Reliability of a modified medication appropriateness index in community pharmacies

BACKGROUND: The medication appropriateness index (MAI) has demonstrated reliability in selected outpatient clinics where medical data were easily accessible from medical charts. However, its use in the community setting where patient data may be limited has not been examined. OBJECTIVE: To evaluate the usefulness of a modified MAI for use in the community pharmacy setting by testing interrater reliability using 3 different rating schemes. METHODS: Two raters evaluated 160 medications for 32 elderly ambulatory patients. Patient information was acquired using community pharmacist-collected medication histories. A summated MAI score, percent agreement, K, positive agreement, negative agreement, and intraclass correlation coefficient were calculated for each criterion using 3 scoring schemes. A paired samples t-test (95% CI) was used to test interrater reliability. RESULTS: The kappa statistics were >0.75 for indication and effectiveness, but good (0.41-0.66) for the remaining criteria using the Hanlon scoring scheme. The intraclass coefficients (0.82, 0.86, 0.87) and overall kappa (0.65, 0.66, 0.61) were similar for the 3 schemes. CONCLUSIONS: This study suggests that the modified MAI has the potential to detect medication appropriateness and inappropriateness in the community pharmacy setting; however, it is not without limitations. Because the MAI has the most clinimetric and psychometric data available, the instrument should be studied further to increase its reliability and generalizability.