Measuring brain natriuretic peptide in suspected left ventricular systolic dysfunction in general practice: cross-sectional study

Participants, methods, and results

General practitioners were invited to refer patients with suspected heart failure to our clinic. The results of transthoracic echocardiography were reported by a single, experienced observer (IA). Ischaemia was diagnosed if Q waves, bundle branch block, T wave inversions, or left ventricular hypertrophy were present on an electrocardiogram. Evidence of heart failure on a chest radiograph was defined as the presence of pulmonary oedema or cardiomegaly. Concentrations of brain natriuretic peptide were measured by immunoradiometric assay (Shionoria assay, Shionogi, Osaka, Japan) of plasma stored at −70°C. A concentration >17.9 pg/ml was considered abnormal based on the results of a large study of left ventricular systolic dysfunction.3

Altogether, 126 patients (68 men) with a mean age of 74.4 (SD 8.9) years were included in the study. Concentrations of the peptide were raised in the 40 patients with left ventricular systolic dysfunction (median concentration 79.4 pg/ml, interquartile range 35.9-151.0) compared with those with normal ventricular systolic function (26.7 pg/ml, 12.2-54.3; P<0.001). A concentration >17.9 pg/ml had a sensitivity of 88% and specificity of 34%. Choosing different cut points did not improve the predictive characteristics: at 10 pg/ml sensitivity was 92% but specificity was 18%, and at 76 pg/ml sensitivity was 66% and specificity 87%.

The prior probability that a disease exists (its prevalence) and the extent to which a test result alters the chance of the disease existing determine whether further investigation is needed; this is the likelihood ratio of positive and negative tests. In the case of heart failure it is unlikely that a single positive test result will remove the need for further cardiac imaging before treatment is started. In contrast, a negative result may give a low posterior probability of disease so that further investigations are unnecessary.

The prevalence (or prior probability) of left ventricular systolic dysfunction in this study was 32%; this is consistent with that reported in other studies.4 The likelihood ratio for a patient without a history of myocardial infarction, with negative results on chest radiography and electrocardiography, and with concentrations of brain natriuretic peptide below the cut off, individually and in combination, are shown in the table. Measuring the concentration of brain natriuretic peptide is the test with the lowest likelihood ratio for a negative test; thus it is the most useful. However, to be useful in clinical practice, this test must provide additional diagnostic information over that given by investigations that are more readily available, which in combination yield a minimum posterior probability of 20%. Adding a test for brain natriuretic peptide to the determination of a patient's history of myocardial infarction in the diagnostic screening process reduces the posterior probability to 15%.

Comment

There seems to be a small diagnostic advantage to measuring brain natriuretic peptide in addition to performing routine investigations. However, given the therapeutic and prognostic importance of correct diagnosis, most clinicians would find a 1 in 7 chance of left ventricular systolic dysfunction unacceptably high in a patient who has not been referred for echocardiography.

Recruitment to this study relied on the general practitioners making a provisional diagnosis of suspected heart failure, and results may be different in other settings, such as population based screening for asymptomatic left ventricular systolic dysfunction. Nevertheless this study suggests that introducing routine measurement of the plasma concentration of brain natriuretic peptide would be unlikely to improve the diagnosis of symptomatic left ventricular systolic dysfunction in the community.