To determine the accuracy of the Glasgow Coma Scale (GCS) compared to the revised Coma Recovery Scale (CRS-r) at detecting the conscious state.

Trained technicians, blinded to clinical course, performed simultaneous CRS-r and GCS scores among neurologic intensive care patients. Sensitivity, specificity and receiver operating curves were estimated to assess the ability for GCS to detect the conscious state. Conscious state was defined by a score of 2 on the communication or a score of 6 on the motor subscale of the CRS-r. All estimates calculated using logistic regression with variance clustered by person and optimal cut-off determined via the Liu method. Four categorizations of the GCS were validated against CRS-r: 1) total score 2) motor sub-score of six  (binary) 3) eye sub-score of ≥3 (binary) 4 ) motor sub-score of six and eye sub-score of ≥3 (binary).

511 GCS and CRS-r scores reflecting 202 patients from December  2013 through September 2021 were identified. 41(8%) of scores demonstrated the conscious state as defined CRS-r. The optimal cut-off for total GCS score (95% CI) was estimated to be 9.5(7.9-11.1), corresponding to sensitivity 90% (81 -100), specificity 88%(78 -98), and AUC 0.95(0.91-0.99). For GCS motor sub-score of six, the estimates of sensitivity were 85%(70-100), specificity 92% (87-97), AUC 0.79(0.6-0.9), and 92% correctly classified. For GCS eye sub-score ≥3, estimates of sensitivity were 95%(88-100), specificity 69%(61-76), AUC 0.65(0.57-0.73), and 71% correctly classified. For the combination of GCS motor sub-score of six and eye sub-score ≥3, estimates of sensitivity were 80%(65-96), specificity 93%(89-97), AUC 0.75(0.56-0.88) and 92% correctly classified.

A total GCS score greater than 9.5 and motor sub-score of 6 demonstrated good diagnostic accuracy for conscious state.