Dividing this product by the f/Vt ratio, we can detect those selleck chem patients who will or will not be able to maintain unassisted breathing. Cst,rs and SaO2 are generally directly proportional and inversely proportional to f/Vt ratio, as Cst,rs and SaO2 gets higher, f/Vt possibly gets lower. The higher Cst,rs and SaO2, the lower f/Vt ratio and IWI tends to be higher.In order to evaluate the predictive performance of IWI, we compared it with the f/Vt ratio, which has been shown to be the most accurate predictor of failure and success in weaning from mechanical ventilation in the study by Yang and Tobin [11].We also compared IWI with the following parameters: PaO2/FiO2 ratio, which represents an important index to evaluate oxygenation; tracheal P 0.

1, which is an estimate of neuromuscular drive and is considered an important indicator of successful weaning, mainly in patients with chronic obstructive pulmonary disease (COPD) [13]; and the product of tracheal P 0.1 and f/Vt ratio (P 0.1 �� f/Vt), which has shown more specificity compared with its components [14,17], f, Vt and Cst,rs.The study was divided into two sets: the first set was derived from the data concerning 115 patients. In this phase, data were used to select the cut-off value for weaning parameters. The selected values were those that resulted in the fewest false classifications. The second set was derived from the data concerning the other 216 patients.Statistical analysisContinuous variables were presented as mean and standard deviation, categorical variables as frequencies and percentages.

Student’s t test was used to compare parametric variables and Mann-Whitney test to compare non-parametric ones.P values less than 0.05 were considered significant. The statistical analysis was performed using SAS software (SAS software package, version 9.0; SAS Institute Inc, Cary, NC, USA).Sensitivity (SE = true positive/true positive + false negative), specificity (SP = true negative/true negative + false positive), positive predictive value (PPV = true positive/true positive + false positive), negative predictive value (NPV = true negative/true negative + false negative) and diagnostic accuracy (DA) = (true positive + true negative)/(true positive + true negative + false positive + false negative) were used to evaluate each index.

The predictive performance of each index was also evaluated by calculating the area under the receiver operator characteristic (ROC) curves [11,18]. The area under the ROC curves for each index was calculated by the nonparametric method of Hanley and McNeil [19] Carfilzomib and compared through a technique developed by the same authors [19]. Classified according to the guideline proposed by Swets [20]: area under the curve of 0.5 is a non-informative result; area under the curve of more than 0.5 and 0.7 or less is less accurate; area under the curve of more than 0.7 and 0.