Session: Hospital Report Cards
Room: Hollister McManus Lounge
Time: Mon 08:30-10:00

Presenter: Ilaria Mosca (Tilburg University. Economics)

Discussant: Yi LuLehigh University

Abstract

Hospital report cards have become increasingly popular, with seven states and many private organizations ranking hospitals on the basis of surgical mortality and other metrics.
In a previous study, we showed that hospitals that receive lower-than-expected report card scores experience a significant drop in market share (Dranove & Sfekas 2008). Given that market share is at stake, low-scoring hospitals have a financial incentive to improve their report card scores. Hospitals could do this in several ways. First, they could work to improve the quality of surgical care. Second, they could refuse care to high-risk patients. Third, they could change the way they assign patients to
medical or surgical treatment, potentially treating some high-risk patients medically even if the patients would benefit more from surgical treatment.
Although the first two possibilities have received some attention in the literature, the issue of within-hospital sorting has not.
We study whether hospitals who receive a poor report card score in a particular year engage in patient shifting during the following year. We focus on the report card program in New York State. Specifically, we examine patients who presented with acute myocardial infarction (AMI), and ask whether the poorly ranked hospitals considered
the patients’ potential effect on the hospital’s report card score when assigning them to
medical or surgical treatment (e.g. a patient with a high risk of death who would benefit
more from surgical than from medical treatment is assigned to medical treatment
instead). We also examine whether patient sorting changed between the pre-report card
period (1989-1990) and the post-period. Our outcomes of interest are patient severity
and patient mortality. We expect that hospitals receiving a poor report card score will
respond by sorting sicker patients into medical, rather than surgical, treatment. We
use hospital discharge records to test this hypothesis using three measures: patient
severity (measured as a composite of several comorbidities and as the raw number of
comorbidities); in-hospital mortality; and length of stay in the hospital. Sorting would
result in increases in all three of these measures in the time period following a bad
report card score.

Key Terms

Quality report cards, patient shifting, within-hospital sorting

Authors:

Ilaria Mosca (Tilburg University. Economics) and Andrew Sfekas (Northwestern University. Economics)