Medicare Advantage Enrollees Are Admitted to Hospitals of Different Quality Than Traditional Medicare Enrollees
Discussant: Laurence Baker
Using Medicare Provider Analysis and Review (MedPAR) data, we first compared quality of care, as measured by CMS star-rating, readmissions rates, and mortality rates, in hospitals that serve MA and traditional Medicare enrollees using linear probability models adjusting for age, gender, race, dual enrollment with Medicaid, ICU use, and ZIP Code fixed effects. We stratified MA enrollees by the star rating of their plan (less than 4 stars, 4 or more stars) and whether the hospitalization was an elective or emergency admission.
Second, in order to account for factors that simultaneously affect the hospital that an enrollee is admitted to, we estimate a McFadden random utility maximization choice model of hospital choice. We create a choice set of every hospital within 50 miles of the patient’s ZIP Code, the 15 closest hospitals, and every hospital that any patient from the same zip code was admitted to. We then estimate the association of different factors with selection of a hospital using a conditional logit model. We conducted stratified analyses according to elective vs emergency hospitalization, dual eligibility for Medicaid, rural vs urban residence, the presence of 2+ chronic conditions, and for the 3 most common discharge diagnoses.
MA enrollees in lower quality plans (<4 stars) were 2.1 percentage points less likely to enter a 4 or 5 star rated hospital, 1.1 percentage points less likely to enter a hospital in the lowest quintile of readmissions, and 1.5 and 0.4 percentage points less likely to enter hospitals in the lowest quintiles for mortality following acute myocardial infarction or stroke, respectively. The differences were smaller, but still significant for MA plans with a star rating of 4 or more. We also find MA enrollees are 1.7 percentage points less likely to be admitted to 1-2 star hospitals. The choice model analysis confirmed the linear probability model results. All sensitivity analyses find similar results.
Enrollees in MA plans, particularly those in lower rated plans, are more likely to be admitted to average quality hospitals than patients who live in the same Zip Codes that are enrolled in traditional Medicare. These findings may be indicative of narrow hospital networks available in the MA program. As 33% of the Medicare population is now enrolled in MA, policymakers may wish more closely to monitor the quality of MA plan networks so beneficiaries can be fully informed of their options.