BACKGROUND: HIV prevalence varies between 0.8 and 20.2% in Kenya's various ethnic groups. The reasons underlying these variations have not been evaluated before.
METHODS: We used data from seven national surveys spanning the period 1989 to 2008 to compare the prevalence of a range of risk factors in Kenya's ethnic groups. Spearman's and linear regression were used to assess the relationship between HIV prevalence and each variable by ethnic group.
RESULTS: The ethnic groups exhibited significant differences in a number of HIV related risk factors. Although the highest HIV prevalence group (the Luo) had the highest rates of HIV testing (Men 2008 survey: 56.8%, 95% CI 51.0-62.5%) and condom usage at last sex (Men 2008∶28.6%, 95% CI 19.6-37.6%), they had the lowest prevalence of circumcision (20.9%, 95% CI 15.9-26.0) the highest prevalence of sex with a non-married, non-cohabiting partner (Men: 40.2%, 95% CI 33.2-47.1%) and pre-marital sex (Men 2008∶73.9%, 95% CI 67.5-80.3%) and the youngest mean age of debut for women (1989 SURVEY: 15.7 years old, 95% CI 15.2-16.2). At a provincial level there was an association between the prevalence of HIV and male concurrency (Spearman's rho = 0.79, P = 0.04). Ethnic groups with higher HIV prevalence were more likely to report condom use (Men 2008 survey: R2 = 0.62, P = 0.01) and having been for HIV testing (Men 2008 survey: R2 = 0.47, P = 0.04).
CONCLUSION: In addition to differences in male circumcision prevalence, variation in sexual behavior may contribute to the large variations in HIV prevalence in Kenya's ethnic groups. To complement the prevention benefits of the medical male circumcision roll-out in several parts of Kenya, interventions to reduce risky sexual behavior should continue to be promoted.