The Glasgow Coma Scale as a predictor of mortality in intensive care: relevance of underlying cause of impaired consciousness

BACKGROUND: The Glasgow Coma Scale-based score (GCS) is used to describe the level of consciousness in critically ill patients. In risk-prediction models, an abnormal GCS carries a heavy weight, without considering the cause of impaired consciousness. We assessed the association between low GCS and mortality across individual diagnoses. METHODS: In this retrospective register-based study, we examined 58,982 admissions to 20 ICUs in Finland, Estonia, and Switzerland between 2015 and 2017. We divided the patients into two categories using Acute Physiology And Chronic Health Evaluation (APACHE) III diagnoses: Neurologic and Other diagnoses. We used the lowest GCS recorded during the first 24 h after intensive care unit (ICU) admission. We defined GCS below 9 as low. We analysed the association between low GCS and the risk of death in each APACHE III diagnosis and in the Neurologic and Other diagnoses categories using hierarchical logistic regression. RESULTS: There were 17,275 (29%) patients in the Neurologic category and 41,707 (71%) in the Other diagnoses category. Low GCS was recorded in 4266 (25%) patients in the Neurologic category and 6078 (15%) in the Other diagnoses category. Low GCS was associated with increased risk of death in the Neurologic category [odds ratio (OR) 11.3, 95% confidence interval (CI) 8.0-15.9] and in the Other diagnoses category (OR 4.8, 95% CI 3.9-5.8). The impact of low GCS varied widely between individual diagnoses within both categories. CONCLUSION: Low GCS was a strong predictor of death, but its relative weight was highly dependent on the diagnosis.