Evaluation of elderly specific pre-hospital trauma triage criteria: a systematic review

Background Pre-hospital identification of major trauma in elderly patients is key for delivery of optimal care, however triage of this group is challenging. Elderly-specific triage criteria may be valuable. This systematic review aimed to summarise the published pre-hospital elderly-specific trauma triage tools and evaluate their sensitivity and specificity and associated clinical outcomes. Methods MEDLINE and EMBASE databases were searched using predetermined criteria (PROSPERO: CRD42019140879). Two authors independently assessed search results, performed data extraction, risk of bias and quality assessments following the Grading of Recommendations, Assessment, Development and Evaluation system. Results 801 articles were screened and 11 studies met eligibility criteria, including 1,332,300 patients from exclusively USA populations. There were eight unique elderly-specific triage criteria reported. Most studies retrospectively applied criteria to trauma databases, with few reporting real-world application. The Ohio Geriatric Triage Criteria was reported in three studies. Age cut-off ranged from 55 to 70 years with ≥ 65 most frequently reported. All reported existing adult criteria with modified physiological parameters using higher thresholds for systolic blood pressure and Glasgow coma scale, although the values used varied. Three criteria added co-morbidity or anti-coagulant/anti-platelet use considerations. Modifications to anatomical or mechanism of injury factors were used by only one triage criteria. Criteria sensitivity ranged from 44 to 93%, with a median of 86.3%, whilst specificity was generally poor (median 54%). Scant real-world data showed an increase in patients meeting triage criteria, but minimal changes to patient transport destination and mortality. All studies were at risk of bias and assessed of “very low” or “low” quality. Conclusions There are several published elderly-specific pre-hospital trauma triage tools in clinical practice, all developed and employed in the USA. Consensus exists for higher thresholds for physiological parameters, however there was variability in age-cut offs, triage criteria content, and tool sensitivity and specificity. Although sensitivity was improved over corresponding ‘adult’ criteria, specificity remained poor. There is a paucity of published real-world data examining the effect on patient care and clinical outcomes of elderly-specific triage criteria. There is uncertainty over the optimal elderly triage tool and further study is required to better inform practice and improve patient outcomes. Supplementary Information The online version contains supplementary material available at 10.1186/s13049-021-00940-z.


Introduction
The proportion of elderly people experiencing major trauma is increasing as healthcare systems globally cope with an ageing population [1][2][3][4][5]. Despite advances in trauma management the elderly population has worse outcomes compared to younger adults with similar injury severity [6,7]. Fewer elderly trauma patients are initially triaged to major trauma centres (Level 1 equivalent centres), are less likely to trigger trauma team activation, and more likely to be initially assessed by junior doctors [5]. Early and accurate identification of the elderly major trauma patient is essential to guide appropriate care pathways [8,9]. Improvements in the triage of elderly trauma patients have the potential to significantly improve clinical outcomes, whilst also being resource efficient [5,8].
Current pre-hospital triage tools are ineffective in identifying elderly major trauma patients [5,[10][11][12][13]. Developing a pre-hospital triage tool with acceptable under-and over-triage rates, whilst also being useable in the pre-hospital environment, is challenging. Field triage tools use a combination of vital signs, and anatomy and mechanism of injury factors [14]. The physiological response to trauma alters with age and vital signs may not change as in younger patients [11,15,16]. Additionally, co-morbidities and concurrent medications may make presentation of major trauma less obvious in elderly patients and mask underlying instability. The traditional paradigm of major trauma triage emphasises identification of high-energy transfer mechanisms [17]. However, in elderly patients major trauma frequently occurs after lower energy transfer mechanisms [4,5]. Falls have become the greatest single cause of trauma in the USA, accounting for almost half of trauma mechanisms [1]. A fall from < 2 m height has been the most common injury mechanism for major trauma in the UK since 2008 with its proportion continually growing [4,5]. Globally, falls have been identified as a major public health concern and the WHO report that falls are the second leading cause of unintentional injury-related death [18]. These challenges result in high levels of under-triage of elderly trauma patients and the inability to quickly and effectively direct ongoing care [5,13]. Considering the mechanistic and physiological differences, major trauma triage tools specifically developed for the elderly patient population may be more efficacious. Such triage tools must apply to all pre-hospital elderly trauma patients if this high under-triage rate is to be improved [5].
There is currently no national or global consensus of how best to identify elderly patients with major traumatic injuries in the pre-hospital environment. There is a timely need to identify an effective pre-hospital elderly major trauma triage tool to optimise care delivery and improve mortality and morbidity. This systematic review aimed to summarise the published elderly-specific criteria for prehospital triage following trauma, evaluate criteria sensitivity and specificity, and examine the associated clinical outcomes.

