Validity of a prescription claims database to estimate medication adherence in older persons

Background: Prescription claims data have been used to estimate refill medication adherence through calculations of cumulative medication acquisition (CMA) and cumulative medication gap (CMG) values. Few studies have assessed the validity of these calculated rates. Objectives: We sought to assess the validity of CMA and CMG calculated from the Manitoba prescription claims database (DPIN) against pill count medication adherence, targeting overall medications and angiotensin converting enzyme inhibitors (ACEIs). Methods: Using a survey of a convenience sample of subjects recruited through community pharmacies, subjects who were eligible for study (ie, 65 years or older, noninstitutionalized, taking 2 or more "discrete" prescribed medications, including an ACEI, and willing to provide informed consent) were studied. Pill counts were conducted on all prescribed medicines during 3 home inter-views over the course of 4 months. Ten months of DPIN data also were collected on each subject. Results: The concordance between CMA and pill count for overall medications was 411/522 (79%) and for ACEIs was 89/101 (88%) with no systematic: differences (McNemar's P = 0.68 and P = 0.097, respectively). CMG and pill count showed even better concordance of 438/514 (85%) for overall medications and 96/101 (95%) for ACEIs, although systematic differences were noted for overall medications (McNemar's P = 0.0012) but not for ACEIs (McNemar's P = 0.500). Spearman's rank correlations were weak for all comparisons. Conclusions: The high concordance between prescription claims database and pill counts suggested that the rate with which patients refill their medications usually is consistent with the rate they consume them. DPIN is not accurate for nondiscrete dosage forms or medications prescribed for "as-required" use.