Michel Pompeu Barros de Oliveira SáI; Marcus Villander Barros de Oliveira SáII; Ana Carla Lopes de AlbuquerqueII; Belisa Barreto Gomes da SilvaII; José Williams Muniz de SiqueiraII; Phabllo Rodrigo Santos de BritoII; Frederico Pires VasconcelosII; Ricardo de Carvalho LimaIII
OBJECTIVE: The aim of this study is to evaluate the applicability of GuaragnaSCORE for predicting mortality in patients undergoing heart valve surgery in the Division of Cardiovascular Surgery of Pronto Socorro Cardiológico de Pernambuco - PROCAPE, Recife, PE, Brazil. Methods: Retrospective study involving 491 consecutive patients operated between May/2007 and December/2010. The registers contained all the information used to calculate the score. The outcome of interest was death. Association of model factors with death (univariate analysis and multivariate logistic regression analysis), association of risk score classes with death and accuracy of the model by the area under the ROC (receiver operating characteristic) curve were calculated.
RESULTS: The incidence of death was 15.1%. The nine variables of the score were predictive of perioperative death in both univariate and multivariate analysis. We observed that the higher the risk class of the patient (low, medium, high, very high, extremely high), the greater is the incidence of postoperative AF (0%; 7.2%; 25.5%; 38.5%; 52.4%), showing that the model seems to be a good predictor of risk of postoperative death, in a statistically significant association (P <0.001). The score presented a good accuracy, since the discrimination power of the model in this study according to the ROC curve was 78.1%.
CONCLUSIONS: The Brazilian score proved to be a simple and objective index, revealing a satisfactory predictor of perioperative mortality in patients undergoing heart valve surgery at our institution.
OBJETIVO: O objetivo deste estudo é avaliar a aplicabilidade do GuaragnaSCORE na predição de mortalidade perioperatória em pacientes submetidos à cirurgia cardíaca valvar na Divisão de Cirurgia Cardiovascular do Pronto Socorro Cardiológico de Pernambuco - PROCAPE, Recife, PE, Brasil.
MÉTODOS: Estudo retrospectivo envolvendo 491 pacientes consecutivos operados entre Maio/2007 e Dezembro/2010. Os registros continham todas as informações utilizadas para calcular a pontuação. O desfecho de interesse foi óbito. A associação de fatores do escore com óbito (análise univariada e análise de regressão logística multivariada), associação de classes de risco do escore com óbito e acurácia do modelo através da área sob a curva ROC (receiver operating characteristic) foram calculados.
RESULTADOS: A incidência de óbito foi de 15,1%. As nove variáveis do escore foram preditoras de morte em análise univariada e multivariada. Observamos que, quanto maior a classe de risco do paciente (baixa, média, alta, muito alta, extremamente alta), maior é a incidência de óbito (0%; 7,2%; 25,5%; 38,5%; 52,4%), demonstrando que o modelo parece ser um bom preditor de risco de óbito, em uma associação estatisticamente significativa (P<0,001). O escore apresentou boa acurácia, levando em consideração que a área sob a curva ROC foi de 78,1%.
CONCLUSÕES: O escore brasileiro demonstrou-se um índice simples e objetivo, revelando-se um preditor satisfatório de óbito no período perioperatório em pacientes submetidos à cirurgia cardíaca valvar em nossa instituição.
CABG: coronary artery bypass graft
CI: confidence interval
CPB: cardiopulmonary bypass
EF: ejection fraction
NYHA: New York Heart Association criteria
OR: odds ratio
PAH: pulmonary arterial hypertension
ROC: curve receiver operating characteristic curve
SPSS: Statistical Package for Social Sciences
Currently, a total of 275.000 cardiac valve replacement surgeries are carried out worldwide , with operative mortality ranging from 1 to 15% [2,3]. The reported mortality in Brazil is 8.9% for heart valve surgeries, according to administrative register from DATASUS .
Guaragna et al.  recently proposed a Brazilian risk score for prediction of surgical risk after heart valve surgery - we baptized the model as GuaragnaSCORE. However, several studies show that risk prediction scores tend to have inferior performance when applied to different groups of patients which have been used to development of the original model . So the external assessment in population of patients with new data from other institutions is always important for the score has wide clinical application [7-9]. Previously, we tested EuroSCORE in coronary artery bypass graft (CABG) surgery at our institution, and this proved to be a simple and objective index, revealing a discriminating satisfactory postoperative outcome, so we showed the importance of validating risk prediction models in local institutions in order to verify its applicability .
