Objectives: To systematically identify and summarize evaluations of measures of circulatory and respiratory compromise, focusing on measures that can be used in field assessment by emergency medical services to inform decisions about the level of trauma care needed. We identified research on the ability of different measures to predict whether a patient was seriously injured and thus required transport to the highest level of trauma care available.
Data sources: We searched Ovid MEDLINE®, CINAHL®, and the Cochrane databases from 1996 through August 2017. Reference lists of included articles were reviewed for additional relevant citations.
Review methods: We included studies of individual measures and measures that combined circulatory, respiratory, and level of consciousness assessment. Evaluations included diagnostic accuracy (sensitivity and specificity) and area under the receiver operating characteristic curve (AUROC). We used data provided to calculate values that were not reported and pooled estimates across studies when feasible.
Results: We identified and included 138 articles reporting results of 134 studies. Circulatory compromise measures evaluated in these studies included systolic blood pressure, heart rate, shock index, lactate, base deficit, and heart rate variability or complexity. The respiratory measures evaluated included respiration rate, oxygen saturation, partial pressure of carbon dioxide, and need for airway support. Many different combination measures were identified, but most were evaluated in only one or two studies. Pooled AUROCs from out-of-hospital data were 0.67 for systolic blood pressure (moderate strength of evidence); 0.67 for heart rate, 0.72 for shock index, 0.77 for lactate, 0.70 for respiratory rate, and 0.89 for Revised Trauma Score combination measure (all low strength of evidence); and were considered poor to fair. The only AUROC that reached a level considered excellent was for the Glasgow Coma Scale, age, and arterial pressure (GAP) combination measure (AUROC, 0.96; estimate based on emergency department data). All of the measures had low sensitivities and comparatively high specificities (e.g., sensitivities ranging from 13% to 74% and specificities ranging from 62% to 96% for out-of-hospital pooled estimates).
Conclusions: Physiologic measures usable in triaging trauma patients have been evaluated in multiple studies; however, their predictive utilities are moderate and far from ideal. Overall, the measures have low sensitivities, high specificities, and AUROCs in the poor-to-fair range. Combination measures that include assessments of consciousness seem to perform better, but whether they are feasible and valuable for out-of-hospital use needs to be determined. Modification of triage measures for children or older adults is needed, given that the measures perform worse in these age groups; however, research has not yet conclusively identified modifications that result in better performance.