The famous estimate that 44,000 to 98,000 patients die every year in the United States due to preventable adverse events, which was widely publicized in the Institute of Medicine's seminal report, is challenged as a significant overestimate in this British study. The investigators reviewed 1000 randomly selected deaths from 10 hospitals, using a standardized protocol based on prior classic studies of preventable inpatient mortality and found that only 5.2% of deaths were considered preventable. The majority of preventable deaths occurred in patients whose life expectancy was considered to be less than 1 year even if optimal care was provided. As in prior studies, reviewers' agreement on whether a death was preventable was only moderate. The reviewers did find evidence of significant quality problems among both preventable and non-preventable deaths.