Abstract
We explored the utility of routine HIV testing data from clinical services for estimating HIV prevalence. A systematic review identified 28 eligible publications, covering concentrated epidemics (16 of 28) and generalized epidemics (12). Of the 16 papers from concentrated epidemics, five presented estimates by risk group and four by testing history with a median HIV prevalence of 1.8% in first-time testers compared with 3% in repeat testers. Two reports from generalized epidemics restricted estimates to asymptomatic clients and three included breakdowns by reason-for-test, with the median HIV prevalence higher in symptomatic clients (62%) than others (24%). Two papers from generalized epidemics showed prevalence estimates based on routine HIV testing data were slightly higher than estimates derived from other surveillance methods, but did not restrict estimates to asymptomatic patients. We conclude that routine HIV testing data may be a supplementary data source for HIV surveillance provided careful analyses are conducted.