Methods
This systematic review was prospectively registered on PROSPERO (Reference: CRD42019140879) and the PRISMA guidelines were followed in its conduct and reporting [19].

Research questions
The systematic review sought to investigate the following research questions: a) Which elderly-specific criteria are included in prehospital trauma triage tools? b) What is the reported sensitivity and specificity of elderly-specific trauma triage criteria used within the pre-hospital setting? c) What are the associations between pre-hospital elderly-specific trauma triage criteria and clinical outcomes?

Eligibility criteria
Eligibility was determined using PICO [20]. The population was elderly patients with traumatic injuries. Our initial scoping reviews revealed a wide range of age-cut-offs and therefore a conservative age cut-off of ≥ 50 years was used to ensure all relevant studies were captured. The intervention was the development or implementation of elderly-specific triage criteria. No specific comparators were predefined. The primary outcome was criteria sensitivity and specificity. Secondary outcomes were pre-defined clinical outcomes as detailed below. Original articles written in English were included. Conference abstracts and abstract-only articles where full text was not available were excluded, as were preclinical/animal studies.

Search strategy
The search strategy utilised OVID to search MEDLINE from 1946 and MEDLINE In-Process & Other Non-Indexed Citations and EMBASE from 1974. The search terms were a combination of keywords and MeSH terms: "trauma", "Wounds & Injuries", "Emergency Medical Services", "pre-hospital care" "Triage", "Geriatrics", "older adults", "elderly", "Humans". These were combined with operators OR and AND to expand and then refine the search. An example of the search is shown in Additional File 1.

Study selection
Search results were independently reviewed by two authors (AJB and DP). Titles and abstracts of all returned articles were screened for eligibility and full texts obtained for those meeting eligibility criteria. The reference lists of included studies were screened in the same manner. All duplicate studies were removed. Any discrepancies were discussed and where consensus could not be reached, this was resolved with a third author (UR).

Data extraction and analysis
Two authors (AJB and DP) independently extracted data using predetermined criteria. Study-level data included author and publication year, study design, country of origin, eligibility criteria, and description of elderly-specific triage criteria. Patient-level data included demographics (age and gender), injury severity, and mechanism of injury (MOI). The outcome measures to be extracted were sensitivity and specificity by Injury Severity Score (ISS) > 15 and ISS > 9, in-hospital mortality, length of critical care stay, length of hospital stay, time to theatre, life-saving interventions, interhospital transfers, and reported compliance with triage criteria. Pooling of studies was not appropriate due to the high methodological and clinical heterogeneity.

Assessment of risk of bias and quality of evidence
Risk of bias was assessed using the Newcastle-Ottawa scale [21]. The quality of the evidence of included studies was assessed using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) methodology [22].

