Trauma triage system tends to overestimate injury severity
Research Activities, February 2010, No. 354
First responders at disaster scenes have used the simple triage and rapid treatment (START) system since the 1980s to quickly sort casualties, by severity, into four categories: red, yellow, green, and black. Red represents patients who are in the most dire need of care, yellow means the patient transport to the hospital can be delayed, and green indicates the patient has only minor injuries. Black indicates the patient is dead. Christopher A. Kahn, M.D., M.P.H., and colleagues at the University of California, Irvine, found that the START system overestimated injuries of 79 victims involved in a 2002 train crash and underestimated the condition of 3 patients.
The authors examined 148 patient records at 13 hospitals and found that just 66 patients were triaged into the correct category, for an accuracy rate of 45 percent. For example, of the 22 patients triaged as red, only 2 patients truly belonged in that category. For the 68 patients triaged as yellow, only 26 met that category's standard. And, for the 58 triaged as green, 120 actually belonged in that category.
The authors found that while the system does do a good job of identifying patients with minor injuries, it does not excel at differentiating between patients who need immediate care and those who have significant but stable injuries. During disaster responses, this mismatch could result in a hospital system being overwhelmed by noncritical patients who use up resources needed for truly critically injured individuals. The START was effective in ensuring patients who were put in the red category got to the hospital faster than those in the yellow and green categories. This study was funded in part by the Agency for Healthcare Research and Quality (HS15768).
See "Does START triage work? An outcomes assessment after a disaster," by Dr. Kahn, Carl H. Shultz, M.D., Ken T. Miller, M.D., Ph.D., and Craig L. Anderson, Ph.D., in the September 2009 Annals of Emergency Medicine 54(3), pp. 424-430.