A Clinical Prediction Model for Selective NT‐proBNP Testing in Hypertension

ABSTRACT

N‐terminal pro‐B‐type natriuretic peptide (NT‐proBNP) testing is recommended for pre‐heart failure (pre‐HF) screening. However, universal NT‐proBNP testing is infeasible and inefficient. We analyzed data from 4642 U.S. adults with hypertension in NHANES 1999–2004. Heart stress, a biomarker‐defined pre‐HF state, was defined using age‐specific NT‐proBNP cutoffs. Multivariable logistic regression was used to identify factors associated with prevalent heart stress, and a point‐based score was constructed. Model discrimination and calibration were internally validated using 1000 bootstrapped samples. Heart stress was present in 31% of individuals with hypertension. Factors associated with heart stress included older age, current smoking, prior coronary heart disease/myocardial infarction, higher systolic blood pressure, lower diastolic blood pressure, and reduced estimated glomerular filtration rate. The derivation model showed acceptable discrimination (AUC 0.71; 95% CI 0.70–0.73) and good calibration. An optimal cutoff of 6 points was identified using the Youden index and was internally validated (AUC 0.71; 95% CI 0.69–0.74). Model performance was consistent across race/ethnicity and body mass index subgroups, with relatively lower discrimination in females than in males (AUC 0.66 vs. 0.78). When stratified by a predicted heart stress probability threshold of 50%, individuals in the higher‐probability group had higher cumulative risk of all‐cause (hazard ratio 3.90; 95% CI 3.55–4.29) and cardiovascular mortality (hazard ratio 5.18; 95% CI 4.44–6.04) than those in the lower‐probability group. These findings suggest that a simple model using readily obtainable clinical variables can help identify individuals with hypertension who are likely to have prevalent heart stress, offering a practical strategy to guide selective NT‐proBNP testing.