Background
HIV and risky alcohol use are intertwined public health issues in sub-Saharan Africa. Research supports the association between alcohol and unprotected sex, but there is limited data using event-level analysis to examine this relationship.

Methods
Using data from Demographic Health Surveys and AIDS Information Surveys collected in 8 sub-Saharan African countries (Kenya, Lesotho, Mozambique, Rwanda, Swaziland, Tanzania, Zambia, and Zimbabwe) drunkenness (reporting male partner or both male and female partner being drunk during last sexual intercourse) at last sex was tested as a predictor of unprotected last sex among the male (n = 24,512) and female (n = 28,229) participants. Partner type, HIV test results, and the other variables were evaluated as effect modifiers of this relationship.

Results
Drunkenness at last sex had a negative effect on the likelihood of condom use among men (AOR 0.84, 95% CI 0.72-0.99) and a marginally significant effect among women (AOR 0.87, 95% CI 0.59-1.02) in Southern Africa. However, for men in Southern Africa, this effect was primarily observed with steady partners. Contrary to predictions, in both Southern and Eastern Africa, for men, drunkenness during sex with casual partners increased the odds of condom use.

Conclusions
These data indicate a need to implement HIV prevention efforts that consider the role of alcohol use in precipitating unprotected sex and how it varies based upon partner type.