Short-term outcome and characteristics of critical care for nontrauma patients in the emergency department

Emergency medical care for critically ill nontrauma patients (CINT) varies between different emergency departments (ED) and healthcare systems, while resuscitation of trauma patients is always performed within the ED. In many ED CINT are treated and stabilized while in many German smaller hospitals CINT are transferred directly to the intensive care unit (ICU) without performing critical care measures in the ED. Little is known about the resuscitation room management of CINT regarding patient characteristics and outcome although bigger hospitals perform ED resuscitation of CINT in routine care. Against this background we conducted this retrospective analysis of CINT treated by an ED resuscitation room concept in a German 756 bed teaching hospital. The collective of CINT treated within the ED resuscitation room (1 October 2018 to 31 March 2019) was analyzed after ethical approval. After each resuscitation room operation, the team leader filled out a standardized paper-based questionnaire and qualified the patient as a resuscitation room patient this way. Only patients who underwent invasive procedures and were admitted to ICU or died in the ED were included. Patient characteristics, performed critical care measures, short-term outcomes and the comparison of admission characteristics between survivors and non-survivors were evaluated. Additionally, the accordance of ED admission diagnoses and discharge diagnoses were analyzed. Overall, 243 of 19,854 ED patients (1.22%) were treated in the resuscitation room. After exclusion of trauma patients, 193 (0.97%) CINT were included. Overall mortality was 29% (n = 56), 24‑h mortality was 13% (n = 25). Patient characteristics (vital signs, blood gas analysis) differed significantly between survivors and nonsurvivors except for respiratory rate and pain scale. An excerpt of conducted resuscitation room measures was as follows: arterial line n = 78 (40%); noninvasive ventilation n = 60 (31%); endotracheal intubation n = 56 (29%); cardiopulmonary resuscitation n = 19 (10%), central venous line n = 8 (4%). The number of conducted measures differed between survivors and nonsurvivors (median and interquartile range, IQR): 4 (IQR 2) vs. 4 (IQR 3) p = 0.0453. The length of ED stay was 148.2 ± 202.7 min until the patient was admitted to an ICU or died within the ED. ED admission diagnoses matched with hospital discharge diagnoses in 78%. The observed mortality was high and was comparable to patient collectives with septic shock. Nonsurvivors showed significantly more impaired vital parameters and blood gas analysis parameters. Vital parameters together with blood gas analysis might enable ED risk stratification of CINT. Resuscitation room management enables immediate stabilization and diagnostic work-up of CINT even when no ICU bed is available. Furthermore, optimal allocation to specialized ICUs can probably be enabled more accurately after a first diagnostic work-up; however, although a first diagnostic work-up including laboratory tests and computed tomography in many cases was performed, ED admission and hospital discharge diagnoses matched only in 78%.


Introduction
In emergency departments (ED) initial patient care is mostly based on clinical symptoms and vital parameters. Many patients show unspecific symptoms, such as reduced level of consciousness [6]. More detailed diagnoses are often possible after laboratory tests and radiology results. Emergency patients can only be allocated to the correct medical specialty with these results. In larger hospitals with different specialty-based intensive care units (ICU) a correct allocation of intensive care patients is often only possible after a first diagnostic work-up. The number of ICU beds is limited and not always available, therefore correct allocation is crucial to ensure hospital operability. While in many countries all patients transported via emergency medical service (EMS) are admitted in the ED, in other countries critically ill nontrauma patients (CINT) are also admitted directly to the ICU. In trauma patients admission via the resuscitation room became standard and led to improved outcomes but systems of care differ especially for CINT with a need for immediate critical care measures [28]. In Germany concepts also vary widely between resuscitation room management and direct ICU admission for CINT. In many German hospitals CINT are transferred directly via EMS or rapidly to the ICU without performing critical care measures in the ED. A direct admission from the EMS to ICU, which is standard in many German hospitals, might lead to wrong patient allocation. Additionally, the correct ICU bed is not always available at this time point. Furthermore, it is well known that a structured interdisciplinary resuscitation room work-up is beneficial in major trauma and is one key element that led to mortality reduction [28]. Consequently, direct admission of trauma patients to ICU has already been discontinued for more than two decades. In CINT it is unknown if resuscitation room care is beneficial and concepts vary not only from country to country but also from hospital to hospital. Even mortality and severity of CINT patients are not well studied in contrast to multiple scientific evaluations in trauma trials and trauma registers. Against this background we evaluated our patient collective of CINT patients treated with our resuscitation room concept. Patient characteristics, short-term outcome, conducted critical care measures, medical severity between survivors and nonsurvivors and diagnoses were analyzed.

