Sexual seroadaptation: lessons for prevention and sex research from a cohort of HIV-positive men who have sex with men

Abstract

Background: Surveillance data on sexually transmitted infections (STIs) and behavioral characteristics identified in studies of the risk of seroconversion are often used as to track sexual behaviors that spread HIV. However, such analyses can be confounded by "seroadaptation"--the restriction of unprotected anal intercourse (UAI), especially unprotected insertive UAI, to seroconcordant partnerships.

Methods: We utilized sexual network methodology and repeated-measures statistics to test the hypothesis that seroadaptive strategies reduce the risk of HIV transmission despite numerous partnerships and frequent UAI.

Principal findings: In a prospective cohort study of HIV superinfection including 168 HIV-positive men who have sex with men (MSM), we found extensive seroadaptation. UAI was 15.5 times more likely to occur with a positive partner than a negative one (95% confidence interval [CI], 9.1-26.4). Receptive UAI was 4.3 times more likely in seroconcordant partnerships than with negative partners (95% CI, 2.8-6.6), but insertive UAI was 13.6 times more likely with positives (95% CI, 7.2-25.6). Our estimates suggest that seroadaptation reduced HIV transmissions by 98%.

Conclusion: Potentially effective HIV prevention strategies, such as seroadaptation, have evolved in communities of MSM before they have been recognized in research or discussed in the public health forum. Thus, to be informative, studies of HIV risk must be designed to assess seroadaptive behaviors rather than be limited to individual characteristics, unprotected intercourse, and numbers of partners. STI surveillance is not an effective indicator of trends in HIV incidence where there are strong patterns of seroadaptation.

Figure 1. A network “sexiogram” differentiates sexual linkages that may or may not have caused new infections.

The individuals represented by the labels A–B and C–D were couples enrolled in the study. The other nodes are sexual partners described by the enrollees. The partnerships were not necessarily concurrent and included all those reported during the 3 months before A's enrollment and those of D, who enrolled 6 months later. Thus, the partnerships spanned a 9-month interval altogether. In some analyses, all partnerships of these HIV-positive individuals known to practice UAI would be counted as potential transmission linkages. This diagram illustrates the preponderance of low-risk partnerships (solid lines) compared with potential transmission linkages (broken lines).

Figure 2. Sexual behavior by risk of HIV transmission and serostatus of partner.

Sexual partnerships and episodes of intercourse reported by 168 seropositive individuals during the last 3 months reveal strong patterns of seroadaptation: Partnerships involving unprotected intercourse are predominantly seroconcordant, and 88.6% of all unprotected intercourse occurred with seropositive partners.

Figure 3. General estimating equation models predicting UAI in partnerships.

The HIV-positive status of partner was a strong predictor of sexual practices in partnerships and for individual episodes of intercourse; partners of unknown status were not treated very differently from HIV-negative partners. The analysis also revealed strategic positioning by showing that positive participants were several times more likely to practice insertive rather than receptive UAI in serodiscordant partnerships. ^A HIV-negative partners are the reference group in all cases. ^B Designates the sexual position of the HIV-positive participant.

Figure 4. Estimates of potential new infections vary by analytic strategy and detail of data.

Sexual behavior reported by this cohort could potentially contribute to as many as 227 infections in a year (I). Using the partnership rather than the individual as the unit of analysis and fully utilizing most of the exposure data, we estimate that as few as 4.2 new cases might occur (IV). The sexual choices of these HIV-positive individuals reduced potential new infections by 98.1%—a finding that would be missed in some analyses.