Reintubation in the ICU, a non-solved problem. What can a decade of automated data show us?: Reintubación en UCI, un problema sin resolver. ¿Qué nos puede decir una década de datos automatizados?

Abstract

Objective: To identify risk factors for reintubation in critically ill patients after planned extubation using clinical data automatically extracted from the clinical information system.

Design: A retrospective observational study was conducted.

Setting: Intensive Care Unit (ICU) of Joan XXIII University Hospital (Tarragona, Spain) (2014-2024).

Patients: Adult patients admitted to the ICU who required invasive mechanical ventilation (IMV) for more than 48 h.

Interventions: Automated extraction of clinical, ventilatory, hemodynamic, and laboratory test data (Centricity Critical Care®). Median, maximum, and minimum values were calculated within predefined time windows of 6 and 24 h prior to extubation.

Main variables of interest: Reintubation rate, pressure support (PS), heart rate (HR), prior administration of antibiotics, ICU stay, incidence of ventilator-associated pneumonia (VAP), and mortality.

Results: The overall reintubation rate was 11%. Multivariate analysis showed that higher mean PS (OR: 1.36; 95%CI: 1.10-1.67) and higher HR (OR: 1.22; 95%CI: 1.00-1.49) in the previous 6 h were independent predictors of reintubation. Antibiotic administration in the previous 24 h showed a protective effect (OR: 0.55; 95%CI: 0.37-0.82). The model achieved an AUC of 0.63 (95%CI, bootstrap: 0.52-0.72). Reintubated patients had a longer ICU stay, a higher incidence of VAP, and greater mortality.

Conclusions: Higher heart rate and pressure support values before extubation were associated with an increased risk of reintubation, whereas antibiotic therapy within 24 h before extubation appeared to have a protective effect. The automated analysis of longitudinal data obtained from electronic health records allowed the identification of clinically plausible associations related to reintubation.