AF and in-hospital mortality in COVID-19 patients

Background There are conflicting data on whether new-onset atrial fibrillation (AF) is independently associated with poor outcomes in COVID-19 patients. This study represents the largest dataset curated by manual chart review comparing clinical outcomes between patients with sinus rhythm, pre-existing AF, and new-onset AF. Objective The primary aim of this study was to assess patient outcomes in COVID-19 patients with sinus rhythm, pre-existing AF, and new-onset AF. The secondary aim was to evaluate predictors of new-onset AF in patients with COVID-19 infection. Methods This was a single-center retrospective study of patients with a confirmed diagnosis of COVID-19 admitted between March and September 2020. Patient demographic data, medical history, and clinical outcome data were manually collected. Adjusted comparisons were performed following propensity score matching between those with pre-existing or new-onset AF and those without AF. Results The study population comprised of 1241 patients. A total of 94 (7.6%) patients had pre-existing AF and 42 (3.4%) patients developed new-onset AF. New-onset AF was associated with increased in-hospital mortality before (odds ratio [OR] 3.58, 95% confidence interval [CI] 1.78-7.06, P < .005) and after (OR 2.80, 95% CI 1.01-7.77, P < .005) propensity score matching compared with the no-AF group. However, pre-existing AF was not independently associated with in-hospital mortality compared with patients with no AF (postmatching OR: 1.13, 95% CI 0.57–2.21, P = .732). Conclusion New-onset AF, but not pre-existing AF, was independently associated with elevated mortality in patients hospitalised with COVID-19. This observation highlights the need for careful monitoring of COVID-19 patients with new-onset AF. Further research is needed to explain the mechanistic relationship between new-onset AF and clinical outcomes in COVID-19 patients.


Introduction
Over 750 million cases of COVID-19 have been reported worldwide. 1The World Health Organization has reclassified COVID-19 from pandemic to endemic status, indicating that it is likely to remain an ongoing global issue. 2 With endemicity remains the ability for viral evolution that can be rapid and give rise to more virulent strains as occurred with the Delta and Omicron variants, highlighting the need for ongoing research of the COVID-19 process.
Atrial fibrillation (AF) has been observed as the most common arrhythmia in the context of COVID-19, with the prevalence rate reportedly as high as 16.5% and linked to hemodynamic compromise in patients with severe illness. 33][14] Of these, 3 studies have employed manual chart review in samples of 160 to 673 patients but present conflicting data as to whether new-onset AF is an independent marker of mortality in patients with COVID-19. 7,12,13All of these studies include small sample sizes, and larger studies are therefore needed to validate these findings. 15,16ritically, only one large study has been performed. 14his study was reliant on natural language processing methods for AF categorization.While machine learning techniques have shown promise in rapidly assessing large quantities of data, their accuracy has been questioned. 15urthermore, a previous study comparing incidence, predictors, and outcomes of patients with AF in COVID-19 highlighted significantly higher rates of AF diagnosis using expert physician manual chart review compared with automated data collection. 10These findings demonstrate the clear need for manual data collection above automatic methods in this subject area.
The primary aim of this study was therefore to compare patient outcomes in those with pre-existing AF, new-onset AF, and sinus rhythm when hospitalized with COVID-19 infection, using manual chart review.The secondary aim was to evaluate predictors of new-onset AF in patients with acute COVID-19 infection.

Study design and population
A single-center, retrospective cohort study was performed including all adult patients with a completed attendance/admission to Guy's and St Thomas' Hospital who tested positive for SARS-CoV-2 by reverse transcription polymerase chain reaction on at least 1 occasion over the period of March 1 to September 31, 2020 (Figure 1).Ethical approval was granted by Health Research Authorities and the South London Research Ethics Committee (REC: 20/SC/0292).Patients were excluded if they were ,18 years of age on admission or had an unconfirmed COVID-19 diagnosis (eg symptoms consistent with COVID-19 in the absence of a positive test result), or where COVID-19 was not the primary reason for admission.

KEY FINDINGS
-New-onset atrial fibrillation (AF) is the most common cardiac arrhythmia complication in patients hospitalized with COVID-19.
-Pre-existing AF is not associated with all-cause inhospital mortality in patients with COVID-

Patient and public involvement
This study was performed under the Control of Patient Information Notice declared by the Secretary of State for Health and Social Care to support the national response to COVID-19.Informed patient consent was therefore not obtained.The study conforms to the principles outlined in the Declaration of Helsinki.Verbal and written feedback was received from members of the public and incorporated into the study design.

