Validation of the model medication adherence questionnaire to measure the adherence to oral medication among patients with type 2 diabetes mellitus in Sri Lanka

Abstract

Background: The model medication adherence questionnaire (MMA) measures oral medication adherence among patients with type 2 diabetes (DM). The validity of the questionnaire needs to be objectively demonstrated. Methods: A descriptive cross-sectional study was carried out among adult patients with type 2 DM who attended clinics in District General Hospital (DGH) Gampaha. A sample of 150 patients was recruited consecutively to establish the criterion and construct validity. The criterion was the composite index of pill count, recital dosage and regular clinic attendance. Sensitivity and specificity with 95% confidence-intervals (CI) were established. Construct validity: convergent and discriminative, was assessed. Results were analyzed by Spearmen correlation and Man-Whitney-U test with p values. Results: The accuracy of the test was denoted by the area under the curve which is 0.87 (95% CI 0.81–0.93) with p<0.0001. The best cut-off point was 70. Patients who score >70 have good adherence to the medication while ≤70 have average adherence. This has a sensitivity of 92.9% (95% CI 85.5–96.9%), a specificity of 72.5% (95% CI 58–83.7%) with a misclassification rate of 14%. The positive likelihood ratio (LR) is 10.2. High positive LR (>10) represents that adherence is very likely in a person with a score >70. The negative likelihood ratio is 0.3 and is in the intermediate range. It suggests that a score <70 is suggestive but insufficient to rule out the non-adherence. The ideal cutoff to predict the optimum HbA1C of MMA is 67. Conclusions: MMA is a valid questionnaire to measure medication adherence.