Characteristics and Risk Factors for Intensive Care Unit Cardiac Arrest in Critically Ill Patients with COVID-19—A Retrospective Study

The severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) causing the coronavirus disease 2019 (COVID-19) led to an ongoing pandemic with a surge of critically ill patients. Very little is known about the occurrence and characteristic of cardiac arrest in critically ill patients with COVID-19 treated at the intensive care unit (ICU). The aim was to investigate the incidence and outcome of intensive care unit cardiac arrest (ICU-CA) in critically ill patients with COVID-19. This was a retrospective analysis of prospectively recorded data of all consecutive adult patients with COVID-19 admitted (27 February 2020–14 January 2021) at the University Medical Centre Hamburg-Eppendorf (Germany). Of 183 critically ill patients with COVID-19, 18% (n = 33) had ICU-CA. The median age of the study population was 63 (55–73) years and 66% (n = 120) were male. Demographic characteristics and comorbidities did not differ significantly between patients with and without ICU-CA. Simplified Acute Physiological Score II (SAPS II) (ICU-CA: median 44 points vs. no ICU-CA: 39 points) and Sequential Organ Failure Assessment (SOFA) score (median 12 points vs. 7 points) on admission were significantly higher in patients with ICU-CA. Acute respiratory distress syndrome (ARDS) was present in 91% (n = 30) with and in 63% (n = 94) without ICU-CA (p = 0.002). Mechanical ventilation was more common in patients with ICU-CA (97% vs. 67%). The median stay in ICU before CA was 6 (1–17) days. A total of 33% (n = 11) of ICU-CAs occurred during the first 24 h of ICU stay. The initial rhythm was non-shockable (pulseless electrical activity (PEA)/asystole) in 91% (n = 30); 94% (n = 31) had sustained return of spontaneous circulation (ROSC). The median time to ROSC was 3 (1–5) minutes. Patients with ICU-CA had significantly higher ICU mortality (61% vs. 37%). Multivariable logistic regression showed that the presence of ARDS (odds ratio (OR) 4.268, 95% confidence interval (CI) 1.211–15.036; p = 0.024) and high SAPS II (OR 1.031, 95% CI 0.997–1.065; p = 0.077) were independently associated with the occurrence of ICU-CA. A total of 18% of critically ill patients with COVID-19 suffered from a cardiac arrest within the intensive care unit. The occurrence of ICU-CA was associated with presence of ARDS and severity of illness.


Introduction
The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) emerged in 2019 in China and caused a worldwide pandemic [1,2]. Although the majority of patients have an asymptomatic or mild course of disease, about 5% suffer from severe illness complicated by acute respiratory distress syndrome (ARDS) or other end-organ failure [3]. During the pandemic, an increase in out-of-hospital cardiac arrest (OHCA) cases was observed in different regions [4,5]. Relatively little is known about the occurrence of in-hospital cardiac arrest (IHCA) during the coronavirus disease 2019 (COVID-19) pandemic [6,7]. One study reported a shift of cardiac arrest (CA) characteristics during the pandemic [6]. However, different studies reported on characteristics of IHCA in patients with COVID-19 [8][9][10][11]. In most studies, a devastating mortality ranging from 88-100% has been reported [8,[10][11][12]. A recent multicenter study from the US showed large variations in outcome after IHCA between centers [13]. A large number of IHCA in the recent literature occurred within the intensive care unit (ICU) [8,12,13]. Cardiac arrest in the ICU, also known as ICU-CA, represents a less investigated subgroup of IHCA. Due to clinically important differences, ICU-CA should be considered separately from IHCA in a general ward [14]. The incidence of ICU-CA varies greatly in the literature (4-78/1000 admissions) [14,15]. Survival rates to discharge after ICU-CA are diverse and range from 2 to 79%, mainly depending on the ICU population studied [14,15]. However, one recent study in critically ill patients with COVID-19 reported a mortality rate of 88% [11]. Mortality after CA is mainly triggered by post-CA shock and brain injury [16]. Further, it has been shown that organ failure before and after ICU-CA is common and the severity of illness after ICU-CA and the occurrence of liver failure are independent predictors of mortality [17]. To date, most studies focused on the characteristics of IHCA without comparing clincical characteristics to other patients with COVID-19. Further, studies focusing on critically ill patients and risk factors for ICU-CA are scarce.
However, data on ICU-CA within critically ill patients with COVID-19 are limited. In the present study, we aimed to investigate the occurrence, risk factors and outcome of patients with COVID-19 suffering from ICU-CA. Further, we investigated factors associated with favorable outcome within ICU-CA.

