A Simplified Risk Scoring System to Predict Mortality in Cardiovascular Intensive Care Unit

BACKGROUND: Cardiovascular intensive care unit (CICU) is an area with high mortality rates globally. The prediction of inpatients mortality risk at CICU needs a simplified scoring systems. Hence, this study aims to analyze the predictors for in-hospital mortality of patients whom hospitalized at CICU of Sardjito General Hospital Yogyakarta and to create a mortality risk score based on the results of this analysis. METHODS: Data were obtained from SCIENCE (Sardjito Cardiovascular Intensive Care) registry. Outcomes of 595 consecutive patients (mean age 59.92 ± 13.0 years) from January to November 2017 were recorded retrospectively. Demography, risk factor, comorbidities, laboratory result and other examinations were analyzed by multivariate logistic regression to create two models of scoring system (probability and cut-off model) to predict in-hospital mortality of any cause. RESULTS: A total of 595 subjects were included in this research; death was found in 55 patients (9.2%). Multiple logistic regression analysis showed some variables that became independent predictor of mortality, i.e. age ≥ 60 years, pneumonia, the use of ventilator machine, and increased of serum glutamate-pyruvate transaminase level, an increased of creatinine level and an ejection fraction < 40%. Receiver operating characteristic (ROC) curve analysis showed a cut-off model scoring system with score 3 to 9 predicting mortality compared to score 0 - 2. This model yielded sensitivity of 80% and specificity 74%. While the probability scoring system (score 0 to 9) showed that the higher the score, the higher the mortality probability (e.g. the mortality of patient with score 2 is 5.27%; while the mortality of patient with score 8 is 87.5%). CONCLUSIONS: Scoring system derived from this study can be used to predict the in-hospital mortality of patients whom hospitalized in our CICU and show a favorable sensitivity and specificity result.