Early prediction of mechanical ventilation length of stay (MV-LOS) forpatients submitted to heart surgery
- 709 Downloads
KeywordsChronic Obstructive Pulmonary Disease Sofa Score Vasoactive Agent Postoperative Level Left Ventricle Systolic Function
Among clinical cardiologists, intensivists and cardiac surgeons, weaning from mechanical ventilation is usually considered a major determinant of short course outcome for patients submitted to heart surgery.
To determine factors collected preoperatively, perioperatively and within the first 24 h postoperative that could predict the MV-LOS.
We retrospectively analyzed 207 patients submitted to heart surgery from June 2000 to February 2001 at a tertiary care cardiology hospital, correlating data obtained before, during, and within first 24 h postoperative to LOS-MV. Data were treated applying linear regression and R-R2 goodness-of-fit statistics.
Under univariate analysis one could note significant predictive value of the variables age (P = 0.001), body mass index (P = 0.004), reoperation (P < 0.001), hemotransfusion (P < 0.001), pre-existing chronic obstructive pulmonary disease (P = 0.02) and diabetes mellitus (P = 0.005), preoperative (P = 0.004) levels of creatinine, immediate postoperative levels of arterial bicarbonate (P = 0.004), oxygen alveolar-arterial gradient (P = 0.002), first posoperative day (FPD) MODs (P = 0.018), and SOFA scores (P = 0.015) and pressure-adjusted heart rate (PAR; P < 0.001). Multivariate analysis (R=0.791, R2=0.625, R2 adjusted=0.594) revealed significant predictive value of the variables age (P < 0.001), PAR (P = 0.006), preoperative creatinine (P = 0.003), hemotransfusion (P < 0.001), arterial bicarbonate (P = 0.003) and reoperation (P < 0.001).
In our sample only the variables age, reoperation, pre- and postoperative creatinine levels, total hemotransfusion, and arterial bicarbonate level at immediate postoperative and PAR in FDP demonstrated significant correlation to LOS-MV. Even under univariate analysis, variables like previous left ventricle systolic function, length of extracorporeal circulation, Cleveland Clinic score, and use of vasoactive agents demonstrated no relevant correlation to MV-LOS.