CROI 2017 Abstract e-Book

Abstract eBook

Poster and Themed Discussion Abstracts

Poster and Themed Discussion Abstracts

187 ESTIMATING THE EFFECT OF COMPLETENESS OF HIV SEQUENCING DATA ON TRANSMISSION NETWORKS Sharoda Dasgupta 1 , Anne Marie France 1 , Mary-Grace Brandt 2 , M Cheryl B. Ocfemia 1 , Tianchi Zhang 1 , Nivedha Panneer 1 , Angela Hernandez 1 , Alexandra Oster 1 1 CDC, Atlanta, GA, USA, 2 Michigan Dept of Community Hlth, Detroit, MI, USA Background: Molecular HIV surveillance data can be used to understand transmission patterns. Completeness of HIV sequencing data may influence findings. We examined how completeness of reporting HIV sequences affects transmission patterns. Methods: We analyzed National HIV Surveillance System data for the state with highest sequence completeness for diagnoses during 2008–2014 (Michigan, 73%). We took 100 random samples of the dataset with replacement at each pre-determined level of completeness, in increments of 5%, from 5%–70%. We aligned partial HIV-1 pol sequences and conducted pairwise comparisons of sequences to identify pairs with a genetic distance of ≤1.5%. We described transmission network characteristics, such as the number of clusters and links and mixing by race/ethnicity and transmission category. We estimated the median and 5% and 95% quantiles for these characteristics at each completeness level, and presented results for 70% (high) and 5% (low) completeness levels. Results: Of 4,040 sequences reported, 2,458 were from blacks/African Americans (blacks), 210 from Hispanics/Latinos (Hispanics), 2,794 frommen who have sex with men (MSM), and 619 from heterosexual women. The percent of sequences linked to ≥1 other sequence decreased from a median of 54% (53–54%) at high completeness to 18% (13–22%) at low completeness. The number of clusters decreased from a median of 400 (393–406) at high to 18 (13–23) at low completeness, and median number of links was 8267 (8041–8434) at high and 43 (24–64) at low completeness. Assortative pairing among blacks was consistent (median 86–87%) regardless of completeness, with narrower ranges of estimates with higher completeness. Among Hispanics, Hispanic/Hispanic pairings ranged from a median of 17% (15–19%) at high to 0% (0–2%) at low completeness. Pairings of MSM with other MSM remained constant (median 82–83%), with narrower ranges of estimates with higher completeness. The proportion of heterosexual women linked to MSM was 20% (19–21%) at high and 14% (0–100%) at low completeness. Conclusion: Detection of clusters and links was sensitive to data completeness. Inferences about mixing between groups remained robust at lower completeness for larger populations, but was sensitive to change with smaller populations (Hispanics and heterosexual women). The results suggest that, with sufficient numbers of observations, inferences about transmission patterns are informative even with low coverage, though the number and size of clusters may be underestimated.

73

CROI 2017