Objective: Medicare Part D claims indicate medication purchased, but people who are not fully adherent may extend prescription use beyond the interval prescribed. This study assessed concordance between Part D claims and medication possession at a study visit in relation to self-reported medication adherence.
Materials and methods: We matched Part D claims for 6 common medications to medications brought to a study visit in 2011-2013 for the Atherosclerosis Risk in Communities study. The combined data consisted of 3027 medication events (claims, medications possessed, or both) for 2099 Atherosclerosis Risk in Communities study participants. Multinomial logistic regression estimated the association of concordance (visit only, Part D only, or both) with self-reported medication adherence while controlling for sociodemographic characteristics, veteran status, and availability under Generic Drug Discount Programs.
Results: Relative to participants with high adherence, medication events for participants with low adherence were approximately 25 percentage points less likely to match and more likely to be visit only (P<0.001). The results were similar but smaller in magnitude (approximately 2-3 percentage points) for participants with medium adherence. Compared with females, medication events for male veterans were approximately 11 percentage points less likely to match and more likely to be visit only. Events for medications available through Generic Drug Discount Programs were 3 percentage points more likely to be visit only.
Conclusions: Part D claims were substantially less likely to be concordant with medications possessed at study visit for participants with low self-reported adherence. This result supports the construction of adherence proxies such as proportion days covered using Part D claims.