Measuring population transmission risk for HIV: an alternative metric of exposure risk in men who have sex with men (MSM) in the US

Abstract

Background: Various metrics for HIV burden and treatment success [e.g. HIV prevalence, community viral load (CVL), population viral load (PVL), percent of HIV-positive persons with undetectable viral load] have important public health limitations for understanding disparities.

Methods and findings: Using data from an ongoing HIV incidence cohort of black and white men who have sex with men (MSM), we propose a new metric to measure the prevalence of those at risk of transmitting HIV and illustrate its value. MSM with plasma VL>400 copies/mL were defined as having 'transmission risk'. We calculated HIV prevalence, CVL, PVL, percent of HIV-positive with undetectable viral loads, and prevalence of plasma VL>400 copies/ml (%VL400) for black and white MSM. We used Monte Carlo simulation incorporating data on sexual mixing by race to estimate exposure of black and white HIV-negative MSM to a partner with transmission risk via unprotected anal intercourse (UAI). Of 709 MSM recruited, 42% (168/399) black and 14% (44/310) white MSM tested HIV-positive (p<.0001). No significant differences were seen in CVL, PVL, or percent of HIV positive with undetectable viral loads. The %VL400 was 25% (98/393) for black vs. 8% (25/310) for white MSM (p<.0001). Black MSM with 2 UAI partners were estimated to have 40% probability (95% CI: 35%, 45%) of having ≥1 UAI partner with transmission risk vs. 20% for white MSM (CI: 15%, 24%).

Discussion: Despite similarities in other metrics, black MSM in our cohort are three times as likely as white MSM to have HIV transmission risk. With comparable risk behaviors, HIV-negative black MSM have a substantially higher likelihood of encountering a UAI partner at risk of transmitting HIV. Our results support increasing HIV testing, linkage to care, and antiretroviral treatment of HIV-positive MSM to reduce prevalence of those with transmission risk, particularly for black MSM.

Figure 1. HIV exposure model equation: The estimated the probability of having ≥1 UAI partner with HIV transmission risk (i.e. HIV viral load >400 copies/ml).

In this equation, k = number of UAI partners with transmission risk (k≤m); m = number of male UAI partners (m = 0 to ∞); r = black or white non-Hispanic race/ethnicity (r = 1, 2); i = race/ethnicity, same as r, and including a third level for ‘other’ non-black or non-white race/ethnicity (i = 1,2,3); T_r,T_i = race-specific %VL400 among male UAI partners; E_r = proportion of participants reporting exclusively same-race UAI partners, among race r; A_{r, i} = the proportion of UAI partners who were race/ethnicity i, among participants of race r reporting inter-racial UAI partners.

Figure 2. The continuum of HIV care for black and white MSM in the InvolveMENt study.

Complete data on linkage to and retention in HIV care are not available for this study.

Figure 3. Viral load distribution functions among black and white MSM aware of their HIV infection (Community Viral Load; panel a), among all HIV positive MSM inclusive of those unaware of HIV infection (Population Viral Load; panel B), and among all MSM inclusive of HIV negative and positive (panel c) in the InvolveMENt study demonstrating similarities in CVL and PVL due to underlying similarities in the continuum of HIV care.

Large differences are evident in the distribution for black and white MSM of MSM at risk of transmitting HIV as this metric accounts for differences in HIV prevalence. Panel D presents the Kolmogorov-Smirnov test for differences in the distributions between black and white MSM for all three distributions, the Wilcoxon Rank Sums test for differences in the median CVL and PVL, and the chi-square test for differences in the prevalence of viral load>400 copies/ml for all three distributions.

Figure 4. HIV exposure model: The estimated probability of having ≥1 UAI partner with HIV transmission risk (i.e. HIV viral load >400 copies/ml) for black (circles) and white (squares) MSM.

Shaded bands represent 95% confidence intervals. This figure demonstrates the differences in HIV exposure based on number of UAI partners between black and white MSM in the InvolveMENt study. This model does is not intended to represent true HIV risk of HIV transmission for a given sexual encounter as necessary covariates such as sexual frequency, sexual practices (i.e. insertive vs. receptive partner), sexually transmitted infections, partner viral load, etc. are not accounted for.