Real-world patterns of NT-proBNP ordering by heart failure status: determinants in a 723-patient cohort

Background

NT-proBNP is recommended for heart failure (HF) diagnosis and risk stratification. However, its use in routine care may vary by clinical context.

Purpose

To identify factors associated with NT-proBNP testing in hospitalized patients admitted for cardiovascular reasons, stratified by the presence or absence of diagnosed HF.

Methods

In this single-center study, we analyzed 723 patients hospitalized for cardiovascular reasons (including uncontrolled hypertension) and stratified them into diagnosed HF (n=421) and no HF (n=302). NT-proBNP was measured in 234 (32.4%) patients overall: 149/421 (35.4%) with diagnosed HF and 85/302 (28.1%) without HF. Within each stratum, logistic regression was performed with NT-proBNP testing (yes/no) as the dependent variable. Univariate and multivariable models included demographics and admission measurements of blood pressure (BP), heart rate, left ventricular ejection fraction (LVEF), eGFR <60 mL/min/1.73 m², HbA1c, and established cardiovascular disease (CVD) (excluding hypertension-only diagnosis).

Results

In patients with diagnosed HF, NT-proBNP testing was independently associated with lower systolic BP (OR 0.98 per 1 mmHg; p=0.001), higher diastolic BP (OR 1.05 per 1 mmHg; p<0.001), higher LVEF (OR 1.04 per 1%; p<0.001), and higher HbA1c (OR 1.30 per 1%; p<0.001), while established CVD was inversely associated with testing (OR 0.57; p=0.035). Model performance: Nagelkerke R²=0.147, accuracy 71.2%. Age, sex, heart rate, and eGFR <60 mL/min/1.73 m² were not independent determinants (all p>0.05) (Fig. 1A).

In patients without diagnosed HF, NT-proBNP testing was independently associated with higher diastolic BP (OR 1.08 per 1 mmHg; p=0.001), higher LVEF (OR 1.16 per 1%; p<0.001), and higher HbA1c (OR 1.23 per 1%; p=0.027), whereas established CVD (OR 0.46; p=0.018) and eGFR <60 mL/min/1.73 m² (OR 0.45; p=0.035) were inversely associated; model performance: Nagelkerke R²=0.306, accuracy 77.3%. Age, sex, systolic BP, and heart rate were not independent determinants (all p>0.05) (Fig. 1B).

Conclusions

NT-proBNP ordering follows distinct real-world patterns by HF status. In patients with diagnosed HF, testing appears selective and is associated with hemodynamic vulnerability, higher HbA1c, and more frequent testing at higher LVEF, consistent with using NT-proBNP to support the diagnosis of HF with preserved ejection fraction and to address diagnostic uncertainty in less overt systolic dysfunction. In patients without HF, testing is associated with higher diastolic BP, higher LVEF, and higher HbA1c, yet is paradoxically less frequent in those with established CVD and impaired renal function, suggesting clinical selection and missed opportunities for risk refinement. These findings support protocolized, risk-guided NT-proBNP use tailored to HF status.Fig. 1For image description, please refer to the figure legend and surrounding text.