Algorithm for assessing patients’ adherence to oral hypoglycemic medication

Clinically useful and accurate medication adherence measures are necessary for optimal diabetes mellitus management in routine practice. While electronic monitoring caps (EMCs) have been considered state-of-the-art in measuring medication adherence,1–,4 their use is limited because they are cumbersome and costly (i.e., $75 per EMC),5 may alter the patient’s behavior (i.e., Hawthorne effect),6 and are often impractical for patients who rely on pillboxes or other reminder systems.7–,10 Therefore, more practical and clinically useful adherence measures are desirable for assessing medication adherence among patients with diabetes mellitus in busy practices because they minimize respondent burden. A number of indirect medication adherence measures, such as self-reported assessments and pharmacy refill data, are available.1,11–,13 The validity of patient-reported adherence data has been demonstrated,14 especially when self-completed,11–,15 yet patient report alone might overestimate adherence.1,16 Provider reporting of data, while previously discounted,17–,20 has gained some recognition in predicting patient nonadherence.21 Pharmacy refill data have also been previously validated in measuring adherence,22 are often readily available from claims data, and may present a more realistic picture of long-term adherence (i.e., patient unaware of monitoring)23; yet, pharmacy data primarily capture medication refill patterns.24 Hence, these three data sources (patient report, provider report, and pharmacy data), when applied separately to measure adherence, have their limitations but could be combined to form an adherence algorithm that is relatively easy to apply in routine practice, as patient report, provider report, and pharmacy refill data can be collected much more easily than EMC data.