Intensive care uses a disproportionate quantity of US health care resources, accounting for 13% of hospital costs, 15% of hospital beds, and 4% of National Health Expenditures.¹ With the COVID-19 pandemic placing tremendous strain on intensive care unit (ICU) capacity,² optimizing critical care utilization has faced increasing scrutiny. A predictive model for ICU admission such as the one presented by Rozeboom et al³ may help hospital leadership with daily elective surgical schedule smoothing and reduce undesirable downstream effects of planned ICU admissions on emergency department diversion.

Variability in the demand for critical care beds presents a significant challenge to efficient distribution of resources, leading to emergency department overcrowding, operative delays, case cancellations, and threatening emergency surge preparedness.⁴ As emergency cases cannot be predicted or controlled, the real benefit of the Surgical Risk Preoperative Assessment System (SURPAS) model with regard to resource allocation lies in its ability to identify elective cases that require postoperative ICU admission. Reducing variability in complex surgical caseload, or load leveling, decreases strain on ICU capacity and may improve throughput.⁴ Rozeboom and colleagues³ developed a tool to predict postoperative ICU admission with excellent performance using just 7 clinical variables. The online calculator and parsimony of required data provide a convenient instrument for practical use, but we noted the composite variable missed 85% of patients who ultimately were admitted to the ICU after surgery. This highlights the need to better understand other drivers of ICU admission not included in the physiologic parameters of the model, such as surgeon preference, use of certain medication infusions, or hourly nursing interventions. These may account for a substantial portion of ICU utilization and represent an opportunity to better align a scarce resource with patient needs. For example, critical care units in some hospitals have up to 50% of beds occupied by low-risk patients who do not benefit from this level of care.⁵ Adherence to data-driven ICU admission and discharge criteria may reduce unindicated utilization by patients with low preoperative risk scores. Caring for select patients in step-down units rather than ICU beds, such as those undergoing free-flap reconstructions⁶ or those requiring insulin infusions,⁷ may also reduce bottlenecks for patient flow without compromising outcomes.