3.1 Literature Retrieval
The database search yielded 966 articles. After removal of duplicates, 546 articles were screened on the basis of their abstracts. After screening, 122 papers were sought for retrieval, of which 14 articles could not be retrieved. The resultant 108 full-texts were reviewed, and 20 identified as meeting the selection criteria. The study selection process and reasons for excluding the 88 excluded studies were illustrated in the PRISMA-P 2020 Flow Diagram (Fig. 1).
3.2 Characteristics of Studies and Risk of Bias
The included studies originated from 10 countries (Australia, France, Italy, Korea, Singapore, Spain, Sweden, The Netherlands, United Kingdom, and United States of America). All studies were retrospective in study design.
There were a total of 67815 patients included in the 20 studies, comprising 38855 patients in the pre-COVID-19 period and 28960 patients in the COVID-19 period. Study sample sizes ranged from 101 to 19303 patients. Study characteristics and the summary of overall findings were summarized in Table 1 and Table 2 respectively.
Table 1
Characteristics of Included Studies
Study | Location | Study Design# | Study Population | Time Period (i) Pre-COVID-19|| (ii) COVID-19|| | Sample Size (i) Pre-COVID-19|| (ii) COVID-19|| |
Baert et al, 202023 | France | Registry-based study | Adult and pediatric cases of presumed medical etiology (EMS*-treated NR§; Received resuscitation NR§) | (i) March 1 – April 31, 2019 (ii) March 1 – April 31, 2020 | (i) 1620 (ii) 1005 |
Baldi et al, 202024 | Lombardy, Italy | Registry-based study | Adult and pediatric cases regardless of etiology (EMS*-treated NR§; Received resuscitation NR§) | (i) February 21 – April 20, 2019 (ii) February 21 – April 20, 2020 | (i) 321 (ii) 490 |
Ball et al, 20207 | Victoria, Australia | Registry-based study | Adult cases regardless of etiology; EMS*-treated and received resuscitation | (i) March 16 – May 12, 2017–2019 (ii) March 16 – May 12, 2020 | (i) 1218 (ii) 380 |
Cho et al, 202025 | Daegu, South Korea | Registry-based study | Adult cases of presumed medical etiology; EMS*-treated and received resuscitation | (i) February 17 – March 31, 2018 (ii) February 17 – March 31, 2020 | (i) 158 (ii) 171 |
Elmer et al, 202026 | Pennsylvania, USA‡ | Registry-based study | Adult cases regardless of etiology; EMS*-treated (Received resuscitation NR§) | (i) January – February 2016–2020 (ii) March 1 – May 25, 2020 | (i) 12252 (ii) 683 |
Lai et al, 202027 | New York City, USA‡ | Non-registry-based study | Adult cases regardless of etiology; EMS*-treated and received resuscitation | (i) March 1 – April 25, 2019 (ii) March 1 – April 25, 2020 | (i) 1336 (ii) 3989 |
Marijon et al, 20206 | Paris, France | Registry-based study | Adult cases of non-traumatic etiology; EMS*-treated (Received resuscitation NR§) | (i) Weeks 12–17, 2012–2019 (ii) March 16 – April 26, 2020 | (i) 3052 (ii) 521 |
Ortiz et al, 202028 | Spain | Registry-based study | Adult and pediatric cases regardless of etiology; EMS*-treated (Received resuscitation NR§) | (i) April 1–30, 2017 and February 1 – March 31, 2018 (ii) February 1 – April 30, 2020 | (i) 1723 (ii) 1446 |
Paoli et al, 202029 | Province of Padua, Italy | Non-registry-based study | Adult and pediatric cases regardless of etiology; EMS*-treated (Received resuscitation NR§) | (i) March 1 – April 30, 2019 (ii) March 1 – April 30, 2020 | (i) 206 (ii) 200 |
Sayre et al, 202030 | Seattle and King County, USA‡ | Registry-based study | Adult and pediatric cases regardless of etiology; EMS*-treated (Received resuscitation NR§) | (i) January 1 – February 25, 2019 (ii) February 26 – April 15, 2020 | (i) 530 (ii) 537 |
Semeraro et al, 202031 | Bologna, Italy | Registry-based study | Adult cases regardless of etiology; EMS*-treated and received resuscitation | (i) January 1 – June 30, 2019 (ii) January 1 – June 30, 2020 | (i) 563 (ii) 624 |