Study Population, Design and Ethics
This was a retrospective analysis of data prospectively recorded at the University Medical Centre Hamburg-Eppendorf (Germany). All consecutive adult patients with COVID-19 admitted (27 February 2020-14 January 2021) to the Department of Intensive Care Medicine were included. The study complied with the Declaration of Helsinki. The Ethics Committee of the Hamburg Chamber of Physicians was informed about the study (No.: WF-60/21). The requirement for informed patient consent was waived due to the use of only anonymized data collected during routine clinical care. The last day of follow-up was 15 March 2021.

Inclusion and Exclusion Criteria
We included all consecutive critically ill adult patients (≥18 years) with confirmed COVID-19 who were admitted to the Department of Intensive Care Medicine during the study period. Patients <18 years or with a prior event of OHCA/IHCA event before ICU admission, as well as cardiac re-arrest after ICU admission, were not considered as ICU-CA and were therefore excluded. Patients without confirmed COVID-19 were excluded.

Study Definitions and Patient Management
Cardiac arrest was defined as cessation of circulation, and therefore, an indication for chest compression and/or cardiac defibrillation in patients who had a pulse and circulation at the time of ICU admission. The sustained return of spontaneous circulation (ROSC) was defined as stable circulation for at least 20 min. The assessment of neurological outcome was performed within routine clinical practice using cerebral performance categories (CPCs) after ICU-CA and during follow-up. A CPC score of 1-2 was defined as a favorable neurological outcome, and a score of 3-5 was defined as an unfavorable neurological outcome. Survival was assessed through the end of the ICU stay. Cardiopulmonary resus-citation and post-CA care were performed in accordance with the European Resuscitation Council guidelines [18,19]. Data were collected according to Utstein-style guidelines [20]. Hypoxic liver injury (HLI) was diagnosed according to established criteria [21]. Confirmed COVID-19 was defined as a positive result on a reverse transcriptase-polymerase chain reaction, and only laboratory-confirmed cases were counted as COVID-19. ARDS was defined using the PaO 2 /FiO 2 ratio (Horowitz index) according to the Berlin definition [22,23]. The severity of illness was evaluated by the sequential organ failure assessment (SOFA) score [24] and simplified acute physiology (SAPS II) [25] score. The Charlson comorbidity index (CCI) [26] was calculated in all patients.

Data Collection
Data were collected through electronic patient data management systems (PDMS, Integrated Care Manager ® (ICM), Version 9.1-Draeger Medical, Luebeck, Germany; Soarian Clinicals, Version 4.3.200-Cerner Health Service, Inc. (North Kansas City, MO, USA) and consisted of age, sex, comorbidities, admission diagnosis, length of ICU stay, treatment modalities, organ support (mechanical ventilation, vasopressor, renal replacement therapy (RRT), blood transfusions, antibiotics, antivirals, etc.), laboratory parameters and further clinical parameters of interest through the end of ICU stay. Pre-existing medication was recorded based on known regular medications and medications on admission. Laboratory assessment was performed daily as part of the clinical routine.

Statistical Analysis
The results are presented as counts and relative frequencies or medians and 25-75% interquartile ranges (IQRs). Binary variables were compared via chi-square analysis/Fisher's exact test, as appropriate. Metric variables were compared via the Mann-Whitney U test. We used multivariable logistic regression to investigate risk factors associated with occurrence of ICU-CA. Factors of clinical relevance were selected and included. A stepwise backward elimination approach that gradually reduces the initial model was used; variables that caused a change in estimates >10%/statistically significant remained in the model. Statistical analysis was conducted using IBM SPSS Statistics Version 24.0 (IBM Corp., Armonk, NY, USA). The study was prepared in accordance with the STROBE (STrengthening the Reporting of OBservational studies in Epidemiology) recommendations.

Study Population
During the study period (1 March 2020 to 14 January 2021), a total of 194 patients with confirmed COVID-19 were treated at the Department of Intensive Care Medicine at the University Medical Centre Hamburg-Eppendorf. After the exclusion of cases who were admitted more than one time in the aforementioned time frame, we could include 183 patients in the final cohort. Within the cohort, we could identify 33 (18%) patients who had suffered from an ICU-CA and 150 (82%) patients who did not suffer from an ICU-CA (see study flow chart, Figure 1).

Characteristics of Intensive Care Unit Cardiac Arrest
The characteristics of ICU-CA are shown in detail in Table 2 and Supplementary Table  S2. Thirty-three (18%) patients of the cohort suffered from cardiac arrest during the ICU stay. The median stay before ICU-CA was 6 (1-17) days. A total of 33% (n = 11) of ICU-CAs occurred during the first 24 h of ICU stay. The initial cardiac rhythm was non-shockable (PEA/asystole) in 91% (n = 30). Defibrillation was performed in 15% (n = 5) of the patients.

