Structured clinical question

In children with an asymptomatic heart murmur [patient] does a chest x ray and/or ECG examination [intervention] assist in the diagnosis or exclusion of congenital heart disease [outcome]?

Search strategy and outcome

Secondary sources: none

Primary sources: Medline 1966 to October week 2, 2001: (heart murmurs OR (heart murmur$ OR cardiac murmur$).tw.) AND (electrocardiography OR ECG.mp) AND (radiography, thoracic.mp. OR chest xray.mp, OR chest x-ray.mp OR chest radiograph.mp) AND (heart defects, congenital/ OR congenital heart disease.mp OR heart defects congenital/ra); filters children <0–18years> & English language.

Serendipity: 1 article.

Search results—132 articles found; 10 articles relevant to clinical question; four excluded due to poor quality. See table 1.

Commentary

Paediatric cardiologists have undertaken most of the research investigating the assessment of the child with a heart murmur, with and without ECG and chest x ray examination. However, the Birkebaek et al study evaluates the general paediatricians’ assessment of a heart murmur and the Rajakumar et al study compared academic general paediatricians and paediatric cardiologists. I could find no studies comparing trainees and consultants.

In the study by Rajakumar et al, general paediatricians and paediatric cardiologists each evaluated the patient referred with a heart murmur (blind to the others’ assessment) and classified them innocent, possibly pathologic or pathologic murmur. They then had a chest x ray and ECG examination and were reclassified. An echocardiogram was then performed, which gave them a definitive diagnosis. The paediatricians classified more innocent murmurs as pathologic and the cardiologists identified more innocent murmurs correctly. After ECG and chest x ray examination paediatricians revised five diagnoses, three incorrectly. That is, for the vast majority ECG and chest x ray examination did not help in the diagnosis, and in those cases where it was thought helpful it was often misleading.

The likelihood ratio of a test, calculated from the sensitivity and specificity, gives an estimate of increased probability of correctly identifying a condition (positive likelihood ratio) or excluding a diagnosis (negative likelihood ratio) when using the diagnostic tool in question. A reasonable pretest probability is assumed and then, using Fagan’s likelihood ratio nomogram, the post-test probability is calculated (see Archimedes in January 2003). For example, if the pretest probability of a pathological heart murmur was 5%, an abnormal chest x ray examination (with a likelihood ratio of 2.36 (Birkebaek et al)) would make the post-test probability of cardiac pathology only about 10%. It was only possible to calculate likelihood ratios for chest x ray examination in one paper and the other likelihood ratios were calculated for clinical evaluation. Interestingly in the Rajakumar et al study the likelihood ratios after ECG and chest x ray examination were very similar to those after clinical evaluation—that is, little was added by doing these tests.

Birkebaek et al evaluated the accuracy of the paediatric radiologists in their interpretation of chest x rays of children with heart murmurs. This paper is relevant as most paediatricians will rely on the report from the radiologist. The six radiologists were each asked to report on all the films blind to the result of the echocardiogram, and six months later the chest x rays were re-evaluated by the same radiologists. The results showed only poor to moderate agreement between radiologists, and more surprisingly poor agreement when the same radiologist reviewed the films six months later.

Overall, it appears from the above research that ECG and chest x ray examination add little to the clinical evaluation of the child with an asymptomatic heart murmur. Concerns about a pathological cause after clinical examination should prompt a referral to a paediatric cardiologist for further assessment.