Meeting Brochure and registration form      SMDM Homepage

Tuesday, 17 October 2006


Eric B. Hollingsworth, BS1, Wei Xiong, PhD1, John Delaney2, and Nathaniel Hupert, MD, MPH1. (1) Weill Medical College, Cornell University, New York, NY, NY, (2) New York Presbyterian Hospital, New York, NY

Purpose: Pre-hospital triage of mass casualty event victims is a critical resource allocation process that has not been extensively studied.  We created a theoretical model of triage as a diagnostic test to quantify the relationship between triage, casualty load, hospital capacity, and mortality.

Methods:  We developed a discrete event simulation of mass casualty triage by first creating a population of critically and non-critically injured patients, applying a triage test designed to distinguish between critical and non-critical casualties, and then evaluating the consequences of the triage decision.  Hypothetical ROC curves were generated from available data.  We found conflicting reports of time-dependent mortality for critical casualties, and chose three representative curves: late (post-6 hours), linear (steady decline of ~12% per hour), and exponential (rapid decline after "golden hour").  We examined casualty loads ranging from 100 to 500 patients, percentage of critical patients ranging from 20-33%, and available emergency department/operating room tandems (EDORs – a measure of available treatment capacity) ranging from 1-9.

Results:  We present results for the 7-EDOR, 100-patient, 25% critical patient scenario with linear time-dependent mortality across a range of sensitivities for the triage algorithm.  Worst case critical mortality ratio is 55% (critical patients are treated last), best case is 34% (critical patients are triaged perfectly), and with no triage is 48%.  High sensitivity for identifying critical casualties (and correspondingly lower specificity) leads to a critical mortality ratio (CM) of 41%, low sensitivity leads to CM of 42%, and standard triage (favoring neither high nor low sensitivity) leads to CM of 39%.  This relationship can be developed across a range of numbers of assumptions including: EDOR number, patient load, time dependent mortality rate, and triage accuracy.

Conclusion:  The number of total casualties and ratio of casualties to treatment resources determines whether a higher or lower sensitivity triage process yields lower mortality.  However, for a wide range of mass casualty scenarios, it appears that implementing triage at typical sensitivity levels (ie, 80% to 90%) minimizes critical patient mortality.  Implementing a triage strategy will reduce critical patient mortality, regardless of whether a high or low sensitivity test is used.

See more of Poster Session IV
See more of The 28th Annual Meeting of the Society for Medical Decision Making (October 15-18, 2006)