Introduction
Despite limited evidence, epinephrine (adrenaline) has been an integral part of international guidelines for cardiopulmonary resuscitation (CPR) for decades [1 et al . [2 2 3 , 4 3 4
The aim of this study was first to develop a statistical methodology to be able to compare data from a randomized controlled trial (RCT) with registry data and second to use this model framework to find explanations for the discrepancies between the PARAMEDIC2 trial and GRR.
Methods
As a first step, we provided a statistical modelling framework of outcome parameters after OHCA, based on demographic, physiological, and medical predictors typically available in publications of trials on CPR in the out-of-hospital setting. This model allowed the computation of probabilities and comparison of outcomes at each step of the patient pathway after OHCA including restoration of spontaneous circulation (ROSC), survival to hospital admission/discharge, and favourable neurological outcome. Based on this modelling framework, we tried to identify factors contributing to the substantial differences in outcomes in the epinephrine group of the PARAMEDIC2 trial (which reflects the group of patients treated according to the international standards and guidelines for CPR) and the GRR.
The German Resuscitation Registry group
The GRR was started in 2007 by the German Society for Anaesthesiology and Intensive Care Medicine as an ongoing international, prospective, multicentre registry. This registry covers more than 23 million inhabitants of Germany. As of August 2021, 228 650 patients with OHCA and 137 215 patients with attempted resuscitation after OHCA have been documented. The German EMS system is based on paramedic-staffed ground ambulances and separate EMS physician vehicles, which rendezvous at the site of the emergency. For a more detailed description of the system see reference [5 6
Inclusion criteria: all patients who suffered OHCA during the period 1 January 2011 to 31 December 2020 and who received treatment from an EMS in any region participating in the GRR were eligible for inclusion in the study. We excluded all patients who received no vasopressor (generally epinephrine) during CPR. This was mainly the case if ROSC was achieved or if resuscitation was withheld (e.g. due to a ‘do not resuscitate’ order) before the temporal sequence of the CPR algorithm according to the guidelines recommended by the administration of epinephrine. Further exclusion criteria were inhospital cardiac arrest, trauma, and age less than 18 years. In order to minimize selection bias in terms of survival and as a check for data quality, only reference EMS systems with a return rate of more than 60% for the inhospital clinical treatment dataset on an annual basis were included. This resulted in a calculated return rate of more than 87%. These reference EMS systems are known to sufficiently represent Germany as a whole. A flowchart indicating both the patient selection from the GRR as well as data availability for each logistic regression model of the resuscitation pathway is shown in Fig. 1 .
Fig. 1.: Flowchart of patient selection from the German Resuscitation Registry (GRR) for the development of the modelling framework. EMS, emergency medical service; ROSC, restoration of spontaneous circulation.
In the GRR, neurological outcome is documented as a cerebral performance category (CPC) ranging from 1 (no or mild neurological deficits) to 5 (brain death), with categories 1 and 2 defined as favourable outcomes.
The PARAMEDIC2 trial
The PARAMEDIC2 trial was a randomized, double-blind trial involving a total of 8014 patients with OHCA in the UK [2 2
We did not have access to individual data of the PARAMEDIC2 trial.
Ethics
Since only primarily anonymised data were processed and evaluated, criteria for nonhuman subjects research were met. Therefore, the responsible institutional review board (Landesärztekammer Stuttgart) waived the requirement of a specific ethics vote. The study was approved by the scientific advisory council of the GRR (study identifier 2020-01).
Development of the model
In terms of patient characteristics, baseline, and treatment-related predictors of outcome after OHCA, the distribution of continuous variables was examined with Q–Q plots. Results are presented with mean and SD when normally distributed, and with median and interquartile range otherwise. Count data are presented as numbers and proportions.
We focussed on four chronologically related outcomes after CPR: ROSC, hospital admission with ROSC, survival to hospital discharge, and favourable neurological outcome at hospital discharge. For better readability, we refer to these as the ‘resuscitation pathway’.
