Is higher-level trauma center care associated with better outcomes in patients injured by low-energy trauma?

Background In high-income countries trauma patients are becoming older, more likely to have comorbidities, and are being injured by low-energy mechanisms, chiefly ground-level falls. It is currently unknown whether existing trauma systems improve the outcomes of these patients. This systematic review investigates the association between higher-level trauma center care and outcomes of adult patients who were admitted to hospital due to injuries sustained following low-energy trauma. Methods A pre-registered systematic review (CRD42020211652) of subject databases and grey literature, supplemented by targeted manual searching, was conducted in January 2021. Studies were eligible if they reported outcomes in adults admitted to hospital due to low-energy trauma. Studies were excluded if participants were not adults or were not admitted to hospital. Studies in lower- and middle-income settings were excluded due to differences in demographics and healthcare systems. Risk of bias was assessed by independent reviewers using the Robins-I tool. In the presence of study heterogeneity a narrative synthesis was pre-specified. Results Three observational studies were included from 2,898 unique records. The studies' risk of bias was moderate-to-serious due to potential residual confounding and selection bias. All studies compared outcomes among adults injured by ground-level falls treated in trauma centers verified by the American College of Surgeons in the USA. The studies reported divergent results. One demonstrated improved outcomes in level 3 or 4 trauma centers (Observed: Expected Mortality 0.973, 95%CI 0.971-0.975), one demonstrated improved outcomes in level 1 trauma centers (Adjusted Odds Ratio 0.71, 95%CI 0.56-0.91), and one demonstrated no difference between level 1 or 2 and level 3 or 4 trauma center care (Adjusted Odds Ratio 0.91 (0.80-1.04). Conclusions There is currently no strong evidence for the efficacy of major trauma centers in caring for adult patients injured by a ground-level fall. Further studies at lower risk of bias and studies conducted outwith the USA are required.


Introduction
Trauma has previously been considered a disease of the young. Epidemiological studies have emphasized the loss of life and prevalence of disability in working-age people due to road traffic injuries, self-inflicted injuries and interpersonal violence. 1,2 Consequently, trauma systems have been developed in many countries to afford rapid access to resuscitation facilities and specialized care. 3 In general, trauma systems consist of a higher-level center with full facilities and specialist services; nearby lower-level centers with less comprehensive facilities; and pre-hospital services which transport injured patients to hospital based on field triage criteria. A description of trauma center levels in the USA is given in Box 1. Although the structure and function of trauma systems is heterogenous there is good evidence that they are associated with lower rates of patient mortality. [4][5][6][7][8] It is essential that trauma systems are continuously monitored, and timely, relevant information about their performance returned to the different groups that care for injured patients. 9 However, in high income countries the epidemiology and etiology of trauma is changing. The mean age of trauma patients is increasing, and so is the prevalence of comorbidities and frailty. Furthermore, high-energy mechanisms of injury such as road traffic collisions are increasingly becoming displaced by low-energy mechanisms, chiefly ground-level falls. 10 It is well-recognized that this emerging population of trauma patients is vulnerable, with increasing age and comorbidity associated with greater risk of mortality. There is evidence that existing trauma systems may not be ideally suited to care for this changing patient group.
In the prehospital environment there are high rates of undertriage, 11 and older patients are less likely to receive major trauma center care at any point in their care journey. 12,13 Current . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

Review Question
In adults requiring hospital admission for the management of injuries sustained through lowenergy trauma, is higher-level trauma center care associated with improved patient outcomes compared to care in a lower-level center?

Review Management
A detailed review protocol was written and registered with PROSPERO (CRD42020211652).
The advice of an information specialist was sought during development of the search strategy.

Information Sources and Search
Keyword and database heading searches were performed on terms related to trauma centers, low-energy trauma and falls. The searches were re-run immediately prior to the final analysis run on 12 January 20201. The complete search strategy is available in Supplemental Table 1.
The databases searched were MEDLINE, EMBASE and CINHAL, the Cochrane Database of Systematic Reviews and the Cochrane Central Register of Controlled Trials. Searches for grey literature were conducted on the websites OpenGrey and the Grey Literature Report.
The reference lists of included studies were hand-searched to identify antecedent studies.
Online citation data was reviewed to identify potentially relevant subsequent studies.
Searches were limited from the year 1987 to present because the American College of Surgeons (ACS) Verification, Review and Consultation (VRC) Program, which established a benchmark for different 'levels' of trauma center, commenced in 1987. 19 In addition, . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 4, 2021. ; https://doi.org/10.1101/2021.06.01.21258178 doi: medRxiv preprint searches were limited to English language articles due to the lack of facility for translation, and to studies in humans.

