Early Predictors of Ventilator-Free Days and Sustained Ventilatory Dependence in Patients With Sepsis: A Retrospective Cohort Study Using 24-Hour Admission Data

Background: Sustained ventilatory dependence is associated with increased morbidity, mortality, and healthcare resource utilization in critically ill patients with sepsis. However, the early identification of patients at risk remains challenging, particularly when using variables available at intensive care unit (ICU) admission. This study aimed to identify potential early markers of ventilatory trajectory, assessed primarily using ventilator-free days through day 28 (VFD-28) and secondarily using sustained ventilatory dependence, based only on data obtained within the first 24 hours of ICU admission.Methods: In this single-center retrospective cohort study, 137 adult patients with sepsis requiring invasive mechanical ventilation were included. The primary outcome was VFD-28, a mortality-inclusive continuous outcome, whereas sustained ventilatory dependence (≥7 days from intubation) was a secondary outcome. Demographic, clinical, and laboratory data obtained within the first 24 hours of ICU admission were univariably screened (25 candidates), and a clinically selected multivariate covariate set (age, admission Glasgow Coma Scale (GCS) score, Clinical Frailty Scale score, lactate, albumin, and prior stroke history) was modeled using Tobit regression for VFD-28 and logistic regression for the secondary binary outcome. Admission GCS score was retained as the sole severity measure after a component-overlap analysis indicated that in separate models evaluating each severity score, Sequential Organ Failure Assessment (SOFA) and Acute Physiology and Chronic Health Evaluation (APACHE) II did not remain independently associated with the outcome, whereas GCS did. Because GCS is itself a shared neurological component of both composite scores, this pattern likely reflects statistical overlap among correlated variables rather than evidence that GCS is a clinically superior severity measure. Cause-specific Cox regression was used to model liberation from ventilation and death while ventilated as competing risks.Results: Higher admission GCS was independently associated with higher VFD-28 values (Tobit coefficient: 4.58, 95% confidence interval (CI): 1.95–7.20; p < 0.001). Sustained ventilatory dependence (a secondary, exploratory binary outcome that cannot by itself distinguish early death from early successful weaning, ≥7 days) occurred in 54.0% of patients. Higher admission GCS was independently associated with lower odds of this outcome (adjusted odds ratio (OR): 0.81, 95% CI: 0.72–0.92; p = 0.001). This association was also observed in an exploratory sensitivity analysis restricted to the smaller subgroup of patients surviving to day 7 (adjusted OR: 0.59, 95% CI: 0.43–0.80). Neither SOFA nor APACHE II scores remained independently associated with the outcome once component overlap with GCS was accounted for. Cause-specific competing-risks analysis showed that admission GCS was associated with the hazard of liberation from mechanical ventilation (hazard ratio (HR): 1.31, 95% CI: 1.15–1.48; p < 0.001) but not with the hazard of death while ventilated (p = 0.851), indicating a weaning-specific rather than mortality-related effect. Patients with sustained ventilatory dependence had significantly higher rates of tracheostomy, percutaneous endoscopic gastrostomy (PEG) placement, and prolonged ICU stay.Conclusions: Admission GCS scores, evaluated in separate models from APACHE II and SOFA, were consistently associated with both ventilator-free days and the secondary binary outcome of sustained ventilatory dependence in patients with sepsis, and they may serve as a potential early marker of ventilatory trajectory in this population. Competing-risks analysis suggested that this association reflects an effect on the speed of weaning rather than on mortality. Given the modest sample size, internal-only validation, and the exploratory nature of several sensitivity analyses, these findings should be regarded as hypothesis-generating rather than confirmatory. A mortality-inclusive ventilator-free-days outcome alongside a single, non-overlapping severity measure may help address known methodological limitations of composite severity scores in this setting, but external validation is needed before these findings could inform individualized ventilatory management or resource allocation.