Background: To assess the factors associated with reintubation in patients who had successfully passed a spontaneous breathing trial.
Methods: We used logistic regression and recursive partitioning analyses of prospectively collected clinical data from adults admitted to ICUs of 37 hospitals in eight countries, who had undergone invasive mechanical ventilation for > 48 h and were deemed ready for extubation.
Results: Extubation failure occurred in 121 of the 900 patients (13.4%). The logistic regression analysis identified the following associations with reintubation: rapid shallow breathing index (RSBI) [odds ratio (OR), 1.009 per unit; 95% confidence interval (CI), 1.003 to 1.015]; positive fluid balance (OR, 1.70; 95% CI, 1.15 to 2.53); and pneumonia as the reason for initiating mechanical ventilation (OR, 1.77; 95% CI, 1.10 to 2.84). The recursive partitioning analysis allowed the separation of patients into different risk groups for extubation failure: (1) RSBI of > 57 breaths/L/min and positive fluid balance (OR, 3.0; 95% CI, 1.8 to 4.8); (2) RSBI of < 57 breaths/L/min and pneumonia as reason for mechanical ventilation (OR, 2.0; 95% CI, 1.1 to 3.6); (3) RSBI of > 57 breaths/L/min and negative fluid balance (OR, 1.4; 95% CI, 0.8 to 2.5); and (4) RSBI of < 57 breaths/L/min (OR, 1 [reference value]).
Conclusions: Among routinely measured clinical variables, RSBI, positive fluid balance 24 h prior to extubation, and pneumonia at the initiation of ventilation were the best predictors of extubation failure. However, the combined predictive ability of these variables was weak.