key: cord-0802157-ggrhfpo0
authors: McCullough, S. Andrew; Goyal, Parag; Krishnan, Udhay; Choi, Justin J.; Safford, Monika M.; Okin, Peter M.
title: Electrocardiographic Findings in COVID-19: Insights on Mortality and Underlying Myocardial Processes
date: 2020-06-13
journal: J Card Fail
DOI: 10.1016/j.cardfail.2020.06.005
sha: 19ac02b701f9f1c841ebcba6eed529f63cb0350f
doc_id: 802157
cord_uid: ggrhfpo0

INTRODUCTION: Coronavirus disease 2019 (COVID-19) is a respiratory syndrome with high rates of mortality, and there is a need for easily obtainable markers to provide prognostic information. We sought to determine whether the electrocardiogram (ECG) on hospital presentation provides prognostic information, specifically related to death. METHODS: We performed a retrospective cohort study in patients with COVID-19 who had an ECG at or near hospital admission. Clinical characteristics and ECG variables were manually abstracted from the electronic health record and first ECG. Our primary outcome was death. RESULTS: 756 patients who presented to a large New York City teaching hospital with COVID-19 underwent an ECG. The mean age was 63.3 ± 16 years, 37% were women, 61% of patients were non-white, and 57% had hypertension; 90 (11.9%) died. In a multivariable logistic regression that included age, ECG, and clinical characteristics, the presence of one or more atrial premature contractions (APC)s (odds ratio (OR)=2.57, 95% confidence interval (CI) 1.23-5.36, p=0.01), a right bundle branch block (RBBB) or intraventricular block (IVB) (OR=2.61, 95% CI 1.32-5.18, p=0.002), ischemic T-wave inversion (OR=3.49, 95% CI 1.56-7.80, p=0.002), and nonspecific repolarization (OR=2.31, 95% CI 1.27-4.21, p=0.006) increased the odds of death. ST elevation was rare (n=5, 0.7%). CONCLUSION: We found that patients with ECG findings of both left sided heart disease (APCs, IVB, repolarization abnormalities) and right sided disease (RBBB) have higher odds of death. ST elevation at presentation was rare.

SARS-Coronavirus 2 (SARS-CoV2) is a highly infectious virus that causes coronavirus disease (COVID-19), a respiratory syndrome associated with high rates of critical illness and mortality. 1, 2 As a consequence, there is a need to identify prognostic markers that can aid clinicians in the rapid triage of patients, guide clinical decision making, and inform patients and their families about the anticipated disease trajectory. Electrocardiography (ECG) is a broadly available diagnostic test that can be quickly performed without exposing a large number of personnel to SARS-CoV2. ECG has demonstrated incremental prognostic value in populationbased studies and in patients with a variety of underlying cardiovascular conditions, including hypertension [3] [4] [5] [6] [7] , and thus offers a particularly appealing modality during the current pandemic. We thus sought to determine whether findings on the first presenting ECG provide prognostic information, specifically related to death, and subsequently provide insights on myocardial processes underlying a poor prognosis.

Our retrospective observational cohort study included patients with confirmed COVID-19 who presented to Weill Cornell Medicine/New York-Presbyterian Hospital, a quaternary referral center and 862-bed teaching hospital, from March 3, 2020 through April 9, 2020. All cases of COVID-19 were confirmed by real-time reverse-transcriptase polymerase chain reaction (RT-PCR) on nasopharyngeal swabs. We included consecutive patients who presented directly to our hospital and underwent an ECG at or near their initial presentation to the hospital. We excluded patients who had complete ventricular pacing, since these ECGs were otherwise non-diagnostic. Using a standardized protocol and REDCap tool 8 , patient data were manually abstracted from the electronic health record using methods that have been previously described. 1 These data were used to develop a COVID-19 registry database.

ECGs were personally reviewed and interpreted by two electrocardiographers (SAM and PO, together responsible for the interpretation of >100,000 ECGs per year) who were blinded to the clinical status of the patients. Any disagreement in interpretation between readers was resolved by consensus. No formal testing of between or within reader variability of interpretation was performed for this study. Data extracted from each ECG included heart hate (HR, in beats 

The primary outcome was death occurring through April 23, 2020, assuring at least two weeks of outcome data for all included patients. We ascertained death based on review of discharge summaries and death notes in the electronic health record.

Demographics (age, sex, and race) and pre-existing comorbid conditions (smoking status, history of immunosuppression, diabetes mellitus (DM), hypertension (HTN), chronic obstructive pulmonary disease (COPD), end-stage renal disease (ESRD), coronary artery disease (CAD), chronic heart failure (CHF), stroke and active cancer) were abstracted from the electronic health record. Additionally, the clinical decision to start supplemental oxygen therapy due to hypoxemia within three hours of presentation was abstracted from the respiratory flowsheets.

