Validation of Amharic versions of medication adherence tools among outpatients with cardiovascular diseases: a cross-sectional study

Abstract Objectives This study aims to validate the Amharic version of the Adherence to Refills and Medication Scale (ARMS-Am) and Morisky Green Levine Scale (MGLS-Am) and compare their psychometric properties among outpatients with cardiovascular diseases (CVD) in Ethiopia. Methods Outpatients with CVD attending an Ethiopian general hospital were recruited for this cross-sectional study. Exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) on split samples, internal consistency (IC), convergent validity, known group validity (KGV), and clinical validity using area under the curve (AUC curve) based on blood pressure (BP) control were evaluated. Key findings There were 613 participants with a mean age of 57.6 ± 14.1; 56.6% were female. A stepwise-EFA and first-order CFA yielded a 9-item ARMS with a 3-factor solution (explaining 50% of total variance). Second-order CFA confirmed ARMS-9Am is unidimensional (overall IC, α = 0.74). There was a strong correlation (ρ = 0.74) between ARMS-9Am and MGLS-Am scores. Pill count had moderate correlation with ARMS-9Am (ρ = −0.42) and MGLS-Am (ρ = −0.32). Both tools had good KGV for discriminating between groups based on control of BP, lipids, and heart failure symptoms. Based on BP control MGLS-Am (AUC = 0.74) had lower discriminatory power than ARMS-9Am (AUC = 0.81). ARMS-9Am had good clinical validity with 87.8% specificity and 70.4% sensitivity for a score of less than 10. Conclusion ARMS-9Am with a cut-off point of < 10 for adherence is a valid tool for measuring medication adherence for people living with CVD and had more robust psychometric properties than MGLS-Am.