Background: Heart murmurs are found in more than 50% of children, yet less than 1% of children are born with congenital heart disease. The auscultation skills of the general practitioners responsible for routine child care are suboptimal, however, leading to frequent referral to pediatric cardiologists for evaluation. Pediatric cardiologists can quickly and accurately diagnose innocent murmurs with physical examination only, thereby avoiding further diagnostic testing. Patients living in remote regions without pediatric cardiology support require evacuation to medical centers, which can delay diagnosis, lead to family stress, and result in significant financial expenditure for travel. Using a digital recording stethoscope and our store-and-forward telemedicine system, we developed a pediatric telecardiology system to allow for remote cardiac auscultation. We hypothesized that such a system could accurately classify auscultatory findings as normal/innocent murmur or pathologic, thereby reducing the need for many evacuations and allowing more timely evaluation of patients with cardiac pathology.

Methods: Patients undergoing evaluation in the pediatric cardiology clinic at Tripler Army Medical Center underwent standard physical examination and complete echocardiography to establish a definitive cardiac diagnosis. Using a commercially available digital stethoscope attached to a handheld PDA, study participants also had 20-second digital heartsound recordings acquired from standard cardiac auscultation areas while both upright and supine. Heartsounds were uploaded to our store-and-forward telemedicine system, allowing for playback in a manner similar to standard clinical auscultation via a custom graphical user interface. Pediatric cardiologists, blinded to all other clinical information, evaluated each heartsound dataset and classified the case as either normal or pathologic.

Results: To date, 41 pediatric patients have been evaluated (24 with normal/innocent murmurs, 17 with cardiac pathology), each of which interpreted by 3 pediatric cardiologists (total of 123 cases). When compared to echocardiographic results, 86% (106/123) of the cases were accurately classified as either normal/innocent murmur or pathologic. Nine cases with pathology were misclassified as normal/innocent murmur (sensitivity 84%). Eight cases were classified as pathologic (specificity 88%) when the findings were normal/innocent murmur.

Conclusions: Digital heartsound recordings evaluated in our store-and-forward telecardiology system can determine normal from pathologic auscultatory findings with a high degree of accuracy. Such a system has the potential to significantly decrease travel expenditures and reduce diagnostic delays for patients requiring pediatric cardiology evaluation. Further refinements to our heartsound system are in progress and are expected to further improve the accuracy of remote cardiac auscultation.