Cost-effectiveness of alternative test strategies for the diagnosis of coronary artery disease.

BACKGROUND The appropriate roles for several diagnostic tests for coronary disease are uncertain. OBJECTIVE To evaluate the cost-effectiveness of alternative approaches to diagnosis of coronary disease. DESIGN Meta-analysis of the accuracy of alternative diagnostic tests plus decision analysis to assess the health outcomes and costs of alternative diagnostic strategies for patients at intermediate pretest risk for coronary disease. DATA SOURCES Studies of test accuracy that met inclusion criteria; published information on treatment effectiveness and disease prevalence. TARGET POPULATION Men and women 45, 55, and 65 years of age with a 25% to 75% pretest risk for coronary disease. TIME HORIZON 30 years. PERSPECTIVE Societal. INTERVENTIONS Diagnostic strategies were initial angiography and initial testing with one of five noninvasive tests--exercise treadmill testing, planar thallium imaging, single-photon emission computed tomography (SPECT), stress echocardiography, and positron emission tomography (PET)--followed by coronary angiography if noninvasive test results were positive. Testing was followed by observation, medical treatment, or revascularization. OUTCOME MEASURES Life-years, quality-adjusted life-years (QALYs), costs, and costs per QALY. RESULTS OF BASE-CASE ANALYSIS Life expectancy varied little with the initial diagnostic test; for a 55-year-old man, the best-performing test increased life expectancy by 7 more days than the worst-performing test. More sensitive tests increased QALYs more. Echocardiography improved health outcomes and reduced costs relative to stress testing and planar thallium imaging. The incremental cost-effectiveness ratio was $75,000/QALY for SPECT relative to echocardiography and was greater than $640,000 for PET relative to SPECT. Compared with SPECT, immediate angiography had an incremental cost-effectiveness ratio of $94,000/QALY. RESULTS OF SENSITIVITY ANALYSIS Qualitative findings varied little with age, sex, pretest probability of disease, or the test indeterminancy rate. Results varied most with sensitivity to severe coronary disease. CONCLUSIONS Echocardiography, SPECT, and immediate angiography are cost-effective alternatives to PET and other diagnostic approaches. Test selection should reflect local variation in test accuracy.