False Negative ECG Device Results May Increase the Risk of Adverse Events in Clinical Oncology Trials

Abstract

Abstract Background Sites participating in clinical trials may not have the expertise and infrastructure to accurately measure cardiac intervals on 12-lead ECGs and rely heavily on the automated ECG device generated results for clinical decision-making. Methods Using a dataset of over 260,000 ECGs collected in clinical oncology studies, we investigated the mean difference and the rate of false negative results between the digital ECG machine QTc and QRS measurements compared to those obtained by a centralized ECG core lab. Results The mean differences between the core lab and the automated algorithm QTcF and QRS measurements were + 1.8 ± 16.0 ms and − 1.0 ± 8.8 ms, respectively. Among the ECGs with a centralized QTcF value > 450 or > 470 ms, 39.5% and 47.8% respectively had a device reported QTcF value ≤ 450 ms or ≤ 470 ms. Among the ECGs with a centrally measured QTcF > 500 ms, 55.8% had a device reported value ≤ 500 ms. Automated QTcF measurements failed to detect a QTcF increase > 60 ms for 53.9% of the ECGs identified by the core lab. Automated measurements also failed to detect QRS prolongation, though to a lesser extent than failures to detect QTc prolongation. Among the ECGs with a centrally measured QRS > 110 or 120 ms, 7.9% and 7.3% respectively had a device reported QRS value ≤ 110 ms or ≤ 120 ms. Conclusion Relying on automated measurements from ECG devices for patient inclusion and treatment (dis)continuation decisions poses a potential risk to patients participating in oncology studies.