AHFM score, a predictive model of in-hospital and long-term mortality in heart failure. (14th October 2021)

• Record Type: Journal Article
• Authors: Santos, J; Pires, I; Neto, V; Correia, J; Goncalves, L; Almeida, I; Correia, E
• Is Part Of: European heart journal. Volume 42(2021)Supplement 1
• Publication Date: 2021-10-14
• Subjects: Epidemiology, Prognosis, Outcome; Cardiology -- Periodicals; Heart -- Diseases -- Periodicals; 616.12005
• DOI: 10.1093/eurheartj/ehab724.0810
• Languages: English
• ISSNs: 0195-668X

Abstract

Abstract: Introduction: Patients hospitalized due to heart failure (HF) constitute a heterogeneous population whose prognosis is difficult to forecast. The purpose of this study was to create a model based on simple bedside recordable echocardiographic, analytical and objective clinical parameters that could accurately predict mortality and/or rehospitalization risk in different stages of HF course. Methods: A retrospective analysis of 347 patients admitted to a Cardiology ward due to decompensated HF was performed. The echocardiographic variables pulmonary artery systolic pressure (PSAP) and E/e' ratio, and the analytical/clinical variables systolic blood pressure (SBP), urea and brain natriuretic peptide (BNP) were selected for inclusion. Subgroups were created for each variable and an odds ratio (OR) for the risk of in-hospital mortality (IHM) was calculated. A numerical value proportional to the OR was attributed to each subgroup. A score was created, ranging from 0–47 points, corresponding to the sum of the classification attributed to each variable. ROC curve analysis was used to assess predictive value of the score for IHM. Kaplan-Meyer and Cox-regression plots were used to assess mortality (24MM) and the composite endpoint of HF rehospitalization or death at 24 months (24HM). Results: Mean patient age was 78 (±9) years; 51% were men. Score variable means were - PSAP: 47 (±15) mmHg; E/e': 16.8 (±7.8); SPB: 138 (±31) mmHg; Urea: 71 (±35) mg/dl; BNP: 911 (±995) pg/ml. Mean ejection fraction (EF) was 48% (±16). 35% of patients had EF<40%. IHM, 24MM and 24HM were 3.5%, 17.1% and 63.6%, respectively. A statistically significant association between IHM and PSAP, E/e', BNP, urea and SBP (p<0.05) was found on univariate analysis. ROC curve analysis of AHFM revealed an AUC of 0.785 (p=0.001) for IHM risk prediction. The cut-off point with most sensitivity (S) and specificity (E) obtained using the Youden index (0.4246) was 18 (S≈75%; E≈67%), associated with significant difference in IHM (1.3% vs 7.6%). IHM by score interval was 1.3%, 3.1% and 25%, respectively, for the intervals 0–18, 19–29 and ≥30. ECHO-AHF score <13 predicted in-hospital survival in all patients. Kaplan Meyer survival analysis by subgroup revealed significant differences in 24MM according to AHFM risk category (13.8% vs 21.9% vs 30.8%, respectively, χ 2= 17.217 p<0.001), but not for 24 MH. Cox-regression analysis demonstrated that AHFM score remained a significant independent predictor of 24MM (HR: 1.067, p=0.05), even after adjustment for other variables, such as coronary disease, chronic kidney disease, atrial fibrillation, EF and diabetes. Conclusion: AHFM score is an accurate model for predicting IHM and long-term risk of HF death. Its use may help to identify patients with high risk of mortality, in need of specialized care, and those with lower risk of death, who might be candidates for early discharge or lenient follow-up. Funding Acknowledgement: Type of funding sources: None.