The proportion of days covered (PDC) is a widely used indicator of medication adherence, estimating the share of time during which an individual has access to their prescribed therapy within a specified observation period. This study sought to modify the traditional PDC calculation to better reflect realistic prescription refill behaviours when using data obtained from online pharmacy providers. Medication adherence was estimated using three PDC-based algorithms applied to real-world dispensing data from an online pharmacy: the standard method (PDC1) and two alternative approaches (PDC2 and PDC3). These methods differ in their denominator definitions and represent progressively more nuanced assumptions. PDC1 defines the denominator as the total duration between the initial dispensation and a pre-specified end date. PDC2 limits the denominator to the period extending to the end of the last recorded medication supply. PDC3 further refines this approach by excluding pre-defined extended gaps between refills that may plausibly reflect appropriate treatment discontinuation rather than nonadherence. The distributions of the three PDC measures were compared across four different follow-up periods. The analysis included individuals receiving angiotensin-converting enzyme (ACE) inhibitors (n = 65,905), statins (n = 100,362), and/or thyroid hormone therapies (n = 30,637). Among users of ACE inhibitors, the proportion achieving a PDC of at least 0.8 ranged from 50% to 74% using PDC1, 81% to 91% using PDC2, and 86% to 100% using PDC3, with estimates varying by medication class and duration of follow-up. Comparable patterns were observed among individuals prescribed statins and thyroid hormones. The proposed PDC adaptations provide researchers and healthcare professionals with practical tools to evaluate medication adherence and pharmacy service performance using real-world data, particularly in contexts where individuals may obtain medications from multiple suppliers. In such settings, dispensing records from a single provider may contain temporary yet clinically appropriate interruptions in medication supply. Improved identification of adherence-related issues offers opportunities to enhance patient experience, support sustained medication use, improve health outcomes, and reduce medication waste. Further research involving patients and prescribers is needed to assess and validate the assumptions underlying these algorithms.
