Predictors of prolonged mechanical ventilation in a cohort of 5123 cardiac surgical patients

Background Prolonged mechanical ventilation (PMV) after heart surgery is associated with increased patient morbidity and mortality. Methods In this prospective observational cohort study the aim was to assess PMV predictors and its impact on ICU, hospital length of stay and survival in cardiac surgical patients admitted to our eight-bed ICU from January 2000 to December 2006. All perioperative patient variables were put into an electronic database. Five thousand one hundred and twenty-three patients were divided into two cohorts: early extubation, undergoing a successful extubation for 12 h or less, and delayed extubation, needing a mechanical ventilation for more than 12 h. Results A logistic regression model identified the following as PMV predictors: age more than 65 years [odds ratio (OR), 1.296; 95% confidence interval (CI), 1.017–1.069; P = 0.016], chronic renal failure (OR, 1.571; 95% CI, 1.566–2.466; P = 0.011), chronic obstructive pulmonary disease (OR, 1.453; 95% CI, 1.695–2.454; P = 0.006), redo surgery (OR, 2.010; 95% CI, 1.389– 2.114; P = 0.001), emergency surgery (OR, 1.622; 95% CI, 1.515–2.494; P = 0.016), New York Heart Association/Canadian Cardiovascular Society class higher than 2 (OR, 1.491; 95% CI, 1.704–2.321; P = 0.001), left ventricular ejection fraction of 30% or less (OR, 2.125; 95% CI, 1.379–1.991; P = 0.000), red blood cell (OR, 5.430; 95% CI, 3.636–8.130; P = 0.000) and fresh frozen plasma transfusion units more than four (OR, 3.019; 95% CI, 1.808–5.050; P = 0.000) and cardiopulmonary bypass time more than 77 min (OR, 2.030; 95% CI, 1.248–2.174; P = 0.002). Early extubation group patients showed a higher probability of being discharged from ICU to cardiac surgical ward (log-rank = 1108.951; P = 0.000) and from cardiac to rehabilitation ward (log-rank = 598.005; P = 0.000) and higher hospital survival (log-rank = 53.215; P = 0.000). Conclusion This review allowed us to assess predictors, helping us to identify ‘a priori’ patients more likely to undergo PMV.