Mortality of severely injured patients after ICU discharge: An underestimated problem?

Background Most trauma patients admitted to the hospital alive and die later on, decease during the initial care in the emergency department or the intensive care unit (ICU). However, a number of patients pass away after having been discharged from the ICU during the initial hospital stay. On �rst sight these cases could be seen as “failure to rescue” of potentially salvageable patients. A low rate of such patients might be a potential indicator of quality for trauma care on ICUs and surgical wards. Methods Retrospective analysis of the TraumaRegister DGU® with data from 2015-2017. Patients that died during the initial ICU stay were compared to those who were discharged from the initial ICU stay for at least 24 hours but died later on. Results A total of 82,313 trauma patients were included in the TraumaRegister DGU®. In total, 6,576 patients (8.0%) died during their hospital stay. Out of those, 5,481 were admitted to the ICU alive and 972 patients (17.7%) were discharged from ICU and died later on. Those were older (mean age: 77 vs. 68 years), less severely injured (mean ISS: 23.1 vs. 30.0 points) and had a longer mean ICU length of stay (10 vs. 6 days). A limitation of life-sustaining therapy due to a documented living will was present in 46.1% of all patients who died during their initial ICU stay and in 59.9% of patients who died after discharge from their initial ICU stay. Conclusions 17.7% of all non-surviving severely injured trauma patients died within the hospital after discharge from their initial ICU treatment. Their death can partially be explained by a limitation of therapy due to a living will. In conclusion, the rate of such late deaths may represent patients that died of potentially avoidable or treatable complications.


Background
Traumatic injuries are a major cause of death and disability.Worldwide, almost 10% of all deaths are related to trauma.(1) Over the last decades, improvements in trauma care in the elds of prehospital care, emergency department (ED) care, operation techniques and intensive care led to a constant reduction of mortality rates.(2,3) Most established quality indicators focus on the initial trauma care before and during admission (4)(5)(6) In general and elective surgery a number of studies suggested that a relevant proportion of patients decease within the hospital after discharge from their initial (post-operative) stay in the ICU.This group may represent potentially preventable deaths (7)(8)(9)(10).
The failure to rescue (FTR) rate was described as a more precise characterisation of post-operative care quality than mortality or complication rates in elective surgery (11).It was rst de ned in 1992 as the mortality rate among patients with complications.However it has since been rarely applied to trauma and acute care surgery (12) To detect and treat patients on general wards with major complications in an early stage, different strategies have been developed to lower both readmission rates to the ICU and mortality (13) Early detection of patients at risk may be achieved by use of warning scores and other systematic tools of monitoring or implementation of medical emergency teams (14,15).Whether these aspects of improved care for surgical patients are also relevant for the care of severely injured patients is not well known.In trauma patients it can be di cult or even impossible to distinguish between preventable complications and adverse events due to "natural course of the trauma".Thus, reliable rates of failure to rescue are still lacking.(16,17).
One approach could be to analyse severely injured trauma patients who have been discharged to the general ward and die after an initial treatment on an intensive care unit.The aim of this study was to determine the rate of late mortality, characteristics of patients and to compare late deaths with nonsurvivors who died during their initial ICU stay.We expect that this approach is a simple way to identify a cohort of potentially preventable fatal complications and therefore eligible for quality management and benchmarking, especially in large trauma registries.

