Reliability of Reported Sexual Partnership Dates and Measures of Concurrency

Abstract

The timing of sexual partnerships constrains the potential flow of sexually transmitted infections (STIs). It is crucial to determine the quality of reported data on sexual partnership timing for both scientific and practical purposes. Temporal data on partnerships are used in contact tracing for STIs to identify sex partners at risk, prioritize partners for intervention, and classify source and spread cases.We observed fairly good test-retest (consistency of repeated reports) and interpartner (agreement between partners in a dyad) reliability of reported partnership timing in six contact tracing data sets and one study of sexual networks. Interviewees reported dates of last sex more reliably than dates of first sex. Heaping in the differences between dates reported in repeated interviews corresponded to the temporal units used in reporting dates (week, month, year). We found modest forward telescoping (reporting date of first sex in the second interview as having occurred more recently than reported in the first interview) in one data set involving primarily young adults whose intervals between interviews tended to be several months or more. Recent events, shorter intervals between interviews, and having the same interviewer across interviews were independently associated with higher test-retest reliability. Also, females were more perfectly reliable (identical reports in repeated interviews) than males, but not more reliable overall. Reliability of reported partnership duration, ordering, and concurrency was moderately high to high.