Comparison of PARAMEDIC2 with GRR
We grouped the available predictors of outcome after OHCA into two groups. The so-called baseline variables included all available predictors given before the arrival of professional EMS personnel: age, sex, initial cardiac rhythm, cause of cardiac arrest, witness of cardiac arrest, CPR performed by a bystander, and the interval between emergency call and arrival of the ambulance at the scene. The so-called baseline plus treatment variables included all previously mentioned predictors plus all available predictors of treatment-related measures influenced by the EMS team: type of vascular access (intravenous or intraosseous), devices for airway management (supraglottic airway only or tracheal tube) and the amount of time between the emergency call and the administration of epinephrine. We modelled the conditional probabilities of each outcome using a logistic regression model. We used baseline and baseline plus treatment variables as covariates to predict the success probability for each outcome during the resuscitation pathway, conditional on the previous outcome. As an example, to model the survival probability at discharge, we conditioned the GRR cohort to only those patients who were successfully admitted to the hospital (who in turn were conditioned to only those patients with a successful ROSC). We ran the models 100 times (each with n = 4015 patients) for each of the four outcomes along the resuscitation pathway.
Goodness-of-fit and discrimination ability of the models were assessed by the area under the receiver operating characteristic curve (AUROC) and Nagelkerke’s pseudo R-squared statistic. Calibration plots are presented in the Supplementary Material (Supplementary Figure SM 1, Supplemental Digital Content 1, https://links.lww.com/EJEM/A346
Comparability between the two cohorts
While the PARAMEDIC2 trial used the mRS, the GRR uses CPC. However, as favourable neurological outcome is defined as mRS ≤ 3 (mRS 3: moderate disability, requiring some help, but able to walk without assistance) and CPC ≤ 2 (CPC 2: moderate cerebral disability: conscious, sufficient cerebral function for independent activities of daily life, able to work in a sheltered environment), the presentation of neurological outcomes in both studies can be considered comparable. The GRR documents, the time interval from the alarm to the administration of the first vasopressor, and the PARAMEDIC2 trial report the time interval from the emergency call to administration of epinephrine. Therefore, in order for the time intervals to be comparable, we added 2 min to the time interval given in the GRR as a reliable estimate the call taker needs to alert the EMS crew [7
Statistical software
All computations were performed using R version 4.0.2 (The R Foundation for Statistical Computing, Vienna, Austria) (The R Foundation for Statistical Computing, Vienna, Austria) [8
Results
Model framework
Cohort description
Patients’ characteristics, baseline and treatment-related characteristics of CPR, and observed outcomes in the PARAMEDIC2 trial’s epinephrine group and all 26 019 patients in the GRR group are shown in Table 1 . The two cohorts display similar characteristics in their age and sex distribution, whereas differences are seen in the proportion of patients with shockable cardiac rhythm, proportion of patients with bystander CPR, type of vascular access, type of airway devices, and the time interval between emergency call and administration of epinephrine.
Table 1. -
Patients’ characteristics, baseline and treatment-related characteristics, and outcomes in the Epinephrine group of the PARAMEDIC2 trial and the German Resuscitation Registry
PARAMEDIC2 Epinephrine Group
German Resuscitation Registry (GRR)
Baseline model
Baseline plus treatment model
n = 4015
n = 26 019
Baseline and treatment-related
Age (year) [mean (SD)]
69.7 (±16.6)
72.4(±14.6)
x
x
Sex (male) (%)
65.0
66.1
x
x
Only supraglottic airway (yes) (%)
70.8
20.6
x
Intraosseous access (yes) (%)
33.4
17.0
Intravenous access (yes) (%)
68.2
94.2
x
Tracheal tube (yes) (%)
30.0
74.2
Interval between emergency call and ambulance arrival at scene (min) [median (IQR)]
6.7 (4.3–9.7)
8.0 (6.0–10.0)a
x
x
Interval between emergency call and administration of epinephrine (min)
21.5 (16.0–27.3)
16.0 (13.0–20.0)a
x
Initial cardiac rhythm (%)
x
x
Shockable (yes)
19.2
23.7
Not shockable (yes)
78.4
75.0
Asystole (yes)
53.5
53.9
Cause of cardiac arrest (%)
x
x
Medical cause (yes)
91.1
82.6
Asphyxia (yes)
2.9
12.7
Drowning (yes)
0.2
0.4
Witness of cardiac arrest (%)
x
x
Bystander (yes)
50.1
45.2
Paramedic/EMS physician or paramedic (yes)
11.3
7.4
CPR performed by bystander (yes) (%)
59.3
34.9
x
x
Outcomes
ROSC (yes) (%)
36.3
46.5
Hospital admission with ROSC (yes) (%)
23.8
36.8
Survival at discharge (yes) (%)
3.2
9.9
Favourable Neurologic outcome at hospital discharge (yes) (%)
2.2
6.5
The distribution of missing data is shown in the Supplementary Material, (Supplemental Digital Content 2,
https://links.lww.com/EJEM/A347 ). Further, the variables of the two types of prediction models – a baseline model and a baseline plus treatment model – are indicated as crosses.
