Risk factors for failing endotracheal extubation in neurocritical patients: a retrospective cohort study

Objective To identify risk factors of failing endotracheal extubation among neurocritical care patients with endotracheal intubation for more than 48 h and passing the autonomous breathing test (SBT) and establish a prediction model accordingly. Methods This study included the clinical data of patients who received standardized monitoring and treatment in the neurocritical care unit of the First Affiliated Hospital of Anhui Medical University from April 2020 to August 2024. Based on the outcomes of extubation after 5 days, data were divided into the success group and the failure group. Clinical features of two groups were compared and accordingly taken into multivariate logistic regression analysis, eventually generating a scoring model with its receiver operating characteristic curve (ROC). The area under the curves (AUC) of other previous scores was compared by Z-test. Results Of 116 recorded cases, 92 (79.3%) were successfully extubated, while 24 (20.7%) required re-intubation within 5 days. Univariate analysis revealed significant differences between two groups in state of consciousness, Glasgow Coma Scale (GCS) total score, GCS motor score (GCS-M), muscle strength, swallowing ability, coughing response, body temperature, oxygenation index, Apache II score, and APS score (all p < 0.05). Multivariate analysis was further carried out, and a scoring model was established accordingly (including GCS-M, coughing ability, and oxygenation index) with a total score of 4 points. The model demonstrated good predictive value, with a cut-off ≥1 distinguishing extubation success with 79.2% sensitivity and 69.6% specificity according to ROC (AUC = 0.79; 95% CI, 0.68–0.90). Conclusion This clinical predictive scoring model could provide guidance for extubation decisions in neurocritical care units but requires further external validation.