Abstract
STUDY OBJECTIVE: To compare the association between different measures of condom use and prevalent HIV infection. DESIGN: Randomised cross sectional study to assess association between HIV infection and different measures of self reported condom use. Female sex workers were randomised to one of five different face to face questionnaires on condom use. Three questionnaires used always to never scales to measure use but differed in the reference period for use; a fourth asked about use in the last 10 coital acts; and the fifth was a retrospective log of coital acts in the past two weeks. Use was assessed with new clients, repeat clients, and non-clients. SETTING: Yaoundé and Douala, Cameroon. PARTICIPANTS: 2266 female sex workers. MAIN RESULTS: The association between condom use and prevalent HIV infection varied for different measures of condom use. None of the five level measures showed a dose response protective effect of condom use. Measures aimed at reducing recall bias (measures based on the past 10 coital acts or a coital log) showed little or no association with prevalent infection. Measures based on the past month or six months had a stronger association with prevalent infection. Regardless of the type of measure or reference period, the strongest association between use and infection was for use with partners who were not clients. CONCLUSION: These findings underscore challenges described by others of measuring condom use and interpreting the association between use and prevalent infection.