Reliability of partner reports of sexual history in a heterosexual population at a sexually transmitted diseases clinic

Background: In epidemiologic research, information about sexual frequency and condom use is by necessity based on self-reports. This study investigated the reliability of self-reported sexual behavior in 162 heterosexual partnerships. Methods: Subjects were part of a larger study of condom use and sexually transmitted diseases (STD) conducted in two Baltimore STD clinics from 1990 to 1992. Partners were enrolled on the same day and were interviewed separately. Information about sexual activity and condom use was collected using a retrospective calendar for the 30 days before enrollment. Results: Participants were predominantly young, unmarried African-Americans. Based on Spearman's correlation coefficients and kappa statistics, the authors found only fair agreement (K = 0.43; r = 0.51) between partner reports of overall condom use for the 30-day period before the interview. Correlation coefficients ranged from 0.43 for frequency of any sexual activity to 0.56 for number of days on which vaginal intercourse occurred. Conclusions: Partner agreement for condom use and frequency of sexual activity decreased as the recall period increased. Higher partner agreement was observed for questions with definite answers compared to the more open-ended sexual behavior questions. These findings indicate potential reporting bias in self-reports of sexual behavior in a population at high risk for STDs.