NAS2H score, a novel predictive score of 1-year all cause mortality in Acute Coronary Syndromes

Abstract Introduction In patients admitted for Acute Coronary Syndromes (ACS), mortality is influenced by several clinical and therapeutical factors, and management of these patients should be guided by an estimate of individual risk. Objective To develop a simple predictive model of 1-year mortality in patients admitted for ACS. Methods The authors present a retrospective, descriptive and correlational study including all patients admitted for ACS in a Cardiology department between the 1st of October 2010 and the 1st of October 2017. A 1-year (1y) follow-up was made through registry consultation and phone call by a Cardiologist. Patients with 1y mortality (1yM) events were studied regarding baseline demographic and clinical characteristics, risk factors and hospitalization data, and a correlational analysis with Chi-square test for categorical variables and t-Student test for continuous variables (confidence level of 95%) was performed. Independent predictors of 1yM were identified through binary logistic regression analysis, using a significance level of 0,05. A discriminatory function was applied, and the Wilks lambda test was used to determine the discriminant score for the studied groups. The authors used SPSS 24,0 for statistical analysis. Results A total of 3251 patients were included, 826 (25,4%) of which were female, with a mean age of 65,5±13,4 years. In the studied sample, 268 patients (8,2%) died in the year following hospital discharge; this group had a mean age of 65,6±13,2 years, and 80 (29,9%) were female patients. There was a significant association between 1yM and multiple clinical, therapeutical and laboratorial variables, but after multivariate analysis only age greater than 65 years old (yo) [p=0,001], previous stroke [p=0,005], haemoglobin (Hb) <10mg/dL [p<0,001], brain natriuretic peptide (BNP) >100pg/mL [p=0,001], and left ventricular ejection fraction (LVEF) <50% [p <0,001] proved to be independent predictors of the studied outcome. Using these variables, the authors developed a scoring model to predict 1yM in patients admitted for ACS with the following formula = 0,002 + (0,736 x Age >65yo) + (0,91 x previous stroke) + (2,562 x Hb <10) + (0,63 x BNP >100) - (1,207 x FEVE >50%). In this function, variables should be substituted by 1 or 0, depending on wheter they are present or not. The discrimination cutoff was 0,57, with a 70,6% sensibility and 75,9% specificity, and a discriminant power of 75,4%. Conclusion Defining the mortality risk of ACS patients after discharge represents a real challenge and demands a careful evaluation of multiple factors in an attempt to achieve an accurate estimation of risk. The authors developed a predicting model for 1yM in ACS patients, with a good discriminant power, based on simple variables. The present score will require validation in a larger cohort of ACS patients before it can be applied in a clinical context.