Background:: Cuff leak test was developed to predict the occurrence of post-extubation stridor (PES). This study evaluated the diagnostic performance of this test in unselected critically ill patients.
Methods:: Multicenter prospective study including unselected ventilated patients at the time of their first planned extubation. The diagnostic performance of 4 different cuff leak tests was assessed.
Results:: Post-extubation stridor occurred in 34 (9.4%) of 362 included patients. Compared to patients without PES, patients with PES required more frequently reintubation (6 [17.6%] vs 26 [7.9%], P = .041), prolonged duration of ventilation (6 [3-13] vs 5 [2-9] days, P = .029), and longer intensive care unit (ICU) stay (12 [6-17.5] vs 7.5 [4-13] days, P = .018). However, ICU mortality was similar in both groups (1 [2.9%] vs 23 [7.0%], P = .61). The 4 cuff leak tests display poor diagnostic accuracy: sensitivities ranging from 27% to 46%, specificities from 70% to 88%, positive predictive values from 14% to 19%, and negative predictive values from 92% to 93%.
Conclusion:: Post-extubation stridor occurs in less than 10% of unselected critically ill patients. The several cuff leak tests display limited diagnostic performance for the detection of PES. Given the high rate of false positives, routine cuff leak test may expose to undue prolonged mechanical ventilation.
Keywords: cuff leak test; endotracheal intubation; evaluation; post-extubation stridor; sensitivity.