Background Agreement between sexual behaviour recall measures among gay, bisexual and other men who have sex with men (GBM) in Vancouver, Canada was examined.
Methods: Study participants were sexually active GBM aged ≥16 years recruited via respondent-driving sampling (RDS). Participants completed a computer-assisted self-interview survey, including individual-level period prevalence (last 6 months) and sexual event-level (last sex with each of the five most recent partners) measures. RDS-weighted kappa statistics assessed the agreement between these types of data across five different sexual risk outcomes, stratified by self-identified HIV status and other demographic factors (age, education, race/ethnicity).
Results: Of 719 participants, 195 (RDS-weighted 23.4%) were HIV-positive. For HIV-negative GBM (n=524, RDS-weighted 76.6%), there were moderate agreements between period prevalence and event-level data for any anal intercourse (AI), any condomless AI and any fisting, but weak agreement for any discordant/unknown AI and any sex toy use. For HIV-positive GBM, there was moderate agreement for any AI, any condomless AI, any discordant/unknown AI and any fisting; there was weak agreement for any sex toy use. Agreement between measurement types was generally higher for GBM who were living with HIV, who were older and who completed secondary school; there was little difference in agreement levels by race/ethnicity.
Conclusions: We observed moderate agreement between sexual behaviour recall through event-level and period prevalence questions. Each method had differential advantages and ideal circumstances for use, which should be guided by one's research question and outcome measure of interest.