BackgroundTo evaluate the predictive value of N-terminal prohormone B-type natriuretic peptide (NTproBNP) for weaning failure among patients undergoing major surgeries during spontaneous breathing trial (SBT), compared to traditional weaning parameters.MethodsThe observational cohort study retrospectively included postsurgical patients who received IMV and underwent a 2 h SBT. According to weaning outcome, NTproBNP level at initiation (NTproBNP1) and at end of 2 h SBT(NTproBNP2), the ΔNTproBNP%, RSBI and MV were compared between weaning failure and weaning success group. Multiple logistical regression and ROC curve were used to evaluate the capability of NTproBNP to predict weaning failure.ResultsOut of the 323 included postsurgical patients, 45 (13.9%) patients had failed weaning. The ΔNTproBNP% was a better predictor for weaning failure (AUC 0.744;95%CI,0.693–0.791) than NTproBNP1(AUC 0.639; 95%CI,0.580–0.694)), NTproBNP2(AUC 0.742, 95%CI,0.688–0.792) and other traditional weaning index such as RSBI (AUC 0.651; 95%CI, 0.597–0.703) and MV (AUC 0.552; 95%CI,0.496–0.607). The cutoff value of ΔNTproBNP% for predicting weaning failure was 23.3% with the sensitivity75.76% and specificity73.38%. The multiple logistic regression analysis found that ΔNTproBNP%>23.3% was an independent predictor of weaning failure.ConclusionΔNTproBNP% may be a useful marker for predict weaning failure for postsurgical patients, and it’s better to be more careful to withdraw from invasive mechanical ventilation for those postsurgical patients with ΔNTproBNP% >23.3%. The corresponding interventions to optimize cardiac function should be actively given to these patients.