Hurt and colleagues observed that primary HIV infection (PHI) in men who have sex with men (MSM) was associated with selecting older sex partners. Specifically, MSM with PHI (median age, 24.5 years) tended to choose partners 5 years older (median, 29.8 years) compared with uninfected MSM who were much closer in age (22.5 years; partners, 23.9 years). This result is not surprising, because older MSM are more likely to be infected. In Colorado Springs, for example, rigorously sought, community-wide sexually transmitted disease/HIV surveillance data show that whereas chlamydia is an infection of the late teens and early 20s, and gonorrhea of the early to mid-20s, the average age for HIV seroconversion is late 20s (median, 27 years; mean, 27.8 years), an average age approaching that of older partners of men with PHI in Hurt and colleagues’ study.

What is remarkable is Hurt and colleagues’ and Coburn and Blower’s interpretation of the observed association between partner age disparity and PHI. They mislabel sex with an older man as a risk factor for HIV and correspondingly call for prevention messages to be refocused on this behavior. Sex with an older MSM cannot be a risk factor for HIV infection, although it is a risk marker. The fact that the age of sex partners remained an independent correlate of recent HIV infection in Hurt and colleagues’ multivariate model reflects mismeasurement of exposure to HIV. It appears objective data on partners’ HIV statuses were unavailable for 90% (18 of 20) of men with PHI and 46% (25 of 54) of uninfected men. It is unclear whether all such partners were classified as “serostatus unknown” (and thus regarded as having exposed participants to HIV) or their serostatuses were estimated from participants’ perceptions (the authors did not describe such procedures). In the former scenario, uninfected men’s exposure to HIV may be overestimated. In the latter scenario, men with PHI may have incorrectly perceived infected partners as uninfected or not reported partners during the period when they acquired infection because the authors collected data on the three most recent partners only.

Other characteristics of partners can also serve as markers of HIV risk in MSM such as race and injection drug use. The social contexts in which MSM form sexual partnerships may be even better indicators of HIV risk. Although local data on risk markers might inform ancillary prevention messages, the educational focus in MSM should remain on avoiding the most direct risk factors-anal (especially receptive) intercourse without a condom and sex between serodiscordant men-as their own data clearly show.