Significance and Usefulness of High‐Resolution 24‐h Electrocardiographic Parameters in Patients With Coronavirus Disease 2019 After Admission: A Prospective Cohort Study Authors: Motohiro Kimata, Kenichi Hashimoto, Naomi Harada, Yusuke Kawamura, Yoshifumi Kimizuka, Mayuko Kaneko, Nobuaki Kiriu, Yasumasa Sekine, Natsumi Iwabuchi, Tetsuro Kiyozumi, Susumu Matsukuma, Yuji Tanaka, Nobutaka Hirooka Published: 2026 ABSTRACT ABSTRACTBackgroundThis study aimed to explore the predictors of disease severity in patients with coronavirus disease 2019 (COVID‐19) using ambulatory electrocardiographic markers (AECG‐Ms), which are prognostic predictors of cardiac death, as well as echocardiography and laboratory parameters.MethodsA prospective cohort study was conducted using data from 70 patients diagnosed with COVID‐19 and admitted to the National Defense Medical College Hospital between January 2021 and December 2023. A high‐resolution Holter electrocardiogram recorder was attached to all patients for 24 h within 48 h of admission, and AECG‐Ms were measured. Laboratory tests and echocardiography were also performed.ResultsLogistic univariate analysis revealed that the T‐wave alternans (TWA) positivity rate was significantly higher in the fatal event group (all‐cause mortality and onset of sustained ventricular tachycardia or ventricular fibrillation) than in the non‐fatal event group. Kaplan–Meier curves at 30 days after admission showed a significant worsening in the survival rate of the TWA and late potential positive groups (log‐rank test, p < 0.0001) compared with the other groups.ConclusionIn patients with COVID‐19 without obvious cardiac disease, AECG‐Ms were highly positive. The study findings may allow easy and rapid prediction of prognosis at the bedside of patients with COVID‐19. Introduction More than seven million patients have died from coronavirus disease 2019 (COVID‐19) globally [1]. Severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2), the causative agent of COVID‐19, enters host cells via the angiotensin‐converting enzyme 2 (ACE2) receptor and mainly causes pneumonia. However, cardiovascular events, such as acute myocardial damage, acute coronary syndrome, cardiac arrhythmias, heart failure, hypercoagulable states, and fatal arrhythmias (such as ventricular tachycardia [VT] and ventricular fibrillation [VF]), can also occur in COVID‐19 [2, 3]. The increased risk of cardiac death in patients with COVID‐19, whether or not they have pre‐existing cardiovascular disease, has also been reported [4]. Furthermore, the risk of cardiac death is particularly high in the presence of underlying risk factors for atherosclerosis [5]. SARS‐CoV‐1, similar to SARS‐CoV‐2 in causing human infection and severe disease, also infects cardiomyocytes via ACE2 receptors and causes myocardial damage [6, 7, 8]. Although electrocardiogram (ECG) changes in COVID‐19 have been reported to include QT prolongation and ST‐T changes [9, 10, 11], surrogate markers to predict prognosis have not yet been established. Therefore, we explored ECG indices other than QT prolongation and ST‐T changes that could accurately predict the prognosis of patients with COVID‐19. Recently, T‐wave alternans (TWA) and late potential (LP) indices suggestive of myocardial organic abnormalities, heart rate variability (HRV) and heart rate turbulence (HRT) indices suggestive of autonomic dysfunction, and non‐sustained ventricular tachycardia (NSVT) have been reported to be useful ECG indicators for predicting fatal arrhythmias or sudden cardiac death [12, 13, 14]. These five ECG indices (TWA, LP, HRV, HRT, and NSVT) are ambulatory electrocardiographic markers (AECG‐Ms), which can be assessed simultaneously using high‐resolution Holter ECG. A multicenter study reported that AECG‐Ms can accurately predict the incidence and prognosis of fatal arrhythmias in patients with organic heart disease and chronic renal failure and that combining multiple AECG‐Ms, such as TWA and LP, is particularly useful for risk stratification and has a high level of evidence [15, 16]. Recently, we reported in three case series that LP, TWA, HRV, and HRT were potential risk factors for COVID‐19 pneumonia in the early phase of admission [17]. Therefore, we hypothesized that cardiovascular complications in patients with COVID‐19 could be assessed with AECG‐Ms and that this would be useful for predicting prognosis. In this study, we aimed to assess whether AECG‐Ms in patients with COVID‐19 in the early phase of hospitalization are prognostic factors and whether AECG‐M results are more useful than other clinical markers. We also evaluated whether the observed changes in AECG‐Ms were specific to COVID‐19 and reversible during the acute and convalescent phases of COVID‐19. Methods Study Population The study population comprised patients with COVID‐19 admitted to the National Defense Medical College Hospital between January 2021 and December 2023. Patients who presented to the National Defense Medical College Hospital between January 2023 and May 2024 with a febrile inflammatory response to diseases other than COVID‐19 were also included in the study as a comparison group (non‐COVID‐19 group). The inclusion criteria for the comparison group were absence of a history of COVID‐19, presence of fever or an elevated inflammatory response, with or without underlying cardiac disease. The exclusion criteria were right or left bundle branch block, persistent atrial fibrillation/flutter, II/III atrioventricular block, psychiatric disease, age < 18 years, and an Acute Physiology and Chronic Health Evaluation II (APACHE) score less than 3 points. We expected the severity of illness to be standardized between the COVID‐19 and non‐COVID‐19 groups by selecting patients with an APACHE II score of > 4 points in both groups. The study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Board of the National Defense Medical College (approval number 4772). Written informed consent was obtained from all the patients. The observation period was from the date of admission to the date of death, or 90 days after admission. Study Protocol A high‐resolution Holter ECG recorder was attached to the patient for 24 h within 48 h of admission, and the AECG‐Ms were evaluated. Concurrently, laboratory tests and echocardiography were performed to identify prognostic clinical markers. Fatal events were defined as all‐cause mortality or onset of sustained VT or VF. On the other hand, serious cardiac events were defined as the onset of sustained VT or VF and significant cardiac