CPR, cardiopulmonary resuscitation; EMS, emergency medical service; IQR, interquartile range; ROSC, restoration of spontaneous circulation.
a For comparison with the PARAMEDIC2 trial, 2 min are added to the raw values of the GRR (see Methods).
Model evaluation
We evaluated the goodness-of-fit and predictive performance of the logistic regression models in predicting the outcomes in the resuscitation patient pathway of the GRR. Table 2 presents the ORs and associated CIs for the logistic regression model for the short-term outcomes of ROSC and hospital admission with ROSC. The models featured moderate AUROC and Nagelkerke pseudo R-square values of 0.70 (95% CI, 0.69–0.70; pseudo-R2 , 0.15) for ROSC and 0.62 (95% CI, 0.61–0.64; pseudo-R2 , 0.05) for hospital admission with ROSC, respectively. The model fit for the outcomes ‘Survival at 30 days’ and ‘Favourable Neurological Outcome’ are shown in Table 3 with corresponding AUROC and Nagelkerke pseudo R-square values of 0.75 (95% CI, 0.74–0.77; pseudo-R2 , 0.21) and 0.77 (95% CI, 0.75–0.79; pseudo-R2 , 0.21), respectively. The calibration plots in the Supplementary Material (Supplementary Figure SM 1, Supplemental Digital Content 1, https://links.lww.com/EJEM/A346 https://links.lww.com/EJEM/A346
Table 2. -
Model coefficient and associated confidence intervals for the logistic regression model for the outcomes ROSC and Hospital Admission with ROSC
ROSC
Hospital Admission with ROSC
Baseline plus treatment modela
Baseline plus treatment modelb
OR (95% CI)
P -value
OR (95% CI)
P -value
Interval between emergency call and ambulance arrival at scene (min)
0.98 (0.97–0.99)
<0.001
0.99 (0.97–1.01)
0.4
Age (year)
1.00 (1.00–1.01)
0.020
0.99 (0.99–1.00)
<0.001
Sex
Female
–
–
Male
0.76 (0.70–0.82)
<0.001
0.91 (0.80–1.03)
0.2
Initial cardiac rhythm
Shockable
–
–
Nonshockable
0.31 (0.28–0.34)
<0.001
0.57 (0.49–0.65)
<0.001
Other
0.40 (0.28–0.58)
<0.001
1.95 (0.90–5.11)
0.13
Cause of cardiac arrest
Medical
–
–
Asphyxia
2.11 (1.90–2.34)
<0.001
2.06 (1.73–2.48)
<0.001
Other
1.43 (1.21–1.68)
<0.001
1.38 (1.04–1.84)
0.029
Witness of cardiac arrest
Bystander
–
–
Paramedic/Rescue-Team
1.15 (0.98–1.35)
0.087
0.65 (0.52–0.83)
<0.001
Other
0.51 (0.48–0.55)
<0.001
0.75 (0.66–0.86)
<0.001
CPR performance
Bystander
–
–
Not further identified
1.22 (1.13–1.32)
<0.001
1.20 (1.06–1.37)
0.005
Only supraglottic airway
No
–
–
Yes
0.53 (0.48–0.58)
<0.001
0.73 (0.62–0.86)
<0.001
Intravenous access
No
–
–
Yes
1.24 (1.03–1.48)
0.021
1.31 (0.96–1.76)
0.087
Interval between emergency call and administration of epinephrine (min)
1.00 (1.00–1.00)
>0.9
1.00 (0.99–1.01)
0.5
Coefficients are shown for the models including both baseline and treatment-related predictors (Baseline plus treatment model,
Table 1 ). Goodness-of-fit and discrimination ability of the models are assessed by the area under the receiver operating curve (AUROC) and Nagelkerke’s pseudo R-squared statistic. Numbers of patients for each logistic regression model are provided in the flowchart (
Fig. 1 ). Note that data availability naturally decreases over the course of the resuscitation pathway as we model probabilities conditional on the previous outcome.
