Persistence and adherence to oral antidiabetics: a population-based cohort study.

Aims: A population-based cohort study design was used to estimate persistence rate, re-initiation rate after discontinuation, and adherence level among incident users of oral antidiabetics (OADs), and to investigate predictors of non-persistence and non-adherence.

Methods: Incident OAD users were identified using healthcare databases of residents covered by the public drug insurance plan of the Province of Quebec, Canada. Patients initiated OAD therapy between January 2000 and October 2009 and were aged 45-85 years at cohort entry. Persistence rate, re-initiation after discontinuation, and adherence level were assessed over 2 years. Predictors of non-persistence and non-adherence were analyzed using Cox and logistic regression models, respectively.

Results: The cohort included 160,231 incident OAD users at entry. One year after OAD initiation, persistence rate was 51 % and adherence level 67 %. Among those deemed non-persistent, 80.6 % re-initiated OAD therapy within 12 months of discontinuation; a proportion increasing with primary persistence duration. The 1-year persistence rate varied according to OAD classes; being the highest for thiazolidinediones (62 %) and the lowest for alpha-glucosidase inhibitors (30 %). The likelihood for non-persistence was 39-54 % higher when drug copayments were required. Conversely, OAD discontinuation was least likely for patients with schizophrenia [hazard ratio 0.70 (95 % CI 0.67-0.73)], dyslipidemia [0.85 (0.84-0.87)], anticoagulation [0.86 (0.83-0.88)], hypertension [0.87 (0.85-0.88)], and ≥7 medications [0.90 (0.88-0.91)]. Predictors of non-adherence were similar.

Conclusions: Non-persistence and non-adherence to OAD therapy were common, although re-initiation rate was high. OAD classes, drug copayments, comorbidities and co-medications may help identifying those who were more likely to benefit from counseling.