The objective of this study is to evaluate the ability of the score of Guaragna et al.  in predicting surgical risk in our institution, specifically in the group undergoing heart valve surgery.
After approval by the ethics committee, in accordance with Resolution 196/96 (National Board of Health - Ministry of Health - Brazil) [10,11], we reviewed the records of patients undergoing consecutive isolated heart valve surgery (replacement or repair) or combined with CABG surgery at the Division of Cardiovascular Surgery of Pronto Socorro Cardiológico de Pernambuco (PROCAPE), Recife, PE, Brazil, from May 2007 to December 2010. We excluded the following: patients whose records did not contain the necessary data concerning the variables to be studied; patients undergoing surgery for tricuspid and/or pulmonary valves (when isolated, due to small number of patients undergoing these procedures); age < 18 years.
It was a retrospective study of exposed and nonexposed to certain factors (independent variables) with outcome (dependent variable) followed by assessment of a model (the score of Guaragna et al. ).
The independent variables were: gender (male/female), age (years), surgical priority (emergency/urgency surgery considered as a single variable and defined as the need to undergo surgical intervention in up to 48 hours, due to imminent risk of death or unstable clinical-hemodynamic condition), heart failure functional class according to New York Heart Association criteria (NYHA I, II, III, IV), ejection fraction (EF%, measured by echocardiography), serum creatinine (mg/dL), pulmonary arterial hypertension (PAH, detected at the echocardiogram, defined as systolic pressure in pulmonary artery > 30 mmHg according to the Brazilian Guideline of Pulmonary Arterial Hypertension of 2005), combined CABG surgery.
The dependent variable was perioperative death (considered in the transoperative period and throughout the entire hospitalization period).
Each patient was evaluated for the presence or absence of the nine risk factors established by Guaragna et al. , respecting the definition of each of them and giving them the correct score (Table 1). Depending on the final score, each patient was placed in one of the five risk groups (Table 2).
Data were analyzed using percentage and descriptive statistics measures. The following tests were used: chisquare test or Fisher's exact test (as appropriate, for nonparametric variables). In the study of univariate association between categorical variables, the values of the odds ratio (OR) and a confidence interval (CI) for this parameter with a reliability of 95% were obtained.
Multivariate analysis was adjusted to a logistic regression model to explain the proportion of patients who died that were significantly associated to the level of 5% (P <0.05) by a backward elimination procedure. The calibration of multivariate model was evaluated by the Hosmer-Lemeshow goodness-of-fit test.
The accuracy (discrimination ability of the score) was calculated using the area under the ROC curve (receiver operating characteristic curve), built on correct prediction of death (among high, very high and extremely high risk categories) and correct prediction of survival (among low and medium risk categories).
The level of significance in the decision of the statistical tests was 5%. The program used for data entry and retrieval of statistical calculations was SPSS (Statistical Package for Social Sciences) version 15.0.
Incidence of death and population characteristics
Taking into account the inclusion and exclusion criteria, we analyzed 491 patients undergoing heart valve surgery with a mean age of 44.6 ± 17.9 years, being 51.5% female. In-hospital death occurred in 15.1% (n=74) patients.
Analyzing the variables proposed in the score with the occurrence of death, we observed that all of them were significantly associated with this complication (Figure 1).
Applying a multivariate logistic regression model, associations of clinical variables of the score remained strongly associated with death (Table 3). The model was well accepted (P<0.001) and showed a degree of explanation of 88.4%. The Hosmer-Lemeshow goodness-of-fit was also well accepted (P=0.811), indicating a good model calibration.
Analysis of the score and prediction of death
The incidence of death according to the risk score classification is showed in Figure 2. We observed that the higher is the risk category, the higher is the incidence of death, in a statistically significant association (P<0.001).
Accuracy of the proposed risk score
According to the results presented in the area under the ROC curve (overall capacity of the measure used to discriminate individuals who died or survived), measured by 78.1%, the score shown a good measure to identify patients with risk of death (Figure 3).