death, such as heart failure. The observation period was from the date of admission to the date of death or 90 days after admission (Study 1). The same examinations were performed in the non‐COVID‐19 group (Study 2). To monitor changes in AECG‐Ms from the acute to the recovery phase, they were also assessed in patients with COVID‐19 during the recovery phase, starting 90 days after the illness, using a high‐resolution 24‐h electrocardiograph (Study 3). Routine Holter ECG Analysis Holter ECG data were acquired on an FM‐1500 (terminal) (1.5 V, 1000 Hz; Fukuda Denshi Co., Tokyo, Japan). Analyses were performed using a dedicated Holter ECG analysis system SCM‐8000 (Fukuda Denshi Co., Tokyo, Japan). Holter ECG recordings were performed during hospitalization, and the patients in both the COVID‐19 and control groups were free to move. However, some severely ill patients required oxygen and were forced to undergo recording while in bed. After automatic analysis, manual reanalysis was performed by a clinical laboratory technician specializing in cardiology. Parameters analyzed as routine Holter ECG items included total heart rate; maximum, minimum, and mean heart rates; number of premature ventricular contractions (PVCs); and presence of NSVT. NSVT is a prognostic ECG indicator of sudden cardiac death and mortality [18]. Similar to the definition by Ikeda et al. [18] and de Sousa et al. [19], we defined NSVT as three or more consecutive PVCs of ≥ 100 bpm. AECG‐M Analysis The TWA, LP, HRT, and HRV were evaluated. A dedicated Holter ECG analysis system, SCM‐8000 (Fukuda Denshi, Tokyo, Japan), was used for routine AECG‐M analysis. Double‐blind AECG‐M analysis was performed by a physician specializing in cardiology and a clinical laboratory technician specializing in cardiology. All ECG recordings were visually inspected, and arrhythmias, such as ectopic beats, artifacts, arrhythmic segments, supraventricular tachycardia, and atrial fibrillation, were excluded by manual editing before HRV analysis. TWA was positive when the reference value was exceeded; the reference value was 19.9 μV when the noise level was less than 10 mV and 23.6 μV when the noise level was between 10 and 20 mV (Figure 1) [20]. LPs were positive when at least two of the following three criteria were met: (1) corrected‐filtered QRS duration (c‐fQRS) > 105 ms, (2) root‐mean‐square voltage of the signals in the last 40 ms (RMS40) < 15 μV, and (3) duration of the low‐amplitude signal after the voltage decreased to less than 40 μV (LAS40) > 39 ms (Figure 1B) [21]. For HRV, the standard deviation of normal NN intervals (SDNN) was used as a main parameter, and the cutoff value for poor prognosis was set as SDNN < 75 ms [22]. Furthermore, we analyzed the coefficient of variation of NN intervals (CVNN), proportion of consecutive NN intervals differing by 50 milliseconds or more (pNN50), low‐frequency (LF), high‐frequency (HF), and LF/HF. To evaluate circadian variability, we also assessed HRV by dividing it into the following time periods: 00:00–06:00, 06:00–12:00, 12:00–18:00, and 18:00–24:00. To determine the HRT, the turbulence onset (TO) and turbulence slope (TS) were used as parameters. TO was calculated using the following equation: TO=RR1+RR2−RR−2+RR−1RR−2+RR−1×100% RR1 and RR2 are the two RR intervals immediately after PVC. RR−1 and RR−2 were the two RR intervals immediately before the PVC. TS was defined as the maximum positive slope of the first post‐PVC RR interval plotted over five consecutive times. A patient was determined to have abnormal HRT if the TO was ≥ 0% and the TS was ≤ 2.5 milliseconds per RR interval (Category 2) (Figure 1C) [23]. (A) Kaplan–Meier curves for 90 days from admission with all‐cause mortality as the endpoint. The survival curves for the groups (1) negative for both LP and TWA, (2) positive for LP only, (3) positive for TWA only, and (4) positive for both LP and TWA. (B) Kaplan–Meier curves for 90 days from admission with the endpoints defined as cardiac death and fatal arrhythmia. The survival curves for the groups (1) negative for both LP and TWA, (2) positive for LP only, (3) positive for TWA only, and (4) positive for both LP and TWA. LP, Late potential; TWA, T‐wave alternans. 12‐Lead ECG Examination The QT and corrected QT (QTc) intervals were measured automatically from the body‐surface 12‐lead ECG using the tangential method. QTc was calculated using Fridericia's correction formula because Bazett's correction formula under‐corrects tachycardia, resulting in a longer QT [24]. Laboratory Tests General biochemistry and blood counts were measured within 48 h of admission. Blood samples were collected and analyzed at the Department of Laboratory Medicine of the National Defense Medical College Hospital. Laboratory assessments included complete blood count and creatinine, estimated glomerular filtration rate (eGFR), C‐reactive protein (CRP), B‐type natriuretic peptide (BNP), D‐dimer, and plasma troponin I concentrations. Soluble fibrin monomer complex (SFMC) levels were measured using a latex agglutination assay in a certified commercial laboratory. COVID‐19 is often complicated by disseminated intravascular coagulation [25]. Therefore, SFMC, which is considered a sensitive indicator of disseminated intravascular coagulation [26], was measured. Echocardiography Echocardiography was performed using the Affiniti (Philips Medical Systems, Andover, MA, USA) system to evaluate contractility and the diastolic capacity of the left and right cardiac chambers. The left ventricular ejection fraction (LVEF) was calculated during sinus rhythm using Simpson's method. The right ventricular fractional area change and tricuspid annular plane systolic excursion were measured to assess right ventricular function. The ratio of early diastolic filling velocity divided by atrial filling velocity (E/A) was also calculated to assess diastolic capacity. Statistical Methods Data are presented as means ± standard deviation for normally distributed continuous variables and medians (interquartile range) for non‐normally distributed variables. The normal distribution of continuous variables was assessed using the Shapiro–Wilk test. Patient characteristics were compared using the χ2 test for categorical variables, the Student's t‐test for continuous data, and the Mann–Whitney U test for non‐normal distributions. Univariate analyses using the Cox proportional hazards model were performed during the 90‐day observation period to examine the association between fatal events and clinical