Cohort study of the characteristics and outcomes in patients with COVID-19 and in-hospital cardiac arrest

Objective We studied characteristics, survival, causes of cardiac arrest, conditions preceding cardiac arrest, predictors of survival and trends in the prevalence of COVID-19 among in-hospital cardiac arrest (IHCA) cases. Design and setting Registry-based observational study. Participants We studied all cases (≥18 years of age) of IHCA receiving cardiopulmonary resuscitation in the Swedish Registry for Cardiopulmonary Resuscitation during 15 March 2020 to 31 December 2020. A total of 1613 patients were included and divided into the following groups: ongoing infection (COVID-19+; n=182), no infection (COVID-19–; n=1062) and unknown/not assessed (n=369). Main outcomes and measures We studied monthly trends in proportions of COVID-19 associated IHCAs, causes of IHCA in relation to COVID-19 status, clinical conditions preceding the cardiac arrest and predictors of survival. Results The rate of COVID-19+ patients suffering an IHCA increased to 23% during the first pandemic wave (April), then abated to 3% in July, and then increased to 19% during the second wave (December). Among COVID-19+ cases, 43% had respiratory insufficiency or infection as the underlying cause of the cardiac arrest, compared with 18% among COVID-19– cases. The most common clinical sign preceding cardiac arrest was hypoxia (57%) among COVID-19+ cases. OR for 30-day survival for COVID-19+ cases was 0.50 (95% CI 0.33 to 0.76), compared with COVID-19– cases. Conclusion During pandemic peaks, up to one-fourth of all IHCAs are complicated by COVID-19, and these patients have halved chance of survival, with women displaying the worst outcomes.


INTRODUCTION
The COVID-19 pandemic has, as of 6 November 2021, infected over 249 million individuals and lead to the death of over 5 million individuals. 1 COVID-19 is now the thirdleading cause of death in Sweden. 2 3 Multiple studies have demonstrated that in-hospital cardiac arrest (IHCA) among patients with COVID-19 is associated with poor survival. [4][5][6][7] A recent study demonstrated that hypoxia was the main cause of cardiac arrest among 40% of patients with COVID-19 and IHCA. 6 We have previously reported on COVID-19 and IHCA in the Swedish Registry for Cardiopulmonary Resuscitation (SRCR), showing a 2.3-fold increase in 30-day mortality among cases with COVID-19, compared with prepandemic cases. This was mainly driven by a ninefold increase in mortality among women with COVID-19. At the time, no case of IHCA with COVID-19 had been discharged alive. 8 The current study expands our previous investigation, including more patients, longer follow-up and emphasises on the causes of cardiac arrest, predictors of survival, coexisting conditions and trends in the prevalence of COVID-19 among IHCA cases.

Data sources
This study is a registry-based observational study with data obtained from the SRCR during the time period 15 March 2020 to 31 December 2020. The SRCR is a national quality registry and has included IHCA cases Strengths and limitations of this study ► A major strength of our study is that it includes all in-hospital cardiac arrests (IHCAs) in Sweden which were reported to the Swedish Registry for Cardiopulmonary Resuscitation. ► The sample recorded in the Swedish Registry for Cardiopulmonary Resuscitation is unbiased since all hospitals participate in the registry and all hospitals report data on COVID-19 status. ► A limitation is that we do not know the severity of the COVID-19 infection, and we do not know if COVID-19 was the main reason for admission to hospital. ► Our study only includes IHCAs receiving cardiopulmonary resuscitation which leaves out all other patients with IHCA, for example, with a Do Not Attempt Resuscitation order.
Open access since 2005. The data is collected by trained nurses who report patient data using a web-based protocol. The registry has previously been described in detail. 9 Vital status was obtained from the Swedish Population Registry and the last day of follow-up was 31 December 2020.

Variable definitions
In SRCR, a patient with cardiac arrest was defined as an unconscious patient with no or abnormal breathing, in whom resuscitation or defibrillation was attempted. IHCA was defined as cardiac arrest in patients admitted to the hospital. With regard to previous coexisting conditions, heart failure was defined as any heart failure described before cardiac arrest. Kidney failure was defined as estimated glomerular filtration rate below 60 mL/min/1.73 m 2 , calculated using the highest creatinine before cardiac arrest with Chronic Kidney Disease Epidemiology Collaboration formula. The SRCR records data on the highest creatinine levels analysed up to 6 months prior to CA. Diabetes was defined as any diabetes diagnosis, regardless of type. Cancer was defined as any previously known cancer. Acute myocardial infarction (MI) was defined as an MI within 72 hours of CA. Previous MI was defined as MI occurring earlier than 72 hours preceding the CA.
Regarding clinical conditions 1 hour prior to CA, arrhythmia was defined as any arrhythmia, hypoxia was defined as an oxygen saturation below 90%, hypotension was defined as systolic blood pressure below 90 mm Hg, seizure was defined as any seizure with loss of consciousness, and heart failure was defined as any heart failure with pulmonary oedema or severe shortness of breath with rales.
Wards with monitoring included the coronary care unit (CCU), intensive care unit, operating room (OR), emergency room (ER), high-dependency unit or the catheterisation laboratory.

