Do We Need to Personalize Renal Function Assessment in the Stratification of Patients Undergoing Cardiac Surgery?

BACKGROUND: Renal dysfunction is an independent predictor of morbidity and mortality in cardiac surgery. For a better assessment of renal function, calculation of creatinine clearance (CC) may be necessary. OBJECTIVE: To objectively evaluate whether CC is a better risk predictor than serum creatinine (SC) in patients undergoing cardiac surgery. METHODS: Analysis of 3,285 patients registered in a prospective, consecutive and mandatory manner in the Sao Paulo Registry of Cardiovascular Surgery (REPLICCAR) between November 2013 and January 2015. Values of SC, CC (Cockcroft-Gault) and EuroSCORE II were obtained. Association analysis of SC and CC with morbidity and mortality was performed by calibration and discrimination tests. Independent multivariate models with SC and CC were generated by multiple logistic regression to predict morbidity and mortality following cardiac surgery. RESULTS: Despite the association between SC and mortality, it did not calibrate properly the risk groups. There was an association between CC and mortality with good calibration of risk groups. In mortality risk prediction, SC was uncalibrated with values > 1.35 mg /dL (p < 0.001). The ROC curve showed that CC is better than SC in predicting both morbidity and mortality risk. In the multivariate model without CC, SC was the only predictor of morbidity, whereas in the model without SC, CC was not only a mortality predictor, but also the only morbidity predictor. CONCLUSION: Compared with SC, CC is a better parameter of renal function in risk stratification of patients undergoing cardiac surgery.

Do We Need to Personalize Renal Function Assessment in the Stratification of Patients Undergoing Cardiac Surgery? / É Necessário Personalizar a Avaliação da Função Renal na Estratificação de Pacientes Submetidos à Cirurgia Cardíaca?

Abstract Background: Renal dysfunction is an independent predictor of morbidity and mortality in cardiac surgery. For a better assessment of renal function, calculation of creatinine clearance (CC) may be necessary. Objective: To objectively evaluate whether CC is a better risk predictor than serum creatinine (SC) in patients undergoing cardiac surgery. Methods: Analysis of 3,285 patients registered in a prospective, consecutive and mandatory manner in the Sao Paulo Registry of Cardiovascular Surgery (REPLICCAR) between November 2013 and January 2015. Values of SC, CC (Cockcroft-Gault) and EuroSCORE II were obtained. Association analysis of SC and CC with morbidity and mortality was performed by calibration and discrimination tests. Independent multivariate models with SC and CC were generated by multiple logistic regression to predict morbidity and mortality following cardiac surgery. Results: Despite the association between SC and mortality, it did not calibrate properly the risk groups. There was an association between CC and mortality with good calibration of risk groups. In mortality risk prediction, SC was uncalibrated with values > 1.35 mg /dL (p < 0.001). The ROC curve showed that CC is better than SC in predicting both morbidity and mortality risk. In the multivariate model without CC, SC was the only predictor of morbidity, whereas in the model without SC, CC was not only a mortality predictor, but also the only morbidity predictor. Conclusion: Compared with SC, CC is a better parameter of renal function in risk stratification of patients undergoing cardiac surgery.

Resumo Fundamentos: Disfunção renal é preditor independente de morbimortalidade após cirurgia cardíaca. Para uma melhor avaliação da função renal, o cálculo do clearance de creatinina (CC) pode ser necessário. Objetivo: Avaliar objetivamente se o CC é melhor que a creatinina sérica (CS) para predizer risco nos pacientes submetidos à cirurgia cardíaca. Métodos: Análise em 3285 pacientes do Registro Paulista de Cirurgia Cardiovascular (REPLICCAR) incluídos de forma prospectiva, consecutiva e mandatória entre novembro de 2013 e janeiro de 2015. Foram obtidos valores de CS, CC (Cockcroft-Gault) e do EuroSCORE II. Análise de associações da CS e do CC com morbimortalidade foi realizada mediante testes de calibração e discriminação. Por regressão logística múltipla, foram criados modelos multivariados independentes com CS e com CC para predição de risco de morbimortalidade após cirurgia cardíaca. Resultados: Apesar da associação entre a CS e morbimortalidade, essa não calibrou adequadamente os grupos de risco. Houve associação entre o CC e morbimortalidade com boa calibração dos grupos de risco. Na predição do risco de mortalidade, a CS ficou descalibrada com valores >1,35 mg/dL (p < 0,001). A curva ROC revelou que o CC é superior à CS na predição de risco de morbimortalidade. No modelo multivariado sem CC, a CS foi a única preditora de morbidade, enquanto que no modelo sem a CS, o CC foi preditor de mortalidade e o único preditor de morbidade. Conclusão: Para avaliação da função renal, o CC é superior que a CS na estratificação de risco dos pacientes submetidos a cirurgia cardíaca.