Systemwide Provider Performance in a Medicaid Program

OBJECTIVES: This study illustrates how claims data can be applied to examine cost and clinical performance of providers in the Medicaid program. METHODS: The authors conducted a cross-sectional analysis of Medicaid beneficiaries in Maryland with diabetes mellitus, hypertension, and asthma treated on an ambulatory basis by hospital-based outpatient departments, physician office-based providers, and community health centers. The study year was July 1987 to June 1988. The authors defined the cost performance (high, medium, or low) of providers in the management of each of the three chronic illnesses, both before and after casemix adjustment, using a classification system based on ambulatory diagnoses (ambulatory care groups). The authors constructed claims-based clinical performance indicators for each of the three conditions. These included the number of patients admitted to acute-care hospitals for any and specific (diabetes mellitus, hypertension, and asthma) causes, the number of patients without a follow-up visit within 30 days of being discharged from the hospital, and the number of patients with consecutive emergency room visits during the study period. RESULTS: The ambulatory care group casemix classification system explained 23%, 33%, and 36% of the variation in total payments for patients with hypertension, diabetes, and asthma, respectively. Without adjustment for casemix, 35% to 50% of providers would be misclassified regarding their cost performance. Forty-one (19.4%) of 211 providers who treated all three illnesses were in the same cost group for all three illnesses and 95 (43%) of 223 providers who treated two of the three illnesses were in the same cost group for both illnesses. Among office-based physicians, for all three chronic illnesses, high-cost providers had more admissions (P 0.05) difference in the clinical performance indicators between high-cost and low-cost hospital outpatient department providers of primary care for hypertensive and diabetic Medicaid beneficiaries. For the other clinical performance indicators, the results were not consistent across the three illnesses or across the different types of providers. CONCLUSIONS: Without adjustments for casemix, a large number of providers are misclassified regarding to cost performance. In addition, most providers are not equally efficient in managing different chronic illnesses. Provider cost performance is not associated consistently with clinical performance, although severity differences not captured by the casemix adjustment may account for these observations. These measurement methods and relationships between provider performance measures may be useful to state Medicaid programs that seek to contain costs, enhance coordination of care, and improve health.