AUTHOR=Bansal Vikas , Smischney Nathan J. , Kashyap Rahul , Li Zhuo , Marquez Alberto , Diedrich Daniel A. , Siegel Jason L. , Sen Ayan , Tomlinson Amanda D. , Venegas-Borsellino Carla P. , Freeman William David TITLE=Reintubation Summation Calculation: A Predictive Score for Extubation Failure in Critically Ill Patients JOURNAL=Frontiers in Medicine VOLUME=Volume 8 - 2021 YEAR=2022 URL=https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2021.789440 DOI=10.3389/fmed.2021.789440 ISSN=2296-858X ABSTRACT=Objective: To derive and validate a multivariate risk score for the prediction of respiratory failure after extubation. Patients and Methods: We performed a retrospective cohort study of adult patients admitted to the intensive care unit from January 1, 2006, to December 31, 2015, who received mechanical ventilation for ≥48 hours. Extubation failure was defined as the need for reintubation within 72 hours after extubation. Multivariate logistic regression model coefficient estimates generated the Reintubation Summation Calculation (RISC) score. Results: The 6,161 included patients were randomly divided into 2 sets: derivation (n=3,080) and validation (n=3,081). Predictors of extubation failure in the derivation set included body mass index <18.5 kg/m2 (odds ratio [OR], 1.91; 95% CI, 1.12-3.26; P=.02), threshold of Glasgow Coma Scale of at least 10 (OR, 1.68; 95% CI, 1.31-2.16; P<.001), mean airway pressure at 1 minute of spontaneous breathing trial <10 cmH2O (OR, 2.11; 95% CI, 1.68-2.66; P<.001), fluid balance ≥1,500 mL 24 hours preceding extubation (OR, 2.36; 95% CI, 1.87-2.96; P<.001), and total mechanical ventilation days ≥5 (OR, 3.94; 95% CI 3.04-5.11; P<.001). The C-index for the derivation and validation sets were 0.72 (95% CI, 0.70-0.75) and 0.72 (95% CI, 0.69-0.75). Multivariate logistic regression demonstrated that an increase of 1 in RISC score increased odds of extubation failure 1.6-fold (OR, 1.58; 95% CI, 1.47-1.69; P<.001). Conclusion: RISC predicts extubation failure in mechanically ventilated patients in the intensive care unit using several clinically relevant variables available in the electronic medical record but requires a larger validation cohort before widespread clinical implementation.