Statistical analysis was performed using SPSS version 25 software (IBM, Inc., Armonk, NY). Data are presented as mean ±SD for continuous variables and proportions for categorical variables. The relationship between clinical and ECG characteristics and death was examined using univariate and multivariable logistic regression analyses. Statistically significant univariate predictors of death were entered into forward selection multivariable models with a p value ≤0.05 required for entry: an age and ECG variables model; age and clinical variables model; and finally, a model incorporating age, clinical, and ECG variables. The relative overall accuracy of age and these models was compared by plotting receiver operating characteristic (ROC) curves of age and the predicted probabilities from each multivariable logistic regression model and calculating the area under each curve with its standard error.

This study was approved by the Weill Cornell Medicine Institutional Review Board, which waived informed consent.

From March 3, 2020 through April 9, 2020, 945 patients presented to our hospital and were confirmed by PCR to have Sars-CoV2. Of these patients, 768 underwent an ECG at or near the time of their admission; 12 patients were excluded from analysis for complete ventricular pacing (Figure 1 ). Among the remaining 756 patients, the median time between presentation to the ED and initial ECG was <1 day and 94.3% had their ECG within 1 day of presentation.

Clinical characteristics of the population are shown in Table 1 . The mean age was 63.3 ± 16 years, 37% were women, 61% of patients were non-white, 37% had obesity, and 29% had diabetes mellitus. Cardiovascular conditions were common: 57% had hypertension, 14% had coronary artery disease, 7% had heart failure, and 7% had a prior stroke. Over half (55%) required supplemental oxygen within the first three hours of presentation.

Baseline electrocardiographic characteristics ( Table 2 ) included mean HR 90 ± 19 bpm and mean Bazett-corrected QT interval of 449 ± 144 ms. The overwhelming majority were in normal sinus rhythm (94.4%), while 5.6% of patients had AF. Atrioventricular block was rare (2.6%): 19 (2.5%) patients had a first degree block and 1 patient (0.1%) was in sinus rhythm with complete heart block and a junctional escape rhythm. APCs occurred in 7.7% and VPCs in 3.4%. A significant proportion (19.3%) had an abnormal axis-13.8% had left axis deviation and 5.5% had a right or right superior axis deviation. Abnormal intraventricular conduction was found in 11.8%, with RBBB in 7.8%, LBBB in 1.5% and nonspecific IVB in 2.5%. Left ventricular hypertrophy (15.5%) was more common than right ventricular hypertrophy (4.0%) and evidence of a previous Q-wave myocardial infarction was present in 13.9%. Repolarization abnormalities were common (40.2%): 0.7% had localized ST elevation, 10.5% localized T-wave inversion, and 29.0% had nonspecific repolarization abnormalities ( Table 2) .

A total of 90 patients (11.9%) died during the follow-up period. The relationship of clinical and ECG variables to all-cause mortality is shown in Tables p=0.007). Additional ECG findings associated with mortality were left and right ventricular hypertrophy and previous Q-wave myocardial infarction, but not heart rate or Bazett-corrected QT interval. Notably, localized ST elevation on presentation was a rare event (n=5) limiting any precise estimate of its association with death in our cohort.

In a forward-selection multivariable logistic regression analysis that included age and statistically significant univariate clinical predictors of mortality from Table 4 ).

This analysis of over 750 patients with ECGs is the largest study of noninvasive diagnostic testing in patients with confirmed COVID-19 to date. Our findings revealed that APCs, RBBB/IVB, localized T-wave inversion, and nonspecific repolarization abnormalities were associated with an increased odds of death after accounting for age, and other important clinical characteristics. ST segment elevation, previously described as an important complication of this disease 9 , was rare on presentation. Our findings underscore the potential for ECGs to serve as a valuable tool to inform prognosis, even before many patients develop respiratory failure requiring invasive mechanical ventilation; and provide insights on myocardial processes at play.

APCs occurred in 7.7% of patients, and were associated with a 2.57-fold increased odds of death. Historically, the presence of APCs have correlated with increased left ventricular (LV) filling pressures, especially after acute MI. Whether the presence of APCs in this clinical scenario is indicative of elevated filling pressures and/or increased myocardial stiffness is not known. [10] [11] [12] However, cytokine hypersecretion, a common finding in COVID-19, has also been linked to transient cardiac systolic and diastolic dysfunction. 13 Nonspecific repolarization abnormalities were also associated with death, an observation that has been made in the general population. 5 However, the striking aspect of our observation is that it predicted death in a very short time period-during hospitalization. Meanwhile, myocardial injury as defined by localized ST segment elevation was rare, occurring in just 5 patients, only one of whom died. ST elevation has been described in the setting of COVID-19 9,16 , however our data suggest that this is not a common finding on hospital presentation, and suggest that other findings, that may otherwise be overlooked, merit greater attention up front by clinicians caring for patients with COVID-19.

