Causes and outcome of coma: A nationwide cross-sectional analysis.

BACKGROUND: The Neurocritical Care Society Curing Coma Campaign has renewed interest in coma management. Large-scale cross-sectional data on the relative frequency of coma etiologies and factors influencing survival remain limited. METHODS: We analyzed the National Inpatient Sample (NIS) database and extracted all cases of coma treated with mechanical ventilation (MV) between 2010 and 2019. A hierarchical classification system assigned each patient a primary coma etiology, prioritizing (1) major structural/cellular brain injuries, (2) laboratory-confirmed diagnoses, and (3) organ failure etiologies. RESULTS: Of 1,904,453 MV admissions, 146,577 (7.6%) involved coma. Mortality was 47% in comatose versus 29% in non-comatose patients (P < 0.001); 30% of comatose patients who died had a do-not-resuscitate (DNR) order. The most common etiologies were cardiac arrest (20%), traumatic brain injury (17%) and intracerebral hemorrhage (ICH, 16%). Highest mortality occurred after cardiac arrest (75%), cardiogenic shock (71%), ICH (63%), subarachnoid hemorrhage (56%), AIS (51%), and hepatic coma (51%); lowest mortality was observed with drug overdose (13%). Multivariable analysis identified DNR status as the strongest mortality predictor (AOR 7.1), followed by high- (AOR 5.0) or intermediate-risk (AOR 1.7) primary diagnosis and age > 55 years (AOR 1.6; all P < 0.001). Non-white and Medicaid patients had lower mortality and were significantly less likely to carry a DNR order. CONCLUSION: Almost half of comatose patients die in the hospital and 30% of these patients are DNR. Given withdrawal of life support is a major determinant of survival, better tools are needed to accurately predict neurological outcomes in disorders of consciousness.