Concordance of self-reported hormonal contraceptive use and presence of exogenous hormones in serum among African women

Objectives: Studies that rely on self-report to investigate the relationship between hormonal contraceptive use and HIV acquisition and transmission, as well as other health outcomes, could have compromised results due to misreporting. We determined the frequency of misreported hormonal contraceptive use among African women with and at risk for HIV. Study design: We tested 1102 archived serum samples from 664 African women who had participated in prospective HIV prevention studies. Using a novel high-performance liquid chromatography–mass spectrometry assay, we quantified exogenous hormones for injectables (medroxyprogesterone acetate or norethisterone), oral contraceptives (OC) (levonorgestrel or ethinyl estradiol) and implants (levonorgestrel or etonogestrel) and compared them to self-reported use. Results: Among women reporting hormonal contraceptive use, 258/358 (72%) of samples were fully concordant with self-report, as were 642/744 (86%) of samples from women reporting no hormonal contraceptive use. However, 42/253 (17%) of samples from women reporting injectable use, 41/66 (62%) of samples from self-reported OC users and 3/39 (8%) of samples from self-reported implant users had no quantifiable hormones. Among self-reported nonusers, 102/744 (14%) had ≥1 hormone present. Concordance between self-reported method and exogenous hormones did not differ by HIV status. Conclusion: Among African women with and at risk for HIV, testing of exogenous hormones revealed agreement with self-reported contraceptive use for most women. However, unexpected exogenous hormones were identified among self-reported hormonal contraceptive users and nonusers, and an important fraction of women reporting hormonal contraceptive use had no hormones detected; absence of oral contraceptive hormones could be due, at least in part, to samples taken during the hormone-free interval. Misreporting of hormonal contraceptive use could lead to biased results in observational studies of the relationship between contraceptive use and health outcomes. Implications: Research studies investigating associations between hormonal contraceptive use and HIV should consider validating self-reported use by objective measures; because both overreporting and underreporting of use occur, potential misclassification based on self-report could lead to biased results in directions that cannot be easily predicted.