Comparing clinical and administrative data for profiling hospitals on postdischarge medication use by patients with acute myocardial infarction.
Am Heart J. 2008 Sep;156(3):595-605
Authors: Austin PC, Donovan L, Yun L, Tu JV
BACKGROUND: Postdischarge medication use is frequently used as a measure of hospital performance, with many reports produced using clinical data obtained from chart reviews. The introduction of a drug benefit program administered by the Centers for Medicare and Medicaid Services presents an opportunity to use administrative data for routine reporting on hospital performance. We determined the concordance between hospital-specific prescribing rates of evidence-based medical therapies obtained from clinical and administrative data in Ontario, Canada. METHODS: This was a retrospective cohort study using data on patients discharged from 102 hospitals in Ontario, Canada with acute myocardial infarction (AMI) between April 1, 1999, and March 31, 2001. We compared hospital-specific rates of discharge prescribing in AMI patients, determined using clinical data obtained using retrospective chart review with hospital-specific rates of prescriptions filled within 30 days of hospital discharge in elderly patients using administrative data. RESULTS: There was a moderate agreement between hospital-specific rates of discharge prescriptions written for AMI patients in clinical data with hospital-specific rates of prescriptions filled using administrative data. Although differences in rates were, on average, small between the 2 data sources, there was moderate variation in the differences between these 2 rates across hospitals. There was very strong agreement between rates of discharge prescribing in all patients and in ideal patients with no contraindications, both determined using clinical data. CONCLUSIONS: Post-AMI discharge prescribing in all patients determined using clinical data is an excellent proxy for prescribing in ideal patients using clinical data. However, there is weaker agreement between administrative and clinical data.
PMID: 18760146 [PubMed - as supplied by publisher]