A nomogram for risk prediction in patients with heart failure and diabetes: Development and validation

Heart failure (HF) commonly coexists with type 2 diabetes mellitus (T2DM), and this combination is linked to a heavier symptom burden and less favorable clinical outcomes. In this retrospective single-center study, a total of 958 consecutive patients were included, among whom 453 had T2DM, with a mean age of 68.23 ± 5.76 years. The diagnosis of HF was confirmed by a multidisciplinary team in accordance with the European Society of Cardiology criteria, and 232 patients were found to have both T2DM and HF. Relative to diabetic patients without HF, those with HF more often presented with dyspnea or fatigue, paroxysmal nocturnal dyspnea/orthopnea, and ankle swelling or nocturia (all P  < .001). They also showed higher rates of wheezing/rhonchi ( P  = .021), fluid and sodium retention ( P  = .008), ST–T abnormalities ( P  = .033), abnormal Q-waves ( P  = .001), and NT-proBNP levels ≥ 15 pmol/L (approximately 125 pg/mL; P  < .001). Potential predictors were first selected using least absolute shrinkage and selection operator (LASSO) regression, after which multivariable logistic regression was performed to construct a nomogram for predicting HF risk in patients with T2DM. The multivariable model indicated that ST–T abnormalities, NT-proBNP, ischemic heart disease, and atrial fibrillation were independently related to HF (all P  < .05). The nomogram exhibited strong apparent discriminatory ability together with satisfactory calibration, suggesting that NT-proBNP-based risk stratification may be useful for HF screening in individuals with T2DM.