Statistical analyses
Patient characteristics are reported in means and medians, along with SD and IQRs, respectively. The Kaplan-Meier estimator was used for describing survival distributions; the log rank test was used to test for differences in survival. Trends in rates of COVID-19 were assessed on a monthly basis during the entire study basis.
Logistic regression was used to calculate ORs for 30-day survival. These models assessed the association between COVID-19 status and 30-day survival, adjusting for age, sex and initial rhythm (shockable or non-shockable). We performed subgroup analyses in relation to sex, age and coexisting conditions (heart failure, cancer, diabetes, kidney failure and MI). These subgroup analyses served to clarify whether the association between COVID-19 status and survival was modified by age, sex or coexisting conditions.
In order to obtain estimates of overall survival, we used Cox proportional hazards model with hours since CA as the time scale. The proportional hazards assumption was fulfilled for all variables.
We used the multiple imputation By chained equations) algorithm to impute missing values 10 11 (online supplemental figure 1). The imputed data set was used to calculate ORs for 30-day survival in the overall group, as well as in COVID+ and COVID-cases. These models included age, sex, initial rhythm, time to start of CPR, time of CA, previous MI, type of ward, heart failure, ECG monitoring, diabetes and acute MI.

Patient and public involvement statement
No patients were involved.

RESULTS
A total of 2227 patients were enrolled in the SRCR between 1 January 2020 and 31 December 2020. After excluding patients <18 years (n=68) and prepandemic cases (n=546), 1613 cases remained from 15 March 2020 to 31 December 2020 and constituted the final study population (online supplemental figure 2). There was a high rate of information on COVID-19 status during the study period among patients registered in the registry (online supplemental figure 3).

Baseline characteristics
The overall mean age was 70.8 years, and the proportion of women was 37.6%. At the end of follow-up, 341 (32.7%) patients were alive. The mean age was similar in the three groups: 70.9 years in COVID+, 71.0 years in COVID-cases and 70.2 years in cases with UNA (online supplemental figure 4). The proportion of women was Regarding comorbidities, acute MI was observed in 12.0% of COVID+and 23.6% of COVID-cases. Previous MI was observed in 11.7% of COVID+, 20.8% of COVIDand 11.7% of UNA cases. The prevalence of heart failure, cancer and diabetes was similar across all groups (table 1).

Follow-up
Return of spontaneous circulation after initial resuscitation, was less common in COVID+ cases, as compared with COVID-cases. Also, angiography, PCI, pacemaker and implantable cardioverter-defibrillator implantation postcardiac arrest were less common in COVID+ cases.

Sex-specific characteristics
Acute MI was observed in 21.2% of COVID +women and 7.6% of COVID +men. Previous MI was observed in 4.7% of COVID +women and 16.2% of COVID +men. The prevalence of previous stroke, renal failure, heart failure, cancer and diabetes was similar among men and women, as was location at the time of cardiac arrest. COVID + men were more likely to have a shockable rhythm (20.8%) compared with COVID +women (11.5%), and to be defibrillated (26.4% in men vs 16.9% in women) (online supplemental table 1).

Open access
Monthly trends in COVID-19 associated IHCA In March, April and May 14%, 23% and 20% of patients suffering IHCA were COVID+ (data from 16 March). The proportion of COVID + cases diminished rapidly during June to July. From September onwards, the COVID +cases increased again to reach 19% in December. In figure 1A, additional details regarding monthly variations are presented.

Aetiology of IHCA
The most common cause of IHCA among COVID+cases was respiratory insufficiency (24%, n=24), and the second most common cause was sepsis or other infection (19%, n=19). Respiratory insufficiency and sepsis/other infection were less common in the other groups (figure 1B), which instead displayed higher rates of acute MI.
Clinical conditions one hour prior to IHCA As evident in figure 1C, which describes the clinical conditions preceding (up to 60 min) the cardiac arrest, hypoxia was more common among COVID+ cases (57%), as compared with COVID-cases (34%).

Survival analysis
The Kaplan-Meier plots (figure 2) show that COVID+ cases generally had a lower probability of survival compared  Cox adjusted survival curves are presented in online supplemental figure 5; COVID+ cases displayed the lowest probability of survival, whereas there was no material difference between COVID-and UNA cases.
Odds ratios for 30-day survival When adjusted for age, sex and initial rhythm the ORs for 30-day survival, comparing COVID+vs COVID-, were 0.50 (0.33-0.76) overall, 0.53 (0.31-0.88) for men and 0.44 (0.20-0.88) for women. In the subgroup of patients with heart failure, MI and cancer, we found no statistically significant associations, whereas in the subgroup of COVID+ patients with kidney failure, OR for 30-day survival was 0.43 (0.16-0.99), when compared with COVID-cases (figure 3).

