An optimally designed trauma system requires a patient-centred approach to care. The patient needs a prehospital response that delivers them the most appropriate level of care as quickly as possible after their injury with rapid case identification, tasking of the clinical team with the most appropriate skills that will reach the patient most quickly and a clear focus on both undertaking time critical interventions and moving the patient quickly through their treatment path. The HIRT system appears to display advantages in all these areas.
Patients treated by the HIRT HEMS model had significantly longer total prehospital times than patients treated by paramedics, but significantly shorter times from ED arrival to CT scan. The overall effect was non-significantly shorter times from injury to CT scan in the HIRT HEMS group. The longer prehospital time in the HIRT HEMS group was however associated with a 41% greater prehospital intubation rate. This suggests that well-rehearsed and efficient interventions carried out by a highly trained physician and paramedic team enables earlier critical care treatment of severely injured patients during the prehospital phase. Importantly this can be achieved without increasing the time between injury and definitive diagnosis via CT scan which allows planning for urgent surgical intervention if required.
It has been postulated that medical intervention by doctors on scene results in delay to hospital admission and hence delay to definitive care. Definitions of the point at which definitive care is achieved have varied in previous studies [9]. For patients with a head injury not necessitating surgical drainage of a haematoma, control of secondary factors that may exacerbate the primary injury constitutes definitive care. Advance intervention teams can often achieve this in the prehospital phase. For those that do require surgical intervention time of CT indicates the point at which the head injury patient has been resuscitated and imaged, enabling planning for surgery. A previous retrospective analysis also suggested that despite longer scene times the total resuscitation time is not prolonged when physicians perform prehospital intubation [10] and that prehospital interventions reduce the primary survey time in the emergency department [11]. In our regression analysis prehospital intubation by a physician team was also not independently associated with longer scene times. Time spent in the emergency department prior to CT varied with physician model however.
As the NSWA paramedic protocols exclude the use of sedative and neuromuscular blocking drugs the difference in intubation rates between the paramedic team and both physician teams is an expected finding. However the time intervals for the GSA HEMS were significantly longer in almost every interval examined when compared with both the paramedic and HIRT HEMS models and GSA HEMS treatment was associated with a 13 minute longer scene time independent of injury severity, interventions performed or entrapment when compared with HIRT HEMS in a regression model. The difference between the physician teams was not expected as both the skills of the physicians and the rates of intubation (the most common major intervention) were similar. A recent study [12] comparing physician staffed HEMS in Germany and the Netherlands found significantly different overall prehospital times between physician staffed HEMS models in the two countries and counter intuitively the faster service had the higher prehospital intervention rate. Many complex differences in prehospital strategy between the countries were demonstrated but the implications remained unclear. Our study however has the advantage of being performed in a single trauma system with the physician teams being dispatched in parallel to the same patient in many cases.
We attempted to remove the potential confounding effects of entrapment and differing GCS scores by analysing subgroups of non-trapped and intubated patients and by conducting a regression analysis but large differences in scene time between physician groups were still observed. The seniority of the physicians in the GSA HEMS also did not appear to explain the observed difference as times were similar for specialists and registrars. It is possible that differences in policies and/or processes resulted in the observed differences but further studies would be required to examine this.
Time from ED arrival to CT scan for the HIRT HEMS patients is significantly faster than both the paramedic and GSA HEMS groups. That the HIRT HEMS was faster than the paramedic group is probably due to the lower number of interventions required in the ED. Again however the difference between physician models is unexpected as the ED intervention rates are similar and after ED arrival all patients are treated under the policies and procedures of the receiving trauma centre rather than the prehospital team. Differences in handover practices could perhaps have caused the observed difference but again this finding requires further investigation.
There are significant differences in the scope of operations between the HIRT and GSA HEMS teams. The HIRT HEMS only conducted prehospital responses, responses that did not require hoisting operations, and in an area less than 60 nm of their base. The GSA HEMS however conduct both interfacility and prehospital transports as well as hoisting operations from anywhere within NSW. There is evidence from European studies that mobilisation time for mountain rescue missions is shorter in dedicated HEMS services when compared with operations that include military and police duties [4]. Although the GSA HEMS is dedicated to HEMS operations it is likely that the degree of specialisation of the HIRT HEMS model simplified procedures resulting in shorter mobilisation times. The operator that provided both the HIRT helicopter and physician team (CareFlight) previously provided a multirole HEMS operation in Sydney with mobilisation times that were similar to those reported here for the GSA HEMS teams. This combined with the previous European data suggest that role specialisation is an important factor in lowering mobilisation times. Other operational contributing factors may have been:
-
Base location: the GSA HEMS operations base is located at a busy general aviation airport whereas the HIRT HEMS base is stand alone and outside of controlled air space. This may partially explain the differences in observed mobilisation times for responses by helicopter.
