Predictive model for optimizing prehospital transfusions in an urban EMS system.

Abstract

Background: Prehospital transfusions might provide a survival benefit for injured patients. Because blood products are a scarce resource, their optimal deployment requires careful consideration. A computer model was built to explore different deployment scenarios for two blood-carrying ambulances (mobile blood banks, MBBs) in the City of Pittsburgh.

Study design and methods: Mixed integer programs were used to determine the optimal locations for the bases of the two MBBs from amongst the City's 14 ambulance bases. Then, a discrete-event simulation of dispatching MBBs to attend to patients who would have qualified for prehospital transfusions due to having hypotension following injury was performed using data from one year of calls to the City's emergency services hotline (911 calls).

Results: Over the one-year period, there were 238 ambulance dispatches to injured patients with hypotension for their age. The average time to transfusion was significantly lower when the MBB attended to the patient compared with receiving their transfusion at the hospital (average 7.2 ± 0.1 min vs. 36.7 ± 0.2 min, respectively). However, there were diminishing returns when more than four deployed MBBs were simulated; with two MBBs, up to 73% of qualifying patients could be serviced, and when four MBBs were deployed, up to 95% of patients could be serviced. Deploying >4 MBBs did not increase the number of serviced eligible patients. There was minimal improvement in MBB efficiency when the restocking and cleaning time after patient delivery was reduced from 40 to 15 min.

Conclusion: Computer modeling can help optimize resources when planning prehospital transfusion programs.