Purpose: To determine whether a group of experienced clinicians can predict intensive care unit (ICU) length of stay (LOS) following cardiac surgery.
Methods: A cohort of 265 adult patients undergoing cardiac surgery at St. Michael's Hospital, Toronto, Ontario, between January 2, 1992, and June 26, 1992, were seen preoperatively by the clinicians participating in the study and ICU length of stay was predicted based on the clinicians' preoperative assessment and/or information recorded in the patient's chart.
Results: Five hundred and ten ICU length of stay predictions were obtained from a group of eight experienced clinicians (anaesthetists/intensivists, cardiologists, nurses). The clinicians predicted the exact ICU length of stay (in days) correctly 51.2% of the time and were within +/- 1 day 84.5% of the time. The clinicians correctly predicted short ICU stays (< or = 2 days) for 87.6% of the patients who had short ICU stays but only predicted long ICU stays (> 2 days) in 39.4% of the patients who had long ICU stays.
Conclusions: Experienced clinicians can predict preoperatively with a considerable degree of accuracy patients who will have short ICU lengths of stay following cardiac surgery. However, many patients who had long ICU stays were not correctly identified preoperatively. Unidentified preoperative risk factors or unanticipated intraoperative/postoperative events may be causing these patients to have longer than expected ICU stays.