Implications of the log-normal distribution for updating estimates of the time remaining until ready for phase I post-anesthesia care unit discharge

Abstract Background : Multiple managerial decisions on the day of surgery depend on estimating how much longer each patient will remain in the phase I post-anesthesia care unit (PACU). We apply statistical methods developed for the time remaining in long-running surgical cases to the PACU. Methods : We use a previously published historical cohort of N = 765 patients from one large teaching hospital who underwent gynecological surgery with time from operating room entrance to end of surgery of at least 4 hours and were admitted to the PACU. Results : The times until patients were ready for discharge followed a two-parameter log-normal distribution (e.g., Lilliefors’ test P = 0.18 and best three-parameter fit with no shift). Because the estimated coefficient of variation was 42%, there was an approximately constant percentage of the patients ready for discharge for each period longer than average. For example, half the patients stayed longer than 2.0 hours. Twenty minutes later, approximately 1/3rd of the patients were ready for discharge, and 2/3rd were not. Twenty minutes later, among the 2/3rd originally remaining, there were 1/3rd since ready for discharge, and so forth. The expected time remaining in the PACU for patients staying longer than the median time was relatively unchanging, independent of the amount of time that they had already been in the PACU. Conclusions : It is imprudent to rely on a certain expected duration of time until PACU discharge because any one time is unreliable. What is dependable is the probabilistic interpretation that, during any one period, there is a relatively unchanging percentage of the patients who will be ready for discharge during the period. Use this mathematical relationship for contacting hospital wards in advance of transfer, optimizing the use of porters, answering queries from families, and forecasting PACU occupancy in real-time.