The Allocation and Rationing of High-Cost Services-Reply

In Reply.—  The letter by Franklin et al concerns an important caveat to our conclusions and one that is clearly identified in our article. Our data were insufficient to comment on the effect of ICU bed availability on decisions about patients never admitted to the ICU. The ideal study on this subject would evaluate all admissions to the ICU and, as stated in their letter, "every patient who could be an ICU candidate." This latter group of patients is potentially very large, including those hospitalized in a non-ICU setting and patients and potential patients elsewhere in the community.Although their analysis suggests that hospital mortality drops when more ICU beds are made available, the findings are inconclusive. Their use of a historical control group, combined with the statement that "other treatment variables did not change," makes the results difficult to interpret. The changing nature of medical practice in the 1980s,