Background: Concurrency data are important for understanding STI transmission in populations. However, as a socially-sensitive behaviour, some studies measure concurrency by examining whether the dates of recent partnerships overlap, rather than asking a direct question. This paper compares these 2 methods in terms of: (i) the prevalence of concurrency; (ii) the extent of discrepancy and associated demographic and behavioural factors; and (iii) their association with reporting STI diagnosis/es.
Methods: Complex survey analyses of probability sample data collected from 15,162 men and women aged 16-74y resident in Britain 2010-2012 using computer-assisted personal-interviewing, including computer-assisted self-interview for the more sensitive questions, including the dates of participants’ (max.) 3 most recent partners (MRPs), a direct question about concurrency, and STI diagnosis/es (timeframe for these variables: past 5 years).
Results: Among participants reporting 2+ partners in the past 5 years (2,828 men; 3,202 women), the proportion of men reporting concurrency at the direct question was similar to that estimated from MRP dates (40.6% vs. 40.0%); but differed among women: 31.0% vs. 38.2%, respectively. The direct question resulted in less missing data than using the MRP dates (0.0% vs. 8.7% among men; 0.1% vs. 7.9% among women). A discrepancy between the measures existed for one-third of men and women, and increased with increasing partner numbers but few other factors. In logistic regression, the direct question was more strongly associated with reporting STI diagnosis/es than using MRP dates: odds ratios adjusted for age and partner numbers: 2.20 (95% CI:1.54-3.15) vs. 1.63 (95%CI:1.17-2.28) among men; 2.26 (95%CI:1.71-2.98) vs. 1.60 (95%CI:1.21-2.12) among women.
Conclusions: Similar levels of concurrency were observed for the two measures, at least for men, although within-individual discrepancy between the measures was common. Behavioural surveys should use a direct question as this requires fewer assumptions, yields fewer missing data, and better predicts reported STI diagnosis/es.