BACKGROUND: To establish a risk score for heart surgery allows the assessment of preoperative risk, informing the patient and defining care during the intervention. OBJECTIVE: To assess preoperative risk factors for death in cardiac valve surgery and construct a simple risk model (score) for in-hospital mortality of patients candidate to surgery at Hospital São Lucas of Pontifícia Universidade Católica do Rio Grande do Sul (HSL-PUCRS). METHODS: The study sample included 1,086 adult patients that underwent cardiac valve surgery between January 1996 and December 2007 at HSL-PUCRS. Logistic regression was used to identify risk and in-hospital mortality factors. The model was developed in 699 patients and its performance was tested in the remaining data (n = 387). The final model was created using the total study sample (n = 1,086). RESULTS: Global mortality was 11.8%: 8.8% of elective cases and 63.8% of emergency cases. At the multivariate analysis, 9 variables remained independent predictors for the outcome: advanced age, surgical priority, female sex, ejection fraction < 45%, concomitant myocardial revascularization (CABG), pulmonary hypertension, NYHA functional class III or IV, creatinine levels (1.5 to 2.49 mg/dl and > 2.5 mg/dl or undergoing dialysis). The area under the ROC curve was 0.83 (95% CI: 0.78-0.86). The risk model showed good capacity for observed/predicted mortality: the Hosmer-Lemeshow test was x² = 5.61; p = 0.691 and r = 0.98 (Pearson's coefficient). CONCLUSION: The variables predictive of in-hospital mortality allowed the construction of a simplified risk score for daily practice, which classifies the patient as having low, moderate, high, very high and extremely high preoperative risk.
Probability; risk; preoperative care; thoracic surgery; heart valves
