Reliability of a modified medication appropriateness index in primary care

Objective: To evaluate the inter-group and intra-group reliability of a modified Medication Appropriateness Index (MAI) for use in primary care. Methods: Elderly (>65 years) polypharmacy (>= 5 drugs) patients in Copenhagen County participated in the study. Information concerning their medical history and information regarding each drug taken by them was provided by their own general practitioners. A MAI was scored by two groups of evaluators for every drug. To evaluate inter-group agreement, 211 drugs taken by 30 patients were rated according to the ten criteria making up the MAI. Both evaluator groups provided summaries of comments on the medication of each patient. Intra-group agreement was determined from MAI ratings performed twice at two different times on 86 drugs taken by ten patients. Agreement and chance-adjusted agreement were determined, the latter through kappa statistics. The proportion of positive (ppos) and negative (pneg) agreement was also determined. Results: The overall chance-adjusted inter-group agreement (kappa) was moderate. The agreement was good on the criteria practical directions and drug-disease interaction, moderate on the criteria dosage and duration, fair on the criteria indication, effectiveness, duplication and expense, and poor on the criterion drug-drug interaction. The overall chance-adjusted intra-group agreement was good for all criteria and very good for the criteria indication and practical directions. Conclusion: The MAI is used to quantify appropriate and inappropriate prescribing and changes in prescribing quality in intervention studies. However, caution should be used when comparing results across different settings and evaluators. Our study suggests that the index should only be used in intervention studies if the same group rates the appropriateness pre- and post-intervention.