Abstract
BACKGROUND: Medication adherence is essential for diabetes control but difficult to measure. Persistence, the continuous refill of medications, is one measurement that could be integrated into practice, but concordance among definitions of persistence is unclear. OBJECTIVE: To compare persistence patterns in the use of oral hypoglycemic agents by diabetic veterans in the first year of treatment, using 2 separate refill metrics. METHODS: Eligible veterans were 18 years or older and had filled at least one prescription for any nondiabetes drug during the 6 months preceding study entry. Study entry was the date that patients first filled a prescription for oral hypoglycemic agents between January 1,2000, and December 31, 2002. Persistence was defined as (1) total days (days' supply/study period) and (2) continuous fill (days' supply/refill interval). RESULTS: Inclusion criteria were met by 110,554 veterans. They had a median age of 63 years, 97% were male, and most had multiple comorbidities. Most patients received only one oral hypoglycemic agent. When total days was used, median persistence at 1 year was 83%, with 45% of subjects demonstrating poor persistence (<80%). When continuous fill was used, median persistence was 0.99, and 24% had poor persistence. Both metrics decreased over time. Continuous fill produced higher persistence estimates and a substantial proportion of overpersistence (>24% of patients). The 2 measures had a kappa agreement statistic of 0.56 (95% CI 0.55 to 0.57). CONCLUSIONS: Although both metrics showed similar trajectories over time, more conservative estimates, ease of calculation, and transparency favor total days over continuous fill. Continuous fill revealed substantial overpersistence, which should be considered separately from good persistence. Explicit definitions of persistence are necessary to enhance comparisons among refill studies and provide context for clinical applications.