Characteristics Associated With Out-of-Hospital Cardiac Arrests and Resuscitations During the Novel Coronavirus Disease 2019 Pandemic in New York City

RESULTS A total of 5325 patients were included in the main analysis (2935 men [56.2%]; mean [SD] age, 71 [18] years), 3989 in the COVID-19 period and 1336 in the comparison period. The incidence of nontraumatic out-of-hospital cardiac arrests in those who underwent EMS resuscitation in 2020 was 3 times the incidence in 2019 (47.5/100 000 vs 15.9/100 000). Patients with out-of-hospital cardiac arrest during 2020 were older (mean [SD] age, 72 [18] vs 68 [19] years), less likely to be white (611 of 2992 [20.4%] vs 382 of 1161 [32.9%]), and more likely to have hypertension (2134 of 3989 [53.5%] vs 611 of 1336 [45.7%]), diabetes (1424 of 3989 [35.7%] vs 348 of 1336 [26.0%]), and physical limitations (2259 of 3989 [56.6%] vs 634 of 1336 [47.5%]). Compared with 2019, the odds of asystole increased in the COVID-19 period (odds ratio [OR], 3.50; 95% CI, 2.53-4.84; P < .001), as did the odds of pulseless electrical activity (OR, 1.99; 95% CI, 1.31-3.02; P = .001). Compared with 2019, the COVID-19 period had substantial reductions in return of spontaneous circulation (ROSC) (727 of 3989 patients [18.2%] vs 463 of 1336 patients [34.7%], P < .001) and sustained ROSC (423 of 3989 patients [10.6%] vs 337 of 1336 patients [25.2%], P < .001), with fatality rates exceeding 90%. These associations remained statistically significant after adjustment for potential confounders (OR for ROSC, 0.59 [95% CI, 0.50-0.70; P < .001]; OR for sustained ROSC, 0.53 [95% CI, 0.43-0.64; P < .001]).

O n March 1, 2020, the first case of novel coronavirus disease 2019  was diagnosed in New York City, New York (NYC); by April 25, 2020, 17 118 confirmed and probable deaths due to COVID-19 had already occurred. 1 On April 6, 2020, NYC out-of-hospital cardiac arrests peaked at 305 cases, an increase of almost 10-fold compared with April 6, 2019. In Northern Italy, during the COVID-19 pandemic, out-of-hospital cardiac arrests increased by 58% compared with the same time period in 2019 and were associated with lower rates of sustained return of spontaneous circulation (ROSC). 2 Infectious viral epidemics causing severe respiratory infections have long been associated with an increased risk of death. [3][4][5][6][7] For the COVID-19 pandemic, factors independently associated with in-hospital deaths included being older than 65 years, hypertension, diabetes, cardiovascular disease, and chronic obstructive pulmonary disease (COPD). 8 To date, factors associated with out-of-hospital cardiac arrests and successful resuscitation during the COVID-19 pandemic have not been defined. Using data from the NYC 911 emergency medical services (EMS) system, our study compared patients with nontraumatic out-of-hospital cardiac arrest who received resuscitation during the COVID-19 period and their outcomes with patients and outcomes during the same period in 2019. Our goal was to identify COVID-19-associated changes in frequency, risk factors, presenting cardiac rhythm, and out-of-hospital death despite EMS resuscitation.

Methods
This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. The institutional review board of the Montefiore Medical Center, Albert Einstein College of Medicine, Bronx, New York, approved this study and, owing to minimal risk to the participants (ie, no effect on their rights and welfare), waived the need for informed consent.

