Can cardiologists distinguish innocent from pathologic murmurs in neonates?

To determine the sensitivity and specificity of the clinical assessment of murmurs in neonates, as performed by pediatric cardiologists, and to identify clinical features that predict the presence of congenital heart disease (CHD) in this population.

Neonates (n = 201) referred for outpatient evaluation of a heart murmur were enrolled consecutively. After a clinical evaluation, the cardiologist documented whether the murmur was "likely innocent" or "likely pathologic." The cardiologist repeated his/her assessment after an electrocardiogram. Echocardiography served as the gold standard.

The median age was 12 days (range, 2-31 days). CHD was present in 113 of 201 (56%). Clinical assessment alone identified patients with CHD with a sensitivity of 80.5% (95% CI, 73.2-87.8), specificity of 90.9% (95% CI, 84.9-96.9), positive predictive value of 91.9% (95% CI, 86.6-97.3), and negative predictive value of 78.4% (95% CI, 70.4-86.4). The addition of an electrocardiogram did not improve these test characteristics. Features that were predictive of CHD were murmur quality (P < .0001), location (P = .02), and timing (P = .04). No patients requiring catheter or surgical intervention were missed by clinical assessment.

The prevalence of CHD in this referral population was high. Clinical assessment detected all complex CHD, although some simple lesions were missed. Murmur quality, location, and timing were predictive of CHD.