Study design
We conducted a single-centre retrospective observational cohort study. Some data were collected prospectively for quality management purposes in a local resuscitation room registry but the rest of the data were retrieved retrospectively. To analyze resuscitation room data a questionnaire with check boxes about the performed procedures was implemented and used for every patient that was judged to be a critical care patient due to performed critical care measures. For CINT, only patients who underwent invasive procedures and/or were admitted to an ICU were included in this registry via the questionnaire. Data

Setting
The ED in our 756-bed academic teaching hospital runs 3 resuscitation rooms, 1 of which is mainly used for trauma care and 2 are mainly used for nontrauma care. All resuscitation rooms were fitted out with standard equipment according to the airway, breathing, circulation, disability and environment (ABCDE) scheme. Beside this, video laryngoscopy (McGrath, Medtronic, Minneapolis, MN, USA and C-MAC, Karl Storz, Tuttlingen, Germany), point of care blood gas analysis including hemoglobin, electrolytes, glucose and lactate, a mobile sonography unit (Xario 100G, Canon Medical Systems GmbH, Neuss, Germany) and a mechanical chest compression device (LUCAS 2, Jolife AB, Lund, Sweden) were available.

Inclusion criteria and exclusion criteria
After exclusion of trauma patients we included all patients who were initially classified as CINT that showed no stable vital parameters and/or with a need for critical care measures (. Fig. 1). Exclusion criteria were patients with a preknown palliative status. Furthermore, patients with signs of acute stroke and acute coronary syndromes with stable vital parameters and no need for critical care procedures were excluded, even when they were locally treated within one of the three resuscitation rooms.

Outcome parameters
The following outcomes were evaluated: Patient characteristics at admission, 24-h mortality and in-hospital mortality, performed critical care measures within the ED and length of stay in the resuscitation room. Furthermore, it was analyzed if performed measures and initial patient characteristics differed between survivors and non-survivors to allow future risk stratification. The ED admission diagnoses of the resuscitation room protocol and hospital discharge diagnoses and as well as their accordance were evaluated.

Data sources
An electronic health record system (iMe-dOne, Deutsche Telekom Healthcare and Security Solutions GmbH, Bonn, Germany) served as the main data source. For quality management purposes a prospective local registry of critical care resuscitation room management (Excel, Microsoft Corporation, Redmond, WA, USA) using pseudonymized patient data complemented the electronic health records. Pseudonymized data of the aforementioned questionnaire was transferred into this secured and userrestricted database and was amended by secondarily pseudonymized patient data of the electronic health records.

Ethics committee approval and data privacy
Due to the process that only routine data were collected, and all data analyses were performed in a pseudonymized way the ethics committee (University hospital Aachen, Germany) waived personal consent by patients or ED personnel (EK051/20). The study was registered in the Clinical Trials Centre of the RWTH Aachen University (registration number CTC-A 20-131) and meets the criteria of the STROBE statement [9].

Statistical analyses
Performed procedures, short-term outcome and diagnoses were analyzed with descriptive data. The comparison of admission characteristics of survivors and nonsurvivors were performed using the Mann-Whitney U-test. Due to exploratory character of the study p-values < 0.05 were considered to be significant. All statistical analyses were carried out with Prism 8.4.2 GraphPad Software (San Diego, CA, USA).