Study design and setting:
The TraumaRegister DGU ® of the German Trauma Society (Deutsche Gesellschaft für Unfallchirurgie, DGU) was founded in 1993.The aim of this multi-center database is a standardised and pseudo anonymised documentation of severely injured patients.
Data were collected prospectively in four consecutive time phases from accident to discharge from hospital: A) Pre-hospital phase, B) Emergency room and initial surgery, C) Intensive care unit and D) Discharge.The documentation included detailed information on demographics, injury pattern, comorbidities, pre-and in-hospital management, course on intensive care unit, relevant laboratory ndings including data on transfusion and outcome of each individual.The inclusion criterion was admission to hospital via emergency room and subsequent admission to an ICU/ICM care unit or reaching the hospital with vital functions and die before admission to ICU.There were two sets of data: The standard dataset required to document the complete set of data as de ned by the registry.The QM dataset (quality management dataset) required a reduced number of items.It was up to the discretion of the participating hospitals which type of dataset was used.If a hospital decided for either dataset, all patients were documented by the chosen dataset.The infrastructure for documentation, data management, and data analysis was provided by AUC -Academy for Trauma Surgery (AUC -Akademie der Unfallchirurgie GmbH), a company a liated to the German Trauma Society.The scienti c leadership was provided by the Committee on Emergency Medicine, Intensive Care and Trauma Management (Sektion NIS) of the German Trauma Society.The participating hospitals submitted their data into a central database via a web-based application.Scienti c data analysis was approved according to a peer review procedure established by Sektion NIS.The participating hospitals were primarily located in Germany (90%), but a rising number of hospitals of other countries contributed data as well (at the moment from Austria, Belgium, China, Finland, Luxembourg, Slovenia, Switzerland, The Netherlands, and the United Arab Emirates).Currently, approximately about 35,000 cases from almost 700 hospitals were entered into the database per year.
This study was conducted according to the publication guidelines of the TraumaRegister DGU ® (TR-DGU) and registered as TR-DGU Project ID 2019-015.

Selection of patients:
Basis for the analysis were patients from the years 2015 until 2017.In 2015, the data set has been revised and the two relevant parameters "Patient's volition" and "cause of death" were added to the TraumaRegister DGU ® .Only patients from German hospitals were included.All non-surviving patients initially admitted to the ICU were analysed.The target group of patients who died after initial discharge from ICU was de ned as those with length of stay in hospital being more than 1 day longer than the cumulated length of stay on ICU.We could not discriminate patients being less than 24 hours outside the ICU as documentation was performed on a daily basis.
The TR-DGU only collects the total number of days in hospital and on ICU; re-admission to ICU is not coded.Therefore, a discrimination between patients that died after readmission to ICU and patients that died on normal ward was not possible.

Analysis:
Variables were analysed using descriptive statistics (percentages and frequencies) and central tendency measures (mean and median) with standard deviation for metric variables.SPSS version 22 (IBM Inc., Armonk, NY, USA) was used for the statistical analysis.The study was approved by the Ethics Committee of the Ruhr-University Bochum, Germany.

Results
The selection process of the study patients is shown in Figure 1.The excluded surviving patients had a mean age of 50.3 years, a mean ISS of 16.3, mean LOS on ICU was 6.3 days, and mean LOS in hospital was 16.9 days.Overall, 5,481 patients were admitted to the ICU alive and died later in hospital.In total 972 (17.1%) of these patients were discharged from the ICU to normal ward for some days.Differences in demographics, injury pattern, therapy and course in the intensive care unit between patients who died in ICU or later after initial discharge are shown in Table 1.Patients that died before ICU discharge were younger (mean age 68 vs. 77 years), more severely injured (mean ISS 30.0 ± 15 vs. 23.1 ± 12; RISC II prognosis for mortality 52.1 vs. 32.4%); length of stay in ICU was shorter (mean days 6.0 ± 9.4 vs. 10.0 ± 13.4), while living will limiting life-sustaining therapy was documented fewer (46.1 vs. 59.9 %).
Of all non-surviving patients that were initially discharged from ICU, 31.5 % were treated less than three days and 70.9 % less than 8 days outside the ICU (Figure 2).In addition to patient-related factors, we also analysed a potential difference with respect to the level of care of the trauma center (table 2).The rate of patients dying within rst ICU stay was the highest in level 1 trauma centers (84.7 %) and was the lowest in level 3 trauma centers (68.7%).