Study characteristics
Study characteristics are summarised in Table 1. All studies were conducted in the USA. All studies were observational and retrospective in design utilising data from established trauma databases, except for one which used a consensus panel of experts to develop elderly triage criteria [32]. The majority of studies (n = 8) retrospectively applied triage criteria to test triage criteria efficacy [23,[25][26][27][28][29][30]33]. Two studies tested real-world application of triage criteria using a before and after design with retrospective analysis of efficacy [24,31]. Three studies utilised the Ohio Trauma Registry [24,26,33], two accessed data from the National Trauma Data Bank [23,25], and one used the National Automotive Sampling Crashworthiness System Database [31]. A further four studies [27][28][29][30] used local state or system-wide databases and with the exception of one [32], all studies were multi-centred.

Triage criteria
Eight distinct elderly triage criteria were reported, summarised in Table 3. Four studies derived triage criteria from previously collected patient records [25,28,29,33]. The most commonly reported elderly triage criteria was the Ohio Geriatric Trauma Triage Criteria [4,24,33]. This was developed by Werman et al. [33] retrospectively applied to a trauma registry by Ichwan et al. [26] and its real-world application assessed by Caterino et al. using a retrospective interrupted time series [24]. All other triage criteria were reported in individual studies only [23,25,[27][28][29][30][31]. One study developed binary regression equations of vital signs for each MOI [25].
One study reported an increase in patients meeting criteria for triage to a trauma centre (TC) after adoption of new elderly-specific triage criteria (58% vs 44%, 13-15%, p < 0.001), yet only a 1% increase in transport of patients to the TC (0.1-2.0%, p < 0.05) [24].

Clinical outcomes
Mortality was the only reported clinical outcome. Predetermined outcomes for investigation such as rates of interhospital transfer and life-saving interventions were not reported. Only one real-world study assessed the effect implementation of elderly-specific triage criteria had on mortality [24].

Risk of bias and quality assessments
There was variability in the risk of bias (Additional file 2). The highest scores were observed in two interrupted time series studies, however risk of confounding remained despite adjusted analyses [24,31]. Lack of cohort demographics reporting prevented assessment of sample representativeness [25,33]. The quality of evidence according to the GRADE system was "low" [23-29, 31, 33] or "very low" [30,32] for all included studies (Additional file 3).

