| dc.description.abstract | Objectives: To study the clinical profile of adult patients submitted to cardiac surgery supported by the Brazilian Universal Health System (SUS) in the metropolitan region of Belo Horizonte city (RMBH) through the European Risk System in Cardiac Operations (EUROSCORE) and verify if this model of cardiac surgery risk profile was predictor of operative mortality and if there was statistical differences between predicted mortality and observed mortality. Methods: Data from 405 adult patients submitted to cardiac surgeries as coronary artery bypass graft (CAGB), valve and thoracic aorta replacement and repair, congenital defects repair and combined ones since October of 2007 to July of 2008 in six of the seven authorized hospitals for cardiac surgeries by SUS in the RMBH. Data collected according to the variables of EUROSCORE were used for classify patients into scores of low, medium and high risk with their respective predicted mortality. The calibration of the model was measured by comparing the predicted mortality to the observed mortality using the Hosmer-Lemeshow Test of Goodness of Fit and ROC curves. The accuracy of the model was evaluated by means of Statistic-c and adjustment was made using a logistic regression model of operative mortality (response variable)on the EUROSCORE (explanatory variable) Data was collected into a gratuity program of data Epidata® and exported to the statistic program SPSS® version 15. Results: The total mortality was 14, 8% for adult patients submitted to cardiac surgery supported by SUS in the RMBH. This mortality was 13,6% in CAGB,13.5% in valve, 36,3% in aortic, 17,1% in combined ones and 0,6% in congenital. Both models satisfactorily discriminated outcomes. ROC curve for numeric score was 0, 6487 (95%CI 0,572-0,723), and 0,657(95%CI 0,583 -0,731) for logistic score. The hospitals were numbered from 1 to 6 had the following total mortality:1(12,9%) ,2(6,4%), 3(6,4%), 4(19,6%), 5(26,8%) and 6(12,3%).There was statistical difference between predicted and observed mortality in hospitals 3 and 5 (p=0,009 and 0,008, respectively).In multivariate analysis hospitals were reclassified into two groups: I(1,2 and 6) and II(3,4 and 5). Total mortality was 10, 2% and 24, 6%, respectively (p<0, 001). OR for group II hospital variable was 3,4 and for numeric score was 1,26. The model of municipal regulation system was not related to increased mortality in this study. Conclusions: EUROSCORE was predictor of mortality, not only in CRVM, but even in other cardiac surgeries in this study. Both total and CAGBs mortality were higher than expected compared to another nationals and internationals results. The risk scores was really predictor of operative mortality, but did not explain the poor results, neither the difference in predicted and observed mortality found in hospitals 3 and 5. The lethalitys risk in the group IIs hospitals was more than three times that it was in group I. Its not possible to attribute the higher lethality to the municipal regulation system of health in this study. Keywords: EUROSCORE, cardiac surgery, operative mortality, health care benchmarks, predicted mortality, observed mortality. | |
