Purpose: Sudden cardiac death (SCD) is rare but devastating. ECG screening for diseases that predispose to SCD is controversial, and there is currently no standard of care or consensus in the United States. If screening ECGs are performed, patients are often referred to a cardiologist’s office based on abnormal ECG machine readings; these referrals may or may not be warranted. The primary objectives of our study were (1) to assess the accuracy of ECG machine readings in an asymptomatic pediatric cohort when compared to review by a pediatric cardiologist and (2) to describe a unique model of SCD screening and the cardiac diseases detected by this model. Methods: A retrospective study was conducted for 2263 asymptomatic patients ages 10-18 who received a screening ECG between the years 2014-2018 at a private general pediatrics office in NYC offering elective screening ECGs during well child checks. Each ECG was automatically interpreted by the ECG machine and then sent to our institution for review. Pediatric cardiologists reviewed the ECGs using standardized criteria specifically aimed at identifying risks for SCD. Cardiology referral was recommended only for significant findings. We compared the ECG machine reads with cardiologist readings and assessed for accuracy. Results: Screening ECGs conducted on 2263 asymptomatic pediatric patients yielded inaccurate ECG machine readings 28.2% (639/2263) of the time (Figure 1). 27.4% (620/2263) of the readings were false positives, meaning an abnormal automated interpretation deemed to not warrant referral when reviewed by the cardiologist. The most common benign ECG readings that did not prompt a referral were bradycardia (349), right ventricular conduction delay (138), left atrial enlargement (110), and ST elevation (58). 19 readings (0.8%) were false negatives (normal automated reading determined to warrant referral when reviewed by the cardiologist). Out of a total of 34 patients referred for formal cardiac evaluation, 25 patients had normal results, and 2 patients were diagnosed with SCD-related conditions including Brugada syndrome and hypertrophic cardiomyopathy (Table 1). Calculated sensitivity and specificity for automated ECG readings identifying SCD risk are approximately 72.9% and 71.7% in this cohort. Conclusions: ECG machine readings have poor accuracy for screening for SCD risk, with 1 out of 4 total interpretations yielding false positive results. These false positive findings are often concerning to primary care providers, leading to unnecessary specialist referrals and subsequent work-up, increased healthcare costs, and patient and parent concern. Furthermore, false negative interpretations can lead to detrimental consequences for the patients. If screening ECGs are performed, a model in which practitioners familiar with SCD review ECGs will increase the accuracy of ECG interpretations and significantly reduce unnecessary cost potentially contributing to development of a reasonable strategy for universal SCD screening.
Screening ECGs conducted on 2263 asymptomatic pediatric population resulted in inaccurate ECG machine readings 28.2% (639/2263) of the time (Left Image). 27.4% (1573/2263) of the ECG machine readings were false positives and 0.8% (19/2263) were false negatives (Right Image).
Comments