Factors Associated with Unfavorable Outcome in Patients with ICU-CA
Detailed characteristics on ICU-CA with favorable and unfavorable outcome are shown in Table 2 and Supplementary Table S2. Cardiac arrest characteristics were comparable between both groups. A longer total resuscitation time (median 4 vs. 2 min) and a higher rate of cardiac re-arrest (30% vs. 0%, p = 0.049) was observed in patients with unfavorable outcome. Lactate (6.1 vs. 3.4 mmol/L) was higher and pH levels (7.2 vs. 7.4) lower after CA in patients with unfavorable outcome (both p < 0.05). The SOFA score before and after ICU-CA was significantly higher in patients with unfavorable outcome. Interventions in place before ICU-CA were well balanced between both groups. Vasopressor therapy was in place in 96% with unfavorable outcome, as compared to 50% with favorable outcome (p = 0.002). ARDS management, including prone positioning, neuromuscular blockage, corticosteroids and inhaled vasodilatory treatment, was comparable in both groups. Complications during ICU stay were comparable between both groups, and the occurrence of cholestasis was significantly higher in patients with ICU-CA and unfavorable outcome (61% vs. 10%).

Discussion
We investigated the occurrence, characteristics, and outcome as well as risk factors for ICU-CA in critically ill patients with COVID-19. We found that ICU-CA occurred in 18% of critically ill patients with COVID-19 and was related to more advanced stages of ARDS and severity of illness. To our knowledge, this is the most comprehensive study focusing on factors for occurrence of ICU-CA in patients with COVID-19. Furthermore, this is the first study showing detailed post-ICU-CA characteristics in patients with COVID-19.
The reported incidence of ICU-CA in the literature varies greatly, ranging from 4-78/1000 ICU admissions [14,15,27]. However, the incidence declined in past years, probably an expression of advances in the management and treatment of critically ill patients [14]. The highest incidence rates were reported in patients with underlying malignant conditions within a comprehensive cancer center [28]. Mainly, the large variation in incidence is explained by the setting and the cohort studied. Two recent studies in a mixed ICU cohort showed an incidence of 22-23/1000 ICU admissions [17,29]. To our knowledge, only one previous study focused on the occurrence and incidence of cardiac arrest in critically ill patients with COVID-19 treated in the intensive care unit [11]. The study by Hayek and colleagues proposed an incidence of about 80/1000 ICU admissions, whereas we observed an incidence of 180/1000 ICU admissions in the present study. This large difference can potentially be the consequence of different factors. In direct comparison, the median SOFA score on admission was considerably higher in our cohort. Almost three-fourths of patients were mechanically ventilated in our study, which is comparable to the study by Hayek and colleagues [11]. However, only 4% were treated by ECMO as compared to 28% in our cohort, which reflects a considerably higher severity of illness. This would be in line with our finding that the severity of ARDS is the strongest risk factor for the occurrence of ICU-CA accompanied by the high incidence observed. However, whether this finding is unique in patients with COVID-19 has to be further elucidated. Earlier studies in the setting of ICU-CA did not focus on patients with respiratory insufficiency or ARDS and were mainly conducted in mixed ICU settings. Furthermore, our hospital acts as a referral center specialized in the care of critically ill patients with ARDS. Therefore, we are treating patients who probably have a more severe course of disease and our findings may not be transferable to other hospitals in different settings.
Critically ill patients with COVID-19 mainly suffer from respiratory insufficiency and many patients require mechanical ventilation. Patients with a need for oxygen support are at high risk for rapidly worsening and the development of respiratory failure. Of interest, the deterioration of SpO 2 in combination with higher early warning scores have recently been described as predictors for worsening in patients with COVID-19 [30,31]. Therefore, close monitoring and early transfer to a more staffed and monitored setting seems reasonable. However, we strictly followed a protocol for early ICU admission of deteriorating patients with COVID-19 in our institution. We observed a high rate of nonshockable rhythm (91%), mainly owing to a non-cardiac cause, which is in line with earlier reports of cardiac arrest in COVID-19 [6,10,11,32]. We also observed a high rate of severe ARDS in our cohort, highlighted by the fact that 24% of patients had an VV-ECMO in place at the time of ICU-CA. The observed resuscitation times in our cohort were mainly short and sustained ROSC was achieved in 94% of patients. The resuscitation times were slightly longer as compared to an earlier study in the mixed ICU, which may be explained by the need for personal protective equipment and the potential delay in the initiation of CPR. However, compared to other studies on COVID-19, we observed quite low resuscitations times [11]. To date, reported outcomes after IHCA/ICU-CA in patients with COVID-19 are worse, ranging from 88-100%, and have led to discussions about futility and limitation of care [8,[10][11][12]32]. Although we observed high mortality in our cohort, it was substantially lower than previously reported. We observed mainly similar cardiac arrest characteristics, but different factors can contribute to better outcome. Firstly, all patients suffered from CA in the ICU. Due to higher nurse/doctor staffing and continuous monitoring, this could explain faster response times and lower times to ROSC. Secondly, some reports derived from regions with a high excessive patient load during the pandemic, possibly contributing to factors associated with worse outcome.
