Frequency adaption of electrocardiographic QRS duration improves outcome prediction.

Background: Robust and accessible risk stratification is of high interest in cardiovascular patients. Parameters associated with left ventricular function have shown to be promising candidates for risk classification in this context. The ECG derived variable QRS duration is easily obtainable and delivers valid prognostic information. Hence, prolonged QRS duration can be seen as risk marker of interest. Further, heart rate (HR) is known to be associated with risk but also directly impacts the QRS duration. Therefore, we aimed at investigating the prognostic performance of frequency adapted QRS duration to predict mortality in cardiovascular patients.

Methods: The study population is based on an estalished registry that enrolled patients scheduled for an invasive coronary angiography due to suspected chronic coronary syndrome. The present analysis evaluated patients based on the following criteria: available long-term follow-up data on the outcome measure overall mortality, availability of a standard resting 12-lead ECG at admission and availability of all variables needed to calculate the ESC-SCORE2 (German weighted version). This led to a cohort of 822 patients (32% female), of whom 122 died within the median follow-up time of 7 years. QRS duration and HR were measured using the automated computation feature of the ECG device. With these parameters the frequency corrected QRS (QRSc) duration was calculated using Bazett’s formula, analogously to how QTc is calculated. As performance metrics area under the curve (AUC), Harrell’s C-index and their respective 95% confidence interval (CI) have been calculated. Different AUCs have been compared using the DeLong test.

Results: In the evaluated cohort the QRS duration was associated with all-cause mortality with an AUC of 0.57 (95% CI 0.51-0.63), further HR predicted all-cause mortality with an AUC of 0.61 (95% CI 0.56-0.67). In comparison, the ESC-SCORE2 predicted death with an AUC of 0.61 (95% CI 0.55-0.67). If the frequency adjusted QRSc time was considered, a significantly higher AUC of 0.63 (95% CI 0.57-0.68) was observed compared to the raw QRS time (p_DeLong<0.001). If additionally taking time-to-event into accout, similar results were observed with a c-index of 0.55 for QRS duration and c-index of 0.61 for QRSc duration.

Conclusion: The presented data shows that the prognostic performance of a prolonged QRS duration in predicting mortality can be significantly improved by adjusting for heart rate. Further investigation on larger cohorts is needed to evaluate the robustness of the prognostic performance of the QRSc.