Evaluation of a university hospital trauma team activation protocol

Multidisciplinary trauma teams reduce mortality and have become an important part of modern trauma care [1]. Protocols for trauma team activation (TTA) are mainly based on prehospital information and aim at ensuring that the severely injured receive multidisciplinary care immediately upon admission, while limiting the waste of resources caused by excessive team mobilizations. TTA guidelines are widely implemented throughout Scandinavia [2]. Although they vary somewhat, the TTAs of most Scandinavian trauma centers comply with the recommendations of the American College of Surgery - Committee On Trauma (ACS-COT) [3‐8] and rely on parameters of physiologic compromise, anatomic damage, and mechanism of injury (MOI). A substantial overtriage (activation of the trauma team despite minor or moderate injury) is common and may reach 70%, mostly reflecting the limited precision of criteria relating to MOI [5‐7]. Overtriage is mainly a resource problem, as assembly of the multidisciplinary trauma team diverts personnel from other important activities in the hospital. Undertriage delays diagnosis and treatment of severely injured patients, and may compromise clinical outcome and increase trauma mortality [8]. ACS-COT suggests that an overtriage as high as 50% is acceptable if necessary to minimize undertriage [3].

Triage criteria should be adapted to the local case-load and injury pattern, which may vary considerably between geographical regions. The predictive properties of triage criteria depend on the prevalence and spectrum of severe injuries. Typical for most Scandinavian hospitals receiving trauma patients is a predominance of blunt over penetrating injuries [9, 10]. Furthermore, the frequency of severe polytrauma admissions is low [11]. The present study was initiated as we frequently observed TTA for patients with an Injury Severity Score (ISS) <15 and without a need for emergency procedures to stabilize airway, respiration or circulation. The aim was to establish the predictive properties of our TTA protocol and its individual criteria in an effort to improve the protocol's precision.

The precision of TTA criteria is important both for ensuring adequate therapeutic steps for the severely injured as well as for hospital resource utilization. Any TTA protocol should be validated to fit the local case load and trauma pattern.

The present study reveals that the patient records' documentation of the basis for TTA should have been better. Measures to improve this are necessary for continuous system surveillance. In addition, not every criterion is evaluated for each patient; emphasis is apparently laid on vital functions and mechanism of injury. An extent of injury-criterion requires an extensive clinical examination by the prehospital personnel. The first report to the EMDC is often given before this has been accomplished. For this reason, performance parameters in the extent-of-injury-group must be considered with caution.

Ideally, the criteria applied for activating the trauma team should be recorded prospectively [16]. Instead, our study may be biased by some extent of under-reporting. However, to include all available information at the time of TTA, the trauma team's admission note in the patient record was added to the EMDC data for completeness. We believe that we thus have been able to reveal practically all clinical data known to the EMDC prior to admission.

Before the UNN TTA protocol was made mandatory in 2004, overtriage was 58% and undertriage was 50% [9]. At that time, TTA was decided by the trauma leader's assessment alone, based on available prehospital information from the EMDC and a recommended, though not mandatory, set of criteria. The present protocol has successfully reduced the undertriage, but at the cost of an increased overtriage.

We report that MOI TTA criteria have a lower predictive value than those based on extent of injury or physiological compromise. This is consistent with the results from earlier studies [5‐7]. If our findings were to elicit a change to our clinical decision rules, a prospective validation of the revised criteria would be preferable. The local trauma pattern may differ considerably between services, and this affects the applicability of revised TTA criteria in other patient populations than the one they were derived in [17]. However, we believe the data we present may be used for a revision of the TTA criteria in our own university hospital setting.

Undertriage mainly affected patients transferred from local hospitals and patients with head injuries (Table 6). In our hospital the same TTA criteria apply for interhospital transfers as for patients admitted directly from the scene of injury. Thus, transferred patients are included in our analysis. We have previously shown that many trauma patients transferred in the acute phase are not adequately diagnosed and stabilized [18]. Minimization of undertriage requires decision-makers to strictly comply with our TTA protocol in all trauma patients.

There have been various approaches to minimize overtriage. In two Scandinavian level I Trauma centers (Aarhus hospital, Aarhus, Denmark and St. Olav's University Hospital, Trondheim, Norway) MOI was shown to be the set of criteria with lowest performance, which is in line with our findings. Particularly the vehicle speed criterion was shown to be inaccurate [5, 7].

Also in a study from Viborg, Denmark, the MOI criteria was shown to have low positive predictive value. In a revised version of the Viborg TTA protocol, fulfillment of a single MOI criterion would not alone lead to TTA [6]. Based on the results of the present study, a complete removal of the MOI criteria could have left up to 14 (8.7%) out of 161 patients with ISS>15 without TTA. Almost half of these patients also required an emergency procedure. However, criterion nr 23 (Motorcycle accident), 24 (Considerable deformation of vehicle compartment) and 25 (Traffic accident with speed>60 km/h) have limited capacity to identify the seriously injured in this material. When one of these is the only criterion fulfilled, only five (3.1%) severely injured patients out of 161 with ISS > 15 and four (2.8%) out of 142 in need of emergency procedures would be identified. Criterion nr 7 (Convulsions) was only used once in a patients not severely injured. All severely injured patients fulfilling criterion nr 8 (Dilated or not responding pupils) were also identified by criterion nr 6 (Lowered level of consciousness (GCS <13) more than 5 min). For reasons stated, we suggest removing criteria nr 7, 8, 23, 24 and 25.

Criterion nr 10 (Unstable fracture of the pelvis) could be misleading. The pelvis of a patient with a suspected pelvic fracture should not be strained outside the hospital, as this may provoke bleeding. The stability of a fractured pelvis can also be difficult to assess clinically [19], and also for this reason the term "unstable" is of limited value for triage. We therefore suggest to merge pelvic fracture with criterion nr 11 (Fracture in two or more long bones), to maintain the possibility of identifying patients with potentially severely bleeding pelvic fractures.

On the basis of the presented results, we propose a revised TTA protocol (Table 7). If applied to the material studied here, the number of patients admitted with TTA would decrease by 94(25%), of whom five would have ISS >15 and four received an emergency procedure. Accordingly, overtriage would decrease from 71% to 62% with either ISS >15 or emergency procedure as the reference standard. We believe that checking every criterion on all patients, including those transferred from another hospital and those with head injuries, would compensate for the potential increase in undertriage after removal of three MOI criteria. We also believe that the revised protocol will increase the focus on physiologic and anatomical criteria, and decrease the focus on MOI criteria, which also might contribute to improve triage.

Our findings are consistent with the results from similar Scandinavian studies. We advocate a more limited use of MOI criteria in our hospital, and suggest that those criteria with the lowest predictive value and highest contribution to overtriage are removed. Given these modifications, we believe that the revised protocol will reduce overtriage without any substantial increase in undertriage. The revised protocol is implemented in our hospital.