Abstract
Background. Emergency medical care for critically ill nontrauma patients (CINT) varies between different emergency departments (ED) and healthcare systems, while resuscitation of trauma patients is always performed within the ED. In many ED CINT are treated and stabilized while in many German smaller hospitals CINT are transferred directly to the intensive care unit (ICU) without performing critical care measures in the ED. Little is known about the resuscitation room management of CINT regarding patient characteristics and outcome although bigger hospitals perform ED resuscitation of CINT in routine care. Against this background we conducted this retrospective analysis of CINT treated by an ED resuscitation room concept in a German 756 bed teaching hospital. Methods. The collective of CINT treated within the ED resuscitation room (1 October 2018 to 31 March 2019) was analyzed after ethical approval. After each resuscitation room operation, the team leader filled out a standardized paper-based questionnaire and qualified the patient as a resuscitation room patient this way. Only patients who underwent invasive procedures and were admitted to ICU or died in the ED were included. Patient characteristics, performed critical care measures, short-term outcomes and the comparison of admission characteristics between survivors and non-survivors were evaluated. Additionally, the accordance of ED admission diagnoses and discharge diagnoses were analyzed. Results. Overall, 243 of 19,854 ED patients (1.22%) were treated in the resuscitation room. After exclusion of trauma patients, 193 (0.97%) CINT were included. Overall mortality was 29% (n = 56), 24-h mortality was 13% (n = 25). Patient characteristics (vital signs, blood gas analysis) differed significantly between survivors and nonsurvivors except for respiratory rate and pain scale. An excerpt of conducted resuscitation room measures was as follows: arterial line n = 78 (40%); noninvasive ventilation n = 60 (31%); endotracheal intubation n = 56 (29%); cardiopulmonary resuscitation n = 19 (10%), central venous line n = 8 (4%). The number of conducted measures differed between survivors and nonsurvivors (median and interquartile range, IQR): 4 (IQR 2) vs. 4 (IQR 3) p = 0.0453. The length of ED stay was 148.2 ± 202.7 min until the patient was admitted to an ICU or died within the ED. ED admission diagnoses matched with hospital discharge diagnoses in 78%. Conclusion. The observed mortality was high and was comparable to patient collectives with septic shock. Nonsurvivors showed significantly more impaired vital parameters and blood gas analysis parameters. Vital parameters together with blood gas analysis might enable ED risk stratification of CINT. Resuscitation room management enables immediate stabilization and diagnostic workup of CINT even when no ICU bed is available. Furthermore, optimal allocation to specialized ICUs can probably be enabled more accurately after a first diagnostic work-up; however, although a first diagnostic work-up including laboratory tests and computed tomography in many cases was performed, ED admission and hospital discharge diagnoses matched only in 78%.

Outcomes
The 24-h mortality was 13% (n = 25) for CINT. All of these patients died within the resuscitation room or ED and were not admitted to the ICU. From this subgroup n = 11 were already admitted under on-going cardiopulmonary resuscitation by the EMS at ED handover. Over-all hospital mortality for CINT was 29% (n = 56).

Resuscitation and critical care measures
In .

Comparison of survivors vs. nonsurvivors
Patient characteristics at admission differed significantly between survivors and nonsurvivors except for respiratory rate and pain scale. (. Table 3). Between survivors and nonsurvivors the amount of conducted critical care procedures in the ED differed also: median 4 (IQR 2) vs. median 4 (IQR 3) p = 0.0453.

Admission and discharge diagnoses
. Figure 2 displays the ED admission diagnoses and shows in how many cases the hospital discharge diagnosis was in accordance. Respiratory insufficiency, reduced level of consciousness and sepsis were the main ED admission diagnoses. In 77.7% (n = 150) ED diagnoses and hospital discharge diagnosis matched after clinical review by the authors.