CI, confidence interval; CPR, cardiopulmonary resuscitation; OR, odds ratio; ROSC, restoration of spontaneous circulation.
a AUROC: 0.70 (95% CI, 0.69–0.70), Nagelkerke’s Pseudo R-square: 0.15.
b AUROC: 0.62 (95% CI, 0.61–0.64), Nagelkerke’s Pseudo R-square: 0.05.
Table 3. -
Model coefficient and associated confidence intervals for the logistic regression model for the outcomes Survival at 30 days and favourable neurologic outcome
Survival at 30 days
Favourable neurologic outcome
Baseline plus treatment modela
Baseline plus treatment modelb
OR (95% CI)
P -value
OR (95% CI)
P -value
Interval between emergency call and ambulance arrival at scene (min)
0.97 (0.95–0.99)
0.009
0.97 (0.95–0.99)
0.017
Age (year)
0.96 (0.96–0.97)
<0.001
0.97 (0.96–0.97)
<0.001
Sex
Female
–
–
Male
1.20 (1.03–1.39)
0.020
1.16 (0.97–1.39)
0.10
Initial cardiac rhythm
Shockable
–
–
Nonshockable
0.27 (0.23–0.32)
<0.001
0.23 (0.19–0.28)
<0.001
Other
0.37 (0.19–0.69)
0.002
0.21 (0.08–0.46)
<0.001
Cause of cardiac arrest
Medical
–
–
Asphyxia
0.93 (0.75–1.14)
0.5
0.75 (0.57–0.97)
0.032
Other
0.71 (0.50–1.00)
0.054
0.66 (0.42–0.99)
0.054
Witness of cardiac arrest
Bystander
–
–
Paramedic/Rescue-Team
1.96 (1.49–2.58)
<0.001
2.90 (2.13–3.94)
<0.001
Other
0.70 (0.60–0.82)
<0.001
0.66 (0.55–0.80)
<0.001
CPR Performance
Bystander
–
–
Not further identified
0.97 (0.83–1.12)
0.6
0.97 (0.82–1.15)
0.7
Only supraglottic airway
No
–
–
Yes
0.95 (0.78–1.16)
0.6
0.90 (0.71–1.14)
0.4
Intravenous access
No
–
–
Yes
1.38 (0.92–2.14)
0.14
1.52 (0.92–2.67)
0.12
Interval between emergency call and administration of epinephrine (min)
1.00 (0.99–1.01)
0.6
1.00 (0.99–1.01)
0.5
Coefficients are shown for the including both demographic and therapeutic predictors (Baseline plus treatment model,
Table 1 ). Goodness-of-fit and discrimination ability of the models are assessed by the area under the receiver operating curve (AUROC) and Nagelkerke’s pseudo R-squared statistic. Full data set was available for 5270 patients for the two outcomes. Numbers of patients for each logistic regression model are provided in the flowchart (
Fig. 1 ).
CI, confidence interval; CPR, cardiopulmonary resuscitation; OR, odds ratio.
a AUROC: 0.75 (95% CI, 0.74–0.77), Nagelkerke’s Pseudo R-square: 0.21.
b AUROC: 0.77 (95% CI, 0.75–0.79), Nagelkerke’s Pseudo R-square: 0.21.
Modelling the resuscitation patient pathway
To assess the prediction skill of the logistic regression models, we simulated a random sample of 1000 cohorts (each with n = 4015 patients) within the GRR and compared our prediction with outcomes measured in the GRR (Fig. 2 ). Figure 2a illustrates the distribution of the average outcome in these 1000 cohorts. It is important to note that the success probabilities for each outcome refer to the percentage relative to the previous successful outcome. For example, the logistic regression model with baseline and treatment-related variables predicts that on average, 78.3% of those patients with an initial ROSC were admitted to hospital with a sustained ROSC. Similarly, the logistic regression model with baseline and treatment-related variables predicted that 25.8% of all patients who were admitted to hospital with ROSC survived to hospital discharge and that 17.1% of all patients who were admitted to hospital with ROSC showed favourable neurological outcomes. The prediction agrees very well with the true underlying values of the GRR, which are portrayed as dashed lines in Fig. 2 .
Fig. 2.: Predictions of success probability in the patients’ resuscitation pathway in a random sample of 1000 cohorts (each with n = 4015 patients). Predictions are derived using logistic regression models with only baseline variables (blue) and using baseline plus treatment-related variables (red) fitted to the German Resuscitation Registry. (a) Prediction of individual outcomes conditional on the previous outcomes. (b) Multiplicative success predictions where the mean predicted success probabilities for each outcome for each cohort are multiplied.
