partner and couples testing. Methods: In an ongoing cluster RCT (NCT03070600), we offered self-tests for at-home partner HIV self-testing (HIVST) to HIV-uninfected adult women seeking routine ANC services at 10 facilities in Siaya and Homa Bay, Kenya as part of the PrEP Implementation for Mothers in ANC (PrIMA) Study. Women were provided with instructions on how to use self-tests and received at least 2 oral-fluid-based rapid HIV tests (OraQuick Rapid HIV-1/2 Antibody Test, OraSure Technologies). Data on HIVST outcomes were ascertained in-person at one- month follow-up visits. Results: Overall, 1239 pregnant HIV-uninfected women with male partners were offered self-tests for at-home partner HIVST. Median age was 23 years (IQR 20-28) and median gestational age was 24 weeks (IQR 20-28). Most women (75%) were in monogamous marriages; 11%were in polygamous marriages; 14%were unmarried. Overall, 43% reported having a partner of unknown HIV status; 52% had an HIV-uninfected partner and 5% HIV-infected. Among all women, 68% accepted self-tests. Self-test acceptance was 73%, 69%, and 20% among women whose partners’ HIV status was unknown, HIV-uninfected, and HIV-infected, respectively. Among women with partners of unknown HIV status, the most frequently (48%) reported reason for declining self-tests was needing to consult their partner; 10% reported fear of intimate partner violence (IPV). HIVST outcomes were available for 391 (73%) women with partners of unknown HIV status. Among these women, 56% offered self-tests to their male partner; 20% had not seen their male partner since accepting self-tests and 13% feared their partner’s reaction and/or IPV. Among women who offered self-tests to

925 A NEW PREDICTION MODEL FOR CHLAMYDIA AND GONORRHEA SCREENING IN WOMEN WITH HIV

Jodie Dionne-Odom 1 , Dustin M. Long 1 , AndrewWestfall 1 , Barbara Van Der Pol 1 , Michael J. Mugavero 1 , Sonia Napravnik 2 , Heidi M. Crane 3 , Mari Kitahata 3 , Richard D. Moore 4 , Benigno Rodriguez 5 , Maile Y. Karris 6 , Elvin Geng 7 , Kenneth H. Mayer 8 , Jeanne M. Marrazzo 1 1 University of Alabama at Birmingham, Birmingham, AL, USA, 2 University of North Carolina at Chapel Hill, Chapel Hill, NC, USA, 3 University of Washington, Seattle, WA, USA, 4 Johns Hopkins University, Baltimore, MD, USA, 5 Case Western Reserve University, Cleveland, OH, USA, 6 University of California San Diego, San Diego, CA, USA, 7 University of California San Francisco, San Francisco, CA, USA, 8 The Fenway Institute, Boston, MA, USA Background: CDC recommends universal, annual chlamydia (CT) and gonorrhea (GC) screening in sexually active adults with HIV, irrespective of age and gender. The yield of annual CT/GC screening in older women who are engaged in HIV care is low. We applied new epidemiologic techniques using machine learning to develop a more precise prediction model to guide cost- effective STI screening in HIV clinic. Methods: We calculated annual CT/GC testing and positivity rates among US women in HIV care during 2007-2016 as part of the 8 site, CFAR Clinical Network of Integrated Clinical Systems (CNICS) cohort. Data was collected from the electronic medical record and validated surveys for risk behaviors were conducted every six months using patient reported outcomes (PRO). Traditional prediction models using multivariable logistic regression were compared to new prediction techniques for classification using machine learning, random forest algorithms and gradient boosted regression trees, which calculates the importance of each variable in predicting the CT/GC infection outcome and avoids model overfitting. Results: We analyzed data from 5,084 women contributing 158,745 HIV visits from 2007 to 2016. During the most recent year in care, median age was 47 years (IQR 39-55), 62%were Black, 70% had CD4 count >350 and 74% had HIV

CROI 2019 362