Exploring the Agreement Between Self-Reported Medication Adherence and Pharmacy Refill-Based Measures in Patients with Kidney Disease

AIM: To assess the quantitative and categorical agreement between two methods of measuring medication adherence: pharmacy refill-based medication possession rates and self-reported medication adherence scale. BACKGROUND: Categorisation of adherence metrics using empirical cut-off scores can lead to misclassification, which can be overcome by expressing adherence as a continuous variable. Pharmacy refill-based adherence can be reported as actual rates, but the validity of expressing self-reported medication adherence scores as a continuous variable to reflect adherence is unknown and its quantitative agreement with refill-based adherence rates untested. METHODS: Patients with kidney disease, including dialysis patients, from Illawarra Shoalhaven region of New South Wales, Australia were recruited between January 2015 and June 2016 to this cross-sectional study. Medication adherence was assessed using the self-reported Morisky Medication Adherence Scale (MMAS) and two pharmacy refill-based measures, Medication Possession Ratio (MPR) and Proportion of Days Covered (PDC) for antihypertensives and cardiometabolic drugs. Categorical and quantitative agreement between self-reported adherence and pharmacy refill-based adherence were assessed using tests of trend, analysis of covariance (ANCOVA), Cohen’s kappa and Bland-Altman analysis. RESULTS: We recruited 113 patients. There was a significant declining trend of MPR (p < 0.001) and PDC (<0.001 for antihypertensives, p = 0.004 for cardiometabolic) scores among categories with worsening MMAS adherence. Adjusted ANCOVA showed significant association between self-report and pharmacy refill-based adherence (p < 0.001). Weighted Cohen’s kappa statistics showed fair agreement between the self-report and pharmacy refill-based categories. Bland-Altman’s analysis showed less than 5% of cases were outside the limits of agreement (−0.36 to 0.27) and the bias for MMAS was negative (−0.05 to −0.09), indicating MMAS did not overestimate adherence. CONCLUSION: There is modest agreement between pharmacy refill-based measures and self-report MMAS measures when assessed categorically or quantitatively. Assessing adherence as a continuous variable should be considered to overcome the challenges associated with categorization of adherence based on arbitrary thresholds.