Chan et al, 202132 | 27 States and multiple Counties, USA‡ | Registry-based study | Adult cases of non-traumatic etiology; EMS*-treated (Received resuscitation NR§) | (i) March 16 – April 30, 2019 (ii) March 16 – April 30, 2020 | (i) 9440 (ii) 9863 |
de Koning et al, 202133 | Hollands-Midden, The Netherlands | Registry-based study | Adult cases regardless of etiology; EMS*-treated (Received resuscitation NR§) | (i) March 16 – April 27, 2019 (ii) March 16 – April 27, 2020 | (i) 45 (ii) 56 |
Fothergill et al, 20218 | London, UK† | Registry-based study | Adult and pediatric cases regardless of etiology; EMS*-treated (Received resuscitation NR§) | (i) March 1 – April 30, 2019 (ii) March 1 – April 30, 2020 | (i) 1724 (ii) 3122 |
Glober et al, 202134 | Indiana (Marion County), USA‡ | Registry-based study | Adult cases of non-traumatic etiology; EMS*-treated (Received resuscitation NR§) | (i) January 1 – June 30, 2019 (ii) January 1 – June 30, 2020 | (i) 884 (ii) 1034 |
Lim et al, 202135 | Singapore | Registry-based study | Adult cases regardless of etiology; EMS*-treated (Received resuscitation NR§) | (i) January 1 – May 31, 2018–2019 (ii) January 1 – May 31, 2020 | (i) 1280 (ii) 1400 |
Mathew et al, 202136 | Detroit, USA‡ | Registry-based study | Adult cases of non-traumatic etiology; EMS*-treated and received resuscitation | (i) March 10 – April 30, 2019 (ii) March 10 – April 30, 2020 | (i) 180 (ii) 291 |
Nickles et al, 202137 | Detroit (Macomb, Oakland, and Wayne Counties), USA‡ | Registry-based study | Adult and pediatric cases of non-traumatic etiology; EMS*-treated (Received resuscitation NR§) | (i) January 1 – May 31, 2019 (ii) January 1 – May 31, 2020 | (i) 1162 (ii) 1854 |
Sultanian et al, 202138 | Sweden | Registry-based study | Adult and pediatric cases regardless of etiology; EMS*-treated and received resuscitation | (i) January 1 – March 16, 2020 (ii) March 16 – July 20, 2020 | (i) 930 (ii) 1016 |
Uy-Evanado et al, 202139 | Oregon (Multnomah County) and California (Ventura County), USA‡ | Registry-based study | Adult and pediatric cases regardless of etiology; EMS*-treated and received resuscitation | (i) March 1 – May 31, 2019 (ii) March 1 – May 31, 2020 | (i) 231 (ii) 278 |
*EMS, Emergency Medical Services; †UK, United Kingdom; ‡USA, United States of America; §NR, Not Reported; ||COVID-19, coronavirus disease 2019; #Study designs for all included studies were multicentred and retrospective in nature |
Table 2
Summary of Overall Findings
OHCA|| Outcomes and Characteristics | Parameters | Number of Studies | Pooled OR‡ (95% CI†) | P value | I2 Statistic |
Primary Outcomes | Annual Incidence# | 10 | N/A§ | < 0.001 | N/A§ |
Case Fatality Rate# | 11 | N/A§ | < 0.001 | N/A§ |
Mortality | 11 | 1.95 (1.51–2.51) | 0.0002 | 67% |
Secondary Outcomes | Termination of Resuscitation | 5 | 2.46 (1.62–3.74) | 0.0040 | 93% |
ROSC* | 15 | 0.65 (0.55–0.77) | < 0.0001 | 85% |
Survival to Hospital Admission | 10 | 0.65 (0.48–0.89) | 0.0122 | 87% |
Survival to Hospital Discharge | 11 | 0.52 (0.40–0.69) | 0.0004 | 67% |
Characteristics | Shockable Rhythm | 15 | 0.73 (0.60–0.88) | 0.0024 | 70% |
Etiology | Medical | 9 | 0.91 (0.60–1.37) | 0.5922 | 93% |
Traumatic | 7 | 0.68 (0.41–1.13) | 0.1108 | 70% |
Asphyxial | 5 | 1.17 (1.02–1.33) | 0.0317 | 0% |
*ROSC, Return of Spontaneous Circulation; †CI, Confidence Interval; ‡OR, Odds Ratio; §N/A, Not Applicable; ||OHCA, Out-of-Hospital Cardiac Arrest; #P values were obtained from two-proportions z-tests comparing Pre-COVID-19 and COVID-19 pooled values |
All studies achieved a score ranging from 7 to 9 on the Newcastle-Ottawa Scale, signifying high quality and low risk of bias for selection (Supplemental Table 1).
