CROI 2017 Abstract e-Book

Abstract eBook

Poster and Themed Discussion Abstracts

CDC, Atlanta, GA, USA Background: The burden of HIV infection, the range of testing, and health outcomes for people living with HIV vary widely across the United States. Understanding the current status of HIV prevention and care outcomes in states informs efforts to achieve local HIV prevention and care goals and the goals of National HIV/AIDS Strategy. Methods: Data from the National HIV Surveillance System on HIV diagnoses among persons aged ≥13 years and their first CD4 test result after diagnosis were used to produce state-level estimates of HIV incidence, prevalence, and percentage of undiagnosed infections during 2008–2014 for each of the 50 states and the District of Columbia. The indicators were derived from estimates of diagnosis delays based on a CD4 depletion model. Estimated annual percentage changes (EAPCs) in incidence, prevalence, and percentage undiagnosed were calculated and considered as significant if p-value is less than 0.05. Results: During 2008–2014, among 36 jurisdictions with numerically stable estimates (>100 HIV diagnoses per year) there were significant increases (EAPCs 1-4%) in HIV prevalence in 23 jurisdictions and significant decreases (EAPCs 3-8%) in percentages of undiagnosed HIV infection in 7 jurisdictions. Estimated annual numbers of HIV infections decreased (EAPCs 2-10%) in 9 jurisdictions. In 2014, HIV prevalence ranged from an estimated 2,359 persons in Nebraska to 145,916 in New York. The estimated annual number of HIV infections ranged from 68 in Nebraska to 5,082 in California. Estimated percentages of undiagnosed HIV infections ranged from10% in Pennsylvania to 19% in Texas. Five jurisdictions (California, Georgia, Florida, New York, and Texas) accounted for 52% of HIV infections and 51% of undiagnosed infections in 2014. In 2014, by region, states located in the South accounted for 45% of persons living with HIV, 51% of HIV infections, and 50% of undiagnosed HIV infections. Conclusion: Estimates of and changes in HIV incidence, prevalence, and undiagnosed HIV infection varied by state and geographic region. Differences in HIV outcomes between states and regions are due to a complex array of social, demographic, economic, and political factors in addition to the capacity of public health, health care systems, and the community to address HIV. Public health officials in the South and states with high percentages of undiagnosed infection should consider tailoring HIV prevention and testing initiatives to their unique environments. 900 ESTIMATING THE UNDIAGNOSED FRACTION: A COMPARISON OF NEW METHODS Jeanette Birnbaum 1 , Jason Carr 2 , Martina Morris 1 1 Univ of Washington, Seattle, WA, USA, 2 Washington State Dept of Hlth, Olympia, USA Background: Estimating the number of undiagnosed HIV infections is critical for measuring the HIV care cascade but methodologically challenging. Estimates of undiagnosed HIV in the US are typically derived from the CDC’s recently updated back-calculation model, which relies on the AIDS incubation distribution and is not recommended for use at the local level. We developed an alternative approach that leverages testing history data and can be applied at the local level (Fellows et al, 2015). In this paper, we seek to increase the precision of the method by incorporating data on CD4 at diagnosis, and to compare with the original and CDC estimates for WA State. Methods: The “testing history” method (TH) relies on inter-test interval data. For newly diagnosed cases with a previous negative test, the last negative test date and date of diagnosis define a window of possible infection. For individuals diagnosed on their first test, a conservative assumption is made, and missing data are treated as missing at random. The TH “Base Case” assigns uniform probability of infection across the window, and uses the standard convolution equation to back-calculate quarterly HIV incidence and undiagnosed cases. But if individuals test after a risky exposure, this assumption would overestimate the time spent undiagnosed. To address this, we modified the TH method to incorporate data on CD4 count at diagnosis (“CD4 Case”). We reassigned 50% of the probability of infection to lie within the median untreated interval from seroconversion to CD4 established in the literature, for those with longer windows. Results: The overall estimates from all three methods were relatively close, ranging from 9.9-11.0%. Incorporating CD4 had a small impact in the expected direction, decreasing the Base Case estimate by 0.6 percentage points (about 6%). The MSM-only estimates revealed large differences between the CDC method and both TH methods. The TH Base Case estimate of the undiagnosed fraction for MSM was 42% lower than the CDC estimate (Table 1). Conclusion: The TH method relies on observed HIV testing patterns, while the CDC method estimates testing rates using a Bayesian approach. This may partially explain the discrepancy with CDC estimates for MSM, and may allow the TH method to more accurately reflect the impact of successful testing interventions. Incorporating CD4 data into the TH method had minimal impact here because the majority of those with high CD4 counts already had a short inter-test interval.

Poster and Themed Discussion Abstracts

901 HIV TESTING MOTIVATIONS OF US MEN WHO HAVE SEX WITH MEN IN A NATIONAL ONLINE SURVEY

David A. Katz 1 , Patrick Sullivan 2 , Robert C. Sineath 2 , Jennie McKenney 2 , Maria Zlotorzynska 2 , Susan Cassels 3 , Joanne Stekler 1 1 Univ of Washington, Seattle, WA, USA, 2 Emory Univ, Atlanta, GA, USA, 3 Univ of California Santa Barbara, Santa Barbara, CA, USA Background: The Centers for Disease Control and Prevention recommends at least annual HIV testing for sexually active men who have sex with men (MSM) and testing every 3-6 months for those at greatest risk. Understanding reasons for seeking testing may help develop and evaluate interventions to increase frequent, regular testing. Methods: US MSM aged 18-39 were recruited from social networking and MSM-focused online venues to participate in a study of online informed consent strategies. Surveys included questions about reasons for HIV testing. Chi-square and rank-sum tests were used to compare characteristics of never, regular, and non-regular testers and result of last test by reason for testing. Results: Of 1413 MSM with HIV testing data (89% of total), 1106 (78%) reported prior HIV testing, of whom 105 (9%) had tested positive. Among HIV-negative ever testers, 51% reported currently testing on a regular schedule, of whom 1% reported testing monthly, 33% quarterly, 38% every 6 months, 22% annually, 3% every 2 years, and 3% on another schedule. The Table compares characteristics of regular, non-regular, and never testers. Regular testers had tested more recently than non-regular testers (median of 3 v. 10 months since last test; p<.0001). Among ever testers, reasons for last test were: routine testing (31%), HIV-positive partner (5%), other potential exposure (28%), new relationship (8%), healthcare provider recommended (7%), HIV/STD symptoms (6%), or other (14%). Among ever testers, 24% reported ever having tested in response to symptoms they thought might be acute HIV infection. The proportion who reported testing positive at last test differed by reason for last test: positive partner (30%), HIV/STD symptoms (15%), other exposure (7%), provider recommended (5%), routine testing (4%), new relationship (1%), and other (5%) [p<0.0001]. HIV-negative MSM thought they should test on a

CROI 2017 391