Background: Self-report is considered most suitable to measure medication adherence in routine clinical practice. However, accuracy of self-report as a quantitative measure of adherence is not well documented.
Objective: To assess the accuracy of a self-report measure of adherence (Medication Adherence Report Scale [MARS-5]) for identifying nonadherent users of inhalation medication among patients with chronic obstructive pulmonary disease (COPD), compared with medication refill adherence (MRA) as reference.
Methods: We used baseline data from the Pharmaceutical Care for Patients with COPD (PHARMACOP)-trial (n = 734). Patients with incomplete MARS-5 and/or incomplete pharmacy refill records were excluded (n = 121). Internal consistency of MARS-5 (Crohnbach α) and Spearman rank correlation (ρ) with MRA were calculated. Different thresholds for nonadherence were used to calculate sensitivity, specificity, and positive predictive value (PPV), compared with dichotomized MRA (MRA ≥80% = adherent). A receiver operating characteristic (ROC) curve was plotted to determine the goodness of test.
Results: 613 patients were included in the analysis. The mean adherence score by MARS-5 (range = 5-25) was 23.5 (SD = 2.6); mean adherence by MRA was 83.4% (SD = 23.8%). Internal consistency of MARS-5 was high (α = 0.77). Continuous MARS-5 scores correlated poorly with continuous MRA scores (ρ = 0.10; P = 0.011). When lowering the nonadherence threshold stepwise from 25 to 20, MARS-5 did not reach sufficient sensitivity (53% to 13%), specificity (57% to 94%), and PPV (42% to 57%) to detect nonadherers compared with dichotomized MRA. ROC curve plotting resulted in an area under the curve value of 0.56 (95% CI = 0.521-0.616; P = 0.005).
Conclusion: Self-reported adherence measured by MARS-5 is inaccurate in identifying nonadherence to inhalation medication in patients with COPD.
Keywords: COPD; MARS; MRA; adherence; inhalation medication.