Agreement and validity of pharmacy data versus self-report for use of osteoporosis medications among chronic glucocorticoid users

Abstract

Pharmacy and linked claims databases are commonly used to determine medication receipt as a measure of quality of care. However, these data sources have not been previously compared with self-reported data for receipt of medications used for glucocorticoid-induced osteoporosis (GIOP). Using databases from a national managed care organization (MCO), we identified 6282 chronic glucocorticoid users (60+ days in 18 months). We compared self-reported current use of alendronate, risedronate, calcitonin, and raloxifene (reference standard) to different intervals of preceding pharmacy data to determine agreement, sensitivity, specificity, and positive and negative predictive values of the pharmacy data. Survey respondents (n = 2363) were mean +/- SD age 53 +/- 14 years old, 70% women, and 78% Caucasian. Agreement between self-reported and pharmacy data ranged from Kappa = 0.64 (95%CI 0.53-0.75) (calcitonin) to 0.80 (0.76-0.84) (alendronate). The positive predictive value of a filled prescription in the pharmacy database in the prior 6 months exceeded 90% compared to the reference standard of self-reported current bisphosphonate use. However, the 6-month interval of pharmacy data failed to capture >25% of self-reported current bisphosphonate users. The optimal interval of pharmacy data to distinguish between current and past bisphosphonate users was 120-180 days. Among chronic glucocorticoid users enrolled in managed care, underreporting of current osteoporosis medication use was uncommon, and agreement between self-report and pharmacy data was high. Use of pharmacy data alone is unlikely to underestimate quality of osteoporosis care, but different intervals of pharmacy data have important implications on the ability to identify current users of osteoporosis medications.