An interesting difference between our results and those of Dr Wijsenbeek and colleagues has to do with location of death in the hospital. Although in our study, of those who died in our hospital, 33% died in the ICU, in the report by Dr Wijsenbeek and colleagues only 7% died in the ICU. It is hard to assess the reason for this large difference, but it is not surprising considering the difference in ICU bed numbers and their use between different countries.2 As an example, the United States has seven times more ICU beds per capita compared with the United Kingdom. A large study that assessed the impact of these numbers found significant differences in patterns of admissions of medical patients to the ICU between the United States and the United Kingdom and suggested that the threshold for admission of a medical patient to the ICU in the United States is substantially lower than in the United Kingdom.3 Generally, patients in the United States had a significantly lower severity of illness, were significantly more likely to be admitted directly from the ED, and included a much larger proportion of the very elderly. Thus, the location of death within the hospital is probably reflective of local practice patterns and availability of ICU beds.