Eligibility Criteria
The population of interest was adult patients (as defined by the included studies) who were admitted to hospital due to low-energy trauma. Low-energy trauma was defined as a fall from less than 2 meters. Important synonyms included 'low fall', 'ground-level fall' and 'fall from standing height'.
Patients were excluded if they were not adults or were not admitted to hospital. Studies in lower and middle-income settings were excluded due to the differences in demographics, injury patterns and healthcare systems.
The intervention under investigation was higher-level trauma center care. Variation in national systems precludes precise definition of exactly what this care comprises. For the purposes of this review, studies' description of a facility as a 'major trauma center' was accepted, with the explicit assumption that such facilities would be on par with ACS-certified level 1 or level 2 trauma centers, or with Major Trauma Centers in the UK. 20,21 The comparison was care in a lower-level trauma center. Similar difficulties regarding precise definition of a non-MTC apply. Once more, studies' description of the facility was taken at face value, with the assumption that such facilities would be on par with ACScertified level 3 and level 4 trauma centers, or Trauma Units in the UK.
Published and unpublished randomized controlled trials and non-randomized studies were eligible for inclusion. Studies were excluded at full text review if they were at high risk of bias or severely methodologically flawed.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 4, 2021. ; https://doi.org/10.1101/2021.06.01.21258178 doi: medRxiv preprint

Outcomes
The primary outcome of interest was mortality, at the time point specified by the included study.
Secondary outcomes included morbidity, length of hospital stay, length of critical care stay, in-hospital complications and surgical operation rates.

Selection of Studies
All identified records were pooled and duplicates removed. A single reviewer (MT) screened all titles and abstracts to establish potential eligibility. Full texts of potentially eligible studies were reviewed by two independent reviewers. Discrepancies were resolved through discussion and the adjudication of the senior author (FL).

Quality Assessment
All eligible studies were evaluated for methodological quality and overlap of the study populations. If this was found to be the case, or if serious methodological flaws or reporting inconsistencies were identified, then studies were excluded. Risk of bias assessment was conducted by two independent reviewers using the Robins-I tool. 22 Any discrepancies were resolved through consensus.

Data Extraction, Synthesis and Analysis
Study-level data were extracted into a standardized data extraction form (Excel, Microsoft, Redmond) and checked by an independent reviewer. A summary of the variables extracted is in Table 1. If data fields were not presented, but could be calculated, then this was performed in Excel. All corresponding authors were contacted for additional information or clarification.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 4, 2021. ;

Patient and Public Involvement
In 2017 the Emergency Medicine Priority Setting Partnership, involving patients and careers across the UK, established that identifying which patients should receive immediate major trauma center care is the tenth most important emergency medicine research priority. 23

Study Selection
Searches identified 2,851 database records and 30 records from grey literature archives.
After removing duplicates 2,144 unique records were screened for eligibility. Of these, 48 full papers were retrieved and 6 were found to be appropriate to proceed for quality assessment. Secondary searches of these six papers identified a further 17 potentially relevant studies. These full papers were retrieved however none were eligible. Figure 1 illustrates the process of study selection in detail.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 4, 2021.

Global Quality Assessment
Global quality assessment identified three studies that were not suitable to proceed to risk of bias assessment. One study was excluded at this point because it was found to have critical methodological flaws. 24 Two further studies were excluded because they repeated findings in the same patient cohort, 25 or a subset of the same cohort, 26 as a third (included) study. 27 Full details are given in Supplemental Table 2.

