Development and validation of gender specific risk models in prediction of mortality for hospitalized heart failure patients in a multiethnic asian population

Abstract

Abstract Funding Acknowledgements Type of funding sources: None. Background/Introduction Heart failure carries substantial morbidity and mortality. Female has different characteristics compared to male which may affect prognosis and are not represented well in many trials. Multiple risk scores for in-hospital mortality have been created and validated such as ADHERE (Acute Decompensated Heart Failure National Registry) and GWTG (Get With The Guidelines). However, these risk scores were in Western population and not gender specific. Purpose To develop and validate a gender specific risk model in predicting mortality amongst Asian patients admitted for acute decompensated heart failure (ADHF). Methods We analyzed data from our national centre’s ADHF registry. Epidemiological, clinical, laboratory and imaging variables were ascertained. Univariate and multivariate analysis using forward stepwise logistic regression were performed to identify predictors for all-cause in hospital mortality with emphasis on disparities between male and female. The accuracy of risk score was assessed using the concordance statistics while calibration was done using the Hosmer-Lemeshow method. The validity of the risk score was determined using separate gender validation cohorts. Results A total of 10148 patients admitted for ADHF were analyzed. 8262 patients admitted for ADHF from January 2009 to December 2018 (6008 males and 2254 females) were in the derivation cohort where as 1886 patients (1304 males and 582 females) were selected randomly from January 2019 to July 2021 for our validation cohort. Female had better survival (p value =0.04). There were 8 predictive variables for female cohort and 10 predictive variables for male cohort. Similar variables were low systolic blood pressure, loop diuretic use, dialysis, mechanical ventilation and CPR (cardiopulmonary resuscitation). Interestingly, there were significant differences. For female, other variables that had impact on mortality were hypertension (AOR, 0.23; 95% CI,0.09–0.58), high urea (AOR, 3.32; 95% CI, 1.36–8.09) and hyponatremia (AOR, 2.87, 95% CI, 1.21–6.78). However for male, the variables were renal insufficiency (AOR, 1.749; 95% CI, 1.13–2.72), history of stroke (AOR, 2.386, 95% CI, 1.22–4.65), LVEF less than 40% (AOR, 2.81; 95% CI, 1.40–5.63), absence of ARB/ACEi (AOR, 2.31; 95% CI, 1.48–3.60) and absence of beta blockers (AOR, 2.65; 95% CI, 1.73–4.05). The AUC for the female derivation cohort was 0.97 (95% CI, 0.95–0.99) and 0.94 (95% CI, 0.92–0.96) for male derivation cohort. AUC in the female validation cohort was 0.91 (95% CI, 0.83-0.98) and male was 0.90 (95% CI, 0.84–0.97). The risk scores showed good calibration (female; x²=8.88, p = 0.18 and male; x²=13.11, p = 0.07). Conclusion A gender specific risk score for ADHF was developed and validated successfully in our Asian population. This will change our practice in providing valuable prognostic information in both female and male thereby guiding the need for more intensive treatment. Abstract Figure. ROC Curves (Female and Male) Abstract Figure. Key Variables (Female and Male)