Methods
We identified individuals with respiratory failure due to COVID-19 infection who were undergoing telemetry monitoring in the ICU and were planned to undergo evaluation with a 12-lead ECG. Subjects were included in the study if they had 12-lead and 7-lead ECGs obtained within a single 24-hour period and confirmed positive SARS-CoV-2 RT-PCR nasal swab. The study was approved by the Institutional Review Board at Columbia University Irving Medical Center.
Each subject received a 12-lead ECG at standard (25mm/sec) speed and normal voltage, utilizing standard lead placement (GE MAC 5500, GE Medical Systems, Milwaukee, WI). A 10 second 7-lead ECG and single lead (lead II) tracing were also obtained from the central telemetry system (Philips Intellivue, Koninklijke Philips Electronics, Netherlands). This system uses the EASI System that derives up to 7 leads from 5 electrodes positioned on the torso, including all 3 limb leads, 3 augmented leads, and 1 precordial lead (typically lead V1).7
Each 12-lead standard ECG, 7-lead telemetry-derived ECG, and Lead II tracing was interpreted and manually-measured by 2 independent cardiologists. Disagreements in reads were adjudicated by an independent third reader. The qualitative assessment or quantitative measurement that was agreed upon by at least 2 of the cardiologists was recorded as the final interpretation. Diagnoses evaluated included underlying rhythm, QRS axis, cardiac intervals (RR interval, PR interval, QRS interval, and QT interval), conduction abnormalities, the presence of T wave or ST segment abnormalities, and the presence of low voltage complexes. Automated measurements by the ECG computer system for rate and common intervals were also recorded. The QT interval was measured in 3 consecutive cycles on leads II, III, and the precordial leads (V5, V6 when available) with the maximal interval taken. The end of the T wave was defined as the tangent from the steepest last limb of the T wave to its intersection with the baseline. For irregular rhythms, a weighted average RR interval was calculated.8 QT interval correction was calculated using commonly-used formulae: Bazett, Fridericia, Framingham, and Hodges. QT dispersion was calculated as the difference between the longest and shortest measured QT intervals on available leads. Significant ST segment changes were considered to be more than 1mm of elevation or depression. Low voltage was considered any ECG where the QRS was less than 5mm in all of the limb leads or 10mm in the precordial leads. Clinical information was obtained from the electronic medical record including demographics, comorbid conditions, laboratory parameters, and pertinent therapies.
Statistical analysis was performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). For continuous variables, means and standard deviations were calculated. For categorical variables, frequency and percentage were calculated. The differences in QT duration between the 12-lead ECG and both the derived 7-lead ECG and single-lead telemetry were calculated using students t-test. Pearson’s correlation coefficients were calculated between ECG measurement methods and scatter plots were generated. Bland-Altman analyses were performed to evaluate agreement between measurement methods.9