The incidence of death in our study was 15.1%. This is 28% greater than that observed in the original study by Guaragna et al. , which was 11.8%. This is probably because our population has surplus of 17% of female, 25% of patients with NYHA class III/IV, 35.2% of left ventricular dysfunction, 301.6% of pulmonary arterial hypertension, 81.5% of creatinine 1.50-2.49 mg/dL and 213.6% of creatinine > 2.5 mg/dL, which makes our population as higher risk.
We observed that age > 60 years is an independent predictor of death in patients undergoing heart valve surgery. Almeida et al.  already demonstrated that age is associated with the occurrence of death after heart valve surgery regardless of type of prosthesis (biological or mechanics).
We also observed that female is an independent predictor of death in patients undergoing heart valve surgery. Andrade et al.  demonstrated that female gender increases by 2 times the chance of death in patients undergoing heart valve surgery (independent association).
We also observed that ejection fraction < 45% is an independent predictor of death in patients undergoing heart valve surgery. De Bacco et al.  showed that left ventricular dysfunction is associated with increased mortality in patients undergoing heart valve replacement.
We also observed that combined CABG is an independent predictor of death in patients undergoing heart valve surgery. We think it happens for two reasons. First, many studies show that combination of two or more concomitant cardiac surgeries increases the risk of death [5,8,9]. Second, the combination with CABG increases time of cardiopulmonary bypass (CPB). Many studies showed the impact of CPB in increased rates of morbidity and mortality in various situations related to cardiac surgery [15,16].
We also observed that pulmonary arterial hypertension is an independent predictor of death in patients undergoing heart valve surgery. Roques el al.  demonstrated the importance of pulmonary hypertension on outcomes during the development of EuroSCORE.
We also observed that NYHA class III/IV is an independent predictor of death in patients undergoing heart valve surgery. Despite this evidence, we must remember that it is precisely these patients (patients more symptomatic) who benefit from more aggressive strategies (surgical treatment) compared with medical therapy .
We also observed that high levels of creatinine is an independent predictor of death in patients undergoing heart valve surgery. Volkmann et al.  showed that, beyond patients with overt renal insufficiency, patients with hidden renal dysfunction (normal creatinine, but with decreased glomerular filtration rate) had a higher incidence of death after cardiac surgery, demonstrating the impact of renal disease in outcomes.
It is worth mentioning that other scores have been tested in cardiac surgery, although not specifically in heart valve surgery. Recently, Mejía et al.  tested the EuroSCORE and 2000 Parsonnet-Bernstein in Heart Institute of the University of São Paulo Clinics Hospital (InCor-USP). They showed that the similarity between both observed and expected mortality by the scores allows them to confirm that the values given by the scores to the various risk factors could be applied to their patients. Therefore, they concluded that both models were similar and adequate in predicting the mortality of patients undergoing CABG, heart valve, and associated surgeries at their institution. The GuaragnaSCORE stands out in the sense of having been created specifically for heart valve surgery.
The GuaragnaSCORE presented a good accuracy, since the discrimination power of the model in this study according to the ROC curve was 78.1%.
The risk score proposed by Guaragna et al.  seems to be a good model for prediction of death in patients undergoing heart valve surgery.
1. Rabkin E, Schoen FJ. Cardiovascular tissue engineering. Cardiovasc Pathol. 2002;11(6):305-17. [MedLine]
2. Brandão CMA. Avaliação do risco em cirurgia cardíaca valvar. In: Grinberg M, Sampaio RO, eds. Doença valvar. Barueri:Manole;2006. p.199-201.