Organ dysfunction and organ failure after CA is frequently observed [16,[33][34][35][36][37]. The high morbidity and mortality after CA were shown to be mainly triggered by post-CA shock and brain injury [16,38]. The effects of pre-existing organ dysfunction and organ support are less clear. In our cohort, more than 90% of patients had invasive or non-invasive respiratory support at the time of ICU-CA. One large study observed that mechanical ventilation at the time of CA is associated with noticeably decreased survival [39]. However, we did not observe differences regarding survival in patients with ICU-CA. About 80% of patients had vasopressor support in place at the time of ICU-CA; we observed an association with an unfavorable outcome. This is in line with two previous studies which also found an association with the pre-arrest use of vasopressors with unfavorable outcome [40,41]. About one-fourth of the patients received RRT prior to ICU-CA and RRT was newly initiated in 42% of patients after ICU-CA. We did not observe an association with unfavorable outcome as compared to earlier studies in OHCA patients [34]. Of interest, we observed that 21% suffered from hypoxic liver injury and 45% from cholestasis, which is in line with earlier reports in critically ill patients with COVID-19 [42]. Although hypoxic liver injury was shown to be associated with unfavorable outcome in ICU-CA, we did not observe this in the present study [17,33]. However, cholestasis was significantly associated with unfavorable outcome in our cohort. This was previously shown in critically ill patients, but not in patients after CA [43].
We observed a significantly higher severity of illness in patients suffering from ICU-CA compared to patients without ICU-CA. Of interest, these differences were already evident on admission, represented by SOFA and SAPS II, and we found that a high SAPS II score was associated with the occurrence of ICU-CA. Although the median time from admission to ICU-CA was 6 days, about one-third of the cohort suffered ICU-CA within 24 h of ICU admission, highlighting this quite vulnerable patient group. This is also comparable to earlier studies on ICU-CA in mixed ICUs [17]. In our cohort, we observed that severity of illness (SOFA) before and after CA was significantly associated with unfavorable outcome. To our knowledge, only the APACHE III score was evaluated for outcome prediction in ICU-CA before and showed moderate predictive performance [44]. However, SOFA score can be easily and quickly assessed. Future studies have to clarify if there is a value of SOFA score in outcome prediction compared to other scoring systems used after ICU-CA.
This study has several limitations. Firstly, we present data deriving from a single center very experienced in the management of CA and post-CA care as well as ARDS and COVID-19. Thus, our results may not be transferrable to other settings. Secondly, we cannot entirely exclude the possibility that changes in general management and advances in critical care of patients with COVID-19 had influence on the occurrence and outcome of ICU-CA. Thirdly, the data were collected retrospectively from a prospectively documented PDMS filled by trained ICU staff. Fourthly, residual confounding is a matter of concern and cannot be entirely excluded. Fifthly, our cohort of patients with ICU-CA was small. Future studies have to clarify and confirm our findings in a larger setting.

Conclusions
In conclusion, this is the first study in critically ill patients with COVID-19 focusing on the post-CA course of ICU-CA. We found that 18% of critically ill patients suffered from ICU-CA with a corresponding incidence of 180/1000 ICU admissions. The occurrence of ICU-CA was strongly associated with higher stages of ARDS and severity of illness on admission. Although patients with ICU-CA suffered from high mortality, cardiopulmonary resuscitation in this selected patient cohort is not futile.

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/10 .3390/jcm10102195/s1, Table S1-pre-existing comorbidities of patients stratified according to ICU cardiac arrest (ICU-CA) and no ICU cardiac arrest (No ICU-CA); Table S2-detailed characteristics of respiratory support and ICU therapies in patients with ICU-CA stratified according favorable and unfavorable outcome; Table S3-hemodynamic characteristics of in patients with ICU-CA stratified according favorable and unfavorable outcome; Table S4-logistic regression model