PS2-29: Comparison of Self-reported Oral Contraceptive Use and Automated Prescription Fills

Abstract

Background/Aims Self-reported use of oral contraceptives (OCs) may be subject to recall bias. Previous reports comparing self-report OC use and computerized pharmacy data have focused on current use or longer-term use in younger women and have found (adjusted) kappa statistics of 79-85%. The objective of the current analysis was to evaluate the reliability of self-reported OC use obtained from a sample of peri- and early post-menopausal women. Methods Participants were 45-59 year-old women eligible for an ongoing population-based case-control study assessing the association between OC use and incident fractures around the menopausal transition. Cases were all women enrolled in Group Health with an ICD9 osteoporotic fracture code in 2008-2009; age-matched non-fracture controls were randomly selected. Eligible cases and controls (n=535) who reported at interview that they always/usually filled prescriptions at GH pharmacies and agreed to record review were included. Respondents who reported ever using OCs before the reference date (fracture date for cases; randomly assigned dates based on distribution of case dates for controls) were asked for each episode of use their age and length of use. A life events- calendar tool was available to interviewers. Computerized pharmacy information was obtained on OC fills back to 1977. OC use was based on GH pharmacy and First Data Bank OC class key ingredients [e.g. ethinyl estradiol levonorgestrel] and name. OC use was evaluated using >1 fill or >2 fills within a 1-year period. Ever use use since ages 35 38 and 40 and 5 years before reference date were examined. Women were required to be enrolled for the designated time period except for the ever-use category. We calculated kappa (K) the chance-corrected measure of agreement and the prevalence-adjusted bias-adjusted kappa (PABAK). Results The agreement between self-reported OC use and OC fills was highest for more recent use (PABAK=92% for OC use within 5 years of reference date and 55% for use at ages >38). Conclusion In women around the menopausal transition agreement between self-reported OC use and computerized OC prescription fills was moderate for use after age 38 and excellent for OC use within 5 years of the reference date.