Reduced breathing variability as a predictor of unsuccessful patient separation from mechanical ventilation

Objectives. To compare descriptors of the breath-to-breath respiratory variability during a 60-min spontaneous breathing trial in patients successfully and unsuccessfully separated from the ventilator and the endotracheal tube and to assess the usefulness of these predictors in discriminating these two categories of patients. Design: Prospective observational study. Setting: Four general intensive care units in university hospitals. Patients: A total of 51 consecutive patients mechanically ventilated for >24 hrs. Interventions: None. Measurements and Main Results: Tidal volume, respiratory period, inspiratory time, expiratory time, mean inspiratory flow (tidal volume/inspiratory time), and duty cycle (inspiratory time/ respiratory period) were obtained from the flow signal. Breath-by-breath variability was expressed in terms of their coefficients of variation (CV), the number of breaths among which a significant correlation was found (lag), and the autocorrelation coefficient between one breath and the following one. Five patients were excluded because of nonstationarity of the data, leaving 46 cases for analysis. Between-group comparison was conducted with the Mann-Whitney test, and a nonparametric classification and regression tree was used to identify variables discriminating success (n = 32) and failure patients (n = 14). All coefficients of variation were significantly higher in success patients, who also exhibited significantly less respiratory autocorrelation (shorter short memory). The classification and regression tree analysis allocated all success patients to a group defined by a coefficient of variation of tidal volume/inspiratory time of >= 19% and a coefficient of variation of inspiratory time/respiratory period of >= 10% that did not contain any failure patient. All failure patients belonged to a group with coefficient of variation of tidal volume/inspiratory time of < 19%, a lag tidal volume of >= 11, and that contained no success patient. Conclusions. In intensive care unit patients undergoing a spontaneous breathing trial, breathing variability is greater in patients successfully separated from the ventilator and the endotracheal tube. Variability indices are sufficient to separate success from failure cases.