Statistical analysis
Statistical analyses were performed using RStudio (version 1.3.1093).Descriptive statistics were used to characterize the study population.Normally distributed continuous variables were expressed as mean 6 SD, and 2-sample t tests were performed to compare groups.Tests for normality and homogeneity in variances were performed using the Shapiro-Wilk test and F test, respectively.Nonparametric data were expressed as median (interquartile range), and the Mann-Whitney U test was performed for group comparison.
Unadjusted comparisons were made using chi-square test (or Fisher's exact test where sample size was ,5) for categorical data and included all patients from the original dataset.Adjusted comparisons were performed following propensity score matching between pre-existing/new AF and no-AF groups.Propensity score matching was performed using the MatchIt package 16 in RStudio to implement 1:1 nearestneighbor matching using propensity scores generated from a logistic regression model.Remaining patients in the no-AF group that were not matched were excluded from further analysis.The following covariates were included in the matched design: age group, sex, race, and preadmission CHA 2 DS 2 -VASc (congestive heart failure, hypertension, age 75 years, diabetes mellitus, prior stroke or transient ischemic attack or thromboembolism, vascular disease, age 65-74 years, sex category) score.Preadmission CHA 2 DS 2 -VASc score has been as a metric for comorbidity status.Covariate balance was assessed before and after matching.Matched analysis was performed using conditional logistic regression.All covariates were adjusted for in the matched analysis.Inpatient survival probability was measured using Kaplan-Meier analysis.A log rank test was conducted to determine if there were differences in the survival distributions for the different types of intervention.Pairwise comparison was performed to test statistical significance between groups.Post hoc analysis was performed with Bonferroni correction.All tests were 2-sided, and P , .05 was considered statistically significant.

Results
From March 1, 2020, to September 31, 2020, there were 1294 patients diagnosed with COVID-19 at our institution.Of these, 1241 patients were admitted with a primary diagnosis of COVID-19 and included in this study (Table 1).The median age range was 58 to 67 years of age, and 730 (59%) patients were male.Intensive care unit admission was required in 339 (26%) patients, while 272 (22%) patients required mechanical ventilation.

COVID-19 and new-onset AF
AF was the most common arrhythmia in patients with COVID-19 (  2).
There was a statistically significant difference in survival distributions between patients stratified by AF classification (Figure 3).Inpatient survival probability was highest in the no-AF group and lowest in the new-onset AF group.Following pairwise comparison, statistical significance was found between pre-existing AF and no-AF groups (P , .005) and new AF and no-AF groups (P , .005).There was no statistical significance between survival distributions in patients with pre-existing AF and new AF (P 5 .723).
In univariate analysis including all patients from the original dataset, the odds of in-hospital mortality were twice as likely in patients with pre-existing AF compared with patients with no AF (OR 2.18, 95% CI 1.29-3.59,P 5 .002).After propensity score matching, there was no statistically significant difference in the primary outcome between patients with pre-existing AF and patients with no AF (OR 1.13, 95% CI 0.57-2.21,P 5 .732)(Figure 2).

Discussion
This is the largest study to date using manual chart review to study the relationship between sinus rhythm, new-onset AF, pre-existing AF and clinical outcomes in patients with COVID-19.In contrast to previous studies, the size of this study allowed a matched analysis to be performed between cohorts to reduce the effect of confounding variables.We demonstrate that, in patients hospitalized with COVID-19, new-onset AF is the most common cardiac arrhythmia complication and is associated with an increased risk of allcause in-hospital mortality, need for mechanical ventilation, and critical care admission.In contrast, while pre-existing AF is associated with greater prevalence of comorbidities in hospitalized COVID-19 patients, it is not independently associated with all-cause in-hospital mortality after adjusting for age, sex, race, and preadmission CHA 2 DS 2 -VASc score.

