Skip Navigation

Modeling Critical Care Regionalization

Amesh A. Adalja, MD, FACP, FACEP, FIDSA, August 21, 2015 

Regionalization of medical care, in which patients with specific high-risk conditions such as stroke, major trauma, acute coronary syndromes, or burns are shunted away from certain medical centers and toward major referral centers (which may be further away), has proven to be an effective means of improving care. This triaging of patients often occurs rapidly and, if possible, during the prehospital EMS phase of treatment. 

However, one condition that has not been regionalized in the modern era is critical illness, though evidence has been mounting that such a system would improve outcomes. In the absence of an existing full-fledged system to study, my colleagues in the Department of Critical Care Medicine at the University of Pittsburgh recently published in Critical Care Medicine a simulation model that attempts to capture the effects of critical care regionalization. 


Prehospital Scoring

Seymour et al developed a model based on acute care hospitalizations in 2006 in King County, Washington. The model employed EMS and hospital-level data and incorporated such variables as hospital travel time, need for mechanical ventilation, need for ICU-level care, and need for hemodialysis, as well as a prehospital risk score.

The model developed a regionalization plan by incorporating prehospital risk scores, EMS transport time, and hospital occupancy data. If a patient’s risk score exceeded a certain threshold, the patient was triaged to a referral hospital; if the risk score was below the threshold, the patient was triaged to a nonreferral hospital. Several scenarios were modeled, including those that used up triage, up/down triage, and diminished bed availability.


Model Results

Actual baseline data revealed that of those patients whose prehospital risk scores exceeded threshold, just 29% were transported to referral centers (compared to all of them under the regionalization model). With the model, ICU bed occupancy in referral centers increased from a baseline of 61% to, at most, 74%, based on which regionalization scenario was operative. ICU lengths of stay at both referral and nonreferral hospitals was unchanged from baseline in the model, while more patients requiring mechanical ventilation and hemodialysis were concentrated at referral centers with the regionalization model. With the model, EMS travel times increased by between 2% and 13%, translating to less than 12 minutes of increased travel time in most cases.


Pilot Projects Should Occur

Seymour et al’s model has several important implications that should provide a basis for trials of actual pilot projects of critical care regionalization. As evidenced by the results of this model, regionalization does not drastically increase travel times, overburden referral center ICUs, or deplete nonreferral hospitals. On the contrary, by incorporating both up and down triage, patient flow can be optimized.

Regionalization not only holds the potential to improve care during ordinary times; it could prove to be a major asset during emergencies such as an influenza or other infectious disease outbreak in which sophisticated critical care may be essential.



Seymour CW, Alotaik O, Wallace DJ, et al. County-level effects of pre-hospital regionalization of critical illness. Crit Care Med 2015;43:1807-1815.