RBBB/IVB was additionally associated with death, and may reflect a higher incidence of early right ventricular (RV) dysfunction in this population. Given the high rate of respiratory failure in a significant proportion of infected individuals, cardiac dysfunction secondary to increased afterload on the right ventricle (referred to as acute cor pulmonale) is not unexpected.

Although ECG findings are insensitive for the detection of acute cor pulmonale 17 , RBBB has been attributed to acute RV overload and distention in multiple studies of acute pulmonary embolism, with a higher frequency noted in cases with larger clot burdens. 18 In a recent study, RV dysfunction was associated with a higher risk of mortality in COVID-19 patients, independent of all other echocardiographic parameters. 19 To our knowledge, the present study is the first to show that RBBB at the time of presentation, often prior to the development of acute respiratory failure, is associated with worse survival. Importantly, the mechanisms of pulmonary vascular disease in COVID-19 are incompletely understood and may include early vascular endothelial injury, microcirculatory thrombi, and pulmonary venous thromboembolism. 20 Therefore, the ECG may provide early evidence for an "RV at risk" even before the onset of severe hypoxemia, pulmonary vasoconstriction, and overt clinical deterioration.

Our study has several strengths. First, clinical data and outcomes were manually abstracted from the electronic health record using a process that has previously shown high reliability. 1 It is important to note that comorbid conditions manually abstracted from different sections of a medical chart (and often include data from free-text sections of the chart) may be more comprehensive than automated data abstractions from the electronic health record which rely on structured fields and/or billing codes. Second, all ECGs were interpreted by experienced electrocardiographers, which has several advantages including improved accuracy over automated interpretation. 21 Our study also has several limitations. First, as the course of this disease is ongoing in our institution, data on death for all patients are inherently incomplete, which may bias the results. Of note, all patients had at least two weeks of outcome data.

Additionally, we had incomplete lab data for the majority of patients included in our study.

Markers of myocardial injury (namely troponin and B-type natriuretic peptide) were obtained clinically at the discretion of the clinician caring for the patient, and often well after initial presentation to the hospital when these ECGs were obtained. Due to concerns about confounding by indication, we did not include laboratory data in this analysis. Additionally, given we did not have prior ECGs in all patients, we did not compare the presentation ECG to prior ECGs for the presence of new findings, and this should be studied further. We did not formally assess reliability of the two ECG readers. However, these readers review over 100,000 ECG per year at this institution, and have historically been concordant in their interpretation. Finally, given the large number of variables analyzed, it is possible that some of the associations reported occurred by chance. However, our findings are consistent with emerging data that myocardial processes are an important prognostic indicator in COVID-19. 14, 15, 19 In conclusion, our data show that the ECG may be a useful prognostic tool in COVID-19, even among patients who are yet to require invasive mechanical ventilation. Our findings highlight that patients with ECG findings of both left sided heart disease (APCs, IVB, repolarization abnormalities) and right sided disease (RBBB) have a higher odds of death.

Localized ST segment elevation at presentation was rare.

This study received support from NewYork-Presbyterian Hospital (NYPH) and Weill Cornell 

Clinical Characteristics of Covid-19 in

Baseline Characteristics and Outcomes of 1591 Patients Infected With SARS-CoV-2 Admitted to ICUs of the Lombardy Region

Left ventricular hypertrophy by electrocardiogram. Prevalence, incidence and mortality in the Framingham study

Electrocardiographic predictors of incident congestive heart failure and all-cause mortality in post-menopausal women: the Women's Health Initiative

Association of nonspecific minor ST-T abnormalities with cardiovascular mortality: the Chicago Western Electric Study

The Coronary Drug Project Research Group. The prognostic importance of the electrocardiogram after myocardial infarction. Experience from the Coronary Drug Project

Dahlöf B; LIFE Study Investigators. Regression of electrocardiographic left ventricular hypertrophy during antihypertensive treatment and prediction of major cardiovascular events: The LIFE Study

Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support

ST-Segment Elevation in Patients with Covid-19 -A Case Series

Measurement of funny current (I(f)) channel mRNA in human atrial tissue: correlation with left atrial filling pressure and atrial fibrillation

Clinical and pathological studies on incomplete left bundle branch block in the aged

Multiple conduction defects with markedly prolonged ventricular depolarization in cardiomyopathy

Complement-related molecular events in sepsis leading to heart failure

Cardiovascular Implications of Fatal Outcomes of Patients With Coronavirus Disease 2019 (COVID-19)

Elevated Troponin in Patients With Coronavirus Disease

The Variety of Cardiovascular Presentations of COVID-19. Circulation

Relation of electrocardiographic changes in pulmonary embolism to right ventricular enlargement

Appearance of right bundle branch block in electrocardiograms of patients with pulmonary embolism as a marker for obstruction of the main pulmonary trunk

Prognostic Value of Right Ventricular Longitudinal Strain in Patients with COVID-19

Endothelial cell infection and endotheliitis in COVID-19

The diagnostic performance of computer programs for the interpretation of the electrocardiogram