The impact of confounders on the test performance of natriuretic peptides for cardiac dysfunction in subjects aged 80 and older

Abstract

The hypothesis that natriuretic peptides could be used to identify 'pancardiac' damage has been proposed. However, multiple factors are known to influence circulating levels of natriuretic peptides, especially in the very old. Therefore, the impact of confounders on the association between natriuretic peptide levels and cardiac dysfunction was further explored in subjects aged 80 and older. A diagnostic cross-sectional study embedded within the BELFRAIL study (n=567) was performed. Baseline BNP and NT-proBNP levels were measured and echocardiograms were performed at the subject's home. Cardiac dysfunction was defined as systolic dysfunction, valvular heart disease or isolated severe diastolic dysfunction. Several functional and structural echocardiographic parameters were independently related to circulating levels of natriuretic peptides. Cystatin C, BMI, β blockers, diabetes, heart frequency, usCRP, age and sex were identified as confounders. The prevalence of cardiac dysfunction was 17.1% in the subjects without and 30.8% in the subjects with chronic atrial fibrillation (CAF) or pacemaker (PM). Only in subjects with CAF or PM the C statistic for cardiac dysfunction improved after correcting for confounders. The post-test probability for a negative test (PTP-) ranged from 3.7% to 12.2% and the PTP+ ranged from 21.9% to 62.2% in different strata of confounders. According to these data adjusting for identified confounders does not improve the diagnostic accuracy of the natriuretic peptides for cardiac dysfunction, except in subjects with CAF or PM. Stratifying for individual confounders showed that different cut-off values could be used to optimize the diagnostic characteristics of natriuretic peptides.