Background: Violence and rape are believed to fuel the HIV epidemic in countries affected by conflict. We compared HIV prevalence in populations directly affected by conflict with that in those not directly affected and in refugees versus the nearest surrounding host communities in sub-Saharan African countries.

Methods: Seven countries affected by conflict (Democratic Republic of Congo, southern Sudan, Rwanda, Uganda, Sierra Leone, Somalia, and Burundi) were chosen since HIV prevalence surveys within the past 5 years had been done and data, including original antenatal-care sentinel surveillance data, were available. We did a systematic and comprehensive literature search using Medline and Embase. Only articles and reports that contained original data for prevalence of HIV infection were included. All survey reports were independently evaluated by two epidemiologists to assess internationally accepted guidelines for HIV sentinel surveillance and population-based surveys. Whenever possible, data from the nearest antenatal care and host country sentinel site of the neighbouring countries were presented. 95% CIs were provided when available.

Findings: Of the 295 articles that met our search criteria, 88 had original prevalence data and 65 had data from the seven selected countries. Data from these countries did not show an increase in prevalence of HIV infection during periods of conflict, irrespective of prevalence when conflict began. Prevalence in urban areas affected by conflict decreased in Burundi, Rwanda, and Uganda at similar rates to urban areas unaffected by conflict in their respective countries. Prevalence in conflict-affected rural areas remained low and fairly stable in these countries. Of the 12 sets of refugee camps, nine had a lower prevalence of HIV infection, two a similar prevalence, and one a higher prevalence than their respective host communities. Despite wide-scale rape in many countries, there are no data to show that rape increased prevalence of HIV infection at the population level.

Interpretation: We have shown that there is a need for mechanisms to provide time-sensitive information on the effect of conflict on incidence of HIV infection, since we found insufficient data to support the assertions that conflict, forced displacement, and wide-scale rape increase prevalence or that refugees spread HIV infection in host communities.