Aims: We examined strategies to improve the positive predictive value of natriuretic peptides in screening for undiagnosed left ventricular systolic dysfunction (LVSD) in the community.
Methods and results: The value of B-type(BNP), N-terminal proB-type (N-BNP) and N-terminal proAtrial(N-ANP) natriuretic peptides was prospectively assessed in 1360 subjects (45-80 years) together with echocardiography and electrocardiography. Seventeen individuals had definite and 13 had borderline, LVSD. Receiver-operating-characteristic (ROC) curve analysis showed the superiority of BNP (ROC areas 0.942 for definite LVSD, P<0.03; 0.934 for borderline LVSD, P<0.003) compared to N-BNP or N-ANP. Peptide levels, major ECG abnormality and ischaemic heart disease (IHD) history were independent predictors of LVSD. Logistic regression modelling incorporating these factors improved ROC areas for all natriuretic peptides. The specificity of all natriuretic peptides is enhanced by consideration of these factors.
Conclusions: In population screening for definite LVSD, consideration of plasma natriuretic peptide levels together with the presence of major ECG abnormalities and IHD history reduces by a factor of six (in comparison to consideration of plasma natriuretic peptide levels in isolation) the number of subjects requiring echocardiography to detect one case of LVSD (for BNP, 44 falling to seven). Similar improvements were evident for N-ANP and N-BNP. Inclusion of major ECG abnormalities and IHD history improves the performance of any natriuretic peptide used in screening programmes for ruling in undiagnosed LVSD.