Data Sources
The NYC 911 EMS system is the largest in the United States, serving a population of more than 8.4 million and responding to more than 1.5 million medical calls annually. This 3-tiered system consists of firefighter-certified first responders, emergency medical technician basic life support units, and paramedic advanced life support (ALS) units. In the NYC 911 system, cardiac arrests receive the highest response priority and all 3 units (firefighter-certified first responders, basic life support units, and ALS units) are immediately dispatched. Both firefighter-certified first responders and basic life support units are certified in basic cardiac life support and carry automated external defibrillators. Paramedic ALS units can obtain and interpret 12-lead electrocardiograms and are certified in advanced cardiac life support, including advanced airway management and administering cardiac resuscitation medications. Out-of-hospital cardiac arrests are managed by EMS responders using regional prehospital protocols modeled after the American Heart Association Guidelines for Cardiopulmonary Resuscitation and Emergency Cardiovascular Care. 9 Data on out-of-hospital cardiac arrests are collected and managed by the Fire Department of NYC Online Medical Control. For cases in which EMS resuscitation is performed, a postresuscitation telephone interview of paramedics and emergency medical technicians is conducted by Online Medical Control staff. The questionnaire collects data in the Utstein style 10 on age, sex, race/ethnicity, preexisting comorbidities, bystander cardiopulmonary resuscitation (CPR), presenting rhythm, and advanced cardiac life support interventions (airway management and medications). The questionnaire was validated in prior cardiac arrest research within our system. 11 Interviews are supplemented with information from electronic prehospital patient care reports completed by EMS responders. Final call-type of cardiac arrest, response time, and ALS first on-scene were obtained from the Fire Department of NYC's 911 computer automated dispatch system. All data are maintained in a secure data warehouse.

Study Design
This population-based, cross-sectional study included patients 18 years or older with out-of-hospital cardiac arrest who received EMS resuscitation during the COVID-19 period (March 1 to April 25, 2020)

Data Analysis
First, we examined characteristics (demographic and other) of individuals with confirmed, nontraumatic out-of-hospital cardiac arrests who underwent resuscitation during the 2 study periods. The assumption was that excess cases of out-ofhospital cardiac arrests in the COVID-19 period were likely associated with the COVID-19 pandemic, either directly or indirectly. Excess cases of out-of-hospital cardiac arrest resuscitations were calculated by taking the daily difference between the number of calls in 2020 and 2019. The cumulative percentage of EMS calls for fever, cough, dyspnea, and virallike symptoms consistent with COVID-19 and the cumulative percentage of excess out-of-hospital cardiac arrest resuscitations were calculated, and the temporal relationship graphed. Second, we compared the association of COVID-19 with outof-hospital ROSC and ROSC that was sustained until emergency department arrival (hereinafter referred to as sustained ROSC), adjusted for known covariates of ROSC and sustained ROSC. These covariates included age (in 10-year increments), race/ethnicity, sex, medical history, EMS response time, bystander CPR, ALS first on-scene, ALS interventions, and presenting rhythm. For the models, first-unit response time was recoded from a continuous time variable to a binary variable of less than 6 minutes (yes or no). A response time of less than 6 minutes has been shown in multiple studies and national registries 12 to provide the most benefit to out-of-hospital cardiac arrest outcomes. 5,13 Last, presenting rhythm was only captured for cases in which ALS personnel were the first arriving units, because basic life support units or firefighter-certified first responders cannot confirm presenting rhythm when first on the scene. Because this reduced the sample available for our multivariate models, the variable was recoded to include an unclassified category to encompass cases in which presenting rhythm was missing.

