Objective: To determine the diagnostic accuracy of clinical assessment of heart murmurs in children and specific clinical features that are predictive of cardiac disease.
Design: Concurrent case series with pretest-posttest assessment.
Setting: Pediatric cardiology outpatient clinic.
Participants: Five full-time academic pediatric cardiologists.
Measures: For each of 222 consecutive patients who were seen for first-time evaluation of a heart murmur, the clinical findings and diagnostic impressions were recorded after clinical assessment. The results of electrocardiograms and echocardiograms were then reviewed, and changes in diagnostic impressions were recorded and compared with the original impressions.
Results: The prevalence of cardiac disease was 33%. Clinical assessment differentiated those patients with pathologic murmurs with a sensitivity of 92%, specificity of 94%, positive predictive value of 88%, and negative predictive value of 96%. If diagnostic uncertainty was considered an indication for echocardiography, then sensitivity and specificity increased to 97% and 98%, respectively. Missed disease included only trivial or minor lesions. Clinical features that were independently predictive of the presence of disease included murmurs that were pansystolic (odds ratio [OR], 54.0), grade 3 or more in intensity (OR, 4.84), heard best at the left upper sternal border (OR, 4.24) and harsh in quality (OR, 2.37), and the presence of an abnormal second heart sound (OR, 4.09) and an early or midsystolic click (OR, 8.35).
Conclusions: Clinical assessment by a pediatric cardiologist is sufficient to distinguish pathologic from innocent heart murmurs. A genetic approach by using specific clinical features that are independently associated with disease may have some practical utility to noncardiologists.