Discussion
The CINT patients admitted to ED resuscitation room showed a high 24-h and hospital mortality despite immediately performed resuscitation measures. Survivors and non survivors differed regarding admission parameters and the amount of immediately performed emergency medical measures. Although initial diagnostics included laboratory and computed tomography, ED admis- Overall, our study enables an overview about a mixed, unlimited real-life patient collective in a 756-bed academic teaching hospital, serving 4 districts. Most CINT showed a reduced level of consciousness, respiratory problems, metabolic or respiratory acidosis and elevated serum lactate while other vital parameters were often only mildly impaired (. Table 1). The rate of critically ill patients, performed interventions (. Table 2) and leading clinical problems (. Tables 1 and 3; . Fig. 2) were comparable to the nontrauma resuscitation room collective described by Bernhard et al. [6].
The 24-h and in-hospital mortality were comparable to the abovementioned collective described at a German university hospital [6]. Furthermore, they comparable to trials that researched protocolbased resuscitation strategies in septic shock [3,20,24]. Mortality rates of patients with septic shock in those trials were between 18.6% (ARISE) and 24.6% (ProMISE), which shows that the sever-ity of a mixed collective of CINT was comparable to these studies [3,20,24]. The mortality in our collective was much higher compared to major trauma patients in the German Trauma Registry but especially the mean age differed between the trauma registry patients and our collective [2].
All patients in our study that were handed over in the ED under on-going cardiopulmonary resuscitation (CPR) died in the resuscitation room after airway management and exclusion of reversible causes of cardiac arrest. Overall, the rate of survival in patients transported by EMS under cardiopulmonary resuscitation was low, which is supported by current literature, but outcome data differed widely due to different study settings and criteria [11,12,27]. If patients with refractory cardiac arrest are transported early and with high quality CPR this might improve outcome [1]. One reason for 100% fatal outcome in our subgroup was that patients qualified for venoarterial extracorporeal life support (V-A-ECLS) under cardiopulmonary resuscitation bypassed the ED and were directly transferred to our cardiology ICU via EMS. Patients under on-going CPR were selected via telephonic triage to either resuscitation room or V-A-ECLS in the ICU. In some ED ECLS is already implemented and this resuscitation technique is not restricted to ICU [27]. In the last decade survival of selected patients receiving ECLS improved over time [13].
Treatment of respiratory failure, airway management, invasive monitoring and circulatory support were the main critical care measures carried out for CINT in the resuscitation room. Therefore, an interdisciplinary resuscitation room concept for all patients with impaired vital parameters with fast primary survey, (invasive) stabilization, early bedside sonography and computed tomography is probably beneficial compared to primary allocation and admission of critically ill patients to specialty-based ICUs (e.g. neurology ICU), only based on the main clinical symptom. A relevant number of patients-especially the nonsurvivors-showed a reduced level of consciousness ad ED admission as the leading symptom but only a few patients turned out to have neurological diagnoses as the main discharge diagnosis (. Tables 1 and 3; . Fig. 2).
The amount and type of conducted emergency medical measures demonstrate that these techniques are crucial skills for interdisciplinary high-volume ED. Especially airway management, noninvasive and invasive ventilation as well as circulatory support are skills that are necessary within the first minutes after patient arrival. Invasive ventilation after rapid sequence induction in the ED was necessary in 29% of the analyzed patients. This fraction is comparable to another collective in a German university hospital [5]. Therefore, structured concepts are necessary to enable a guidelinebased and rapid medical treatment. For trauma patients certified concepts with courses and certification (e.g. Advanced Trauma Life Support ® ) are current standards in every trauma centre, while for nontrauma patients medical and organizational settings differ widely [10,14,21,28]. In major trauma patients a structured concept and guideline adherence can lead to significant reduction of mortality [25]. Overall, airway management in the ED is a current field of clini-cal research although outcome data are mostly not available [5,15,26]. A mean ED treatment phase of 148.2 min in our collective points out that this concept led to relevant resource consumption. This must be taken into account when implementing such a strategy. The length of stay in the ED was longer than described by Bernhard et al., but we did not differentiate between the first phase in the resuscitation room and the second phase of transport to computed tomography and treatment time until an ICU bed was available [6]. Unfortunately, we did not document the exact use of human resources regarding consultant, resident and nurse deployment, which is one of the major limitations in our study. Overall, data about resuscitation room management of CINT collectives are rare and therefore it remains unclear if an invasive, rapid approach is generally beneficial compared to