parameters. Furthermore, univariate logistic regression analysis was performed during the 90‐day observation period to examine the association between serious cardiac events and clinical parameters. For each AECG‐M, cardiac event‐free survival was calculated using the Kaplan–Meier method, and significant differences in lethal events were calculated using the log‐rank test. Comparison of HRV parameters among each time interval (0:00–6:00, 6:00–12:00, 12:00–18:00, and 18:00–24:00) was performed using Friedman analysis. All statistical procedures were performed using SPSS Statistics version 25 (IBM Corp., Armonk, NY, USA), and statistical significance was set at p < 0.05. Results Study 1Comparisons between groups among patients with COVID‐19. Comparisons between groups among patients with COVID‐19. Patient Characteristics Table 1 shows the patient characteristics, history, medications, echocardiographic findings, and laboratory results at 90 days after admission (without fatal events vs. with fatal events). Seventy patients (mean age: 64.0 ± 14.1 years, 40 males) admitted with a diagnosis of COVID‐19 were included in the study. Twelve patients experienced fatal events within 90 days of admission. Patients in the fatal event group were significantly older than those in the event‐free group among patients with COVID‐19 (p < 0.001). Cox regression analysis (univariate) showed that the number of PVCs and TWA, LP, HRT, and NSVT positivity were significant risk factors for fatal events (p = 0.02–0.002; hazard ratio [HR] 3.26–8.34). TABLE 1Characteristics of study patients with COVID‐19 (without fatal events vs. with fatal events).Without fatal events a (n = 58)With fatal events (n = 12) p Cox proportional hazards analysis (univariate) HR [95% CI] p Age (years)63.0 (53.0, 74.0)76.0 (66.3, 79.5)0.0071.08 [1.01–1.15]0.02Sex: Male, n (%)33 (55.9)7 (63.6)0.990.93 [0.29–2.93]0.90BMI22.8 (19.9, 29.2)19.8 (18.3, 20.8)0.060.89 [0.78–1.02]0.08Vital signs on admissionBody temperature (°C)37.5 ± 1.037.5 ± 1.00.950.98 [0.55–1.73]0.93sBP (mmHg)130.0 (111.8, 144.0)120.0 (107.5, 157.5)0.950.99 [0.98–1.02]0.83dBP (mmHg)79.1 ± 17.579.9 ± 13.60.881.01 [0.97–1.04]0.78Heart rate (/min)89.5 (79.8, 105.3)98.5 (85.3, 112.3)0.271.02 [0.99–1.05]0.08Respiratory rate (/min)21.5 (18.0, 25.8)21.5 (18.5, 24.0)0.910.99 [0.91–1.08]0.86APACHE‐II score9.0 (6.0, 12.3)10.5 (8.0, 14.8)0.191.02 [0.94–1.10]0.68HistorySmoking, n (%)32 (55.2)6 (50.0)0.760.78 [0.25–2.42]0.78Diabetes mellitus, n (%)24 (41.4)6 (50.0)0.751.47 [0.47–4.56]0.50Hypertension, n (%)27 (46.6)9 (75.0)0.113.27 [0.89–12.1]0.08Dyslipidemia, n (%)17 (29.3)4 (33.3)0.991.21 [0.36–4.01]0.76Hyperuricemia, n (%)19 (32.8)3 (25.0)0.520.59 [0.16–2.167]0.59CKD G1‐G3a, n (%)45 (77.6)6 (50.0)0.070.34 [0.11–1.07]0.06CKD G3b‐G5, n (%)13 (22.4)6 (50.0)0.072.91 [0.94–9.03]0.06Cardiac disease, n (%)4 (6.9)3 (25.0)0.093.19 [0.86–11.8]0.08Family history of SCD, n (%)1 (1.7)0 (0.0)0.99—0.77MedicationCCB, n (%)10 (17.2)4 (33.3)0.222.37 [0.69–8.12]0.17RAS inhibitor, n (%)7 (12.1)1 (8.3)0.990.72 [0.09–5.60]0.75β‐blocker, n (%)7 (12.1)0 (0.0)0.59—0.45UCG resultsLVDd (mm)48.2 (44.0, 50.7)50.8 (47.5, 52.6)0.151.05 [0.94–1.16]0.41LVEF (%)66.3 (61.4, 71.0)72.7 (58.2, 78.5)0.251.03 [0.96–1.11]0.43RVFAC (%)48.5 (45.4, 51.5)47.1 (43.0, 51.0)0.490.98 [0.90–1.06]0.59E/A1.00 (0.78, 1.23)1.06 (0.83, 1.36)0.801.01 [0.19–5.36]0.99Laboratory resultsWBC count (/μL)7400 (4600, 10 800)9900 (5700, 12 300)0.111.00 [0.99–1.00]0.1Procalcitonin (ng/dL)0.20 (0.07, 1.29)0.60 (0.27–2.64)0.120.97 [0.84–1.12]0.69BNP (pg/mL)35.9 (13.9, 72.6)53.0 (24.8, 338.0)0.121.00 [0.99–1.00]0.72Troponin‐I (pg/mL)10.0 (0.5, 25.9)15.2 (0.5, 218.9)0.281.00 [0.99–1.00]0.99D‐dimer (μg/mL)1.9 (0.8, 5.3)3.3 (1.5, 6.5)0.230.99 [0.90–1.09]0.77SF (μg/mL)4.6 (3.2, 11.8)3.3 (3.0, 19.2)0.601.00 [0.99–1.02]0.54CRP (mg/dL)5.12 (2.40, 10.11)8.90 (3.83, 18.63)0.141.02 [0.97–1.07]0.46Cr (mg/dL)0.89 (0.64, 1.66)1.16 (0.71–1.99)0.381.03 [0.86–1.24]0.73eGFR (mL/min)60.5 ± 35.348.7 ± 35.00.300.99 [0.98–1.01]0.3ECG resultsQT interval (msec)382.0 (354.0, 404.0)367.0 (331.0, 383.5)0.350.99 [0.98–1.02]0.98QTc interval (msec)419.9 (397.6, 440.8)424.8 (389.8, 438.2)0.781.01 [0.99–1.03]0.34PVC (/24 h)8.5 (1.8–185.5)465.5 (7.5–1757.5)0.261.00 [1.00–1.01]0.03AECG‐M resultsLP positive, n (%)18 (31.0)7 (58.3)0.043.93 [1.02–15.2]0.04TWA positive, n (%)9 (15.5)7 (58.3)0.0028.34 [2.14–32.5]0.002HRT positive, n (%)14 (24.1)7 (58.3)0.043.45 [1.09–10.9]0.04HRV(SDNN) abnormal, n (%)27 (46.6)9 (75.0)0.113.26 [0.88–12.0]0.08NSVT positive, n (%)3 (5.2)4 (33.3)0.016.00 [1.80–20.0]0.004 Note: The values in parentheses are the lower and upper limits of the 95% confidence interval and the other values are the mean and standard deviation.Abbreviations: APACHE, acute physiology and chronic health evaluation; BMI, body mass index; BNP, brain natriuretic peptide; CKD, chronic kidney disease; Cr, creatinine; CRP, C‐reactive protein; dBP, diastolic blood pressure; E/A, rate between early diastolic filling velocity and atrial filling velocity; eGFR, estimated glomerular filtration rate; fQRS, filtered QRS duration; HRT, heart rate turbulence; HRV, heart rate variability; LAS40, low‐amplitude signal below 40 μV; LP, late potential; LVDd, Left Ventricular Dimension diastole; LVEF, left ventricular ejection fraction; NSVT, non‐sustained ventricular tachycardia; NYHA, New York Heart Association; PVC, premature ventricular contraction; RAS, Renin‐Angiotensin‐Aldosterone System; RMS40, root‐mean‐square voltage at last 40 ms; RVFAC, right ventricular fractional area change; sBP, systolic blood pressure; SCD, sudden cardiac death; SDNN, standard deviation of normal RR intervals; SF, soluble fibrin monomer complex; TWA, T‐wave alternans; UCG, ultrasound cardiography; WBC, white blood cell. a Fatal events = all‐cause mortality or onset of sustained ventricular tachycardia or fibrillation. Characteristics of study patients with COVID‐19 (without fatal events vs. with fatal events). Note: The values in parentheses are the lower and upper limits of the 95% confidence interval and the other values are the mean and standard deviation. Abbreviations: APACHE, acute physiology and chronic health evaluation; BMI, body mass index; BNP, brain natriuretic peptide; CKD, chronic kidney disease; Cr, creatinine; CRP, C‐reactive protein; dBP, diastolic blood pressure; E/A, rate between early diastolic filling velocity and atrial filling velocity; eGFR, estimated glomerular filtration rate; fQRS, filtered QRS duration; HRT, heart rate turbulence; HRV, heart rate variability; LAS40, low‐amplitude