Measuring trauma system performance: Right patient, right place-Mission accomplished?

A regional trauma system must establish and monitor acceptable overtriage and undertriage rates. Although diagnoses from discharge data sets can be used with mortality prediction models to define high-risk injury, retrospective analyses introduce methodological errors when evaluating real-time triage processes. The purpose of this study was to determine if major trauma patients identified using field criteria correlated with those retrospectively labeled high risk and to assess system performance by measuring triage accuracy and trauma center utilization.

A statewide database was queried for all injury-related International Classification of Diseases, 9th Revision, code discharges from designated trauma centers and nontrauma centers for 2012. Children and burn patients were excluded. Patients assigned a trauma alert fee were considered field-triage(+). The International Classification Injury Severity Score methodology was used to estimate injury-related survival probabilities, with an International Classification Injury Severity Score less than 0.85 considered high risk. Triage rates were expressed relative to the total population; the proportion of low- and high-risk patients discharged from trauma centers defined trauma center utilization.

There were 116,990 patients who met study criteria, including 11,368 (10%) high-risk, 70,741 field-triage(-) patients treated in nontrauma centers and 28,548 field-triage(-) and 17,791 field-triage(+) patients treated in trauma centers. Field triage was 86% accurate, with 10% overtriage and 4% undertriage. System triage was 66% accurate, with 32% overtriage and 2% undertriage. Overtriage patients more often, and undertriage patients less often, had severe injury characteristics than appropriately triaged patients.

Trauma system performance assessed using retrospective administrative data provides a convenient measure of performance but must be used with caution. Residual mistriage can partly be attributed to error introduced by retrospective high-risk definitions, whereas differences between field and system triage accuracy can be attributed to the trauma center's role as a large community hospital. Given the limitations of the data and methods, these results may represent optimal patient distribution within this mature system.