Discussion
The majority of in-hospital deaths after major trauma occurred during the initial ICU treatment period.This study showed that a relevant group (17.1%) of all non-surviving patients initially admitted to the ICU were discharged from ICU at least once and treated outside the ICU for more than one day.
Multiple studies focussed on in-hospital mortality after discharge from the ICU (18-22) Pearse et al. performed a prospective observational study on mortality after surgery in Europe (9) In emergency surgery, they found a mortality rate of 11.1 % of all patients initially discharged alive from ICU to normal ward.They concluded that ICU resources were not allocated to patients at greatest risk of death.A variety of factors in uencing readmission rate have been identi ed.Several studies showed a relationship between daytime and weekday of discharge from ICU and outcome.They con rmed that unplanned discharges from the ICU during night-time and on weekends were independent risk factors for readmission to the ICU and increased mortality (18, 23,24).Others con rmed older age, severity of illness on the day of admission to the ICU and day of discharge from the ICU as well as comorbidities as independent risk factors (8, 25-27).We did not compare surviving and non-surviving patients after ICUdischarge but non-survivors with or without initial discharge from ICU. Non-survivors after initial ICU discharge were older, less severely injured and the severity of comorbidities as measured by the American Association of Anaesthesiologists Classi cation (ASA)(28) were higher.Interestingly, the injury pattern and the course in the ICU differed in several points: While the percentage of blunt trauma and isolated head injury did not differ relevantly, the proportion of trauma patients without head injury was much higher in non-survivors discharged from the ICU.Percentage of patients after low fall was higher in nonsurvivors discharged from ICU. Regarding the ICU treatment period one hypothesis could be that patients dying after ICU discharge underwent less intensive care as they were less severely injured.They may have been underestimated by the treating physicians and were discharged from the ICU prematurely.However, the mean length of stay on the ICU long (mean days on ICU: 4) as well as the days with ventilatory support (mean days ventilated: 5.8).Low fall is a known surrogate marker for frailty, predicting long-term mortality in older trauma patients (Ref).
In general, fatalities after the initial discharge from the ICU may be explained by either a limitation of lifesustaining intensive care treatment with palliative care outside the ICU due to a living will or by patients who were discharged from the ICU after initial recovery but suffered a late fatal complication.
The causes of mortality of these patients remained unclear.A documented living will to limit sustaining therapy was present in 60 % of patients who died after discharge from the ICU.However, we couldn't state that all of these patients actually underwent palliative care and were discharged from the ICU due to therapy limitation.A living will was also present in 46 % of non-survivors who died during their initial ICU stay.Consequently, a living will to limit life-sustaining treatment did not per se lead to an ICU discharge in this setting.The ICU stay was also rather long so that we cannot assume that there was a signi cant number of short (e.g.24 hours) ICU try-outs for the very old critically ill patients with consecutive withdrawal of intensive care treatment in those patients with failing recovery.
After initial recovery late complications like sepsis (19.5 %) and thromboembolic complications (10.4 %) resulted in death of the patient.Assumed causes of death differed to typical reasons for early mortality, i.e. bleeding, head injury and organ failure in 23.4 % of cases.We can only assume which other causes of death occurred as the registry data gives no more detailed information regarding the parameter "assumed cause of death".However it appears logical that all late deaths after ICU discharge were due to a potentially occult complication.Surprisingly, we found relevant differences between trauma centers of different care levels.About 5% of our study population was treated in a local level 3 trauma center, 24 % in a regional level 2 trauma center, and the vast majority in a supra-regional or level 1 trauma center.The percentage of non-survivors after initial ICU discharge was highest in level 3 centers and lowest in level 1 centers (Table 2).Although it seems intuitive that fewer late fatal complications occur in experienced trauma centers, previous studies focusing on the relationship between center volume and complication rates were inconclusive.Several studies found a correlation between high volume and improved outcome, especially in high-risk surgical and trauma patients (29-31) However, other studies found no association between institution or surgeon volume and survival.(32-34)Bell et al. (35) investigated the relationship between trauma center volume and in-hospital outcome and showed that higher hospital volumes were associated with decreased likelihood of mortality but not for complication or FTR.
We cannot conclude from our data that our ndings were related to differences in availability of ICU beds, other recourses or higher rates of failure to rescue (FTR).Different institutions may have different limitations in resources and, for example, comfort care in the high-dependency unit in one hospital may be equivalent to staying in the ICU for comfort care in another hospital.
Although our study identi ed a cohort of FTR patients, we could not measure and compare FTR rates.Firstly, we did not know whether patients that died during their initial stay in the ICU suffered from complications or what the in uence on mortality might have been.Secondly, we did not know complication rates of surviving patients to show the rate of survived complications.FTR rates are considered as a quality indicator superior to complication or mortality rates as they are more associated with institutional factors.(36-39)The rate was based on elective surgical populations and reclassi ed deaths not caused by adverse events.Holena et al.(36) demonstrated that this approach lacks validity in trauma because patients often die as a direct consequence of injury without any adverse events.They also argued that another common approach to simply exclude deaths without recorded adverse events reduces the reliability of the FTR rate.They proposed to include all deaths but excluding those with an expected mortality >50%.Though the limit is still under discussion and yet to set de nitely.
Moore et al (37) performed a systematic review on complication rates as a trauma care performance indicator.Only three complications (pneumonia, pulmonary embolism and deep vein thrombosis) were recommended to evaluate acute trauma care hospitals.A recent study by Chung et al. (38) showed that 33% of all major complications in trauma patients occurred on normal surgical wards.
Rauf et al. (40) observed that the time distribution of in-hospital mortality in severely injured patients shows a constant decrease.About 61 % of all deaths occur within the rst 48 h after admission.Mean time to death was about 6.5 days with a median of two days.
This might suggest to focus on the initial hours and days after trauma in terms of quality improvement.However, we believe that our study identi ed a relevant eld of potential improvement to decrease mortality rates even further.
Therefore we propose the mortality rate of patients after initial discharge from ICU as a reliable quality indicator for trauma care.Its calculation and extraction from registries is simple and independent from de nition and measurement of adverse events.

