Derivation and validation of a clinical predictive model of NT-proBNP ≥125 pg/mL to detect pre-heart failure

Abstract

BackgroundSeveral guidelines recommend the measurement of N-terminal pro-B-type natriuretic peptide (NT-proBNP) to diagnose heart failure (HF); however, no screening criteria for measuring NT-proBNP in asymptomatic patients exist. We develop/validate a clinical prediction model for elevated NT-proBNP to support clinical outpatient decision-making. MethodsIn this multicenter cohort study, we used a derivation cohort (24 facilities) from 2017 to 2021 and a validation cohort at one facility from 2020 to 2021. Patients were aged ≥65 years with at least one risk factor of HF. The primary endpoint was NT-proBNP ≥125 pg/mL. The final model was selected using backward stepwise logistic regression analysis. Diagnostic performance was evaluated for sensitivity and specificity, the area under the curve (AUC), and calibration. In total, 1645 patients (derivation cohort, n = 837; validation cohort, n = 808) were included, of whom 378 (23.0 %) had NT-proBNP ≥125 pg/mL. Body mass index, age, systolic blood pressure, estimated glomerular filtration rate, cardiothoracic ratio, and heart disease were used as predictors and aggregated into a BASE-CH score of 0–11 points. ResultsInternal validation resulted in an AUC of 0.74 and an external validation AUC of 0.70. ConclusionsBased on available clinical and laboratory variables, we developed and validated a new risk score to predict NT-proBNP ≥125 pg/mL in patients at risk for HF or with pre-HF.