Study Characteristics
The three included studies are summarized in Table 2 All studies used inpatient mortality as the primary outcome, and only Ahmed 2019 reported secondary outcomes.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Risk of Bias
Risk of bias assessment is summarized in Table 3.
Selection of participants into each study was judged to be at moderate risk of bias. This was largely due to the potential for confounding by indication, which may have occurred in the pre-hospital setting, or tertiary referral bias, which may have arisen due to the transfer of patients from lower-level centers to higher-level centers. The handling of patient transfers between centers is also of particular concern because this has been found to be associated with better patient outcomes. 30 In general, the reliance upon trauma databanks may introduce bias due to unmeasured, missing or inaccurate data. 32 Two studies were judged as being at 'serious' risk of bias due to the handling of missing data. Ahmed 2019 acknowledges that missing data could influence their propensity score matching, but provide no account of the nature, quantity or distribution of missingness. Cook 2012 makes no reference to missing data. Roubik 2017 found 1% of patients were missing data for predictor variables and tested the implications of this, finding that it made no significant difference to their results. However the data were not missing at . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 4, 2021. ; https://doi.org/10.1101/2021.06.01.21258178 doi: medRxiv preprint random (level 3 and 4 centers had a higher rate of missing data) and therefore the study was conservatively judged to be at moderate risk of bias.
Finally, all studies were at risk of reporting bias due to the lack of published pre-specified analysis plans.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 4, 2021. ; https://doi.org/10.1101/2021.06.01.21258178 doi: medRxiv preprint Notes on the descriptors of risk of bias in the Robins-i tool: Low: the study is comparable to a well-performed randomized trial with regard to this domain. Moderate: the study is sound for a non-randomized study with regard to this domain but cannot be considered comparable to a well-performed randomized trial. Serious: the study has some important problems. Critical: the study is too problematic to provide any useful evidence on the effects of intervention No information: on which to base a judgement about risk of bias for this domain. .

CC-BY 4.0 International license
It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 4, 2021.

Mortality by Trauma Centre Level
Cook 2012 and Roubik 2017 both report unadjusted mortality by trauma center level, demonstrating higher unadjusted mortality at higher-level trauma centers (Table 4). After adjustment the findings for the primary outcome of each study are divergent. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 4, 2021. ; significant findings, the differential O:E mortality observed for level 3 and 4 centers indicates factors beyond case mix may account for this difference.

Temporal Trends in Mortality
Roubik 2017 reports a decrease in the AOR for mortality in higher-level centers from 2008 (AOR 0.9, 95% CI 0.9 -1.0) to 2014 (AOR 0.7, 95% CI 0.6-0.7), using 2007 as the reference year. There was no significant temporal trend in mortality at any other trauma center level.
Cook 2012 found that the 95% confidence intervals for the AORs all crossed 1, except in the year 2001. They conclude that the adjusted odds ratio for mortality over time remained consistent.

Secondary Outcomes
Ahmed 2019 is the only study to compare secondary outcomes. The median time to patient expiration was 6 days (IQR 3-11 days, p = 0.17) and the median time to patient discharge was 5 days (IQR 4-7 days, p < 0.001) in both higher-level and lower-level care. This is comparable to the length of stay observed in Cook 2012 and Roubik 2017.
Ahmed 2019 demonstrates a statistically significant difference in the discharge disposition.
In higher-level care 21.1% of patients (n = 3,785) were discharged home without help compared to 18.3% (n = 3290) of patients in lower-level care (p<0.001).
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 4, 2021. ;

Summary of Results
This is the first study to systematically review the effect of higher-level trauma center care on outcomes in patients who have sustained low-energy trauma. The results were divergent, with the studies reporting either no difference in outcome; an advantage in higher-level care; and an advantage in lower-level care. All studies were observational and at moderate-tosevere risk of bias so therefore these results should be interpreted with caution.

Interpretation of Findings
Roubik 2017 is the largest study in this review and was judged to be at the least risk of bias. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 4, 2021. ; and the outcomes of 803 patients. It is therefore difficult to draw generalizable conclusions from this result.