Statistical Analysis
Unadjusted outcomes were compared using descriptive statistics. Categorical data were compared using Pearson χ 2 , whereas continuous data were compared using 2-tailed t tests or medians and interquartile ranges. Multivariable logistic regression analyses were performed to identify characteristics of patients with out-of-hospital cardiac arrest in the COVID-19 period as well as to assess the association of the COVID-19 period with ROSC and sustained ROSC, controlling for the above covariates. A 2-sided P < .05 was considered statistically significant for both unadjusted and adjusted analyses.
Two sensitivity analyses using these same outcomes were conducted. The first compared the peak 2-week period of outof-hospital cardiac arrests during the COVID-19 period (March 29 to April 11, 2020) with the same 2-week period in 2019. The second compared the peak 2-week COVID-19 period with that of the 2 weeks just before (March 16-28, 2020) and after (April 12-25, 2020) that peak. Analyses were conducted in SAS, version 9.4 (SAS Institute Inc).  [18] years). Compared with 2019, 2020 had an excess of 2653 patients with out-of-hospital cardiac arrest who underwent EMS resuscitation (3989 in 2020 vs 1336 in 2019, P < .001), an incidence rate triple that of 2019 (47.5/100 000 vs 15.9/ 100 000). No time lag was observed between the proportion of daily NYC 911 EMS calls for fever, cough, dyspnea, and virallike symptoms consistent with COVID-19 and excess out-ofhospital cardiac arrest resuscitations, defined as the difference between 2020 and 2019 counts each day (Figure, A).    Table 2). By contrast, the odds of cardiac disease, asthma/COPD, cancer, and cerebrovascular accidents were not increased in 2020 relative to 2019. During the COVID-19 period, out-of-hospital cardiac arrests were 3.5 times more likely to present in asystole (OR, 3.50; 95% CI, 2.53-4.84; P < .001) and twice as likely to present in pulseless electrical activity (OR, 1.99; 95% CI, 1.31-3.02; P = .001) than Presenting rhythm data were only collected for those out-of-hospital cardiac arrests in which an advanced life support unit was first on the scene. g Ventricular rhythms include ventricular fibrillation and ventricular tachycardia.

Results
in ventricular rhythms (ventricular fibrillation or ventricular tachycardia). Our 2 sensitivity analyses revealed the same associations as did our main model ( Table 2). For ROSC and sustained ROSC, results from our sensitivity analysis were nearly identical to those in Table 4 in the direction of associations, although some factors lost statistical significance for ROSC (being black and having a presenting rhythm of pulseless electrical activity) or for sustained ROSC (history of asthma/COPD, physical activity limitations, and a response time of ≥6 minutes). Return of spontaneous circulation and sustained ROSC were both significantly lower during the 2-week COVID-19 peak period in both sensitivity analyses.

