CROI 2017 Abstract e-Book

Abstract eBook

Poster and Themed Discussion Abstracts

Methods: We use geo-referenced HIV testing data from the 2010 Malawi Demographic and Health survey; a national representative survey of 7,091 women and 6,497 men aged 15-49. We construct gender-stratified HIV prevalence maps. We also construct gender-stratified, community-level, risk maps that show the proportion of the community who have engaged in high-risk behavior; defined as three or more lifetime sex partners for women and four or more for men. We quantify the spatial correlation between areas of high risk and high prevalence by calculating local bivariate Moran’s and plotting cluster maps. We then conduct a regression at the district level, adjusting for the spatial auto-correlation of errors, to determine the extent to which community-level risk behavior explains local prevalence. Results: Average HIV prevalence is 13% in women, 9% in men. Maps show prevalence varies geographically from 1% to 29% in women, and from 1% to 20% in men. Large-scale spatial patterns are apparent. Notably there is a north-south trend. Maps of risk behavior show the proportion in the high-risk group varies from 0% to 40% in women, and from 16% to 58% in men. A north-south trend is apparent; more discernable in women than in men. Cluster maps show a strong, positive, correlation between community-level risk and HIV prevalence. The regression model shows that variation in the size of the women’s high-risk group, within a community, explains 75% of the variation in the prevalence of HIV in women and 65% of the variation in the prevalence of HIV in men. When high-risk women are more than ~15% of the community, the prevalence in women and men will be above average. Conclusion: Our results support our hypothesis. Geographic variation in community-level high-risk sexual behavior in women generates the large-scale spatial patterns in the HIV epidemic in Malawi, and can explain the north-south trend in increasing HIV prevalence. Our study suggests a simple and plausible explanation for the high degree of geographic variation in the severity of HIV epidemics in sub-Saharan Africa. 851 THE EPIDEMIOLOGY OF HIV IN PEOPLE BORN OUTSIDE THE UNITED STATES, 2010–2014 Results: Among 210,888 children and adults with HIV diagnosed from 2010-2014, 36,324 (17.2%) were estimated to be migrants after imputing COB for 37,670 (17.9%). Migrants accounted for 9,700 (22.2%) and 26,624 (15.9%) of female and male cases, respectively. Among 30,043 migrants with known COB, the most frequent regions of birth (ROB) were Mexico/Central America (MCA, 10,778, 35.9%), the Caribbean (6,327, 21.1%), and Africa (5,336, 17.8%). The most frequent ROB among females were Africa (3,193, 39.5%) and the Caribbean (2,185, 27.0%), while among males, these were MCA (9,322, 42.4%) and the Caribbean (4,142, 18.9%). Migrants comprised 19,756/48,914 (40.4%) of Hispanics/ Latinos; 10,571/93,955 (11.3%) of blacks, 2,552/4,031 (63.3%) of Asians/Pacific Islanders, and 2,300/55,852 (4.1%) of whites. Relative to US-born individuals, migrants were more frequently female (26.7% versus 19.5%); however, males outnumbered females among migrants from all ROB except Africa (59.8% female). Compared with US-born persons, a higher percentage of migrants acquired HIV heterosexually (males, 17.7% vs 9.8%; females, 91.2% vs. 82.9%) and had stage 3 disease (AIDS) at HIV diagnosis (34.7% vs. 25.8%). Over one-third of migrants (11,706, 39.0%) resided in the South. African and Caribbean migrants accounted for 13.9% of the 2014 US foreign-born population, but together represented 38.8% of HIV cases with known COB among migrants. Migrants from South America and the Caribbean were the most geographically concentrated, with 60.8% and 69.2% of cases, respectively, residing in the five metropolitan statistical areas with the largest number of cases from each population. Conclusion: African and Caribbean migrants are disproportionately affected by HIV. Characterizing migrants with HIV is essential for development of effective HIV interventions, particularly in areas with large migrant populations. 