Cohort of Circumcised Men Shows No Evidence Of ‘Risk Compensation’

First published online:

During three years of follow-up, a cohort of HIV-negative Ugandan men who underwent circumcision did not appear to compensate for taking this preventive measure by adopting riskier sexual behaviors to a greater degree than uncircumcised men.1 For example, condom use declined in both groups, but the decreases were comparable in magnitude. Similarly, the proportion of both circumcised and uncircumcised men who considered themselves likely to acquire HIV infection grew, but the changes were statistically indistinguishable. The study cohort consisted of men who had participated in a 2003–2006 randomized controlled circumcision trial, were recontacted during posttrial surveillance and were followed up for assessments of "risk compensation" in 2007–2011.

In the original trial, men aged 15–49 who tested negative for HIV, received posttest counseling and provided consent were randomly assigned either to undergo immediate circumcision or to have the procedure two years later. At six-, 12- and 24-month follow-up visits, men received a penile examination, an HIV test, individual-level health education and counseling, and free condoms (which were also available throughout the trial period at community-based locations). In addition, at each visit they completed interviews that covered their socioeconomic characteristics and their sexual and health behaviors. At the 24-month follow-up, uncircumcised men were offered free circumcision services and invited to enroll in the posttrial surveillance study; those opting to undergo the procedure were advised that it is not 100% effective in preventing HIV acquisition. Men who enrolled in the surveillance study returned for two follow-up interviews, roughly 18 months and three years later, and provided information on their sexual behavior during the past year. Researchers used chi-square tests to examine differences between men who chose circumcision and those who did not, and linear model analyses to examine behavior changes during the three years of follow-up.

The surveillance study cohort consisted of 1,297 men who underwent circumcision and 300 who did not. At baseline (before those who opted to be circumcised underwent the procedure), close to half of men in both groups were 25 or younger, and three-fourths had at most a primary school education; the majority were Protestant. Six in 10 men in each group were married; however, in the past year, nine in 10 had had sex, one-third had had two or more partners and one in seven had had multiple nonmarital partners. Alcohol use before sex was common (reported by 61–64% of sexually active men); levels of consistent condom use were low overall (13% in each group), but were higher with respect to nonmarital partners (44–50%). Eight in 10 sexually active men in each group reported no condom use at last sex with any partner; more than two in five of those with nonmarital partners reported nonuse with those partners. Twenty-eight percent of men who chose circumcision and 26% of others considered themselves likely to become infected with HIV. The groups differed by occupation; notably, 13% of men who chose circumcision were students, compared with 7% of others. 

Between posttrial baseline and the final follow-up visit, reported levels of risk-related behaviors changed significantly in both groups. The proportion of participants who were sexually active in the past year (adjusted for loss to follow-up) increased by eight percentage points among circumcised men and by five percentage points among others. Circumcised men became increasingly likely to report just one partner (an increase of five percentage points); uncircumcised men, to report more than one (nine percentage points). Among sexually active men, reports of alcohol use before sex declined in both groups (by 6–8 percentage points), as did reports of consistent condom use with all partners (by 6–7 points) and with nonmarital partners (by 6–14 points among those who had had such partners). Reported nonuse of condoms at last sex became more common overall (increasing by nine percentage points among circumcised men and six points among others) and by men reporting a nonmarital partner (increasing by six and 19 points). The proportion of men considering themselves likely at risk of acquiring HIV infection rose substantially (by 25–28 percentage points) in both groups.

For the most part, the magnitude of changes did not differ significantly between circumcised and uncircumcised men. However, uncircumcised men became significantly more likely than circumcised men to report multiple sex partners in the previous year and nonuse of condoms at last sex with a nonmarital partner.

Results were similar in analyses stratified by age (16–20, 21–25, 26–30, 31–35 and 36–51). Again, almost no differences were evident between circumcised and uncircumcised men in change from baseline to final follow-up. The only exception was that in the youngest age-group, uncircumcised men registered a greater decline than circumcised men in the proportion reporting condom use at last sex with a nonmarital partner.

According to the researchers, their study suggests that male circumcision is not linked to risk compensation. They note that findings elsewhere in the literature have been mixed, but add that their study, unlike others, took place "in a posttrial setting, where the health benefits of circumcision were widely known and men voluntarily chose to accept or decline free circumcision services." At the same time, they acknowledge a number of limitations of their study: Participants represented men who were motivated to enroll in a randomized controlled trial, and the education and counseling they received during its two years were more intensive than what would be provided in routine services; furthermore, most circumcised men in the posttrial cohort had had the procedure at least a year before the final assessment, but risk compensation may occur sooner after the operation. To further explore risk compensation among HIV-negative men who opt for circumcision, the researchers suggest that future studies "focus on the general population of men in programmatic settings and behavior changes within the short term after [the procedure]."
—D. Hollander


1. Kong X et al., Assessment of changes in risk behaviors during 3 years of posttrial follow-up of male circumcision trial participants uncircumcised at trial closure in Rakai, Uganda, American Journal of Epidemiology, 2012, 176(10):875–885.