The Effects of Combining Two Postures and Two Ventilator Flow Waveforms on the Cuff Leak Test.

Abstract

The cuff leak test (CLT) has been shown to have excellent specificity and moderate sensitivity for predicting postextubation stridor (PES). However, the ventilator flow waveform and the subject position are not uniform in current clinical practice. We conducted a prospective cohort study in the respiratory ICU of the Beijing Chaoyang Hospital, Capital Medical University. Prior to extubation, 4 CLTs, combining 2 different postures and ventilator flow waveforms, were conducted, and the diagnostic performance of each test was assessed. Of the 143 included subjects, PES occurred in 13 (9.1%), and 10 (7%) subjects required re-intubation. Initially, an air leak volume of 110 mL was used as the standard to judge performance. The test that involved the square waveform and the subject in semi-recumbent position (test 4) had the best diagnostic performance, with a specificity of 80% and a sensitivity of 67% for predicting PES. After analyzing the receiver operating characteristic curve, an optimal diagnostic threshold of 116 mL for air leak volume was found to result in a specificity of 92% and a sensitivity of 63% for test 4. Additionally, when the air leak ratio of test 4 was 0.32, the area under the curve was 0.76, the specificity was 92%, and the sensitivity was 62%. In this study, performing the CLT with the subject in semi-recumbent position using the square waveform appeared to allow for the best prediction of PES.