Limitations
This study has several limitations: It is a retrospective study of a registry and therefore there is potential for bias or residual confounding from factors we did not measure.We only have information on acute care hospital outcome without any follow-up data on discharged patients.Furthermore, we did not discriminate between patients that died on the normal ward on those readmitted to ICU.We also cannot give information on patients treated less than 2 days outside the ICU.Our study protocol included these patients in the group of patients that died within rst ICU course.We cannot give detailed information whether and how a living will to limit life-sustaining therapy contributed to the observed mortality rate.
The registry does not discriminate between "the patient did not want any more intensive care medicine though intensivists recommended" and "therapy was limited due to grave prognosis".Furthermore, there are patients who are discharged from ICU ("terminal discharge") due to poor prognosis, and patients who die before this terminal discharge can be planned.We included patients from Germany, as differences in health systems, i.e. limited ICU resources or cultural differences i.e. handling of living will statements might have in uenced our results.
However, the strength of this study is the large study population with detailed patient characteristics representing most severely injured patient in one representative developed country and health system.Conclusion 17.7 % of all non-surviving patients after severe injury initially treated on an ICU, have been discharged to the general ward prior to their death.These deaths can only partially be explained by a limitation of lifesustaining therapy due to a living will.They might represent an important group of patients that might have died from potentially avoidable and treatable complications.The incidence of mortality after discharge from the ICU might be a valuable quality indicator for trauma care of the severely injured patients provided that patients undergoing palliative care are excluded reliably.Further studies are required to identify patients at risk of dying after discharge from the ICU at an early stage and characteristics responsible for the adverse outcome.

List Of Abbreviations
Selection process of the study patients

Table 2
Continuous variables are presented as mean ± standard deviation; median.Abbreviations: ASA: American Society of Anastesiologists classification; ISS: Injury Severity Score; AIS: Abbreviated Injury Score; RISC II: Revised Injury Severity Score II; PRBC: packed red blood cells; Non-surviving patients after initial ICU admission depending on Level of trauma center