Effect of different methods for estimating persistence and adherence to new glucose-lowering drugs: results of an observational, inception cohort study in Portugal

Purpose: Several methods have been developed for assessing medication-taking behavior; understanding the determinants and variability in estimates obtained is crucial in interpreting results. We estimated persistence and adherence levels to new glucose-lowering drugs (GLDs) in type 2 diabetes mellitus (T2DM) patients using different methods: through the collection of pharmacy records and combining pharmacy records with self-reported data. Methods: We conducted a prospective observational cohort study of T2DM patients initiating a new GLD. Data were collected at baseline through interviews (demographic and clinical data). Follow-up data included pharmacy records (refill dates and medication possession) and telephone questionnaires (self-declared monitored GLD refill in another pharmacy, reasons for drug withdrawal). The cohort was divided into incident and prevalent new users. Persistence and adherence (proportion of days covered) were estimated for patients using pharmacy records exclusively (Method 1) and ≥1 self-declared statement of being persistent (Method 2). Log-rank tests were used to compare Kaplan-Meier curves of time to nonpersistence. Results: A total of 1,328 patients were recruited. When considering Method 1, 38.7% (95% confidence interval [95% CI]: 36.0-41.5) of patients were persistent, whereas combining with self-reported information, this estimate increased to 65.6% (95% CI: 62.9-68.2). Using Method 1, the risk of persistence failure was associated with using an oral GLD, living alone and living in a suburban/urban setting. Three hundred and twenty-seven (24.8%) patients stopped to use the inception GLD. Conclusion: Regardless of the method used, results indicated low levels of persistence and adherence to a new GLD; however, when combining self-reported information, higher estimates were obtained. Considering pharmacy records exclusively, prevalent new users, who were more complex patients in terms of T2DM disease but more likely to be pharmacy-loyal patients, were significantly more adherent than the incident new users. Barriers and reasons leading to GLD withdrawal, namely adverse drug event management, should be addressed, since they represent half of the reasons for treatment switching or discontinuation