Results
Thirty-three subjects were identified and included in the analysis. Patient demographics, medical history, laboratory data and clinical features are provided in Table 1. Overall, the population was 42.4% female with a mean age of 60.1 ± 14.1 years. Obesity and other cardiovascular comorbidities were common (e.g. 47.1% were obese with BMI>30, 63.6% had hypertension, 48.5% diabetes mellitus, and 12.1% had coronary artery disease). The study population was critically ill with all subjects receiving mechanical ventilation, approximately 66% receiving renal replacement therapy, and nearly 75% requiring vasopressors. The majority of patients were treated with hydroxychloroquine (78.8%) and azithromycin (81.8%).
The median amount of time between obtaining the 12-lead ECG and telemetry-derived ECG was 53 minutes (range 3 minutes to 13 hours and 16 minutes). ECG parameters measured from each ECG method are reported inTable 2 . The mean uncorrected QT interval in milliseconds (msec) for the group as a whole was 398.6 ± 64.8, 401.2 ± 66.1, 388.2 ± 64.6 and 382.1 ± 60.7 by computer measurement on 12-lead, manual measurement on 12-lead, 7-lead derived ECG, and telemetry lead II, respectively. Similarly, the mean corrected QT interval by Bazett formula in msec was 474.2 ± 39.4, 467.5 ± 50.9, 454.1 ± 49.4 and 447.9 ± 47.3 by computer measurement on 12-lead, manual measurement on 12-lead, 7-lead derived ECG, and telemetry lead II, respectively. These measurements for the entire group were statistically similar.Table 3 shows the mean individual differences in QT measurements between the manual 12-lead ECG and both the 7-lead derived ECG and single-lead telemetry. The mean difference in measured QT was significantly shorter on both the 7-lead and single lead derived ECGs compared with the 12-lead ECG (p value for all comparisons <0.05). The mean QT difference was significantly larger between the 12-lead ECG and single lead telemetry than between the 12-lead ECG and 7-lead derived ECG.
Table 4 shows the morphologic features of ECGs. Thirty-one (93.94%) patients were in sinus rhythm while 2 patients were in atrial fibrillation. Twenty-nine (87.88%) subjects were identified as having a normal axis on the 12-lead ECG, compared to 26 (78.79%) on the 7-lead ECG. Five (15.15%) subjects had conduction delay on the 12-lead ECG; 2 had right bundle branch block (RBBB), 1 subject had left bundle branch block (LBBB), and 2 patients had nonspecific intraventricular conduction delay (IVCD). Using both the 7-lead ECG and telemetry lead II, 4 (12.12%) subjects were identified as having conduction delay. Eleven (33.33%) subjects had T wave inversion on the 12-lead ECG, compared with 14 (42.42%) on the 7-lead ECG and 9 (27.27%) on single lead telemetry. One (3.03%) subject had ST depression and none had ST elevation on the 12-lead ECG, compared to 4 (12.12%) subjects with ST depression and 5 (15.15%) with ST elevation on the 7-lead ECG, and 1 (3.03%) subject with ST depression and 2 (6.06%) with ST elevation on telemetry lead II. Seven (21.21%) subjects had low voltage QRS on the 12-lead ECG, compared to 2 (6.06%) on the 7-lead ECG and 1 (3.03%) on telemetry lead II. Figure 1 provides a representative example of a 12-lead and 7-lead ECG of the same patient, with less evident conduction delay and more pronounced ST segment changes on the 7-lead ECG.
Manually-measured QTc from standard 12-lead ECG was strongly linearly correlated with the manually-measured QTc from the 7-lead ECG and telemetry lead II, as well as the automatic computer-generated 12-lead QTc measurement (p<0.0001) (Figure 2 ). Bland-Altman plots show a reasonable amount of agreement among ECG methods for QT measurement (Figure 3 ). There is evidence for a positive bias, consistent with an underestimation of the QT interval when measured on the telemetry-derived ECGs as compared with the standard 12-lead ECG. These differences in measurement were frequently between 25 and 50 msec and were up to 110 msec when using the single lead tracing. Additional correlation and Bland-Altman plots for the remaining ECG parameters are provided in the Supplemental Material. A consistent positive bias was identified when comparing the corrected QT intervals between 12-lead and telemetry-derived ECGs.