Assessing Hospital-Associated Deaths-Reply

In Reply.—  Dr Wagner argues that it would be more appropriate to observe hospital-associated mortality during a period from admission to 30 days after discharge rather than from admission to 30 days after admission. He observes that in states such as New York, with very long stays for certain patients with certain conditions (such as stroke), taking a period of 30 days after admission omits a substantial number of inpatient deaths; he also is concerned that physicians can delay inpatient deaths so that they fall after the 30-day observation period and artificially improve the performance of the hospital.The problem Dr Wagner identifies in data from New York is largely that New York hospitals are serving as skilled nursing facilities for a number of terminally ill patients, a situation that is much less frequent in other states. The statistic he proposes would charge New York hospitals with deaths that would