The prognostic significance of electrocardiography findings in patients with coronavirus disease 2019: A retrospective study

Abstract Background Coronavirus disease 2019 (COVID‐19) has reached a pandemic level. Cardiac injury is not uncommon among COVID‐19 patients. We sought to describe the electrocardiographic characteristics and to identify the prognostic significance of electrocardiography (ECG) findings of patients with COVID‐19. Hypothesis ECG abnormality was associated with higher risk of death. Methods Consecutive patients with laboratory‐confirmed COVID‐19 and definite in‐hospital outcome were retrospectively included. Demographic characteristics and clinical data were extracted from medical record. Initial ECGs at admission or during hospitalization were reviewed. A point‐based scoring system of abnormal ECG findings was formed, in which 1 point each was assigned for the presence of axis deviation, arrhythmias, atrioventricular block, conduction tissue disease, QTc interval prolongation, pathological Q wave, ST‐segment change, and T‐wave change. The association between abnormal ECG scores and in‐hospital mortality was assessed in multivariable Cox regression models. Results A total of 306 patients (mean 62.84 ± 14.69 years old, 48.0% male) were included. T‐wave change (31.7%), QTc interval prolongation (30.1%), and arrhythmias (16.3%) were three most common found ECG abnormalities. 30 (9.80%) patients died during hospitalization. Abnormal ECG scores were significantly higher among non‐survivors (median 2 points vs 1 point, p < 0.001). The risk of in‐hospital death increased by a factor of 1.478 (HR 1.478, 95% CI 1.131–1.933, p = 0.004) after adjusted by age, comorbidities, cardiac injury and treatments. Conclusions ECG abnormality was common in patients admitted for COVID‐19 and was associated with adverse in‐hospital outcome. In‐hospital mortality risk increased with increasing abnormal ECG scores.

Conclusions: ECG abnormality was common in patients admitted for COVID-19 and was associated with adverse in-hospital outcome. In-hospital mortality risk increased with increasing abnormal ECG scores. quickly spread throughout the world and caused a global pandemic. 1 Although clinical manifestations of COVID-19 were mainly respiratory, cardiac injury, and arrhythmias were not uncommon and the association between cardiac injury and poor in-hospital outcome had been determined. 2,3 As a simple and easily obtainable tool to identify patients with acute or chronic cardiac disease, ECG is frequently performed in patients with cardiovascular disease and was also studied in pneumonia, 4 severe acute respiratory syndrome (SARS), 5 and Middle East respiratory syndrome (MERS). 6 However, systematic studies of ECG characteristics in COVID-19 patients were limited 7,8 and its prognostic significance remained to be fully elucidated. 9 The purpose of this study was to describe the electrocardiographic characteristics and to identify the prognostic significance of ECG findings at admission or during hospitalization in patients with COVID-19.

| Clinical data collection
Demographic information (age and sex) and clinical data consisting of disease duration, blood pressure at admission, comorbidities, serum level of high-sensitive cardiac troponin I (hs-cTnI), and treatment data were extracted from medical record. Disease duration was defined as time from symptom onset to admission. Cardiac injury was diagnosed if serum level of hs-cTnI was above the 99th percentile upper reference limit (which was 0.026ug/L in our hospital).

| 12-Lead electrocardiography
Resting standard 12-lead ECGs were performed by trained physicians using PageWriter TC10 (Philips, Amsterdam, Noord-Holland, Netherland) machine or MAC 800 (GE Healthcare, Chicago, IL) machine in all patients at admission or during hospitalization using a paper speed of 25 mm/s and a sensitivity of 1 mV = 10 mm. The initial ECG record of each patient was reviewed. Heart rate, PR interval, QTc interval (corrected by Bazett's formula) and mean frontal plane QRS electrical axis were measured automatically by the ECG machine/computergenerated measurements. ECG parameters (definitions listed in the Table S1) were reviewed and confirmed by the principal investigators (D.Y. and J.L.). Discordances were solved by consensus, with the supervision of the senior expert (F.Z.) in electrocardiography.
Patients were considered to have arrhythmias if they had sinus tachycardia, sinus bradycardia, sinus node arrest, or atrial fibrillation and were considered to have conduction tissue disease if they had right bundle branch block (RBBB), left bundle branch block (LBBB), or left anterior fascicular block(LAFB). 10 ST-segment change included ST-elevation and ST-depression and T-wave change was consisted of inverted T-wave and flat T-wave (Figure S1A-S1B). To evaluate the prognostic significance of abnormal ECG findings, we created a pointbased scoring system, in which 1 point each was assigned for the presence of axis deviation (Figure S1C-S1D), arrhythmias, atrioventricular block, conduction tissue disease, QTc interval prolongation, pathological Q wave, ST-segment change, and T-wave change. We calculated abnormal ECG scores by adding 1 point each for any of the eight ECG findings aforementioned. Such ECG scoring was analyzed as a continuous variable.

| In-hospital outcome
All patients were followed up during hospitalization. The in-hospital outcome comprised incidence of in-hospital death or discharge.
Patients were discharged if they had relieved clinical symptoms, normal body temperature, significant resolution of inflammation as shown by chest radiography, and at least 2 consecutive negative results shown by real-time PCR assay for COVID-19. 11 The vital status of patients was determined by medical record. Discharged patients were censored at the date of discharge.   Table 2 and Figure 1, T-wave change (97 patients, 31.7%) was the most common abnormal finding and QTc interval prolongation, arrhythmias, axis deviation, conduction tissue disease, ST-segment change, atrioventricular block, and pathological Q wave were present in 92 patients (30.1%), 50 patients (16.3%), 34 patients (11.1%), 28 patients (9.2%), 24 patients (7.8%), 12 patients (3.9%), and 6 patients (2.0%), respectively. Table 2

| DISCUSSION
The main findings of our study are threefold: (a) Various abnormal ECG findings were common among patients with COVID-19; COVID-19 had found that any abnormal finding on the ECG was found in 120 patients (37%) 9 . The proportion of abnormal ECG was as high as 93% among critical ill COVID-19 patients. 8  Complications and poor outcomes more frequently occurred in elderly COVID-19 patients. 11,15 Consistent with these findings, increasing age was independently associated with in-hospital death in the present study. An unexpected finding was that treatment with glucocorticoid was also related to mortality in multivariable Cox regression analysis. A possible explanation might be that severer patients tended to receive glucocorticoid therapy 15 and non-survivors also was with a higher proportion of glucocorticoid therapy. 11

| CONCLUSIONS
Our data show that ECG abnormality was common among admitted patients for COVID-19 and was associated with adverse in-hospital outcome. In-hospital mortality risk increased with increasing abnormal ECG scores, suggesting that close observation should be kept on patients with multiple ECG abnormalities during their hospitalization.
ECG might be an easy tool for risk stratification in such patients.

| LIMITATION
Some limitations of our study should be acknowledged. Firstly, only initial ECGs on admission or during hospitalization were reviewed.
Neither previous ECGs nor subsequent ECGs were analyzed for comparison. It is difficult to distinguish pre-existing cardiac disease between acute infective status related ECG abnormalities. Secondly, echocardiographic data was available in only a minority of patients and was not included in our analysis. Thirdly, the follow-up of patients was short and our results might not be representative of the long-term prognosis. Fourthly, the event rate was low and the multivariable Cox regression analysis with many variates may be less reliable. Finally, several individual ECG abnormalities were detected in small numbers of patients and the estimate of their prognostic role should be further verified in larger populations.

ACKNOWLEDGMENT
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

CONFLICT OF INTEREST
All authors declare no conflicts of interest that might be relevant to the contents of this manuscript.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.