Figure 2b illustrates the predictions where the percentages for outcome refer to the entire cohort at the beginning of the pathway. These numbers were derived by multiplying the average outcome prediction at each step of each cohort in the 100-member ensemble with the previous outcome prediction: for example, multiplying the average predictions of two outcomes ROSC and Hospital Admission with ROSC of the logistic regression models resulted in an average prediction rate of 37.3% with successful hospital admission of the entire cohort. Similarly, multiplying the individual predictions of the outcomes ROSC, hospital admission with ROSC, and survival until discharge resulted in an average prediction of 9.6% with successful hospital admission of the entire cohort. Again, the predictions made by the logistic regression models agreed very well with the true underlying values of the GRR. Thus, the modelling framework employed in this study suggests that on a cohort level, the outcome predictions in the resuscitation pathway can be combined in a multiplicative fashion.
Comparison of PARAMEDIC2 with GRR
Using this model framework, we were able to compare two cohorts (the GRR group and the PARAMEDIC2 group) with the aim of answering the question: ‘What would the outcomes in the GRR look like for patients with baseline variables of CPR comparable to those of the patients in the PARAMEDIC2 trial? And further, might the differences in the known treatment-related variables explain the discrepancies in outcome documented in the GRR and the PARAMEDIC2 trial?’
Figure 3 illustrates the modelled outcomes of patients in the GRR fitting the summary measures of patients in the Epinephrine group of the PARAMEDIC2 trial based on logistic regression modelling. Based on baseline variables only, simulation of a patient cohort as in the Epinephrine group of the PARAMEDIC2 trial resulted in an average ROSC success rate of 45.7%, average rates of hospital admissions with ROSC of 35.1%, survival to hospital discharge of 7.7%, and survival with favourable neurological outcome of 5.0%, respectively, significantly better than the observed outcomes in the epinephrine group of the PARAMEDIC2 trial (P < 0.001 for all four outcome parameters). Simulation of patient cohorts that are comparable to the PARAMEDIC2 group in baseline plus treatment-related variables resulted in slightly lower average success rates than with baseline variables only (Fig. 3 ), but still significantly better than in the Epinephrine group of the PARAMEDIC2 trial (P < 0.001).
Fig. 3.: Modelled outcomes of patients in the GRR fitting the summary measures of patients in the Epinephrine group of the PARAMEDIC2 trial-based logistic regression modelling (left panels). Subcohorts with n = 4015 patients that match the reference cohort of the Epinephrine Group in the PARAMEDIC2 trial are repeatedly and randomly drawn from the GRR for 1000 times and the mean success probability for each outcome is computed. The figures in the table present the mean ± SD of these predicted mean success probabilities in the ensemble of random subcohorts.
Discussion
We developed a method to compare outcomes of OHCA patients in the GRR with those in the PARAMEDIC2 epinephrine group. In general, comparison and calculation of (expected) outcome rates between centres according to specific prognosis factors is of great scientific interest [9–11
Simulation of a patient cohort with characteristics and baseline conditions as in the PARAMEDIC2 trial proves that outcomes would have had significantly better along the whole resuscitation pathway as observed in the trial. The reasons for this are speculative and manifold. However, as our baseline model adjusted for patient characteristics and other major predictors of outcome after OHCA before ambulance arrival (in the particular rate of bystander CPR and time interval until ambulance arrival at the scene), the answers may lie partly in treatment-related factors during out-of-hospital CPR but also after ROSC (postresuscitation care and inhospital treatment such as targeted temperature management, access to cardiac catheter lab or extracorporeal CPR, and withdrawal of care during intensive care therapy). Information from the PARAMEDIC2 group is available for only three treatment-related variables during the out-of-hospital setting: type of vascular access (intravenous/intraosseous), type of advanced airway management (supraglottic airway/tracheal tube), and time interval from emergency call to the administration of epinephrine. Adding these treatment-related variables to our logistic regression model, the rate of survival to discharge would decrease significantly from 7.7 to 5.6%, and the rate of survival with favourable neurological outcome from 5.0 to 3.4%. Therefore, comparing simulated outcomes of the baseline model with those of the baseline plus treatment model, our results suggest that patients suffering from OHCA might benefit from a higher rate of use of intravenous access, tracheal intubation, and a shorter time to administration of epinephrine. This is supported by the results from several previous studies. A secondary analysis of the ALPS trial (Resuscitation Outcomes Consortium Amiodarone, Lidocaine or Placebo Study) found that antiarrhythmic drug administration during OHCA favoured improved neurological outcome only with intravenous use [12 13 , 14 15 et al . [16 17 18–20
Limitations
Some limitations of our simulation study have to be discussed. Neurological outcome is defined by the CPC score in GRR and by the mRS in the PARAMEDIC2 trial. The Core Outcome Set for Cardiac Arrest statement considers the mRS as more reliable [21 22
The predictive accuracy of our models is moderate, and they may have slightly overestimated the outcome probabilities. Though, particularly calibration plots for survival to discharge and survival with favourable neurological outcome for baseline plus treatment variables show very good accuracy.
