B-Type Natriuretic Peptide Levels and Mortality in Patients With and Without Heart Failure

Abstract

Background: Circulating B-type natriuretic peptide (BNP) concentrations strongly predict mortality in patients with heart failure (HF). Both cardiac and extracardiac stimuli influence BNP levels, suggesting that BNP might have similar prognostic value in patients without HF.

Objectives: The aim of this study was to compare the prognostic value of BNP between patients with and those without HF.

Methods: Using the Vanderbilt University Medical Center electronic health record, 30,487 patients (median age 63 years, 50% men, 17% black, 38% with HF) who had a first plasma BNP measurement between 2002 and 2013, with follow-up through 2015, were studied. The risk for death according to BNP level was quantified using multivariate Cox proportional hazards models.

Results: BNP levels were lower in patients without HF (median 89 pg/ml; interquartile range: 34 to 238 pg/ml) compared with those with HF (median 388 pg/ml; interquartile range: 150 to 940 pg/ml) (p < 0.0001). Over 90,898 person-years of follow-up, 5,903 patients without HF (31%) and 6,181 patients with HF (53%) died. In multivariate models including demographic and clinical characteristics, BNP and age were the strongest predictors of death in both patients with and those without HF. In acute care settings and even among outpatients with modestly elevated BNP, the risk for death according to BNP was similar between patients with and those without HF. For instance, a BNP level of 400 pg/ml was associated with a 3-year risk for death of 21% (95% confidence interval: 20% to 23%) and 19% (95% confidence interval: 17% to 20%) in patients with and those without HF, respectively.

Conclusions: Among patients without HF, plasma BNP level is a stronger predictor of death than traditional risk factors. The risk for death associated with any given BNP level is similar between patients with and those without HF, particularly in the acute care setting.

Keywords: death; electronic health record; prognosis; retrospective.

Figure 1. Clinical correlates of plasma B-type natriuretic peptide levels in clinically evaluated adult patients according to heart failure status

Panels A and C demonstrate the relative contribution of each of the variables (y-axis) to plasma BNP levels, ranked according to their Chi-squared values (x-axis) from the multivariable-adjusted model. The Chi-squared value is corrected for the degrees of freedom allocated to that covariate in the model, so that the relative contribution for each of the covariates can be compared on the same scale. For example, among patients without heart failure, age, body mass index (BMI), and left ventricular (LV) mass are most strongly associated with BNP levels. Panels B and D show partial effect plots for the associations between selected clinical characteristics and the log odds of higher or lower circulating B-type natriuretic peptide levels. All of the factors listed were included in a multivariable-adjusted ordinal regression model. eGFR = estimated glomerular filtration rate, LVEF = left ventricular ejection fraction.

Figure 2. Ranking of the relative strength of association between demographic and clinical factors with the risk of death according to heart failure status among adult patients with a first B-type natriuretic peptide measure

The panels display the relative contribution of clinical factors (y-axis) and risk of death ranked according to their Chi-squared values from the multivariable-adjusted Cox model. The Chi-squared value is corrected for the degrees of freedom allocated to that covariate, so that the relative contribution for each of the covariates can be compared on the same scale. For example, among patients without heart failure, B-type natriuretic peptide (BNP) is the strongest determinant of the risk of death. All of the factors listed were included in a multivariable-adjusted Cox proportional hazards model. BMI = body mass index, eGFR = estimated glomerular filtration rate, LV = left ventricular.

Central Illustration. Estimated 3-year death probability across the spectrum of plasma B-type natriuretic peptide levels stratified by the presence or absence of heart failure

Three-year probability of death (95% confidence intervals) according to plasma B-type natriuretic peptide (BNP) level and stratified by heart failure status. Cox model adjusted for the following factors (plot settings): age (63 years), race (white), sex (female), CAD (yes), CKD (no), HTN (yes), DM (no), heart rate (83 bpm), pulse pressure (55 mmHg), BMI (29.2 kg/m²), eGFR (68 ml/min/1.73m²), glucose (107 mg/dl), LVEF (55%), LV mass (199 g), and BNP assay (first). Chi-square = 9.07, degrees of freedom = 2, and interaction by heart failure status p = 0.011. HF = heart failure.