Zhuang_2024_ESC.Heart.Fail__

BACKGROUND AND AIMS: As the incidence of heart failure (HF) increases, the need for practical tools to evaluate the long-term prognosis in these patients remains critical. Our study aimed to develop a 48 month prediction model for all-cause mortality in decompensated HF patients using available clinical indicators. METHODS: HF patients (n = 503), 60 years or older, were divided into a training cohort (n = 402) and a validation cohort (n = 101). Data on demographics, comorbidities, laboratory results and medications were gathered. Prediction models were developed using the Prognostic Nutritional Index (PNI), cholinesterase (ChE) and a multifactorial nomogram incorporating clinical variables. These models were constructed using the least absolute shrinkage and selection operator algorithm and multivariate logistic regression analysis. The performance of the model was assessed in terms of calibration, discrimination and clinical utility. RESULTS: The mean age was 77.11 +/- 8.85 years, and 216 (42.9%) were female. The multifactorial nomogram included variables of ChE, lymphocyte count, albumin, serum creatinine and N-terminal pro-brain natriuretic peptide (all P < 0.05). In the training cohort, the nomogram's C-index was 0.926 [95% confidence interval (CI) 0.896-0.950], outperforming the PNI indices at 0.883 and ChE at 0.804 (Z-tests, P < 0.05). The C-index in the validation cohort was 0.913 (Z-tests, P < 0.05). Calibration and decision curve analysis confirmed model reliability, indicating a more significant net benefit than PNI and ChE alone. CONCLUSIONS: Both the ChE- and PNI-based prediction models effectively predict the long-term prognosis in patients over 60 years of age with decompensated HF. The multifactorial nomogram model shows superior performance, improving clinical decision-making and patient outcomes.