The Reliability of Contraceptive Discontinuation Reporting in Burkina Faso, Kenya, and Uganda

Family planning researchers have been studying the discontinuation of contraception-the prevalence and reasons for it-for decades, as it has implications for contraceptive prevalence, total fertility, and unintended fertility. However little is known about the reliability of contraceptive discontinuation reporting: only two studies have examined the reliability of reported discontinuation in low-resource contexts. We use longitudinal and overlapping data from reproductive calendars to test the reliability of women's reporting of contraceptive discontinuation in Burkina Faso, Kenya, and Uganda. We test whether recent discontinuations reported at baseline are reported one year later, and if the same reason for discontinuation is reported. Results reveal moderate reliability at the aggregate level of reporting of the index discontinuation; however, reliability is low when the timing of discontinuation is considered. There is variability across the reliability of reasons for discontinuation; discontinuation due to desired pregnancy is reported reliably while other reasons are less reliably reported. Our findings have important implications for the field, particularly in how the data are used and interpreted; we urge caution around event-level analyses of contraceptive discontinuation and reasons for discontinuation, given low reliability.