Purpose: A model for predicting the prognosis of patients with heart failure with reduced left ventricular ejection fraction (HFrEF) is currently not available. This study aimed to develop an age-biomarker-clinical history prognostic index (ABC-PI) and validate it for the assessment of individual prognosis.
Patients and methods: A total of 5,974 HFrEF patients were enrolled and 1,529 were included in this study after excluding missing values and loss to follow-up. Variables that significantly contributed to prediction of all-cause mortality were assessed by Cox regression and latent trait analysis (LTA) was used to validate discrimination of variables.
Results: After Cox regression, the following seven most significant variables were selected: age, N-terminal pro-B-type natriuretic peptide, renal dysfunction, left ventricular mass index, percutaneous coronary intervention, atrial fibrillation, and New York Heart Association (C-index: 0.801 ± 0.013). After verification by LTA, discrimination of these seven variables was proven. A nomogram was used to form the ABC-PI, and then the total score was set to 100 points. A lower score indicated a higher risk. After verification, the 3-year mortality rate was 34.7% in the high-risk group and only 2.6% in the low-risk group.
Conclusion: Our novel ABC-PI shows a good performance and does not require re-input in the original model. The ABC-PI can be used to effectively and practically predict the prognosis of HFrEF patients.
Keywords: HFrEF; NT-proBNP; latent trait analysis; nomogram.
© 2020 Hao Li et al., published by De Gruyter.