3. Ambler G, Omar RZ, Royston P, Kinsman R, Keogh BE, Taylor KM. Generic, simple risk stratification model for heart valve surgery. Circulation. 2005;112(2):224-31. [MedLine]
4. Ribeiro AL, Gagliardi SP, Nogueira JL, Silveira LM, Colosimo EA, Lopes do Nascimento CA. Mortality related to cardiac surgery in Brazil, 2000-2003. J Thorac Cardiovasc Surg. 2006;131(4):907-9. [MedLine]
5. Guaragna JC, Bodanese LC, Bueno FL, Goldani MA. Proposed preoperative risk score for patients candidate to cardiac valve surgery. Arq Bras Cardiol. 2010;94(4):541-8. [MedLine]
6. Shahian DM, Blackstone EH, Edwards FH, Grover FL, Grunkemeier GL, Naftel DC; STS workforce on evidencebased surgery. Cardiac surgery risk models: a position article. Ann Thorac Surg. 2004;78(5):1868-77. [MedLine]
7. Sá MP, Figueira ES, Santos CA, Figueiredo OJ, Lima RO, Rueda FG, et al. Validation of MagedanzSCORE as a predictor of mediastinitis after coronary artery bypass graft surgery. Rev Bras Cir Cardiovasc. 2011;26(3):386-92. [MedLine] View article
8. Sá MP, Soares EF, Santos CA, Figueredo OJ, Lima RO, Escobar RR, et al. EuroSCORE and mortality in coronary artery bypass graft surgery at Pernambuco Cardiologic Emergency Medical Services [Pronto Socorro Cardiológico de Pernambuco]. Rev Bras Cir Cardiovasc. 2010;25(4):474-82. [MedLine]
9. Mejía OA, Lisboa LA, Puig LB, Dias RR, Dallan LA, Pomerantzeff PM, et al. The 2000 Bernstein-Parsonnet score and EuroSCORE are similar in predicting mortality at the Heart Institute, USP. Rev Bras Cir Cardiovasc. 2011;26(1):1-6. [MedLine] View article
10. Sá MP, Lima RC. Research Ethics Committee: mandatory necessity. Requirement needed. Rev Bras Cir Cardiovasc. 2010;25(3):III-IV. [MedLine]
11. Lima SG, Lima TA, Macedo LA, Sá MP, Vidal ML, Gomes AF, et al. Ethics in research with human beings: from knowledge to practice. Arq Bras Cardiol. 2010;95(3):289-94. [MedLine]
12. Almeida AS, Picon PD, Wender OC. Outcomes of patients subjected to aortic valve replacement surgery using mechanical or biological prostheses. Rev Bras Cir Cardiovasc. 2011;26(3):326-37. [MedLine] View article
13. Andrade IN, Moraes Neto FR, Oliveira JP, Silva IT, Andrade TG, Moraes CR. Assesment of the EuroSCORE as a predictor for mortality in valve cardiac surgery at the Heart Institute of Pernambuco. Rev Bras Cir Cardiovasc. 2010;25(1):11-8. [MedLine] View article
14. De Bacco MW, Sartori AP, Sant'Anna JR, Santos MF, Prates PR, Kalil RA, et al. Risk factors for hospital mortality in valve replacement with mechanical prosthesis. Rev Bras Cir Cardiovasc. 2009;24(3):334-40. [MedLine] View article
15. Anderson AJ, Barros Neto FX, Costa MA, Dantas LD, Hueb AC, Prata MF. Predictors of mortality in patients over 70 years-old undergoing CABG or valve surgery with cardiopulmonary bypass. Rev Bras Cir Cardiovasc. 2011;26(1):69-75. [MedLine] View article
16. Sá MP, Lima LP, Rueda FG, Escobar RR, Cavalcanti PE, Thé EC, et al. Comparative study between on-pump and off-pump coronary artery bypass graft in women. Rev Bras Cir Cardiovasc. 2010;25(2):238-44. [MedLine] View article
17. Roques F, Nashef SA, Michel P, Gauducheau E, de Vincentiis C, Baudet E, et al. Risk factors and outcome in European cardiac surgery: analysis of the EuroSCORE multinational database of 19030 patients. Eur J Cardiothorac Surg. 1999;15(6):816-22.
18. Kassab AK, Kassab KK. Mitral regurgitation: comparison among clinical and surgical treatment medium term in agreement with the functional class. Rev Bras Cir Cardiovasc. 2002;17(2):128-31.
19. Volkmann MA, Behr PE, Burmeister JE, Consoni PR, Kalil RA, Prates PR, et al. Hidden renal dysfunction causes increased in-hospital mortality risk after coronary artery bypass graft surgery. Rev Bras Cir Cardiovasc. 2011;26(3):319-25. [MedLine] View article
Article receive on Tuesday, November 8, 2011
This work was carried out at Division of Cardiovascular Surgery of Pronto Socorro Cardiológico de Pernambuco (PROCAPE). University of Pernambuco (UPE), Recife, PE, Brazil.