signal below 40 μV; LP, late potential; LVDd, Left Ventricular Dimension diastole; LVEF, left ventricular ejection fraction; NSVT, non‐sustained ventricular tachycardia; NYHA, New York Heart Association; PVC, premature ventricular contraction; RAS, Renin‐Angiotensin‐Aldosterone System; RMS40, root‐mean‐square voltage at last 40 ms; RVFAC, right ventricular fractional area change; sBP, systolic blood pressure; SCD, sudden cardiac death; SDNN, standard deviation of normal RR intervals; SF, soluble fibrin monomer complex; TWA, T‐wave alternans; UCG, ultrasound cardiography; WBC, white blood cell. Fatal events = all‐cause mortality or onset of sustained ventricular tachycardia or fibrillation. Table 2 shows the patient characteristics, history, medications, echocardiographic findings, and laboratory results at 90 days after admission (without fatal events vs. with serious cardiac events [sustained VT/VF or death due to heart failure]). Six patients experienced serious cardiac events (sustained VT/VF or death from heart failure) within 90 days of admission. The serious cardiac event group was significantly older than the event‐free group among patients with COVID‐19 (p = 0.04). Logistic regression analysis (univariate) showed that the number of PVCs and TWA and NSVT positivity were significant risk factors for serious cardiac events (p = 0.04–0.01; hazard ratio [HR] 9.17–22.8). TABLE 2Characteristics of study patients with COVID‐19 (without fatal events vs. with serious cardiac events).Without fatal events a (n = 58)With serious cardiac events(n = 6) p Logistic regression analysis (univariate) HR (95% CI) p Age (years)63.0 (53.0, 74.0)76.5 (69.75, 79.5)0.0121.13 [1.00–1.27]0.04Sex: Male, n (%)33 (55.9)4 (66.7)0.640.66 [0.11–3.90]0.65BMI22.8 (19.9, 29.2)20.4 (19.3, 30.1)0.540.96 [0.82–1.12]0.58Vital signs on admissionBody temperature (°C)37.5 ± 1.037.2 ± 1.00.530.48 [0.15–1.53]0.22sBP (mmHg)130.0 (111.8, 144.0)139.0 (109.3, 158.3)0.731.02 [0.98–1.06]0.32dBP (mmHg)79.1 ± 17.577.7 ± 13.50.820.96 [0.89–1.03]0.26Heart rate (/min)89.5 (79.8, 105.3)95.0 (77.3, 126.8)0.781.03 [0.99–1.08]0.14Respiratory rate (/min)21.5 (18.0, 25.8)21.0 (17.3, 26.5)0.970.93 [0.77–1.12]0.42APACHE‐II score9.0 (6.0, 12.3)14.0 (7.8, 15.3)0.161.09 [0.93–1.28]0.27HistorySmoking, n (%)32 (55.2)3 (50.0)0.810.81 [0.15–4.37]0.81Diabetes mellitus, n (%)24 (41.4)4 (66.7)0.232.83 [0.48–16.73]0.25Hypertension, n (%)27 (46.6)6 (100.0)0.015.74 [0.63–52.2]0.12Dyslipidemia, n (%)17 (29.3)3 (50.0)0.302.35 [0.43–12.9]0.32Hyperuricemia, n (%)19 (32.8)3 (50.0)0.401.74 [0.32–9.48]0.52CKD G1‐G3a, n (%)45 (77.6)3 (50.0)0.140.76 [0.13–4.50]0.76CKD G3b‐G5, n (%)13 (22.4)3 (50.0)0.141.46 [0.15–14.4]0.75Cardiac disease, n (%)4 (6.9)3 (50.0)0.00113.5 [2.03–89.8]0.01Family history of SCD, n (%)1 (1.7)0 (0.0)0.76—1.00MedicationCCB, n (%)10 (17.2)4 (66.7)0.0059.60 [1.54–59.8]0.02RAS inhibitor, n (%)7 (12.1)1 (16.6)0.371.46 [0.15–14.4]0.75β‐blocker, n (%)7 (12.1)1 (16.6)0.751.46 [0.15–14.4]0.75UCG resultsLVDd (mm)48.2 (44.0, 50.7)51.0 (39.2, 52.7)0.371.02 [0.87–1.19]0.84LVEF (%)66.3 (61.4, 71.0)61.3 (44.5, 79.6)0.840.99 [0.90–1.08]0.80RVFAC (%)48.5 (45.4, 51.5)45.8 (37.8, 50.8)0.750.98 [0.87–1.10]0.68E/A1.00 (0.78, 1.23)0.97 (0.83, 1.06)0.710.35 [0.01–9.82]0.53Laboratory resultsWBC count (/μL)7400 (4600, 10 800)11 775 ± 53390.201.00 [0.99–1.00]0.09Procalcitonin (ng/dL)0.20 (0.07, 1.29)0.60 (0.33–2.63)0.170.96 [0.73–1.25]0.75BNP (pg/mL)35.9 (13.9, 72.6)193.0 (27.4, 410.2)0.031.00 [0.99–1.00]0.45Troponin‐I (pg/mL)10.0 (0.5, 25.9)122.2 (3.0, 263.0)0.101.00 [0.99–1.00]0.95D‐dimer (μg/mL)1.9 (0.8, 5.3)3.3 (1.2, 8.3)0.360.98 [0.85–1.13]0.81SF (μg/mL)4.6 (3.2, 11.8)3.2 (3.0, 21.0)0.500.98 [0.93–1.05]0.67CRP (mg/dL)5.12 (2.40, 10.11)20.00 (13.48, 22.63)0.181.04 [0.97–1.11]0.27Cr (mg/dL)0.89 (0.64, 1.66)1.46 (0.76–9.72)0.191.10 [0.88–1.39]0.40eGFR (mL/min)60.5 ± 35.341.8 ± 37.40.180.98 [0.96–1.01]0.20ECG resultsQT interval (ms)382.0 (354.0, 404.0)378.0 (358.0, 487.9)0.521.02 [0.99–1.04]0.07QTc interval (ms)419.9 (397.6, 440.8)432.0 (418.2, 495.6)0.081.02 [1.00–1.04]0.04PVC (/24 h)8.5 (1.8–185.5)1020.5 (245.3–10021.0)0.0011.00 [1.0001–1.0004]0.04AECG‐M resultsLP positive, n (%)18 (31.0)5 (83.3)0.019.72 [1.06–89.6]0.05TWA positive, n (%)9 (15.5)5 (83.3)0.000122.8 [2.37–219.4]0.01HRT positive, n (%)14 (24.1)4 (66.7)0.033.43 [0.56–21.2]0.19HRV(SDNN) abnormal, n (%)27 (46.6)4 (66.7)0.352.30 [0.39–13.5]0.36NSVT positive, n (%)3 (5.2)2 (33.3)0.019.17 [1.17–71.7]0.04 Note: The values in parentheses are the lower and upper limits of the 95% confidence interval and the other values are the mean and standard deviation.Abbreviations: APACHE, acute physiology and chronic health evaluation; BMI, body mass index; BNP, brain natriuretic peptide; CKD, chronic kidney disease; Cr, creatinine; CRP, C‐reactive protein; dBP, diastolic blood pressure; E/A, rate between early diastolic filling velocity and atrial filling velocity; eGFR, estimated glomerular filtration rate; fQRS, filtered QRS duration; HRT, heart rate turbulence; HRV, heart rate variability; LAS40, low‐amplitude signal below 40 μV; LP, late potential; LVDd, Left Ventricular Dimension diastole; LVEF, left ventricular ejection fraction; NSVT, non‐sustained ventricular tachycardia; NYHA, New York Heart Association; PVC, premature ventricular contraction; RAS, Renin‐Angiotensin‐Aldosterone System; RMS40, root‐mean‐square voltage at last 40 ms; RVFAC, right ventricular fractional area change; sBP, systolic blood pressure; SCD, sudden cardiac death; SDNN, standard deviation of normal RR intervals; SF, soluble fibrin monomer complex; TWA, T‐wave alternans; UCG, ultrasound cardiography; WBC, white blood cell. a Fatal events = all‐cause mortality or onset of sustained ventricular tachycardia or fibrillation. Characteristics of study patients with COVID‐19 (without fatal events vs. with serious cardiac events). Note: The values in parentheses are the lower and upper limits of the 95% confidence interval and the other values are the mean and standard deviation. Abbreviations: APACHE, acute physiology and chronic health evaluation; BMI, body mass index; BNP, brain natriuretic peptide; CKD, chronic kidney disease; Cr, creatinine; CRP, C‐reactive protein; dBP, diastolic blood pressure; E/A, rate between early diastolic filling velocity and atrial filling velocity; eGFR, estimated glomerular filtration rate; fQRS, filtered QRS duration; HRT, heart rate turbulence; HRV, heart rate variability; LAS40, low‐amplitude signal below 40 μV; LP, late potential; LVDd, Left Ventricular Dimension diastole; LVEF, left