This study is a joint project of the TraumaRegister DGU® (TR-DGU) and the German Resuscitation Registry (GRR).
German resuscitation registry
The German Resuscitation Registry (GRR) is a voluntary registry founded in 2002 and is run by the German Society of Anesthesiology and Intensive Care Medicine (Deutsche Gesellschaft für Anästhesiologie und Intensivmedizin e.V., DGAI). Each EMS or hospital can decide whether they want to participate in GRR. There is no legal obligation to participate. The GRR collects anonymous data on out-of-hospital and in-hospital cardiac arrest patients. The data is collected at different time points “pre-hospital treatment/initial treatment”, “in-hospital treatment”, and “long-term survival”. The datasets adhere to the Utstein recommendations [13, 14], and data are entered into the database via a password-protected online reporting system. The database includes plausibility and completeness checks as well as an analysing tool. Participating EMS and hospitals receive a comprehensive report, including risk-stratification with the ROSC-after-cardiac-arrest-score (RACA) or the CaRdiac-Arrest-Survival-Score (CRASS) [15, 16]. GRR receives information from EMS systems covering approximately 30 million inhabitants, 36% of Germany’s total population of 83 million. This project is approved by the scientific advisory board of GRR (Ref. No.: 20190601_JW).
TraumaRegister DGU®
The TraumaRegister DGU® (TR-DGU) of the German Trauma Society (Deutsche Gesellschaft für Unfallchirurgie e.V., DGU) was founded in 1993. This multi-centre database provides pseudonymised information about severely injured patients. Data are collected prospectively from the site of the accident until discharge from hospital at the following times: A) pre-hospital phase, B) emergency room and initial surgery, C) intensive care unit and D) discharge. The documentation includes detailed information on demographics, injury patterns, comorbidities, pre- and in-hospital management, treatment and care in the intensive care unit (ICU), relevant laboratory findings and data on transfusion and outcome of each individual. Inclusion criteria in the registry are admission to hospital via emergency room with subsequent care in an ICU or admitted to hospital with vital signs but dead before admission to ICU.
Scientific data analysis of this project is approved according to a peer-review process and it is in line with the publication guidelines of the TR-DGU and registered as TR-DGU project ID 2018–043.
Patients
Patients suffering OHCA due to trauma between 01.01.2014 and 31.12.2019, where cardiopulmonary resuscitation (CPR) was started, and the data was registered either in the GRR or in TR-DGU were eligible for inclusion. The data from both registries were not matched or merged; analyses were performed independently and in parallel. Due to data security and confidentiality, only anonymised data was available in both registries, and there was no information available about whether or not a patient was included in both registries. GRR provided information about the out-of-hospital treatment and outcome after cardiac arrest, but the TR-DGU was limited to trauma patients arriving in the hospital. The dataset in GRR focused on resuscitation related items, and the dataset in TR-DGU focused on surgery-related items. In this analysis, we used data from these two large national registries to analyse the midterm outcome of CPR after traumatic cardiac arrest in Germany.
To ensure high-quality data, we only included EMS systems from GRR with:
-
incidence of resuscitation started > 30/100,000 inhabitants per year,
-
any return of spontaneous circulation (ROSC) < 80%,
-
ROSC-After-Cardiac-Arrest-score (RACA-score) available for > 60% of the patients [15],
-
documentation of in-hospital care (in case of hospital admission) for > 30% of the patients.
First, all included cases in GRR were analysed. Primary endpoints were ROSC and hospital admission with ROSC. The secondary endpoints were hospital mortality and discharge with a good neurological outcome (Cerebral Performance Categories (CPC) 1 or 2).
Second, all primary admissions from TR-DGU treated in participating German hospitals with information about outcomes were included (n = 178,141). A total of 4,147 (2.7%) patients had a cardiac arrest and cardiopulmonary resuscitation on scene. Patients with missing data for blood pressure (BP) or heart rate (HR) on hospital admission (n = 637) and those with ongoing CPR (n = 842) were excluded. Patients with missing trauma mechanism (n = 46) and those with a mechanism other than trauma (n = 180) were also excluded, leaving 2,460 cases with OHCA after trauma for analysis (1.4%).
In summary, the included data from GRR consisted of all OHCA cases attributed to trauma in the high data quality group (irrespective of outcome). The TR-DGU data consisted of patients with successful prehospital resuscitation who arrived at hospital with ROSC.
Statistical analysis
Continuous data were presented as mean with standard deviation (SD) or median with quartiles in case of skewed distribution. Statistical comparisons were performed with Mann–Whitney U-test. Categorical data were presented as number of patients with percentages (%), and differences were evaluated with chi-squared test. P-values < 0.05 were considered statistically significant.
We developed two multivariate logistic regression models to identify risk factors for in-hospital mortality (dependent variable). The model was developed mainly on clinical significance of the predictor variables. Experts and statisticians from both registries (CPR and trauma) were involved. The model in the TR-DGU dataset considered classical predictors from trauma research (age, sex, penetrating mechanism, unconsciousness, shock, Injury Severity Score, injured body regions: head, thorax, abdomen, extremities), known risk factors from previous research in traumatic CA [3] (patient found in CA; repeated CA in the emergency room), early interventions in the hospital (blood transfusion, emergency surgery) and hospital level of care. Non-significant predictors (p > 0.05) were deleted from the final model.
The model in the GRR dataset considered classical predictors from CA research [15] (age, sex, bystander CPR, initial ECG, found in CA, location of CA) with relevant impact in univariate analysis (p < 0.5) and status at hospital admission (shock at admission).
In GRR, data were first used to analyse all patients with CPR started and then the subgroup of patients with ROSC on admission only, similar to the TR-DGU approach. TR-DGU data were analysed for patients with ROSC on hospital admission only. Results were presented as odds ratios (OR) with 95% confidence intervals. Analyses were performed using SPSS statistical software (version 26, IBM Inc., Armonk, NY, USA).
The ethics committee at Christian-Albrechts University in Kiel provided ethical approval for this study (Ref. No.: D 497/2019).