Simple modification of trauma mechanism alarm criteria published for the TraumaNetwork DGU^® may significantly improve overtriage – a cross sectional study

With regard to actual over- and undertriage, verifying the use of these ERTTAC based on a definition of ISS > 15 for severe trauma resulted in 76% overtriage and 20% undertriage at our institution. Both rates are higher than recommended by the American College of Surgeons (ACS). In 2014 the ACS revised its 2006 recommendations from 30 to 50% for overtriage and < 5–10% for undertriage to < 25–35% and < 5%, respectively [12]. The over- and undertriage rates found in this study are well within the range of data published from other level I and II trauma centres in the US and Europe, with reported overtriage rates varying between 23% [15] and 91% [16], and undertriage rates between 1% [17] and 32% [18].

For this evaluation, we utilized the most commonly used definition of the ACS for the severely injured with an ISS > 15 to identify who should be treated at a trauma centre [19–21]. Some US-American studies recommend criteria additional to an ISS > 15, e.g. the urgent need for intervention or ICU treatment, to determine trauma centre need [22, 23]. Apart from the exact definition of trauma centre need, there is an ongoing debate as to whether or not all the identified patients automatically require ER trauma team activation. Most studies do not report details on the latter [19, 20, 24], whereby the official 2017 Swiss Trauma Board recommendations combine the indication for trauma centre with subsequent ER trauma team treatment.

With regard to theoretical over- and undertriage, only 15 of 326 patients with an ISS > 15 did not meet the TraumaNetwork DGU^® ERTTAC used in this investigation, yielding a theoretical undertriage rate of 4.6% if the criteria were applied properly in all cases. Such a rate would correspond well with the recent recommendations of the ACS [12]. Of the 80 cases not treated in the ER, even though their ISS was > 15, only 8 (10%) did not fulfil at least one ERTTA criterion. The theoretical undertriage for this group would have been 2.0%. In contrast, the theoretical overtriage rate of 87% at our institution would have been higher than found in reality. A high overtriage rate in daily practice often results from the criterion “gut feeling” [21] or a less accurate evaluation by paramedics compared to anaesthesiologists [25]. A study from France [24] demonstrated improved overtriage by strictly following triage protocols. Fitzharris et al. [26] demonstrated a 74% adherence rate to hospital protocols for triage for about 58,000 urban patients transported to trauma centres in New South Wales, Australia, and an even better adherence (85%) for cases where all ERTTAC were fulfilled. Overall, it seems difficult to uncover the reasons for single cases of non-adherence, which appear numerous and may be related to preclinical statements and/or to first clinical registration. Fitzharris et al. showed that although the triage protocol offered clear instructions, paramedics interpreted injury severity and treatment requirements differently [26]. Additionally, adherence was reported lower for female patients, older patients and less trained paramedics. In paediatric patients Escobar et al. [27] lowered the undertriage rate from 15% to 5% by applying a standardized procedure during patient registration to check whether ERTTAC were fulfilled. Empowering nurses to initiate trauma team activation was a crucial step to diminish undertriage. At our hospital an ED nurse on duty was responsible for ERTTA according to the protocol. Given our preliminary findings, the first step towards achieving the theoretically possible, low undertriage rate is to further increase adherence to the ERTTAC list. If other centres confirm our pilot results and manage to strictly implement these ERTTAC as published for the TraumaNetwork DGU^® in daily routine, they may expect sufficiently low undertriage rates in relation to ACS recommendations.

With regard to the adjustment of the GCS-criterion, to improve overtriage rates we were interested in how far simple adaptations of ERTTAC might lower the overtriage rate without relevantly compromising undertriage. First of all, we hypothesized a major effect due to modifying the GCS from < 14 to < 9, given the better evidence for the latter in the literature as a predictor of severe trauma [3, 28]. Nationwide studies from the US demonstrated that about 40% of undertriage cases for major trauma detection (ISS > 15) were TBIs [29]. Unexpectedly, the specially designed simulation tests on our data revealed only a slight improvement in overtriage expressed as a decrease of 2.2% (n = 28). Concurrently, undertriage increased by 2.5% (n = 8). Newgard et al. found GCS to be the best predictor for an ISS > 16, independent of patient age [30, 31]. With regard to the ACS triage criteria the required triage protocol sensitivity of 95% was not met by comparative testing of varying GCS ERTTAC. In contrast, Norwood et al. [28] strengthened the GCS ERTTAC from< 15 to < 13 and ultimately reported a deterioration of results for both over- and undertriage. The attempt to improve such over- and undertriage rates combined with simultaneous simplification of GCS application by using only its motor component [32] resulted in a mere 1.7% decrease in overtriage with a concurrent increase in undertriage of 1.1%. Given our results in the context of the current literature we saw no sufficient argument as to why we should change our ERTTAC with regard to the GCS.

With regard to the adjustment of the trauma mechanism-criterion, simulation testing revealed a simple way to reduce the relatively high overtriage of about 80% in our setting by modifying only one parameter: Excluding the trauma mechanism criterion for all car and truck passengers was shown to decrease the overtriage rate by 24% (n = 257) without relevantly increasing undertriage (2.5%; n = 8). Several authors identified the problem of high overtriage as a result of relatively non-specific trauma mechanism criteria and reported overtriage rates over 90% if these criteria were used in isolation [16, 33]. In 2006 the ACS restricted their injury mechanism criteria, for example, by excluding the deformity criterion or adapting speed limits. Implementing these stricter criteria in a study including data from three level I trauma centres, Lerner et al. reported on a reduction of overtriage from 34% to 23% with a simultaneous increase of undertriage from 3.3% to 4% [15]. Uleberg et al. found that high energy trauma per se did not justify trauma team activation in asymptomatic patients [33]. Totally omitting injury criteria from the ERTTAC list was shown to result in a decrease of overtriage by 22–35% with a simultaneous increase of undertriage by 1–3% [17, 34]. Lerner et al. reported that ERTTAC such as death of another passenger in the same vehicle, height of fall and pre-hospital rescue time best predicted the need for trauma centre resources [16]. A fall from a height of over 5 m correlated with more severe trauma and resulted in higher undertriage if not used as an ERTTA criterion [20]. In our cohort this was already true for a fall from a height of over 3 m. Davidson et al. found vehicle criteria to correlate with an ISS > 15 mostly in elderly patients > 55 years of age [35]. Brown et al. concluded that even though physiological ERTTAC generally appear to be predictive of an ISS > 15 and anatomic criteria of the need for an operative intervention, the combination of both will result in an undertriage of up to 50% if not restricted by additional injury mechanism criteria [36]. So far there is no consensus in the literature on the use of mechanism criteria, especially regarding traffic injuries. Most authors conclude that if these are used alone, they have little predictive value in determining the need for trauma centre care [17, 34]. An adaptation of these ERTTAC to the progress of security techniques by the automotive industry appears overdue and given our finding that for almost every fourth trauma patient unnecessary ERTTA could be avoided by a simple restriction of injury mechanism criteria, we would opt for a corresponding adaptation of the ERTTAC published for the TraumaNetwork DGU^®. Future improvement might be achieved by the use of advanced automatic crash notification [37] or more complicated specific computer algorithms [38] as indicated by recent studies.