This study include a large, random sample from 25 metropolitan New York hospitals. The sample was drawnearly in the epidemic to include patients with long, complicated, and ongoing hospital stays; In-hospital mortality was used as primary outcome.