Reliability of drug utilization evaluation as an assessment of medication appropriateness.

OBJECTIVE: To test the reliability of drug utilization evaluation (DUE) applied to medications commonly used by the ambulatory elderly. METHODS: A DUE model was developed for four domains: (1) justification for use, (2) critical process indicators, (3) complications, and (4) clinical outcomes. DUE criteria specific to use in the elderly were developed for angiotensin-converting enzyme (ACE) inhibitors and histamine2 (H2)-antagonists, and consensus was reached by an external expert panel. After pilot testing, two clinical pharmacists independently evaluated these medications, applying the DUE criteria and rating each item as appropriate or inappropriate. Interrater and intrarater reliability was assessed by using kappa statistics. RESULTS: In a sample of 208 ambulatory elderly veterans, 42 (20.2%) were taking an ACE inhibitor and 56 (26.9%) an H2-antagonist. The interrater agreement for individual domains, represented by kappa statistics, were 0.10-0.58 and 0-0.83 for ACE inhibitors and H2-antagonists, respectively. The kappa statistic for overall agreement, which considered ratings from all criteria across all domains, was 0.24 for ACE inhibitors and 0.18 for H2-antagonists. Intrarater reliability was assessed 3 months later, and kappa statistics were 0.61-0.65 (0.49 overall) and 0-0.96 (0.81 overall) for ACE inhibitors and H2-antagonists, respectively. CONCLUSIONS: Intrarater reliability for DUE was good to excellent. However, interrater reliability exhibited only marginal reproducibility, particularly where evaluators were required to use subjective judgement (i.e., complications, clinical outcomes). DUE may not be a suitable standard for assessing medication appropriateness in ambulatory elderly patients.

Duke Scholars

Author Kenneth Edwin Schmader Medicine, Geriatrics and Palliative Care

Author Gregory P. Samsa Biostatistics & Bioinformatics, Division of Biostatistics