Discussion
Eight unique elderly-specific criteria for pre-hospital triage following trauma are described and elderly age cut-offs varied from 55 to 70 years. Studies lacked consistency in triage criteria and physiological parameter adaptations for elderly patients. Sensitivity was varied but was improved when compared to corresponding adult triage criteria. Specificity was poor at less than 50% in over half of studies, therefore risking high rates of overtriage. In-hospital mortality was low, which may reflect the low severity of injury across the studies. There was a lack of evaluation of clinical outcomes in real-world application of elderly-specific triage criteria. This review illustrates the uncertainty in how best to identify and triage elderly trauma patients and it remains unclear which elderly trauma triage criteria is superior.
Variation in age cut-offs reflects the continued uncertainty in defining the elderly trauma population [34]. Four of the eleven studies used a cut-off of 55 years, a threshold that would be not considered elderly in many healthcare systems [34,35]. Risk of adverse outcome following trauma increases with age [34,36], therefore although a dichotomous age cut-off is convenient, it may be inappropriate. Age-based markers may support a more nuanced approach [37], however no tools utilised these. These may be useful in development of elderlyspecific triage tools and are an avenue for research [37]. Furthermore, there is an increasing recognition that frailty and sarcopenia are associated with adverse outcome following trauma [38,39]. Indeed frailty assessments appear superior to age in predicting outcome [40], however no published triage tools utilised frailty scoring. Co-morbidities appear to increase risk of adverse outcome, although current evidence is inconsistent [5,  33,34,36,41]. Nevertheless, only one tool included comorbidity assessment using a crude ≥ 2 co-morbidities [27]. Specific pre-existing conditions and their severity may adversely impact outcome and a more refined assessment could be beneficial [34,42]. This review illustrates a growing consensus for elderly-specific triage criteria to include adapted physiological parameters, most commonly by increasing SBP and GCS thresholds. This reflects the altered physiological response to injury and the increased morbidity associated with higher values than in adult patients [15]. Significant traumatic injuries frequently occur in elderly patients after low energy mechanisms [4,5], however only one tool included fall from any height as a component. There is a necessity for tools to be developed in line with the evolving knowledge of how major trauma occurs in elderly patients. Similarly, local guidance and EMS education should be informed by this developing understanding, which may also support improvements in care. The discordance found by Caterino et al. in elderly patients meeting elderly-specific triage criteria and actual transportation rates underlines this importance of continued evidence-based training and education [24]. Triage tools must be accessible to EMS teams in a fast-paced and often austere pre-hospital environment. Most elderly-specific triage criteria were modifications of existing adult field triage tools and adaptations in this way may best support tool implementation [43]. The under-triage of elderly trauma patients and associated harm is well-reported, and improving sensitivity of elderly trauma triage is a priority [5,11,13,44]. The published elderly-specific triage criteria substantially improve upon adult triage tools, although the reported sensitivity is highly variable. This comes with a reduction in specificity. The elderly trauma population is substantial and growing, therefore high rates of over-triage would place significant demand on services. Pre-hospital identification of elderly patients suffering significant traumatic injuries may be used to inform more than just transport location. Trauma units may be the most appropriate location of care and offer multiple advantages over major trauma centres for elderly patients [45], hence measuring triage in terms of transportation location may be overly simplistic. Pre-hospital identification of those at high risk of adverse outcome may direct in-hospital care such as differential levels of trauma activation, earlier imaging and senior clinician assessment, use of multidisciplinary elderly trauma teams, inform hospital care pathways, and support staff guidance and education [46,47]. This has the potential to use finite resources in a more precise and targeted manner. Studies used ISS > 15 as the gold standard to assess sensitivity and specificity.
Whilst this is the traditional definition of major trauma, this may be an imperfect and inappropriate gold standard in elderly patients [48]. ISS is unable to account for the impact of frailty, multiple injuries to one body area, and may not reliably predict the need for life-saving interventions, hence its validity as an arbitrary definition of major trauma is challenged [49,50]. Moreover, adverse outcome may occur at lower ISS in elderly patients and therefore identification of patients with lower ISS may be needed [51]. This review also intended to evaluate criteria efficacy in terms of ISS > 9, along with a range of predetermined clinical outcomes in addition to mortality, including delivery of life-saving interventions, however no study reported these outcomes. This demonstrates the need for future studies to better understand the effect of elderly trauma triage tools on a range of clinical and system-related outcomes.
The average ISS of this review's included studies reflects its included population of all elderly trauma patients. Some reports have only included patients with an ISS > 15 [5]. A broader inclusion is important to allow assessment of the effects of elderly-specific triage criteria across the whole elderly trauma population. The widely reported high under-triage rates of elderly trauma patients highlights the difficulties in identifying these patients [5,11,13,14,24,44]. A clinically implemented elderly-specific triage tool will therefore need to apply to the whole elderly trauma population in order to reduce this substantial under-triage rate.
The low mortality of included studies likely reflects the injury severity. The review reveals the scarcity of realworld outcome data concerning elderly-specific triage criteria. Caterino et al. showed a reduction in mortality after criteria adoption, however despite adjusted analysis, residual confounding is likely to remain with this study design [24]. There was minimal change in transportation rates and so improvements in outcome may reflect other changes in care rather than direct effects of triage criteria adoption [24]. Werman et al. was the only study to concurrently assess predictors of mortality and these were concordant with published literature [33,34]. There is a need to assess the effect of elderly-specific criteria on mortality and other sensitive measures of recovery within other geographical populations, in order to improve outcomes for this vulnerable, high-risk group [5].
The principal limitation of this systematic review is the risk of bias and low quality of the included studies. All studies were USA-based therefore it is unclear how translatable this data is to other systems and transportation geographies. Finally, this systematic review was limited to published data and we were unable to capture local EMS elderly-specific triage criteria not reported in the current