We had no access to individual data of the PARAMEDIC2 trial and were only able to use the summary measures as available in the publication [2
Conclusion
We developed a framework for modelling outcomes after OHCA that enables us to achieve comparability of study results between different EMS systems worldwide.
We demonstrate the multiplicative nature of the resuscitation pathway, which suggests that improvements in any part of the pathway directly propagate to successive outcomes.
Our models comparing the outcomes reported in the PARAMEDIC2 trial with those reported in the GRR demonstrate that the ineffectiveness of epinephrine regarding survival with favourable neurological outcome after OHCA may not be easily transferred to EMS and health care systems organized differently from those in the trial.
Our logistic regression models suggest that a higher rate of tracheal intubation, intravenous access, and a shorter time to administration of epinephrine might improve outcomes in this cohort.
Acknowledgements
The authors would like to thank Ms. Jeannie Wurz (medical editor, Department of Anaesthesiology and Pain Medicine, Inselspital, Bern University Hospital, University of Bern, Switzerland) for proofreading and linguistic correction of our manuscript.
Conflicts of interest
There are no conflicts of interest.
References
1. Panchal AR, Bartos JA, Cabañas JG, Donnino MW, Drennan IR, Hirsch KG, et al.; Adult Basic and Advanced Life Support Writing Group. Part 3: adult basic and advanced life support: 2020 American Heart Association guidelines for cardiopulmonary resuscitation and emergency cardiovascular care. Circulation 2020; 142:S366–S468.
2. Perkins GD, Ji C, Deakin CD, Quinn T, Nolan JP, Scomparin C, et al.; PARAMEDIC2 Collaborators. A randomized trial of epinephrine in out-of-hospital cardiac arrest. N Engl J Med 2018; 379:711–721.
3. Nehme Z, Bernard S, Smith K. Epinephrine in out-of-hospital cardiac arrest. N Engl J Med 2019; 380:397.
4. Wnent J, Bein B, Fischer M; German Resuscitation Registry Steering. Epinephrine in out-of-hospital cardiac arrest. N Engl J Med 2019; 380:396–397.
5. Neukamm J, Gräsner JT, Schewe JC, Breil M, Bahr J, Heister U, et al. The impact of response time reliability on CPR incidence and resuscitation success: a benchmark study from the German Resuscitation Registry. Crit Care 2011; 15:R282.
6. Jacobs I, Nadkarni V, Bahr J, Berg RA, Billi JE, Bossaert L, et al.; International Liaison Committee on Resuscitation; American Heart Association; European Resuscitation Council; Australian Resuscitation Council; New Zealand Resuscitation Council; Heart and Stroke Foundation of Canada; InterAmerican Heart Foundation; Resuscitation Councils of Southern Africa; ILCOR Task Force on Cardiac Arrest and Cardiopulmonary Resuscitation Outcomes. Cardiac arrest and cardiopulmonary resuscitation outcome reports: update and simplification of the Utstein templates for resuscitation registries: a statement for healthcare professionals from a task force of the International Liaison Committee on Resuscitation (American Heart Association, European Resuscitation Council, Australian Resuscitation Council, New Zealand Resuscitation Council, Heart and Stroke Foundation of Canada, InterAmerican Heart Foundation, Resuscitation Councils of Southern Africa). Circulation 2004; 110:3385–3397.