ventricular ejection fraction; NSVT, non‐sustained ventricular tachycardia; NYHA, New York Heart Association; PVC, premature ventricular contraction; RAS, Renin‐Angiotensin‐Aldosterone System; RMS40, root‐mean‐square voltage at last 40 ms; RVFAC, right ventricular fractional area change; sBP, systolic blood pressure; SCD, sudden cardiac death; SDNN, standard deviation of normal RR intervals; SF, soluble fibrin monomer complex; TWA, T‐wave alternans; UCG, ultrasound cardiography; WBC, white blood cell. Fatal events = all‐cause mortality or onset of sustained ventricular tachycardia or fibrillation. Association Between AECG‐Ms and Fatal or Cardiac Events in the COVID‐19 Group Table 3 compares the prediction accuracies of the AECG‐Ms alone and in combination. The LP + TWA assessment had a higher specificity, positive predictive accuracy, negative predictive accuracy, and accuracy of 98.1%, 80.0%, 89.5%, and 88.7%, respectively. TABLE 3Comparison of prediction accuracy for individual and combined AECG‐Ms.SensitivitySpecificityPositive predictive accuracyNegative predictive accuracyAccuracyTWA (+)70.082.743.893.580.6LP (+)70.066.028.084.166.7NSVT (+)33.394.857.187.384.3TWA (+) + LP (+)40.098.180.089.588.7TWA (+) + NSVT (+)98.325.086.475.085.7LP (+) + NSVT (+)16.798.366.785.184.3TWA (+) + LP (+) + NSVT (+)8.3100.0100.084.184.3Abbreviations: LP, late potential; NSVT, non‐sustained ventricular tachycardia; TWA, T‐wave alternans. Comparison of prediction accuracy for individual and combined AECG‐Ms. Abbreviations: LP, late potential; NSVT, non‐sustained ventricular tachycardia; TWA, T‐wave alternans. Figure 1 shows the Kaplan–Meier curves 90 days after admission, with all‐cause mortality as the endpoint. The survival rate was significantly worse in the TWA‐ and LP‐positive groups (log‐rank test, p < 0.001) than in other groups (TWA‐ and LP‐negative, TWA‐positive and LP‐negative, and TWA‐negative and LP‐positive). Figure 1B shows the Kaplan–Meier curves at 90 days after admission, with the endpoints defined as cardiac death and fatal arrhythmia. The survival rate was significantly worse in the TWA‐ and LP‐positive groups (log‐rank test, p = 0.001) than in other groups (TWA‐ and LP‐negative, TWA‐positive and LP‐negative, and TWA‐negative and LP‐positive). Table S1 shows the association between the HRV parameters and the risk of all‐cause mortality and major cardiac events by time of day. None of these parameters were significant predictors of all‐cause mortality or major cardiac events. Clinical Characteristics of Patients Who Experienced a Fatal Event (All‐Cause Mortality or VT or VF Onset) Table 4 shows the clinical characteristics of the 12 patients who experienced a fatal event, including age, sex, positive AECG‐M items, sustained VT or VF, time from admission to the fatal event, and the main cause of death. At least one positive AECG‐M item was present in all patients who experienced a fatal event. The mean time from Holter ECG measurement to the fatal event was 26.3 days; the main cause of death in six cases was respiratory failure; however, all patients had heart failure at the end of life. Five patients presented with sustained VT. Of the 10 deaths, six were associated with mild heart failure at the end of life, although respiratory failure remained the predominant cause of death. Although the fatal events up to day 30 included pneumonia, SVT, and heart failure, none of these events occurred during days 30–90. Instead, cancer‐related deaths accounted for two‐thirds (67%) of all deaths. TABLE 4Characteristics of cases that experienced a fatal event (all‐cause mortality or VT or VF onset).No.Age (years)SexTWA (+)LP (+)HRV (+)HRT abnormalNSVT (+)Both LP and TWA (+)CRP (mg/dL)BNP (pg/mL)S‐VTDays between admission and eventMain cause of death166F(−)(+)(+)(−)(−)(−)14.5927.1S‐VT1(Survived)284M(+)(+)(−)(+)(+)(−)2.482.9S‐VT3(Survived)367M(+)(+)(+)(+)(+)(+)5.8162(−)8ARDS482F(+)(−)(−)(−)(−)(−)14.553(−)9Respiratory failure576M(−)(+)(+)(−)(+)(−)6.434.3(−)10Respiratory failure676M(+)(+)(+)(+)(−)(+)20.320.5S‐VT12Respiratory failure777M(+)(−)(+)(+)(+)(−)11.4434.3S‐VT12Gastrointestinal hemorrhage, VT878F(+)(+)(+)(+)(+)(+)19.7338(−)14 Respiratory failure Heart failure 971M(+)(+)(−)(−)(−)(+)23.447.9S‐VT22Pneumonia1080M(+)(−)(+)(+)(−)(−)5.724.8(−)57Renal cell cancer stage IV1158F(−)(+)(+)(+)(−)(−)15.4(−)(−)79Ovarian cancer stage III1263F(−)(−)(+)(−)(−)(−)3.26.9(−)90Tongue cancer stage IVbAbbreviations: ARDS, acute respiratory distress syndrome; BNP, brain natriuretic peptide; CRP, C‐reactive protein; HRT, heart rate turbulence; HRV, heart rate variability; LP, late potential; NSVT, non‐sustained ventricular tachycardia; Paf, paroxysmal atrial fibrillation; S‐VT, sustained ventricular tachycardia; TWA, T‐wave alternans; VF, ventricular fibrillation; VT, ventricular tachycardia. Characteristics of cases that experienced a fatal event (all‐cause mortality or VT or VF onset). Respiratory failure Heart failure Abbreviations: ARDS, acute respiratory distress syndrome; BNP, brain natriuretic peptide; CRP, C‐reactive protein; HRT, heart rate turbulence; HRV, heart rate variability; LP, late potential; NSVT, non‐sustained ventricular tachycardia; Paf, paroxysmal atrial fibrillation; S‐VT, sustained ventricular tachycardia; TWA, T‐wave alternans; VF, ventricular fibrillation; VT, ventricular tachycardia. Presentation of a Representative Case With Multiple Positive AECG‐Ms (Case 1) A 67‐year‐old man was admitted to our hospital with a diagnosis of COVID‐19 after developing dyspnea and coughing. He had a history of postoperative lung cancer and renal cancer and was undergoing treatment for diabetes mellitus, but had no history of significant cardiac disease. On admission, his white blood cell (WBC) count was 7100 /μL and CRP was 6.5 mg/dL. Echocardiography revealed an LVEF of 64.2%. Chest radiographs and computed tomography showed bilateral ground‐glass opacities. The patient received remdesivir (100 mg/day) and steroid pulse therapy; however, his dyspnea worsened. His brain natriuretic peptide (BNP) level increased from 162 to 611 pg/mL, and heart failure was suspected. On the 8th day of admission, the patient died of respiratory failure concomitant with heart failure. Although it was difficult to predict the prognosis based on laboratory and echocardiography results measured within 48 h of admission, all five AECG‐M parameters were positive (Table 5, Figure 2A–C).Study 2Comparison of each parameter between the COVID‐19 and non‐COVID‐19 groups. Comparison of each parameter between the