Does START triage work? An outcomes assessment after a disaster

Ann Emerg Med. 2009 Sep;54(3):424-30, 430.e1. doi: 10.1016/j.annemergmed.2008.12.035. Epub 2009 Feb 5.


Study objective: The mass casualty triage system known as simple triage and rapid treatment (START) has been widely used in the United States since the 1980s. However, no outcomes assessment has been conducted after a disaster to determine whether assigned triage levels match patients' actual clinical status. Researchers hypothesize that START achieves at least 90% sensitivity and specificity for each triage level and ensures that the most critical patients are transported first to area hospitals.

Methods: The performance of START was evaluated at a train crash disaster in 2003. Patient field triage categories and scene times were obtained from county reports. Patient medical records were then reviewed at all receiving hospitals. Victim arrival times were obtained and correct triage categories determined a priori using a combination of the modified Baxt criteria and hospital admission. Field and outcomes-based triage categories were compared, defining the appropriateness of each triage assignment.

Results: Investigators reviewed 148 records at 14 receiving hospitals. Field triage designations comprised 22 red (immediate), 68 yellow (delayed), and 58 green (minor) patients. Outcomes-based designations found 2 red, 26 yellow, and 120 green patients. Seventy-nine patients were overtriaged, 3 were undertriaged, and 66 patients' outcomes matched their triage level. No triage level met both the 90% sensitivity and 90% specificity requirement set forth in the hypothesis, although red was 100% sensitive (95% confidence interval [CI] 16% to 100%) and green was 89.3% specific (95% CI 72% to 98%). The Obuchowski statistic was 0.81, meaning that victims from a higher-acuity outcome group had an 81% chance of assignment to a higher-acuity triage category. The median arrival time for red patients was more than 1 hour earlier than the other patients.

Conclusion: This analysis demonstrates poor agreement between triage levels assigned by START at a train crash and a priori outcomes criteria for each level. START ensured acceptable levels of undertriage (100% red sensitivity and 89% green specificity) but incorporated a substantial amount of overtriage. START proved useful in prioritizing transport of the most critical patients to area hospitals first.

Publication types

  • Research Support, U.S. Gov't, P.H.S.

MeSH terms

  • California
  • Critical Illness / classification
  • Critical Illness / therapy
  • Disasters*
  • Hospitals / statistics & numerical data
  • Humans
  • Outcome Assessment, Health Care
  • Railroads*
  • Retrospective Studies
  • Sensitivity and Specificity
  • Transportation of Patients / statistics & numerical data
  • Triage / methods*
  • Triage / statistics & numerical data*