Discussion
Using data from the NYC 911 EMS system during the COVID-19 pandemic, we report 2653 excess out-of-hospital cardiac arrests, a number that, by itself, represents double the number of patients with out-of-hospital cardiac arrests who underwent EMS resuscitation during the comparable 2019 period. More than 90% of these excess cases resulted in out-of-hospital deaths, some of which likely contributed to the 17 118 confirmed and suspected COVID-19-related deaths that occurred in NYC during the first 8 weeks of the pandemic. Risk factors for excess COVID-19-related out-ofhospital cardiac arrests included older age and minority race/ethnicity, after adjustment for comorbidities. Importantly, nonshockable presenting rhythms of asystole and pulseless electrical activity were more commonly documented in 2020 compared with 2019 and likely account for the substantial increase in out-of-hospital cardiac arrest mortality.
Conditions associated with COVID-19, including hypoxemic respiratory failure, massive myocardial infarction, and pulmonary emboli, can lead to rapid decompensation and result in cardiac arrest with initial nonshockable rhythms. [14][15][16] Our results were similar to those observed in Northern Italy, where out-of-hospital cardiac arrests increased by 58% from the same time period in 2019. 2 Italy had an increase in out-of-hospital mortality from 67.3% to 82.2% and an increase of initial nonshockable rhythms from 83% to 90%. 2 In Wuhan, China, unsuccessful resuscitation for in-hospital cardiac arrests occurred 86.8% of the time, with 89.7% of patients having asystole as the initial presenting rhythm. 17 Increased out-of-hospital cardiac arrests during influenza are thought to be due to the body's systemic inflammatory response, which destabilizes atherosclerotic plaques that, in turn, produce myocardial infarctions and cardiovascular deaths. 5,7,18 In addition to overwhelming pneumonia, viral sepsis, and acute respiratory failure, 19 COVID-19 causes endothelial injury predisposing to thrombosis in the arterial and venous system with myocardial infarction in the absence of atherosclerosis and increased risk of venous thromboembolism. [20][21][22][23] Declining  oxygenation and biomarkers of tissue injury (elevated levels of cardiac troponins, cytokines, D-dimer, and lactate) are risk factors for death in hospitalized patients with COVID-19. 24,25 Similar to risk factors for death in hospitalized patients, we found that increasing age, hypertension, and diabetes were independent risk factors for patients with out-of-hospital cardiac arrest during 2020. We also observed that patients reported to have physical activity limitations, such as being bed or wheelchair bound, were at increased risk for COVID-19related out-of-hospital cardiac arrests. Immobility may be a marker for frailty and is a risk factor for thromboembolic disease. Although sex, asthma/COPD, prior cardiac disease, and cerebrovascular accidents are known risk factors for inhospital cardiac deaths, 26-29 they were not risk factors in our study of excess out-of-hospital cardiac arrests in 2020. This may be because these comorbidities were risks for out-ofhospital cardiac arrests 30 in the comparison 2019 period and therefore did not contribute significantly to excess out-ofhospital cardiac arrest cases in 2020.
In our study, minority race/ethnicity was a risk factor for COVID-19-related out-of-hospital cardiac arrests even after adjusting for comorbidities that disproportionately affect minority populations. Black, Hispanic, and Asian patients were at increased risk for COVID-19-associated out-of-hospital cardiac arrests and death. Explanations for these disparities are multifactorial, difficult to disaggregate, and range from individual vulnerabilities to social/environmental factors. The disparate burden of out-of-hospital cardiac arrests in minority populations may be a consequence of underlying comorbidities, genetic-environmental interactions, socioeconomic conditions that include increased viral exposure due to crowding and reduced opportunity to work from home, as well as reduced access to health care. 31 Although we observed a temporal association (without time lag) between NYC 911 EMS calls for fever, cough, dyspnea, and viral-like symptoms and out-of-hospital cardiac arrests, that in itself is insufficient to demonstrate that excess out-of-hospital cardiac arrests and deaths after attempted resuscitation were solely owing to sudden cardiopulmonary decompensation from COVID-19 infection. The observed temporal relationship does not preclude other explanations, such as the possibility that delays in seeking or receiving health care may have negatively affected slowly progressive COVID-19 infections or preexisting conditions (eg, cardiopulmonary diseases or cancer), resulting in out-of-hospital cardiac arrests and deaths. During this period, hospitals reported few admissions for other conditions, 32 and in Italy, admission rates for acute coronary syndrome significantly declined. 33 Reasons for such delays may include not only lack of health care access but also purposeful avoidance due to fears of contracting COVID-19. In addition, pandemic-related environmental, emotional, and economic stressors could have indirectly contributed to excess out-of-hospital cardiac arrests and deaths. Because our data cannot address the proportion of out-ofhospital cardiac arrests that was directly or indirectly due to COVID-19, further research is needed. Even before the results of further research are available, the increased COVID-19related out-of-hospital cardiac arrest rates in our study rein-force the need for improved health care outreach during pandemics, especially for vulnerable populations.
Our results agree with established findings of higher rates of sustained ROSC with shorter EMS response time. 12,13,30 With the increased number of patients presenting with COVID-19like symptoms, the median response time of available EMS units to out-of-hospital cardiac arrests was increased by approximately 1 minute; however, this difference was not statistically significant when compared with the same period in 2019. Although the time range was variable, the median response time was less than the 3-minute increase reported in Italy. 2 In contrast, if ALS units arrived first on the scene, we observed significantly higher rates of ROSC and sustained ROSC compared with other units, even during the COVID-19 period. Studies characterizing the association of prehospital ALS management with out-of-hospital cardiac arrest in the pre-COVID-19 era report conflicting results. [34][35][36][37] In our study, ALS interventions (ACLS medications and endotracheal intubation) were associated with significant increases in both ROSC and sustained ROSC (Table 4) in all analyses. We speculate that these ALS interventions were more likely to occur and to be successful when ALS units were first on the scene. In addition, paramedics' higher training and medical knowledge provide critical skills in patient assessment that lead to effective treatment decisions and team-based leadership 35 during resuscitations.
During the COVID-19 study period, less invasive airway management (supraglottic airway or bag-valve-mask ventilation) was associated with lower rates of ROSC and sustained ROSC. Several studies, including a meta-analysis, have shown increased ROSC rates and overall survival to hospital discharge with endotracheal intubation, 38,39 although the mechanism for this improvement has not yet been elucidated. The significant decrease in the use of more invasive procedures, such as endotracheal intubation, in favor of less invasive procedures (supraglottic airways and bag-valvemask ventilation) may be due to EMS responders wanting to reduce exposure to the patient during the COVID-19 pandemic. This may have been a concern despite the availability of personal protective equipment, including fresh N95 masks, eye protection, gowns, and gloves that were supplied to and required of all personnel during resuscitations. This finding was similarly observed in the management of out-ofhospital cardiac arrests in Italy at the height of their response to COVID-19. 2