7. SQR-BW. Quality report emergency medical service Baden-Württemberg. Reporting year 2020. 2021
https://sqrbw.de/fileadmin/SQRBW/Downloads/Qualitaetsberichte/SQRBW_Qualitaetsbericht_2020_Inhalt_web2.pdf . [Accessed 23 October 2021]
8. R Core Team. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2020.
https://www.R-project.org/ [Accessed 2 December 2021]. [Accessed 2 December 2021]
9. Gräsner JT, Meybohm P, Lefering R, Wnent J, Bahr J, Messelken M, et al.; German Resuscitation Registry Study Group. ROSC after cardiac arrest–the RACA score to predict outcome after out-of-hospital cardiac arrest. Eur Heart J 2011; 32:1649–1656.
10. Seewald S, Wnent J, Lefering R, Fischer M, Bohn A, Jantzen T, et al. CaRdiac Arrest Survival Score (CRASS) - a tool to predict good neurological outcome after out-of-hospital cardiac arrest. Resuscitation 2020; 146:66–73.
11. Lefering R, Huber-Wagner S, Bouillon B, Lawrence T, Lecky F, Bouamra O. Cross-validation of two prognostic trauma scores in severely injured patients. Eur J Trauma Emerg Surg 2021; 47:1837–1845.
12. Daya MR, Leroux BG, Dorian P, Rea TD, Newgard CD, Morrison LJ, et al.; Resuscitation Outcomes Consortium Investigators. Survival after intravenous versus intraosseous amiodarone, lidocaine, or placebo in out-of-hospital shock-refractory cardiac arrest. Circulation 2020; 141:188–198.
13. Khera R, Chan PS, Donnino M, Girotra S; American Heart Association’s Get With The Guidelines-Resuscitation Investigators. Hospital variation in time to epinephrine for nonshockable in-hospital cardiac arrest. Circulation 2016; 134:2105–2114.
14. Hansen M, Schmicker RH, Newgard CD, Grunau B, Scheuermeyer F, Cheskes S, et al.; Resuscitation Outcomes Consortium Investigators. Time to epinephrine administration and survival from nonshockable out-of-hospital cardiac arrest among children and adults. Circulation 2018; 137:2032–2040.
15. Benger JR, Kirby K, Black S, Brett SJ, Clout M, Lazaroo MJ, et al. Effect of a strategy of a supraglottic airway device vs tracheal intubation during out-of-hospital cardiac arrest on functional outcome: the AIRWAYS-2 randomized clinical trial. JAMA 2018; 320:779–791.
16. Wang HE, Schmicker RH, Daya MR, Stephens SW, Idris AH, Carlson JN, et al. Effect of a strategy of initial laryngeal tube insertion vs endotracheal intubation on 72-hour survival in adults with out-of-hospital cardiac arrest: a randomized clinical trial. JAMA 2018; 320:769–778.
17. Jabre P, Penaloza A, Pinero D, Duchateau FX, Borron SW, Javaudin F, et al. Effect of bag-mask ventilation vs endotracheal intubation during cardiopulmonary resuscitation on neurological outcome after out-of-hospital cardiorespiratory arrest: a randomized clinical trial. JAMA 2018; 319:779–787.
18. Behrens NH, Fischer M, Krieger T, Monaco K, Wnent J, Seewald S, et al. Effect of airway management strategies during resuscitation from out-of-hospital cardiac arrest on clinical outcome: a registry-based analysis. Resuscitation 2020; 152:157–164.
19. Benoit JL, Gerecht RB, Steuerwald MT, McMullan JT. Endotracheal intubation versus supraglottic airway placement in out-of-hospital cardiac arrest: a meta-analysis. Resuscitation 2015; 93:20–26.
20. Sulzgruber P, Datler P, Sterz F, Poppe M, Lobmeyr E, Keferböck M, et al. The impact of airway strategy on the patient outcome after out-of-hospital cardiac arrest: a propensity score matched analysis. Eur Heart J Acute Cardiovasc Care 2018; 7:423–431.
21. Haywood K, Whitehead L, Nadkarni VM, Achana F, Beesems S, Böttiger BW, et al.; COSCA Collaborators. COSCA (Core Outcome Set for Cardiac Arrest) in adults: an advisory statement from the international liaison committee on resuscitation. Resuscitation 2018; 127:147–163.
22. Rittenberger JC, Raina K, Holm MB, Kim YJ, Callaway CW. Association between cerebral performance category, modified rankin scale, and discharge disposition after cardiac arrest. Resuscitation 2011; 82:1036–1040.
