High impact of Sepsis-related Organ Failure Assessment score on mortality prediction of patients undergoing valve replacement on the first postoperative day

Background

International prognostic models for surgical cardiac valve replacement (SCVR) are rare, elaborated by Edwards and colleagues as the most recent and important score [1]. However, that score was based on patients with North American demographic and epidemiological characteristics. Most scores for SCVR consider only preoperative variables.

Objectives

To create a predictive score for inhospital mortality in patients undergoing SCVR and admitted to a public (A) and a private (B) surgical intensive care unit, considering preoperative, perioperative, and first postoperative day variables.

Case series and methods

A classical cohort with data consecutively collected from June 2000 (group B, 121 patients) and January 2001 (group A, 326 patients) to February 2003. All 46 variables were previously defined according to the major prognostic indices in the literature. The statistical analysis comprised univariate analysis with the chi-square test, the Student t test, the Mann–Whitney test and the Pearson test, followed by logistic regression and stepwise (likelihood ratio) analysis, with the linear trend test and receiver operating characteristic curve.

Results

The score created, shown in Table 1, provides the following risk prediction: 0–4, low risk; 5–8, medium risk; and 9–13, high risk. The results had significance (P < 0.0001) and a linear trend (P < 0.0001). The area under the receiver operating characteristic curve was 0.78.

Table 1

Conclusions

This prognostic score shows the strength of first postoperative day variables, as does the Sepsis-related Organ Failure Assessment score [2], and the need for high doses of amines. The combined valvular surgery was the only perioperative marker. The 50-year cutoff point for age shows the precocity of valve disease in our country.