To the Editor.
—The recent article by Dr Carson and colleagues1 is an interesting and timely study of an important issue—namely, a comparison of the cost-effectiveness of 2 different organizational structures for an intensive care unit (ICU). However, several limitations in the study design and data interpretation make the conclusions suspect. For example, it is stated that all patients admitted to the ICU during the study period were eligible for enrollment, but not that consecutive patients were actually enrolled in the study. Perhaps there was a systematic bias in patient referral, as is alluded to in the paper.It is of note that the actual mortality for patients in the open period was lower than in the closed period, although the difference was not statistically significant. The comparison of actual to predicted mortality based on Acute Physiology and Chronic Health Evaluation II (APACHE II) scores is error prone, and