Limitations
Our study shares several limitations found in recently published COVID-19 in-hospital mortality studies. First, our study population was limited to those who received care, in this case, EMS resuscitation. Second, because postmortem testing to confirm COVID-19 was rarely performed, we cannot distinguish between increased cardiopulmonary arrests directly due to COVID-19 or indirectly due to unattended comorbid diseases during this pandemic. Support for the increase in outof-hospital cardiac arrests being directly COVID-19 related in our study and for a similar trend in the study from Italy is based on comparisons with the prior year. We acknowledge that although cardiovascular disease, asthma/COPD, cerebrovascular accidents, and cancer were not risk factors for out-ofhospital cardiac arrests during the COVID-19 pandemic in our study, patient lack of access to or avoidance of health care leading to acute decompensation of comorbid illnesses may have played a role. We do not believe that reliance on prehospital patient information for comorbid history resulted in differential misclassification, because the same method was used in both periods (2019 and 2020), and the percentage of bystanderwitnessed events was similar. Ultimately, corroboration by death certificates, along with autopsy studies, is required to determine the proportion of out-of-hospital cardiac arrests and deaths that were related to COVID-19. A strength of the current investigation is the longitudinal, system-wide ascertainment of out-of-hospital cardiac arrests and resuscitations in the largest US 911 system during the largest pandemic since the 1918 influenza pandemic. 40 By including data from the entire 911 system and comparing it with the same time period 1 year prior, the potential for differential ascertainment biases was minimized. By choosing the longer period as our main analysis rather than the 2-week COVID-19 peak period, we purposely biased our results toward the null.

Conclusions
The tragedy of the COVID-19 pandemic is not just the number of patients infected, but the large increase in out-of-hospital cardiac arrests and deaths. This catastrophe transpired despite similar rates of bystander CPR, similar EMS response times, and similar durations of resuscitation efforts, compared with 2019. The findings of this cross-sectional study emphasize the importance of intervening early in the course of COVID-19 infection, before acute decompensation. They also speak to the critical need to design better systems for providing health care access to vulnerable, at-risk patients with acute and chronic conditions during a pandemic. Aggressive efforts for identifying outpatient risk factors for out-of-hospital cardiac arrests and death, such as hypoxia and hypercoagulability, especially in minority populations, should be instituted. Further research is needed to determine if early, targeted interventions in the outpatient setting for those at risk, such as regular telemedicine visits and home-based monitoring of vital signs, oxygen saturation, and biomarkers of tissue injury in those that test positive could lead to reductions in out-of-hospital fatalities.