Incremental value of ECG in predicting echocardiographic abnormalities among patients referred for palpitations

Background Transthoracic echocardiography (TTE) for palpitations is frequently requested, yet its diagnostic yield is low. We evaluated whether routinely available ECG data improve prediction of echocardiographic abnormalities to target TTE use. Purpose To determine the incremental value of ECG parameters for predicting abnormal TTE in patients referred for palpitations and to derive a simple bedside risk score. Methods We performed a retrospective cohort study of 3,219 consecutive adults referred for outpatient TTE because of palpitations. Each TTE was paired with the closest 12-lead ECG. The primary outcome was any abnormal TTE; the secondary outcome was severe abnormality (ventricular dysfunction, significant valvular disease or structural cardiomyopathy). Candidate predictors were age, sex, body mass index (BMI), global ECG diagnosis (normal/abnormal), heart rate, PR, QRS and QTc intervals, and P/QRS/T-wave axes. Multivariable logistic regression with nested models (demographics only; +continuous ECG intervals/axes; +global ECG diagnosis) compared discrimination by area under the curve (AUC). A simplified points-based risk tool was created from model coefficients (Figure). Results Any TTE abnormality occurred in 9.6% of patients and severe abnormality in 3.1%. Independent predictors of any abnormality were abnormal ECG (odds ratio [OR] 2.2), male sex (OR 3.3), age per 10 years (OR 1.6), BMI per 5 kg/m2 (OR 1.3), QTc per 10 ms (OR 1.10) and heart rate per 10 beats per minute (OR 0.85). Model discrimination improved stepwise: demographics AUC 0.79; +ECG intervals/axes AUC 0.82; +global ECG diagnosis AUC 0.83. Using a prespecified 10% risk threshold with the final model, sensitivity was 0.75, specificity 0.73, negative predictive value (NPV) 0.97 and positive predictive value (PPV) 0.23. In the bedside score, low-, intermediate- and high-risk strata showed observed rates of any abnormal TTE of 1.9%, 7.3% and 25.8% and severe abnormality of 0.6%, 2.0% and 8.9%, respectively. For severe outcomes overall, NPV was approximately 0.99 across clinically relevant thresholds. Conclusion ECG features provide meaningful incremental discrimination beyond demographics for predicting abnormal TTE in patients with palpitations. Given the low event rate and very high NPV, an ECG-augmented risk model and simple bedside score can help safely defer TTE in low-risk patients while focusing imaging on those most likely to benefit.