Data curation
Manual curation of data allows for verification of data, improved data accuracy, and a reduction in missing data, challenges that are common when using large registry datasets and automatic methods of data collection. 17Difficulties in manual curation of large datasets mainly exist due to labor time in the context of limited resources.Larger studies assessing clinical outcomes of patients with COVID-19 and AF have therefore often been performed using automatic extraction from electronic healthcare records or registry datasets. 5,10,13,14These techniques have been shown to miss important clinical features.Indeed, in an automated study of 3970 patients with COVID-19, manual review of clinical records in a subset of 1110 patients was found to capture a higher incidence of AF/atrial flutter and prevalence of comorbidities compared with automatic extraction from electronic  healthcare records.This highlights significant ongoing limitations of automatic data collection. 10In the largest study to date of employing automatic data curation methods, AF was diagnosed in 1687 of 9564 COVID-19 patients using natural language processing techniques and found to be an independent predictor of in-hospital mortality. 14However, no manual validation was performed in this study.While allowing rapid assimilation of large quantities of data, machine learning techniques such as natural language processing are reliant on correct inference of electronic health record notes, which remains a key challenge in such processing techniques. 18In contrast to automatic methods of data collection, data in this study were manually curated for improved accuracy and data verification, ensuring great confidence in the data obtained. 17

COVID and pre-existing atrial fibrillation
In line with large-scale population-based studies, patients with COVID-19 and pre-existing AF were more likely to be older, have increased frailty, and have pre-existing respiratory or renal disease compared with patients with newonset AF and no AF. 19They were significantly more likely to have additional vascular risk factors including hypertension, diabetes, heart failure, peripheral vascular disease, and coronary artery disease.These data therefore suggest that pre-existing AF is a surrogate marker for morbidity status, rather than an independent marker of mortality in COVID-19.This observation accounts for the present study findings of a statistically significant increase in the risk of in-hospital mortality of patients with pre-existing AF in unadjusted analysis, yet a nonsignificant finding after propensity score matching.Furthermore, lower baseline functional state and increased morbidity in this cohort may have resulted in reduced admissions to intensive care and invasive ventilation compared with patients with new-onset AF due to clinical recommendations and preagreed restrictions on appropriate ceiling of care and resuscitation status.This may provide an explanation for pre-existing AF being protective against intubation and ventilation in COVID-19 infection in unadjusted analysis.

COVID and new-onset AF
In contrast to patients with pre-existing AF, patients with new-onset AF were younger, with fewer comorbidities and a lower CHA 2 DS 2 -VASc score.Furthermore, hypertension and diabetes in particular were more common in the new-onset AF group compared with the sinus rhythm group.It is well established that hypertension and diabetes have specific effects on atrial structure and electrophysiological function, and these effects are frequently documented in experimental models in the absence of sustained AF.As such, it is feasible that the new-onset AF group may highlight a group of patients that are susceptible to AF, which becomes clinically apparent during severe COVID-19 infection.
In keeping with the findings of this study, previous studies have noted an increase in markers of disease severity and need for intensive care admission in new-onset AF patients compared with pre-existing AF and sinus rhythm patients. 13,14Smaller studies have provided conflicting data on whether new-onset AF is an independent markers of disease severity and all-cause mortality in patients with COVID-19.Both Russo and colleagues 7 and Sanz and colleagues 12 demonstrated no difference in acute respiratory syndrome or all-cause mortality in patients with new-onset AF compared with those with no AF.Click or tap here to enter text.However, Sano and colleagues 13 demonstrated significantly worse outcomes in patients with new-onset AF compared with patients in sinus rhythm or those with pre-existing AF.In this study, which is currently the largest manually curated dataset, new-onset AF was found to be independently associated with increased need for mechanical ventilation, critical care admission, and inpatient mortality.The time of AF onset is unknown, as these data were not collected during this study, and therefore it is unclear whether the development of new-onset AF is an early or late marker of severe COVID-19.Further research is needed to investigate whether new-onset AF predicts future clinical deterioration.
In the present study, AF was the most common cardiac arrhythmia present.1][22] While previous studies have reported worse outcomes in patients with AF and COVID-19, this study is the first to disentangle the relationship between outcomes in patients with pre-existing AF and new-onset AF with the analysis certainty brought by manual chart review.This study demonstrates that newonset AF but not pre-existing AF is independently associated with in-hospital mortality.This is of particular importance because only new-onset AF can be a direct consequence of COVID-19 infection.Previous studies have suggested that COVID-19 may have cardiotoxic effects via direct and indirect mechanisms and new-onset AF may therefore be a specific marker of cardiac injury resulting in poorer outcomes. 23Recent data indicate that cardiovascular complications of COVID-19 continue to occur following COVID-19 infection. 24Although the specific pathophysiology of this remains under investigation and is likely multifaceted, possible mechanisms include the effects of COVID-19 infection on angiotensin-converting enzyme 2-related signaling pathways, cytokine storm, changes in fluid balance, hypokalemia, hypoxemia, and activation of the sympathetic nervous system. 23,25,26However, several of these mechanisms are not specific to COVID-19 infection but can be attributed to the physiological response to critical illness.It is recognized that non-COVID acute respiratory viral infection requiring critical care admission is associated with an increased incidence of new-onset AF. 26 The presence of new-onset AF may therefore be a marker of disease severity rather than a specific consequence of COVID-19 infection, although this requires further investigation.Nevertheless, patients diagnosed with new-onset AF in the context of COVID-19 should be monitored closely for acute deterioration and need for advanced care.

