Early recognition and prediction of sudden heart failure (HF) is particularly important due to severe morbidity and mortality. Recently, a research article was published in Diabetes Obesity & Metabolism, an authoritative journal in the field of metabolic endocrine diseases. The purpose of the study was to predict the onset of heart failure in patients with diabetes.
A time-varying Cox model was established based on the ACCORD clinical trial to predict the risk of heart failure, defined as hospitalization for heart failure (HHF). The researchers conducted external validation with patient-level data from the Harmony Outcomes Trial and the Chronic Renal Insufficiency Cohort (CRIC) study and converted the model into an integer-based scoring algorithm with 10-year risk assessment. Researchers used a stepwise algorithm to identify and select predictors from demographic characteristics, physical examinations, laboratory results, medical history, medications, and health care utilization to develop a risk prediction model. The primary outcome of the study was incident HF, defined as HHF, and model performance was evaluated using the C statistic and Brier score.
A total of 9649 patients with diabetes did not develop HF in the study. The median follow-up time was 4 years, and a total of 299 hospitalizations for HF events occurred. The model identified several predictors of the 10-year HF incidence risk score "DM-CURE": socio-demographic (education, age at T2DM diagnosis), metabolic (glycated hemoglobin, systolic blood pressure, BMI, HDL), diabetes-related complications Risk assessment was performed on disease (myocardial infarction, revascularization, cardiovascular medications, neuropathy, duration of hypertension, proteinuria, UACR, ESKD) and health care utilization (all-cause hospitalization, emergency department visits). Among them, age at T2DM diagnosis, health care utilization, and cardiovascular disease-related variables are the strongest factors affecting future HF. The model has good discriminability (C statistic: 0.838, 95%CI: 0.821-0.855) and calibration (Brier score: 0.006, 95%CI: 0.006-0.007) for ACCORD data, and good performance for validation data (Harmony: C statistic is 0.881, 95%CI is 0.863-0.899; C statistic is 0.813, 95%CI is 0.794-0.833). The 10-year risk of developing HF increases in a graded manner, with quintile 1 (score ≤14) ≤1%, quintile 2 (score 15-23) 1-5%, and quintile 3 (score 24-23). 27) 5-10%, quintile 4 (score 28-33) 10-20%, quintile 5 (score > 33) ≥ 20%.
It can be seen that the DM-CURE model and score can be used for risk stratification of HHF in T2DM patients and are easy for clinical application.