Relationship of Findings to Previous Research
The lack of consistent results in these studies and the failure to demonstrate improved mortality among patients in optimally equipped centers warrants further attention.
Firstly, it is possible that there is truly no mortality benefit in higher-level care. This could be explained by the summation of patient-and injury-related factors resulting in a disease process which even optimally resourced facilities are unable to ameliorate. The best support for this hypothesis is provided by Ahmed 2019, who found no mortality benefit in a cohort closely representing the emerging population of low-energy trauma patients in terms of age, mechanism of injury and absence of physiological derangement. 27 However, Ahmed 2019's finding that 2.8% more patients were discharged home from higher-level centers implies the contrary: that this patient population may benefit from higher-level care, but those benefits are realized in functional outcomes rather than mortality.
Secondly, it is possible that current comparisons of injury severity are less accurate in the low-energy trauma population. For example, one flaw of the Injury Severity Score is that multiple injuries in one body region are under-represented in the overall score. 33 This may have a disproportionate impact in the low-energy trauma population, who suffer a high proportion of head injuries, potentially resulting in under-recognition of their injury burden. 13 An alternative definition of major trauma has been proposed based on the receipt of lifesaving interventions and consequently the need for higher-level care. 34,35 Thirdly, the timing or nature of outcome measurement in this patient population may be suboptimal. The standard outcome of 30-day mortality may fail to measure trauma-related . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 4, 2021. ; deaths in older patients: although most deaths among severely injured older trauma patients (ISS >15, age ≥ 75) occur within 20 days, 36 it is possible that many deaths in less severelyinjured patients occur later through the accumulation of complications. In addition, trauma frequently entails morbidity and impaired quality of life among survivors. 37 Patient-reported outcome measures, such as those now collected by TARN, may be at least as appropriate in assessing the efficacy of trauma centers as current standards.
Finally, the findings of this study are supportive of a more general hypothesis: that patients injured by low-energy trauma have distinct care needs, and that specialization of trauma care may not be the most appropriate solution in low-energy trauma. This hypothesis is supported by several studies which demonstrate improved outcomes when extra resources are dedicated to prompt, comprehensive assessment and holistic management of low-energy trauma patients. 16,17,38,39 Implications for Future Research This result demonstrates that the role of higher-level trauma centers in the care of patients suffering low-energy trauma requires further research. Studies at lower risk of bias and studies in non-US trauma systems and populations are required. In addition, further research is also needed into the potential of specific interventions to improve outcomes in this population.
Other areas of importance include the management of low-energy trauma patients with (1) specific injury patterns such as intracranial or thoracic injuries; (2) patients in different injury severity categories, (3) patients who require transfer between lower-and higher-level care, and (4) the examination of outcomes other than in-patient mortality, for example patient quality of life. In relation to the latter, there is also a lack of qualitative research in this field . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 4, 2021. ; exploring the values and attitudes of older trauma patients to receiving care in potentially very distant hospitals compared to their local hospital.

Strengths and Limitations
Comparing trauma centers using their ACS verification level has face validity owing to the "homogeneity of resources and expectations of clinical practice". 20,29 However, it has been demonstrated that significant differences in risk-adjusted mortality outcomes exist between similarly verified ACS trauma centres. 40 . The potential impact of intra-level variation in care is mitigated in this review by the large numbers of hospitals in two of the included studies.
A closely related issue is that the generalizability of the findings of this review to non-ACSverified trauma centers and to hospitals outside the USA is limited. International variation in other aspects of healthcare provision, notably pre-hospital medical services, as well as differences in population demographics and injury patterns are also likely to limit generalizability.
The precision of this review is reduced by the exclusion of studies of patients with specific injury patterns or grades of injury severity. For example, significant bodies of evidence exist concerning the optimum management of patients with traumatic brain injuries or a high total injury burden. [41][42][43] Articles not published in English were excluded from this review which may have introduced bias against evidence from trauma systems in some high-income countries, notably in Europe and South-East Asia. Language bias against lower-to-middle-income countries is likely to be minimal because studies in these settings were explicitly excluded in the review protocol.
Finally, no formal assessment for publication bias was undertaken.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 4, 2021. ; https://www.rcem.ac.uk/docs/Research/FINAL%20ranking%20top%2010%20v2.pdf.
24. Scheetz LJ. Injury patterns, severity and outcomes among older adults who sustained brain injury following a same level fall: a retrospective analysis. Int Emerg Nurs. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 4, 2021. ; https://doi.org/10.1101/2021.06.01.21258178 doi: medRxiv preprint