Objective: To evaluate the feasibility of linking claims-based pattern analysis with medical record review in the assessment of quality of hospital care among Medicare beneficiaries with acute myocardial infarction.
Design: An analysis of risk-adjusted mortality after hospital admission for acute myocardial infarction using the regression model from the Health Care Financing Administration for predicting mortality rates. Hospital records for 300 patients admitted for myocardial infarction were abstracted to evaluate the accuracy of diagnostic coding and the adequacy of claims data-based risk adjustment and to assess process measures of quality care.
Setting: Six Connecticut hospitals in the pilot study of the Medicare Hospital Information Project.
Patients: Medicare beneficiaries 65 years of age or older who were hospitalized with a primary diagnosis of acute myocardial infarction from 1989 to 1991.
Main Outcome Measures: Principal diagnosis code verification rates for acute myocardial infarction; observed mortality rates at 30 and 365 days; 30-day standardized mortality ratios; and utilization rates for thrombolytic agents, aspirin, and β-blockers.
Results: The coding of acute myocardial infarction diagnosis had an overall accuracy of 96%. Little change was noted in relative mortality ratio hospital rank order after the exclusion of 13 patients who did not fulfill criteria for acute myocardial infarction and after additional risk adjustment with Killip class data. Utilization rates for therapies among eligible patients were as follows: aspirin, 73%; β-blockers, 41%; and thrombolytic agents, 43%. The use of thrombolytic agents was associated with a lower 30-day mortality; the use of thrombolytic agents, aspirin, and β-blockers was related to lower mortality rates at 1 year after discharge; and the use of these three therapies was lower in the two hospitals with the highest risk-adjusted mortality.
Conclusions: Medicare principal diagnosis codes for acute myocardial infarction were accurate in the six study hospitals. Therapies that have been endorsed by clinicians in Connecticut were underused in elderly patients. Pattern analysis of Medicare claims data can be useful as a quality-of-care screening tool; however, additional clinical information is required to stimulate quality improvement efforts within hospitals.