Study design and data source
This study was a single-centre, retrospective review of prospectively collected routine data. Data were retrieved from the database of the prehospital emergency physician response system located at a University Medical Centre. This response system is staffed with a prehospital emergency physician, specialised in either anaesthesia, internal medicine or surgery, and a paramedic. It is dispatched around 2000 times a year to take care of approximately 200,000 individuals in the eastern part of the city and its suburban area.
Routine medical documentation based on the minimal dataset for emergencies (MIND) laid out by the German Interdisciplinary Society of Intensive Care and Emergency Medicine (DIVI), which is electronically collected and stored (MEDEA, iLogs, Klagenfurt, Austria), was retrieved. Routine documentation includes, but is not limited to, Glasgow Coma Score (GCS), National Advisory Committee for Aeronautics Score (NACA), and suspected diagnosis15,16.
Ethical approval and need for consent
Ethical approval was sought and granted by the ethics committee of the Medical University of Graz (IRB00002556), decision number 30–373 ex 17/18 before data retrieval and study conduction. The need for informed consent was waived as data were retrieved and analysed retrospectively and in a pseudonymised fashion.
All used methods and performed analyses were carried out in accordance with relevant guidelines (STROBE statement) and regulations (especially the European General Data Protection Regulation).
Patient selection, data extraction and preparation
Data of missions between January 1st, 2010 and December 31st, 2019 were extracted. Cases in which physicians treated unconscious patients (defined as GCS below 9 upon arrival at the scene until the end of the mission) were selected. Patients with cardiac arrest, interhospital transfers, blank datasets (e.g., due to technical difficulties of the documentation system), patients with missing age data, and patients who were already treated by another physician on scene were excluded.
The following variables for analyses were derived: age, GCS, and anonymised physician identifier. If the age was not exactly known or documented, the treating prehospital physician’s estimate was used. These data were checked for plausibility both electronically and manually. Using the United Nations age categories adapted to create comparable group sizes, the patient cohort was divided into the following age groups: under 18 years, 18–29 years, 30–49 years, 50–64 years, 65–75 years, 75–84 years, and 85 years and older.
Further data retrieved were: conduction of RSI, suspected cause of unconsciousness, and care dependency. RSI was considered conducted if anaesthetic drugs were administered and an advanced airway (either an endotracheal tube, a supraglottic airway device, or a surgical airway) was placed. Each case was allocated to one of seven predefined causes (trauma, medical, toxicology, seizure, cerebrovascular, other, and unknown) of unconsciousness. Documented patient history, clinical examination, and prehospital diagnosis were used as a basis, which were evaluated by two researchers independently and discussed in the team in cases of disagreement. Consensus was reached in all cases. Care dependency was considered if explicitly mentioned in free text documentation or if the patient was living in a nursing home. The anonymous dataset used for this study can be found in the electronic supplementary materials attached to this manuscript.
Statistical analysis
Factors possibly influencing the decision to perform RSI were presented as median and interquartile ranges (IQR) or number (n) and percentages (%), as appropriate. Between-group comparisons were performed using a Kruskal–Wallis test or Chi-square test, as appropriate. Bonferroni correction was used to adjust for multiple testing, p-values below 0.05 were considered significant.
A multivariable mixed linear regression model with RSI as the dependent variable and the aforementioned possible factors as independent variables was constructed; these were: age in categories (under 18 years, 18–29 years, 30–49 years, 50–64 years, 65–74 years, 75–84 years, and 85 years and older; 50–64 years, being the median age group, was chosen as the reference group), independent living, GCS, and suspected cause of unconsciousness. The model was further adjusted for anonymised identifiers of physicians making the decision as random effects.
For sensitivity analysis, the model was repeated in patients with non-traumatic suspected causes for unconsciousness only. Receiver Operating Characteristics (ROC) analyses were performed to assess predictive capabilities of these models.
All analyses were performed using IBPM SPSS 26 (IBM Corp, Armonk, NY, USA).
IRB identifier and decision
Ethics committee of the Medical University of Graz (IRB00002556), decision number 30–373 ex 17/18.
