Abstract Body

Three randomized control trials (RCTs) demonstrated at least 60% protection of voluntary medical male circumcision (VMMC) against HIV acquisition, and protection against acquisition of ulcerative STIs and HPV. This protection may be offset by risk compensation (RC). Prior RC studies involved men who were part of or follow-on to the RCTs, with men exposed to safe sex messages each time they were seen. This is the first study of RC among men circumcised in a national VMMC program, who were not exposed to safe sex messages each time they were surveyed.

We assessed change in sexual risk behavior over 2 years among circumcised versus uncircumcised men. We enrolled a cohort of 2,379 HIV-negative men aged 18-40 in 2 urban areas in Zimbabwe: 1,196 circumcised near recruitment, 1,183 eligible for VMMC but declined it. Men were surveyed at baseline, 6, 12, and 24-months with extensive sexual behavior measures including number of partners in last 6-, 12-months, and lifetime, sex with sex workers, condom use, concurrent partnerships, and alcohol and drug use. Longitudinal analyses were performed using generalized estimating equations to test for main and interaction effects of study group and time on sexual risk measures, adjusting for baseline differences.

Cohort groups differed at baseline on marital status, income and education, but not on sexual behavior. Study group was significant for only 3 measures: circumcised men were more likely to have had a partner in past 6 and 12 months, and less likely to drink alcohol before sex. Time was significant for 12 measures, with increases in: had a partner in the previous 6 months, 2+ partners in previous 6 and 12 months, had a partner of unknown HIV status, had concurrent partners, concurrent partners and did not use a condom at last sex, 2+ non-spouse partners, suspected partner had other partners, and being drunk in the past 30 days. There was a significant decrease in using a condom at last sex, consistent condom use with spouse, and having an STI Dx/Sx in the previous 6 months. Group by time interaction was significant only for drinking before sex.

Lack of group by time interaction indicates no evidence for RC after VMMC. Of concern is the strong evidence for increased risk behavior among both groups over time. This study coincided with increased availability of ART in Zimbabwe. Possible emphasis on treatment at the expense of behavioral prevention may lead to viewing HIV as a chronic condition, so greater risk behavior.