3.2.1 Primary Outcomes: Annual OHCA Incidence
Ten studies reported or provided sufficient data to calculate the annual OHCA incidence (per 100000 population) (Supplemental Table 2).6–8, 24,29,30, 33–35,37 Among them, de Koning et al33 reported the lowest annual OHCA incidence of 51 and 63 cases per 100000 population in the pre-COVID-19 and COVID-19 time periods respectively. Meanwhile, Glober et al34 reported the highest annual OHCA incidence of 183 and 214 cases per 100000 population in the pre-COVID-19 and COVID-19 time periods respectively. With the exception of Paoli et al29, all studies reported a trend of higher annual OHCA incidence during the COVID-19 period compared to the pre-pandemic period.
The meta-analysis of proportions showed an annual OHCA incidence of 0.0860% or 86.0 cases per 100000 population (34511 cases out of 40116274 population) in the pre-COVID-19 period (95%CI 0.07–0.11%) with a heterogeneity of I2 = 100%, p < 0.01. In contrast, the annual OHCA incidence was 0.12% or 121.7 cases per 100000 population (48820 cases per 40116274 population) in the COVID-19 period (95%CI 0.08–0.15%) with a heterogeneity of I2 = 100%, p < 0.01 (Figs. 2A-B).
3.2.2 Primary Outcomes: Mortality and Case Fatality Rate
Eleven studies provided sufficient data on mortality.6–8,23−25,28,31,32,38,39 Mortality ranged from 85.3–98.4% across both pre-COVID-19 and COVID-19 time periods. During the pre-COVID-19 period, Uy-Evanado et al39 reported the lowest mortality of 85.3%, while Chan et al32 reported the highest mortality of 96.5%. In contrast, Uy-Evanado et al39 reported the lowest mortality of 92.1%, while Fothergill et al8 recorded the highest mortality of 98.4%, during the COVID-19 period. Overall, all studies reported higher mortality in the COVID-19 period compared to the pre-pandemic period.
The odds of mortality were significantly higher during the COVID-19 period as compared to the pre-COVID-19 period (OR = 1.95, 95%CI 1.51–2.51, p = 0.0002, I2 = 67%) (Fig. 3).
A separate meta-analysis of proportions for the outcome of CFR revealed a pooled rate of 93.09% in the pre-COVID-19 time period (95%CI 91.09–94.86%). Between-study heterogeneity was observed at I2 = 96%, p < 0.01. In the COVID-19 period, CFR was 96.38% (95%CI 95.12–97.47%) with a heterogeneity of I2 = 92%, p = < 0.01 (Figs. 2C-D).
3.2.3 Difference in Pooled Estimates of COVID-19 and Pre-COVID-19 Time Periods for Annual OHCA Incidence and Case Fatality Rate
Comparing the pooled estimates for annual OHCA incidence and CFR for pre-COVID-19 and COVID-19 time periods, two-proportions z-tests revealed significant differences between pre-COVID-19 and COVID-19 periods for the outcomes of annual OHCA incidence (39.5% increase, p < 0.001) and CFR (2.65% increase, p < 0.001), as illustrated in Fig. 4.
3.2.4 Secondary Outcomes
3.2.4.1 Termination of Resuscitation in the Field
Five studies reported the outcome of field TOR.8,24,27,32,36 The percentage of population experiencing field TOR ranged from 35.6–89.5% across intervals of the pre-COVID-19 and COVID-19 time periods. All studies reported a trend of higher percentage of patients with field TOR during the COVID-19 period as compared to the pre-pandemic period.
Meta-analysis showed significantly higher odds of field TOR during the COVID-19 period as compared to the pre-COVID-19 period (OR = 2.46, 95%CI 1.62–3.74, p = 0.0040, I2 = 93%) (Fig. 5A).
3.2.4.2 Return of Spontaneous Circulation
Fifteen studies reported the outcome of ROSC.7,8, 23–29,31,32,35,36,38,39 The percentage of population experiencing ROSC ranged from 1.0–41.1% across intervals of the pre-COVID-19 and COVID-19 time periods. Almost all studies (except Elmer et al26) reported a lower percentage of population with ROSC during the pandemic as compared to before the pandemic.
Meta-analysis showed significantly lower odds of ROSC during the COVID-19 time period as compared to the pre-COVID-19 period (OR = 0.65, 95%CI 0.55–0.77, p < 0.0001, I2 = 85%) (Fig. 5B).
3.2.4.3 Survival to Hospital Admission
Ten studies reported the outcome of survival to hospital admission.6,7,25,26,28,29,31,35,36,39 With the exception of Elmer et al26 and Paoli et al29, all studies reported a lower percentage of survival to hospital admission in the COVID-19 period relative to during the pre-COVID-19 period.