Limitations
Although this was a single-center study, the population served by our institution is diverse as reflected by the demographic and ethnicity variability in the study population.This study included patients hospitalized and therefore excludes asymptomatic patients or those with mild COVID-19 symptoms.Furthermore, patients were not followed up beyond their hospital stay and clinical outcomes therefore represent the acute phase of COVID-19. 27.COVID polymerase chain reaction testing was used to determine COVID status, and while there may be false positive or negative results, it remains the gold standard diagnostic investigation for COVID-19 infection.Finally, new-onset AF patients were defined as such if there was no known history of AF within the community.Without continuous heart rhythm within the community, it is feasible that some of these patients may have had asymptomatic pre-existing AF.Of the 45 patients categorized in the new-onset AF group, 10 patients had historical electrocardiograms that confirmed sinus rhythm prior to admission.While we acknowledge this does not exclude a history of paroxysmal AF, it is noted that this is a frequent limitation present in all large population-based AF studies including the Framingham Heart Study and more recently the Nationwide Finnish anticoagulation in atrial fibrillation study. 27,28The results of this study are therefore interpretable through the same lens as this large body of prior literature.

Conclusion
In patients hospitalized with COVID-19, new-onset atrial fibrillation is independently associated with elevated risk of need for mechanical ventilation, critical care admission, and in-hospital mortality.In contrast, while pre-existing AF is associated with greater prevalence of comorbidities in hospitalized COVID-19 patients, it is not independently associated with all-cause in-hospital mortality after adjusting for age, sex, race, and preadmission CHA 2 DS 2 -VASc score.Patients with new-onset AF in the context of COVID-19 should be closely monitored for acute deterioration and need for escalation of care.This study highlights the need for targeted research to explain the mechanistic relationship between new-onset atrial fibrillation and COVID-19.

Figure 2
Figure 2 Forest plots and tabulation representing prematch and postmatch odds ratios for thromboembolic event, ischemic stroke, ventilation, critical care admission, and in-hospital mortality in patients with pre-existing atrial fibrillation (AF) (A, C) and new AF (B, D) compared with patients with no AF.CI 5 confidence interval; N/A 5 not applicable.

Figure 3
Figure3Kaplan-Meier Curve for all-cause mortality in patients with new-onset atrial fibrillation (AF), pre-existing AF, and no AF demonstrating reduced inpatient survival in the new-onset AF group compared with the no-AF group.

Figure 4
Figure 4 History of atrial fibrillation (AF)and preadmission CHA 2 DS 2 -VASc (congestive heart failure, hypertension, age 75 years, diabetes mellitus, prior stroke or transient ischemic attack or thromboembolism, vascular disease, age 65-74 years, sex category) score.Increased preadmission CHA 2 DS 2 -VAScscore between no AF and new-onset AF, new-onset AF and pre-existing AF, and no AF and pre-existing AF (P , .05).

Table 1
Demographics and clinical baseline characteristics of patients with COVID-19 infection, stratified by diagnosis of atrial fibrillation Values are range or n (%).Values are P values represent those calculated by Kruskal-Wallis test.CHA 2 DS 2 -VASc 5 congestive heart failure, hypertension, age 75 years, diabetes mellitus, prior stroke or transient ischemic attack or thromboembolism, vascular disease, age 65-74 years, sex category; DOAC 5 direct oral anticoagulant; LMWH 5 low-molecular-weight heparin; POC 5 package of care; TIA 5 transient ischemic attack.

Table 2
Subclassification of arrhythmia complications in patients admitted with COVID-19 infection