COVID‐19 and non‐COVID‐19 groups. TABLE 5AECG‐M results in Case 1.Positive reference valueResultsTWA positive(+)TWA (μV)> 19.9 (*1)35.7LP positive(*2)(+)fQRS (ms)> 105160RMS40 (μV)< 1510.5LAS40 (ms)> 3957HRT positive(+)HRV abnormality(+)SDNN (ms)< 7564.0NSVT positive(−) Note: (*1) Noise level < 10. (*2) When two or more c‐fQRS, RMS40, and LAS40 were positive, the patient was classified as having LP (+).Abbreviations: fQRS, fragmented QRS; HRT, heart rate turbulence; HRV, heart rate variability; LAS40, low‐amplitude signal below 40 μV; LP, late potential; RMS40, root‐mean‐square voltage at last 40 ms; SDNN, standard deviation of normal RR intervals. AECG‐M results in Case 1. Note: (*1) Noise level < 10. (*2) When two or more c‐fQRS, RMS40, and LAS40 were positive, the patient was classified as having LP (+). Abbreviations: fQRS, fragmented QRS; HRT, heart rate turbulence; HRV, heart rate variability; LAS40, low‐amplitude signal below 40 μV; LP, late potential; RMS40, root‐mean‐square voltage at last 40 ms; SDNN, standard deviation of normal RR intervals. (A) TWA results using the high‐resolution Holter electrocardiogram in Case 1. TWA: 35.7 μV, noise level: 12.3 mV. When the noise level is between 10 mV and 20 mV, the cutoff value is 23.6 μV, and when the alternans duration is 1 min or more and the TWA is equal to or greater than the cutoff value, TWA is positive. The patient's TWA was positive. TWA, T‐wave alternans. (B) LP results using the high‐resolution Holter electrocardiogram in Case 1. c‐fQRSd (corrected filtered QRS duration): 160 ms (reference value: > 105 ms), RMS40 (root‐mean‐square voltage of the signals in the last 40 ms): 10.4 μV (reference value: < 15 μV), LAS40 (duration of the low‐amplitude signal after the voltage decreased to less than 40 μV): 57 ms (reference value: > 39 ms). The LP of the patient was positive. LP, late potential. (C) HRT results using the high‐resolution Holter electrocardiogram in Case 1. TO (turbulence onset): 0.01 (abnormal value: ≥ 0%), TS (turbulence slope): 1.00 (abnormal value: ≤ 2.5 ms per RR interval). The patient's HRT was Category 2 and was determined to be positive. HRT, heart rate turbulence. Comparison Between the COVID‐19 and Non‐COVID‐19 Groups The non‐COVID‐19 group comprised 24 participants (mean age: 60.5 ± 16.2 years, 12 males) included in the analysis. The main etiology of the background disease in this group was infection (17 patients, 70.8%), with a high proportion of respiratory tract infections (4 patients, 16.7%). Table S2 shows the characteristics of the non‐COVID‐19 and COVID‐19 groups. The COVID‐19 group had a significantly greater left ventricular end‐diastolic dimension than the non‐COVID‐19 group (p = 0.004). The WBC count and CRP levels were significantly higher in the non‐COVID‐19 group than in the COVID‐19 group (WBC count, p = 0.017; CRP level, p = 0.003). Table S3 shows the AECG‐M parameters between the COVID‐19 and non‐COVID‐19 groups, with the COVID‐19 group having significantly higher TWA and LP positivity rates than the non‐COVID‐19 group (TWA, p = 0.009; LP, p < 0.001). Tables S4 and S5 show the circadian variability of HRV indices in the COVID‐19 and control groups, respectively. Although the COVID‐19 group exhibited significant diurnal variations in HRV for LF and HF only, the control group exhibited significant diurnal variations in HRV for SDNN, CVNN, pNN50%, LF, and HF.Study 3Comparison of AECG‐M immediately after COVID‐19 illness and during recovery. Comparison of AECG‐M immediately after COVID‐19 illness and during recovery. Changes in AECG‐M From the Acute Phase to the Recovery Phase AECG‐M was measured in 15 patients with COVID‐19 during the recovery period (mean: 323 ± 149 days after the first measurement). Changes in TWA, c‐fQRS, and the standard deviation of normal RR intervals (SDNN) from the acute phase to the recovery phase are shown in Figure S1C. There was a significant improvement in TWA and c‐fQRS values in the recovery phase compared to those in the acute phase (TWA, p = 0.013; c‐fQRS, p = 0.04) and a significant decrease in the number of patients with TWA positivity (p = 0.01). Eight patients (53.3%) were TWA‐ or LP‐positive. Of the 15 patients in Study 3, the SDNN improved in 13 patients from the acute phase to the recovery phase (p < 0.0001). Six (40.0%) were LP‐positive, three (20%) were TWA‐positive, and three (20.0%) had abnormal HRV (Figure S1C). Presentation of a Representative Case With Improved AECG‐M in Recovery (Case 2) A 66‐year‐old woman was treated for COVID‐19. She had a history of postoperative breast cancer and gastric cancer, but no significant cardiac disease. The gastric cancer was completely cured via endoscopic resection. Following a right breast cancer mastectomy, the patient is currently undergoing chemotherapy and radiation therapy for metastasis to the left femur. On admission, her white blood cell (WBC) and CRP levels were 2300/μL and 6.8 mg/dL, respectively. Echocardiography revealed an LVEF of 59.5%. Chest radiography and computed tomography revealed bilateral ground‐glass opacities and a crazy‐paving pattern. AECG‐Ms assessed within 48 h of admission were positive for LP, HRT, and HRV. Heparin, remdesivir, and dexamethasone were administered. The patient recovered from the COVID‐19 pneumonia and was discharged 17 days after admission. The AECG‐Ms assessed 483 days after the initial AECG‐M assessment showed that the HRT and HRV were negative. The changes in AECG‐Ms during the acute and recovery phases are presented in Table S6. LP parameters, such as c‐fQRS, RMS40, and LAS 40, deteriorated (c‐fQRS: 114 → 121, positive criteria > 105 ms; RMS40:9.6 → 8.5, positive criteria < 15 μV; LAS40:50 → 57, positive criteria > 39 ms). LP was positive during the recovery phase. In contrast, TWA values decreased during the recovery phase (25.3 → 16.5, positive criteria > 19.9 μV), and TWA changed to negative. The HRT also became negative. SDNN increased (48.76 → 115.61, positive criteria < 75.0 ms), and HRV changed to negative. HRT values also changed (HRT: 48.76 → 115.61, positive criteria < 75.0 ms). Discussion The main findings of this study are as follows. First, TWA, LP, and NSVT positivity were significant risk factors for total mortality, with TWA positivity being the risk factor with the highest HR. The group with both TWA and LP positivity had a poorer prognosis than the other groups in Kaplan–Meier curve analysis (Study 1). Second, the COVID‐19 group had significantly higher positivity rates for the myocardial damage indicators TWA and