852 CHANGES IN HIV RISK FACTORS AMONG MEN WHO HAVE SEX WITH MEN AND WOMEN, 2008–2014 Wade Ivy 1 , Thema Chapple 2 , Gabriela Paz-Bailey 1 , for the NHBS Study Group 1 CDC, Atlanta, GA, USA, 2 Morehouse Coll, Atlanta, GA, USA Background: Since 2008, HIV diagnoses have consistently declined among African American (black) women. Molecular surveillance data have suggested that nearly one-third of HIV infections among heterosexual women are linked to HIV infections attributable to male-to-male sexual contact. We used National HIV Behavioral Surveillance (NHBS) data to evaluate changes in behaviors and access to care among men who have sex with men and women (MSMW). We hypothesize that reductions in risk behaviors and improved HIV treatment coverage among MSMWmay contribute to the declining trends in HIV diagnoses among black women. Methods: We used cross-sectional data from NHBS among MSM recruited using venue-based sampling in 20 US cities in 2008, 2011, and 2014. Data frommen who reported sex with both men and women in the past 12 months were analyzed using GEE to determine changes over time in access to healthcare, HIV testing and antiretroviral (ARV) treatment, and HIV risk behaviors, controlling for city of residence and accounting for clustering around venue recruitment. Models were run separately using each of the variables under investigation as outcomes. Results: Among the 3,339 MSMW in this analysis, the percentage who reported a recent (past 12 months) HIV test (55%, 64%, 65%; p < 0.0001) and current health insurance (51%, 59%, 65%; p =<0.0001) increased significantly from 2008 to 2014. High-risk sexual behaviors increased or remained stable over time. Anal sex without a condomwith a man in the past 12 months increased significantly (44%, 46%, 51%; p = 0.0014), and reporting 3 or more male (57%, 61%, 57%; p = 0.4719), or 3 or more female (39%, 34% 39%; p = 0.6176) sex partners in the past 12 months remained stable. Among HIV-positive MSMW, the percentage who reported being aware of their infection (31%, 40%, 55%; p = <0.0001) increased. Among those HIV-positive aware, the percentage on anti-retroviral treatment (47%, 64%, 80%; p =<0.0001) also increased significantly over time. Conclusion: Although risk behaviors among MSMW increased over time in this analysis, there were increases in HIV testing, insurance coverage, HIV status awareness and being on ARV treatment. While HIV transmission fromMSMW to women only comprise a portion of all HIV infections in women, targeting expansion of HIV treatment and access to care to MSMWmay help further drive down HIV infection rates among women. 853 ASSOCIATION BETWEEN HOUSING STABILITY AND NEW HIV DIAGNOSIS IN SEEK AND TEST STUDIES Bridget Whitney 1 , Bridget Kruszka 1 , Jennifer Lorvick 2 , Gregory M. Lucas 3 , Wendee Wechsberg 4 , Irene Kuo 5 , William E. Cunningham 6 , Asa Clemenzi-Allen 7 , Julie Franks 8 , Heidi M. Crane 1 1 Univ of Washington, Seattle, WA, USA, 2 RTI Intl, Berkeley, CA, USA, 3 The Johns Hopkins Univ, Baltimore, MD, USA, 4 RTI Intl, Rsr Triangle Park, NC, USA, 5 George Washington Univ, Washington, DC, USA, 6 Univ of California Los Angeles, Los Angeles, CA, USA, 7 Univ of California San Francisco, San Francisco, CA, USA, 8 Columbia Univ, New York, NY, USA Background: Housing instability can act as a barrier to timely HIV diagnosis and treatment. Studies have found that homelessness and living in marginal housing are associated with a variety of HIV risk behaviors, including higher rates of substance use and injection drug use. Using four studies from the Seek, Test, Treat, and Retain (STTR) HIV consortium, we examined the association between housing stability and HIV diagnosis. We hypothesized that new diagnosis of HIV would be highest in participants reporting marginal housing and homelessness. Methods: We constructed a harmonized measure of housing stability, categorized as stable, marginally housed, and homeless. Participants were considered marginally housed if they reported living in a temporary facility or an informal settlement at baseline and were considered homeless if they self-reported homelessness or living in shelters. The outcome of interest was HIV diagnosis. Random-effects individual patient data (IPD) meta-analysis models, adjusted for age and gender, was used to assess the relationship between housing stability and HIV diagnosis. Roxanne P. Kerani 1 , Anna S. Johnson 2 , Susan Buskin 3 , Deepa Rao 1 , Matthew R. Golden 1 , Xiaohong Hu 2 , Irene Hall 2 1 Univ of Washington, Seattle, WA, USA, 2 CDC, Atlanta, GA, USA, 3 Pub Hlth–Seattle & King County, Seattle, WA, USA Background: Previous estimates of HIV among migrants to the United States (US) preceded data availability from all states. Methods: We analyzed National HIV Surveillance System data for persons with HIV diagnosed 2010-2014. Country of birth (COB) is included on HIV case report forms, but data reporting is incomplete. If COB was missing, nativity (US, including US dependent areas, or outside the US) was imputed with covariates age, race/ethnicity, HIV transmission category, and stage at diagnosis. Transmission category was imputed for cases with missing risk factor information.

Poster and Themed Discussion Abstracts

CROI 2017 368