Background: Extubation failure in the neurocritical care unit (NCCU) is difficult to predict, and is an important source of prolonged intensive care, exposure to morbidity, and increased cost.
Methods: In this observational cohort study in the NCCU of a tertiary care hospital, we examined patients undergoing extubation or tracheostomy with >6 h of intubation. Observational data were collected at the time of the decision to extubate or pursue tracheostomy. The primary end-point was extubation failure within 72 h.
Results: A total of 378 tracheostomy versus extubation decisions were made on 339 individuals, resulting in 93 tracheostomies and 285 extubations. The extubation failure rate was 48/285 (16.8%). Individuals who underwent extubation had similar GCS scores [median 10T (IQR 10-11), P = 0.21]. Extubation failures had similar rates of pneumonia and fever, chest X-ray (CXR) findings, and admission diagnoses (P = NS). Factors associated with success in univariate analysis included intact gag reflex, normal eye movements, ability to close eyes to command, and ability to cough to command (all P < 0.05). In multivariate analysis, the ability to follow four commands (close eyes, show two fingers, wiggle toes, cough to command) was associated with success (P = 0.01). ROC analysis identified a significant difference in favor of a multivariate model incorporating four commands over GCS alone (P = 0.007).
Conclusion: The ability to follow four commands and other examination criteria were strongly associated with extubation success in this observational study. Modeling suggests that specific neurologic examination parameters provide additional predictive information over GCS alone. A prospective, protocol-driven trial is needed to test and expand these findings.