LP than the non‐COVID‐19 group, suggesting that the AECG‐M abnormalities observed in patients with COVID‐19 could be associated with SARS‐CoV‐2 infection (Study 2). Finally, TWA, LP, and HRV abnormalities improved during COVID‐19 recovery compared with the acute phase results (Study 3). Usefulness of ECG Marker in Covid‐19 (Study 1) Study 1 showed that AECG‐Ms, which are risk‐stratification ECG markers for organic heart disease, are also useful for the risk stratification of patients with COVID‐19. The usefulness of various ECG parameters in COVID‐19 has been reported (Supplementary Table 7), with many reports that QT prolongation is a prognostic factor in patients with COVID‐19 [10, 11, 27, 28, 29, 30], followed by several reports that HRV is a prognostic [31, 32, 33]. However, to our knowledge, no previous study has examined multiple AECG‐M indices in patients with COVID‐19. In 2023, Tondas et al. first reported on the usefulness of TWA [34]. The authors measured TWA on admission and at discharge in 76 patients (mild and moderate disease) with positive SARS‐CoV‐2 polymerase chain reaction (PCR) and 34 patients with negative PCR and found an elevated TWA in the PCR‐positive patient group compared with the PCR‐negative patient group at discharge. Although several previous studies have described QT prolongation as a prognostic factor in patients with COVID‐19 [10, 11, 27, 28, 29, 30], and HRV as a prognostic factor [31, 32, 33], in this study, we found no significant difference in either QTc or HRV between the fatal and non‐fatal event groups (Table 1). In contrast, while the aforementioned studies did not indicate TWA as a prognostic factor, in this study, the HR for TWA was 8.34, which had even higher prognostic accuracy when combined with LP. Possible reasons for the lack of significant differences in QTc and HRV between the groups in the present study include the size of the study population and differences in fatal event rates and drugs used compared with those of previous studies. In a study by Al‐Zakhari et al. [10], 99 (27.2%) of 339 study participants died, Akhtar et al.'s study [11] recorded 133 deaths (45.4%) among 293 study participants, and Mol et al.'s study [31] reported a mortality rate of approximately 22% during the observation period. These studies included patients with COVID‐19 from March to June 2020, with a high proportion of highly lethal SARS‐CoV‐2 variants. In our study, 12 (17.1%) of 70 patients in the COVID‐19 group experienced fatal events. Although viral strains could not be studied, omicron strains with low lethality were likely predominant at the time of the study. These results may have differed because previous studies included more alpha and delta variants with higher mortality. Clinical Features of the 12 Cases Leading to Fatal Events Of the 12 patients in the fatal event group, five had positive NSVT results, and three had positive results for four AECG‐Ms. Five patients in the fatal event group had VT; all five had abnormal TWA or HRV, which is consistent with reports that TWA and HRV are prognostically useful in fatal arrhythmias [21, 35]. AECG‐Ms were positive in all cases, even in cases with respiratory failure as the primary cause of death. The reasons for positive AECG‐Ms, which are predictive markers of cardiac death, even in cases with respiratory failure as the primary cause of death, are unknown. Although not validated in this study, the literature suggests that ACE2 receptor‐mediated myocardial remodeling and ischemia may be present in patients with COVID‐19 [36, 37, 38], and may be responsible for the positive AECG‐M results. However, ACE2 receptors are widely present in the lungs and other organs, including the heart, and in severe cases, the heart may be affected in parallel with damage to the respiratory system. The primary cause of death in the three cases was cancer; all occurred in patients with cancerous tumors (Stages III–IV), but with additional treatment planned, such as chemotherapy. One patient died suddenly, two died earlier than expected, one died of cancer within a year of diagnosis, and the other died of cancer within the second year of diagnosis. The detailed mechanism is unclear; however, SARS‐CoV‐2 infection may modify immunological abnormalities and accelerate the time of death in patients with cancer. Pathophysiological Considerations Regarding the High Prevalence of Positive AECG‐M in COVID‐19 (Study 2) Our data suggest that the positive AECG‐M rates in patients with COVID‐19 were significantly higher than those in the control group. The pathophysiological considerations related to this result are as follows: SARS‐CoV‐1, a coronavirus similar to SARS‐CoV‐2 in infecting humans and causing severe illness, also infects cardiomyocytes via ACE2 receptors and causes myocardial damage, as confirmed by experiments in mice [6, 7, 8]. As with SARS‐CoV‐1, SARS‐CoV‐2, which also infects cardiomyocytes via the ACE2 receptor, may cause similar pathology, which may be reflected in AECG‐M abnormalities [6]. However, it is also possible that COVID‐19‐related cardiac and electrophysiological abnormalities are multifactorial. Systemic inflammation, myocardial ischemia, hypoxia, electrolyte imbalance, and autonomic dysfunction can cause ECG abnormalities [39, 40]. Interestingly, while the control group exhibited significant diurnal variations in the HRV time‐domain parameters, the COVID‐19 group showed a tendency toward reduced diurnal variations (Tables S4 and S5). This suggests that COVID‐19 patients may experience more severe autonomic dysfunction than those with fever caused by conditions other than COVID‐19. Pan et al. reported that autonomic dysfunction in COVID‐19 patients is associated with disease severity and outcomes [41]. However, in this study, despite matching the APACHE II scores of the control group to those of the COVID‐19 group, significant differences in outcomes were observed between the two groups. However, the mechanism of autonomic dysfunction caused by COVID‐19 has not yet been clearly elucidated. Further research is needed to characterize the pathology of myocardial damage in this study and its association with AECG‐M. However, the results may be common to coronaviruses that target the ACE2 receptor. Acute and Chronic Changes in AECG‐Ms in Patients With COVID‐19 (Study 3) During the recovery period after COVID‐19, the electrophysiological abnormalities of the myocardium improved to some extent, but the AECG‐M abnormality rates during the chronic period remained high. Unfortunately, in