Can Ranking Hospitals on the Basis of Patients' Travel Distances Improve Quality of Care?

Conventional outcomes report cards-- public disclosure of information about the patient-background-adjusted health outcomes of individual hospitals and physicians -- may help improve quality, but they may also encourage providers to "game" the system by avoiding sick and/or seeking healthy patients. In this paper, I propose an alternative approach: ranking hospitals on the basis of the travel distances of their Medicare patients. At least in theory, a distance report card could dominate conventional outcomes report cards: a distance report card might measure quality of care at least as well but suffer less from selection problems. I use data on elderly Medicare beneficiaries with heart attack and stroke from 1994 and 1999 to show that a distance report card would be both valid -- that is, correlated with true quality -- and able to distinguish confidently among hospitals -- that is, able to reject at conventional significance levels the hypothesis that the true quality of a low-ranked hospital was the same as the quality of the average hospital. The hypothetical distance report card I propose compares favorably to (although does not necessarily dominate) the California AMI outcomes report card.