less invasive ED care and rapid transfer to ICU;however, rapid critical care treatment in the ED is probably beneficial for patients in septic shock. Very early continuous infusion of norepinephrine seems to be beneficial in septic shock and leads to shortened ICU stay and earlier restoration of circulation [16,22,23]. If continuous norepinephrine infusion is restricted to ICU settings, delays in shock therapy are probably inevitable. Severe pneumonia with end-organ dysfunction and sepsis of other origins also accounted for a relevant number in our collective (. Fig. 2). Invasive measurement of arterial pressure seems to be more accurate in septic shock compared to oscillometric standard measurements [19]. Our approach of frequent placement of arterial lines seems to be more accurate and therefore probably beneficial in correct clinical evaluation and risk stratification. If fluid and vasopressor therapy is delayed due to incorrect oscillometric values, this might negatively affect patient outcome.
Nonsurvivors showed a significantly higher rate of conducted critical care measures which is explainable due to more impaired vital functions. Significant baseline differences in vital parameters and blood gas analysis were detected in comparison between survivors vs. nonsurvivors, except for respiratory rate and pain scale; however, respiratory rate showed a median of 19 and 20/min in both groups, which is already a nonphysiological value. Risk stratification of ED patients during triage and initial assessment is crucial for allocation of pa-tients and resources. Several scores were developed and compared to differentiate critical from noncritical patients [7,18,29]. All patients included in this study were classified as critically ill patients initially. Within this group-representing 0.97% of all ED patients-the amount of GCS reduction, acidosis and elevation of lactate were highly significantly associated with fatal outcome. Impaired vigilance and lactate elevation are associated with fatal outcome in ED patients with severe sepsis [3,8,20,24]. Despite several scores to detect patients at risk, failure of vital sign normalization is more strongly associated with mortality than single measures [18]. Therefore, these parameters are useful in risk stratification within the group of already detected critically ill patients. The clinical value of point of care lactate measurement was also shown by Kramer et al., but after exclusion of patients receiving cardiopulmonary resuscitation at ED arrival, initial lactate did not differ between survivors and nonsurvivors [17].
A broad spectrum of ED admission diagnoses was seen in CINT (. Fig. 2). Despite a structured resuscitation room procedure including computed tomography and laboratory tests, the ED admission diagnosis matched with the hospital discharge diagnosis in only 77.7% of the patients. In nonpneumonic sepsis higher rates of mismatch between admission diagnosis and discharge diagnosis were observed, compared to other admission diagnoses and pneumogenic sepsis. Atmna et al. reported a discordant diagnosis in patients diagnosed with pneumonia in the ED in 29% of the cases [4]; however, it remains unclear which factors led to discordant diagnoses in our setting. In supposed gastrointestinal bleedings the rate of discordant diagnosis between ED and hospital discharge was 50%. The main reason for this observation was probably the time point of documentation. The ED diagnosis was documented prior to transport for esophagogastroscopy.
Although it was not the purpose of our study often bridging time was observed until ICU admission was possible. It can be assumed that without invasive ED management, the patients would have received some necessary treatment later. Future (randomized) studies have to compare direct ICU admission via EMS versus initial resuscitation room management regarding patient outcome.
To ensure structured care for these patients at risk we determined three different team compositions for the following groups of CINT as a consequence of our findings: basic team for patients at risk with stable vital parameters (1 ED nurse, 1-2 ED physicians); expanded team for patients with unstable vital functions (1 ED resident, 1 ED consultant, 2 ED nurses); cardiac arrest receiving team (2 ED residents, 1 ED consultant, 2 ED nurses, cardiology consultant and orthopaedic surgery resident/consultant on demand).

Limitations
Our study was a retrospective study with all its limitations, despite the fact that some of the data were collected prospectively for quality management and clinical review. We used our special resuscitation room questionnaires to include the patients. Therefore, we might have missed some patients who were treated invasively due to forgotten questionnaires. We could not differentiate between patients who were already assessed as critical by EMSvia physician-tophysician telephone communication and patients that turned out to be critical after arrival and no procedures like rapid sequence induction were prepared, because the content of the prehospital telephone call was not documented in the health records.

Conclusion
Our mixed real life clinical collective shows that resuscitation room management for CINT is a complex process with a high necessity of invasive procedures. Overall mortality of CINT was high and comparable to collectives of patients in septic shock. Beside vital parameters blood gas analysis and lactate enable early risk stratification for fatal outcome although scores are missing.