Meta-analysis showed significantly lower odds of survival to hospital admission in the COVID-19 period as compared to the pre-COVID-19 period (OR = 0.65, 95%CI 0.48–0.89, p = 0.0122, I2 = 87%) (Fig. 5C).
3.2.4.4 Survival to Hospital Discharge
Eleven studies reported the outcome of survival to hospital discharge.6–8,23−25,28,31,32,38,39 Almost all studies (except Semeraro et al31) reported a lower percentage of survival to hospital discharge in the COVID-19 period relative to during the pre-COVID-19 period.
Meta-analysis showed significantly lower odds of survival to hospital discharge during the COVID-19 period as compared to the pre-COVID-19 period (OR = 0.52, 95%CI 0.40–0.69, p = 0.0004, I2 = 67%) (Fig. 5D).
3.2.5 Clinical Characteristics
3.2.5.1 Shockable Rhythm
Fifteen studies provided data for the characteristic of shockable rhythm.6–8,23−25,27,28,31–33,35,36,38,39 The percentage who had shockable rhythm ranged from 3.6–40% across intervals of the pre-COVID-19 and COVID-19 time periods. Apart from Semeraro et al31, all studies reported a trend of lower percentage with shockable rhythm in the COVID-19 period as compared to the pre-COVID-19 period.
Meta-analysis showed significantly lower odds of shockable rhythm in the COVID-19 period as compared to the pre-COVID-19 period (OR = 0.73, 95% CI 0.60–0.88, p = 0.0024, I2 = 70% (Fig. 6A).
3.2.5.2 Etiology
Nine7,8,24,26,29,31,32,37,38, seven7,8,24,26,29,31,38 and five7,24,29,32,37 studies reported data for medical, traumatic, and asphyxial etiologies of OHCA respectively (Supplemental Table 3). The majority of patients experienced OHCA from medical etiology.
Meta-analysis showed no difference in medical (OR = 0.91, 95%CI 0.60–1.37, p = 0.592, I2 = 93%) and traumatic etiologies (OR = 0.68, 95%CI 0.41–1.13, p = 0.1108, I2 = 70%) in both COVID-19 and pre-COVID-19 periods. There were significantly higher odds of asphyxial etiology (OR = 1.17, 95%CI 1.02–1.33, p = 0.0317, I2 = 0%) for OHCA in the COVID-19 period as compared to the pre-COVID-19 period (Figs. 6B-D). We did not analyze drowning and overdose etiologies of OHCA due to paucity of data.
3.3 Sensitivity Analyses
Sensitivity analyses on the influence of outliers were performed as there was substantial statistical heterogeneity observed in almost all outcomes and clinical characteristics (except asphyxial etiology). Potential outliers were first screened on visual inspection of their confidence intervals, followed by influential analyses using influential diagnostic plots and Baujat plots (Supplemental Figs. 2–25). Applying this approach on each clinical characteristic as well as the primary and secondary outcomes, none of the estimates were substantially changed in direction or magnitude, with the exception of survival to hospital admission where the direction was preserved, but the magnitude of effect was increased by 12%. The revised estimates on sensitivity analyses were: mortality (OR = 1.79 [95%CI 1.46–2.20, p = 0.0001, I2 = 51%] after excluding Sultanian et al38), field TOR (OR = 2.73 [95%CI 1.67–4.47, p = 0.00074, I2 = 65%] after excluding Chan et al32), ROSC (OR = 0.68 [95%CI 0.59–0.80, p = 0.0001, I2 = 72%] after excluding Lai et al27), survival to hospital admission (OR = 0.73 [95% CI 0.59–0.90, p = 0.0085, I2 = 80%] after excluding Mathew et al36), survival to hospital discharge (OR = 0.57 [95%CI 0.45–0.71, p = 0.0003, I2 = 51%] after excluding Sultanian et al38), shockable rhythm (OR = 0.77 [95% CI 0.67–0.89, p = 0.0017, I2 = 57%] after excluding Lai et al27), medical etiology (OR = 1.02 [95%CI 0.76–1.38, p = 0.8768, I2 = 83%] after excluding Sultanian et al38), traumatic etiology (OR = 0.77 [95%CI 0.48–1.24, p = 0.2172, I2 = 62%] after excluding Baldi et al24).
3.4 Publication Bias
The funnel plots were based on chosen primary and secondary outcomes with the highest number of studies (mortality and ROSC respectively). These plots revealed no visual asymmetry, hence suggesting the absence of publication bias (Supplemental Fig. 1). This was supported by non-significant Egger’s regression tests (p = 0.09128 and p = 0.750 respectively).