Objective: To establish ways of defining long-term ICU patients and to develop methods of quantifying their impact on bed utilization.

Design: Retrospective analysis of prospectively collected sample data.

Setting: Two university hospital adult surgical ICUs.

Patients: 6588 consecutive patients admitted over 8 years to ICU 1 and 2913 patents admitted over 5 years to ICU 2.

Interventions: None.

Measurements and results: Patients with uncommon but expected long stays were designated as long-term patients and those with exceptional and unexpected long stays were designated as outliers. Visualization of the length-of-stay frequency distributions revealed that the "tail" of the distributions began at a stay of about 10 days, and this was chosen as the threshold for long-term patients. A threshold of 30 days was chosen as the outlier threshold since only isolated patients stayed longer. The impact (long-term patients/total admissions) of long-term patients on bed utilization was assessed and year-to-year differences examined. Long-term patients staying at least 10 days used 2.7-5.0 bed-years.

Conclusions: Long-term patients can have a major impact on ICU bed utilization. With 23-45% of the beds occupied by long-term patients, the availability of beds for short-stay patients is reduced. This is important in ICUs where efficient patient throughput is vital, so that beds are availability for both elective and emergency admissions.