this study, we could not determine whether the AECG‐Ms were positive before infection or whether they remained positive as sequelae. Distinguishing whether the AECG‐M abnormality lasted for several months or was originally abnormal was difficult. However, organic changes in the myocardium do not fully resolve and may persist for several months. Compared with the immediate aftermath of COVID‐19, the TWA, LP, and HRV parameters improved significantly during the recovery period (mean: 323 ± 149 days after the first assessment) (Figure S1C). In our study, during the recovery phase, a certain number of AECG‐Ms remained positive (LP positive, 40%; TWA positive, 20%; and abnormal HRV, 20%). In a study by Puntmann et al. [42], the recovery period of patients with COVID‐19 (from cardiac magnetic resonance imaging [MRI] performed during the recovery phase [71 days, range 64–92 days]) revealed increased myocardial T2 values (76.9%) and a high myocardial signal on gadolinium contrast in 41.0% of the patients. These results suggested that myocardial damage may persist for several months after infection. Various cardiovascular sequelae, such as ischemic heart disease, arrhythmias, and heart failure, have been reported after COVID‐19 [43]. Although comparing the frequency of abnormal MRI and AECG‐M results is difficult, our data suggest that, although reverse remodeling is ongoing during the recovery period after COVID‐19, the organic changes in the heart may not fully recover and may be prolonged for several months. Clinical Implications In COVID‐19, which remains locally endemic, AECG‐Ms may assist in determining whether a patient can be managed in a general ward, requires intensive care, or can be discharged during the early stages of admission. They may also be useful in clinical practice, as they may allow for easy and rapid prediction of prognosis at the bedside of patients with COVID‐19 who have many limitations in undergoing imaging and other tests. Limitations This study has some limitations. First, it had a relatively small sample size (COVID‐19 group, n = 70; comparison group, n = 70) and was conducted at a single center. In addition, the small number of recovery cases introduced a potential selection bias. Second, the lack of pre‐infection AECG data makes it difficult to determine whether the observed AECG changes were truly caused by COVID‐19 or represented pre‐existing conditions. Third, HRV parameters are influenced by the baseline heart rate, and it is possible that HRV parameter values are influenced by tachycardia‐related physiological stress. Fourth, the non‐COVID control group included patients with febrile inflammatory conditions of heterogeneous etiology, which may have introduced confounding factors. However, in this study, we were unable to adjust for these factors and limited our adjustments to the Acute Physiology and Chronic Health Evaluation II (APACHE II) score, a measure of disease severity. Fifth, many patients were hospitalized in other hospitals, and for many of them, the timing of symptom onset and initiation of treatment prior to admission was unclear. Consequently, we were unable to strictly define the relationship between the timing of symptom onset, treatment initiation, and the timing of Holter monitoring. Sixth, software problems with echocardiography in the COVID‐19 wards prevented tissue Doppler testing, and E/A was used as an indicator of diastolic dysfunction. Furthermore, imaging studies such as coronary computed tomography and myocardial scintigraphy were not performed, and objective proof of myocardial ischemia was not provided. Seventh, the study was conducted when institutions were unable to perform autopsies on patients with COVID‐19; therefore, the pathology was not studied in deceased cases. Eighth, three of the twelve deaths were due to cancer, which have partially influenced the prognosis analysis in all cause of mortality. Therefore, this information should be considered when interpreting the findings of the present study. However, it is noteworthy that the prognostic utility of AECG‐M was preserved in the subgroup analysis of patients with serious cardiac events, yielding results that were consistent with those observed for all‐cause mortality. Nineth, the recovery‐phase analysis included only surviving patients, raising the possibility of selection bias and/or survivor bias. Despite these limitations, the data collected in the context of limited human resources and a high risk of SARS‐CoV‐2 infection may be useful for sustained COVID‐19 control. Conclusion TWA, LP, and NSVT can be predictive of prognosis in the early stages of COVID‐19. In particular, simultaneous LP and TWA positivity in patients with COVID‐19 is a poorer prognostic factor than LP or TWA positivity alone. Compared with the disease groups with fever or a pronounced inflammatory response, patients with COVID‐19 had significantly higher TWA, LP, and NSVT positivity. In addition, patients with COVID‐19 and positive AECG‐Ms tended to partially improve during the recovery phase, although some patients did not recover completely. The human coronavirus, the pathogen of COVID‐19, may periodically mutate to become highly virulent and re‐emerge in the future. AECG‐Ms could be useful during future outbreaks of other human coronavirus infections, such as SARS‐CoV‐1 and SARS‐CoV‐2, which involve ACE2 receptors. Funding This work was supported by Fukuda Foundation for Medical Technology. Ethics Statement The study was conducted in accordance with the Declaration of Helsinki and was approved by the Competent Authorities and Ethics Committees of the participating centers. This study was approved by the Ethics Committee of the National Defense Medical College (Approval number 4772). Consent Written informed consent was obtained from all patients. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Figure S1: (A) Changes in TWA from the acute phase to the recovery phase. (B) Changes in c‐fQRS from the acute phase to the recovery phase. Table S1: Predictive analysis of HRV for all causes of death and serious cardiac events. Table S2: Characteristics of patients in the COVID‐19 and control groups. Table S3: AECG‐M results between the COVID‐19 and non‐COVID‐19 groups. Table S4: Circadian variability of HRV indices in the COVID‐19 group. Table S5: Circadian variability of HRV indices in the control group. Table S6: Changes in AECG‐M parameters during the acute and recovery phases (483 days later) in Case 2. Table S7: Previously reported prognostic factors for COVID‐19.