Predictors of survival
Regarding predictors of 30-day survival among COVID+, we note that CIs were generally wide. Lack of ECG Figure 3 OR for 30-day survival. Forest plot with the adjusted OR for 30-day survival among patients with ongoing infection versus no infection and unknown/NA vs no infection. Stratified on overall, men, women, age <70 years, age ≥70 years, heart failure, kidney failure, diabetes, myocardial infarction and cancer. Myocardial infarction (MI) was defined as acute or previous MI. UNA, unknown/not assessed.

Open access
monitoring and delayed start of CPR showed point estimates below 1.0, although non-significant. OR for patients treated in non-monitored wards was 0.26 (95% CI 0.08 to 0.78) as compared with monitored wards (figure 4). No coexisting condition was associated with survival among COVID+ cases.

DISCUSSION
This study elucidates characteristics and outcomes in patients with COVID-19 who develop IHCA. We show that the prevalence of COVID-19 among patients suffering an IHCA increased to approximately one in four cardiac arrests during the first pandemic wave, and one in five cardiac arrests during the second wave. In IHCA the probability of survival to 30 days is halved by the presence of COVID-19.
Regarding location of CA, we note that the most common location for COVID +patients was regular wards, which are not monitored. This is unfortunate since our Open access analyses showed that type of ward (monitored vs nonmonitored) was significantly associated with survival, such that COVID + cases in non-monitored wards displayed 74% lower probability of survival as compared with COVID +cases in monitored wards. As compared with COVID-cases, cardiac arrest in the ER was more common in COVID +cases. The often rapid deterioration of cardiopulmonary function in patients with COVID-19 may be one of the explanations for this finding. Fewer COVID + cases were located in the CCU, which was an expected finding given that cardiac aetiology was less common among these patients.
We note that the most common cause of cardiac arrest in COVID+cases, as well as the most frequent clinical condition directly preceding the arrest, was respiratory. A total of 57% of cases displayed hypoxia before cardiac arrest. This may highlight an opportunity for improving outcomes; measures to prevent hypoxia and to correct it immediately may reduce the risk of cardiac arrest in patients with COVID-19. The high rate of respiratory aetiology was driven by men (online supplemental figures 6-7).
However, the fact that 43% of cases with COVID-19 did not have hypoxia prior to cardiac arrest suggests that other factors are important as well. Thromboembolism, MI, arrhythmias, etc may all contribute to the development of a cardiac arrest. 12 A previous study from Wuhan showed that 87.5% of COVID+ cases with IHCA had a respiratory aetiology and a study from Southwest Georgia that 53% of the patients with IHCA and COVID-19 had ARDS. 5 7 The survival rates were poor among COVID+patients with an overall 30-days survival of 21%, compared with 36% among COVID-. The survival rate was, however, not as low as in the study from Wuhan, in which 3% (151 patients studied) survived, or in the study from New York with 31 patients or in the study from Southwest Georgia with 63 patients with none surviving. 5 7 13 One reason for the poor survival could be the small number of patients found in shockable rhythm (17% vs 25% for COVID+ and COVID-, respectively) since patients with shockable rhythm have a more favourable outcome. After adjusting for sex, age and shockable rhythm the 30-day survival was still significantly worse among patients with an ongoing infection.
We demonstrate that COVID +women had halved chance of survival at 30 days, compared with COVIDwomen. We find it interesting that COVID +women had acute MI three times as often as men, despite the fact that men exhibited shockable rhythm-and were defibrillated-twice as often as women.

Strengths and limitations
This study includes all IHCAs in Sweden which were reported to SRCR. The sample recorded in the SRCR is unbiased since all hospitals participate in the registry and all hospitals report data on COVID-19 status. However, we do not know the severity of the COVID-19 infection, and we do not know if COVID-19 was the main reason for admission to hospital. With regard to the classification of COVID-19 status, we have performed a misclassification analysis which demonstrated that ORs were not materially affected by misclassification bias. Missingness was prevalent with regards to cause of cardiac arrest, which is due to the difficulties determining this factor. However, we find no reason to believe that missingness differs across COVID-19 status categories, and it should, therefore, not bias our inferences. Our study only includes IHCAs receiving CPR. This leaves out all other patients with IHCA, for example, with a Do Not Attempt Resuscitation order.
Our regression models that included only COVID-19 cases should be interpreted with caution due to the large number of predictors in the model, with relatively few patients (resulting in wide CIs). Further studies are warranted, using a larger study population, and a longer follow-up especially regarding subgroup analyses, neurological outcomes and the quality of life for these patients.

CONCLUSION
During pandemic peaks, up to one-fourth of all IHCAs are complicated by COVID-19, and these patients have halved chance of survival, with women displaying the worst outcomes.
Contributors AH and AR designed the study and were guarantors of the study. AH wrote the first draft of the manuscript, analysed all data and made initial interpretations of data. AR has been supervising. MJ, PS, PL, AR-F, JI, JG and JH revised the article critically for important intellectual content and approved the version of the article to be published. Competing interests None declared.
Patient consent for publication Not applicable.
Ethics approval The study was approved by the Swedish Ethical Review Authority (ID 2020-02017). The data were anonymised before the authors accessed it for the purpose of the study.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement No data are available. No additional data available.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
Open access This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/ licenses/by/4.0/.