A nomogram for predicting the risk of heart failure with preserved ejection fraction

Abstract

BackgroundThis study purposed to design and establish a nomogram to predict the risk of having heart failure with preserved ejection fraction. MethodThe clinical data of 1031 patients diagnosed with heart failure (HF) in the First Affiliated Hospital of Jinan University from January 2018 to December 2022 were retrospectively analyzed, among which 618 patients were diagnosed with heart failure with preserved ejection fraction (HFpEF). Patients were randomly divided into a training set (70%, n = 722) and a validation set (30%, n = 309). The prediction model of HFpEF was established by using clinical characteristic data parameters, and the risk of having HFpEF was predicted by using a nomogram. Single-factor analysis was used to select independent risk factors (P < 0.05), and then binary logistic regression was used to screen predictive variables (P < 0.05). The discrimination ability of the model was evaluated by the ROC curve and calculating the area under the curve (AUC). In addition, the predictive ability of the established nomogram was evaluated using calibration curves and the Hosmer-Lemeshow goodness of fit test (HL test), and the clinical net benefit was evaluated using decision curve analysis (DCA). ResultsThe results of binary logistic regression analysis showed that age, gender, hypertension, coronary heart disease, glycosylated hemoglobin, serum creatinine, E/e’ septal, relative wall thickness (RWT), left ventricular mass index (LVMI) and pulmonary hypertension (PH) were independent influencing factors for the risk of having HFpEF (P < 0.05). Based on the results of logistic regression analysis, a nomogram was established and calibration curves were made. The prediction model showed that the AUC of the training dataset was 0.876 (95%CI, 0.851–0.902), and 0.837 (95%CI, 0.791–0.883) in the validation set. According to the calibration curves and HL test, the nomogram shows good calibration, and DCA shows that our model is clinically useful. ConclusionA nomogram prediction model was constructed to predict the patient's risk of having HFpEF. This prediction model indicated that the combination of creatinine, E/e’, RWT, LVMI and PH may be valuable in the diagnosis of HFpEF.