Markers of in surgical intensive care unit length of stay in patients submitted to heart surgery: the intensivist point of view

Background

Postoperative management of heart surgery (HS) has been changing in the last decade. 'Fast-track strategy' has been proposed, but not for all patients. Markers of length of stay (LOS) in surgical intensive care unit (SICU) are still needed.

Objectives

To analyze data of preoperative, peroperative and postoperative period to identify factors that could be correlated to higher SICU-LOS among patients submitted to HS in our hospital.

Methods

Three hundred and fifty patients submitted to HS in our hospital, from June 2000 to November 2001 were retrospectively analyzed from prospective collected data in a Customized Database System. Applying Cox Regression with stepwise (Likelihood ratio) of 62 variables obtained preoperative, peroperative and postoperative period, we observed its correlation with higher SICU-LOS.

Results

Variables considered significant under multivariate analysis with respectively odds ratio (OR) and P value were: age (OR = 1.0, P = 0.02), type of surgery, Emergency (E) HS (OR = 0.7, P = 0.01), Urgency (U) HS (OR = 0.6, P = 0.01), need of re-operation (OR = 2.8, P = 0.0008), occurrence of acute atrial fibrillation (OR = 1.7, P = 0,0001) and following the first postoperative day (FPD): LOS in mechanical ventilation (MV) (OR = 0.9, P = 0.01), use of pulmonary artery catheter (PAC) (OR = 2.2, P = 0.00001) and intra-aortic balloon (IAB) use (OR = 1.9, P = 0.01).

Conclusion

In our series using preoperative, peroperative and postoperative variables, we found three preoperative markers of longer SICU-LOS (Age, U-HS and E-HS), three FPD occurrence (PAC, IAB, LOS-MV) that could show the severity of hemodynamic disturbance, also two postoperative markers that were HS complications (AAF and need of re-operation). The strongest variables to predict higher SICU-LOS are found in the postoperative period in our hospital and are supported by highest OR. We need further step by step (pre, per and FPD) data analyses to format a predictive SICU-LOS model in HS.