Two methods for assessing appropriate utilization of pediatric ICU (PICU) resources have been proposed, efficiency of bed utilization, and severity-adjusted length of stay (LOS). Efficiency is the proportion of patient days where a unique PICU therapy is used (e.g. mechanical ventilation or vasoactive drug infusions). This has the disadvantage of requiring daily data collection. Severity and diagnosis adjusted LOS can be predicted from the first 24 hours of PICU stay with only PRISM, diagnosis, patient conditions and mechanical ventilation. This study investigates the agreement of the two methods.

Methods: Data from the first 24 hours of PICU stay for consecutive admissions to 32 PICUs (PRISM III, diagnosis, descriptive information) was used to develop and validate a LOS predictor based on PRISM III, diagnosis, and other patient conditions. Patients with LOS ≤ 12 days (± 2 SD) were included (n=9989). Daily data collection included unique PICU therapies in 11 PICUs. Efficiency was defined as number of days where a unique PICU therapy was used divided by the total patient days. The LOS model was used to estimate a LOS ratio as the predicted total days of care/observed total days of care. The association between the two measures was tested in the 11 PICUs by rank correlation analysis.

Results: Efficiency ranged from 0.67 to 0.29. The LOS ratios ranged from 1.26 to 0.93. The rank correlation between the two utilization measures was 0.86, (p<.001), indicating that the two utilization rankings are in good agreement.

Conclusion: The two resource measures are highly correlated. Hence, the LOS model which requires only admission day data is a good indicator of actual efficiency rankings. Results from both methods can only be interpreted relative to other PICUs since a threshold for efficiency or LOS ratios has not been established.