Out-of-hospital personnel worldwide are taught to calculate the Glasgow Coma Scale (GCS) score as a routine part of their assessment of injured patients. This 13-point (counting from 3 to 15) tool is a core component of the Centers for Disease Control and Prevention National Field Triage Guidelines,1 used to evaluate risk, guide treatment, and help determine the most appropriate hospital for care (eg, trauma center). The ubiquitous acceptance of the GCS presumes that it is an optimal tool for this purpose; however, it has multiple potent limitations.2, 3, 4Editor’s Capsule Summary
What is already known on this topic
The Glasgow Coma Scale (GCS) is widely used in out-of-hospital trauma care despite little evidence that it is an optimal scale.
What question this study addressed
Can alternative scales or individual items meet or surpass the performance of the GCS?
What this study adds to our knowledge
This 2-site retrospective study of 47,973 trauma patients showed that several simpler scores performed as well as the GCS. In particular, the single item “patient does not follow commands” was identical to GCS score less than or equal to 13.
How this is relevant to clinical practice
This study provides sufficient justification for out-of-hospital systems to experiment with omitting the GCS for simpler measures.
First, the GCS is unreliable. It includes multiple subjective elements and has demonstrated surprisingly low interrater reliability in a wide variety of settings.5, 6, 7, 8, 9, 10, 11, 12, 13, 14 In a study of independent paired assessments by attending emergency physicians, for example, GCS scores were the same in just 38% and were 2 or more points apart in 33%.5
Second, the GCS is widely perceived as complicated2, 3, 13, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38 and takes more than just a few seconds to evaluate. Feldman et al39 found that out-of-hospital providers could calculate the GCS score correctly only 40% of the time, underscoring the inability of this scale to be consistently remembered and applied. Riechers et al37 noted that only 15% of military physicians could correctly calculate the GCS score, despite all of them being familiar with the scale and most having completed the Advanced Trauma Life Support course. Bassi et al13 observed that less than half (48%) of clinicians correctly scored the GCS in a written clinical scenario, with neurosurgeons correct just 56% of the time. Other studies have demonstrated a similarly low level of accuracy.17, 18
Third, summing the 3 different component scales into its 120 different scoring combinations was never intended by the GCS creators,40 and is statistically unsound in that it inappropriately assumes that each gradation of each subscale exhibits an identical magnitude of clinical importance—a presumption that is neither intuitive nor supported by existing evidence.2, 4, 29, 32, 41, 42
Fourth, the GCS is unnecessarily complex in that its subscales and other simpler derivatives have been reported to demonstrate similar predictive capability.19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 38, 43, 44 The GCS predicts mortality well at its extremes and poorly in its midrange,32 and thus most of its predictive capacity is anchored by the endpoints. Accordingly, only some of the GCS elements appear truly predictive, whereas the others are either redundant or simply “noise.” Simplified derivatives that have been previously reported to demonstrate similar accuracy are the binary assessments of motor GCS score less than 6,26, 30 motor GCS score less than 5,43 and the 3-element Simplified Motor Score (motor GCS score of 6, 5, or <5, corresponding to a Simplified Motor Score of 2, 1, or 0, respectively) (Figure 1).19, 20, 21, 36, 38, 44
Despite the above evidence, in 2011 the Centers for Disease Control and Prevention National Expert Panel on Field Triage declined to replace the GCS with a simplified tool, citing “…the lack of confirmatory evidence, the long standing use of GCSt [total GCS] and its familiarity among current EMS [emergency medical services] practitioners, the inclusion of the motor score within the GCSt, and complications because of the difficulty of comparing scoring systems.”1 They advocate the field triage of patients with total GCS scores of less than or equal to 13 to trauma centers of the highest available level.1
Most recently, Kupas et al33 tested motor GCS score less than 6—or “patient does not follow commands”—in a statewide database of 393,877 trauma patients and noted similar performance in the prediction of trauma outcomes. A subsequent systematic review44 noted “slightly greater” areas under receiver operating characteristic (ROC) curves for the total GCS compared with simplified scales, but acknowledged the differences as “likely to be clinically unimportant.” The differences in ROC area were noted to be at the extremes rather than the central curve area, where clinical decisionmaking occurs, and thus do not appear to be relevant or important.4