-
Road responses: GSA HEMS responses by road vehicle showed significantly longer transit times to the scene compared with helicopter responses when the distance from the base was controlled for. Sydney is a geographically large urban area with problematic road transportation infrastructure. Rapid response by helicopter via the HIRT HEMS model appears to produce faster access times than road response in such an environment, even over short distances. This is supported by the arrival of the HIRT HEMS at the patient prior to the Sydney based GSA HEMS team in all but one occasion of parallel tasking, despite the GSA HEMS responding over short distances by road to locations to which they were closer than the HIRT HEMS team.
The pool of physicians and paramedics is much larger in the GSA-HEMS system compared with HIRT HEMS. GSA HEMS had more than 45 physicians and 20 paramedics resulting in over 900 possible physician-paramedic pairs, compared with 12 physicians and 2 paramedics with 24 possible pairs in HIRT HEMS. Small clinical staff pool sizes also characterise many European HEMS such as in Norway [13]. High levels of team familiarity have been associated with significantly faster theatre turn-around times in orthopaedic surgery [14] and lower surgical and teamwork errors in cardiac surgery [15]. A possible correlation between prehospital team member familiarity and scene times would require further exploration. Whether the specialisation of the HIRT HEMS contributed to shorter scene times and ED arrival to CT scan transit times would also require evaluation with further studies.
The HIRT and RLTC dispatch systems have been compared previously regarding accuracy in identifying severely injured children and the subsequent effect on the prehospital paediatric trauma system in Sydney [6]. In addition to being more accurate, the current study also suggests that the HIRT case identification system is faster than the RLTC with HIRT HEMS having identified the case and become airborne in a shorter time than it took the RLTC to notify GSA HEMS. This difference in dispatch systems was a major contributor to the observed difference between the two HEMS systems in the time to depart the base from FKS.
At cessation of randomisation into the HIRT in March 2011 NSWA withdrew access by the HIRT HEMS team to the CAD screens. Tasking of both HEMS since that time has been exclusively by the RTLC utilising dispatch policies introduced by NSWA without consultation with HIRT HEMS. The current NSWA tasking policy directs that the closest available physician team should be dispatched regardless of differences in operational capability between the services. Additionally a motorway that runs between the HEMS bases is used as a guide by the RLTC to allocate cases. The motorway is 1.5 nm from the HIRT HEMS base and 5.5 nm from the GSA HEMS base. This study indicates that review of both the case identification process and team allocation is warranted.
Limitations
The most obvious limitation of this study is that it is an exploratory analysis of data from a randomised controlled trial that was designed to answer a different question. There is evidence for example, of case selection in the patients identified by the RLTC for GSA HEMS response who were younger, and more likely to be a trapped victim of a transportation incident than patients who were treated by either road paramedics or HIRT HEMS. Patients treated by HIRT HEMS however were more severely injured as measured by the NISS and GCS scores. Attempts were made to control for these factors by analysing subgroups which removed some of the confounding variables such as entrapment and requirement for intubation, and by performing a regression analysis.
Despite these limitations the cases of parallel tasking of the two physician response models to the same cases provides a direct comparison of the total time to patient access for the two end-to-end case identification, mobilisation and response systems, with the proviso that the distance from the respective operations base to the scene is controlled for. The fact that in most parallel tasking cases the GSA HEMS was closer but did not arrive first at the patient suggests that the observed differences are not due to selection as both services were responding to the same patient simultaneously.
The estimates for the additional scene time required to perform thoracic decompression or apply a pelvic compression device by physician teams are unexpectedly long. Absolute numbers of patients requiring these interventions were very small however resulting in wide confidence intervals. A recent study from the German national trauma registry demonstrated an additional 3.2 mins was required on scene by physician teams to establish a chest tube in a sample of more than eleven hundred patients requiring the procedure [16].
As shown by the HIRT study [5], physician staffed retrieval teams are now an established component of standard care in the Sydney prehospital system. The opportunity to answer the key hypothesis posed by the HIRT study in NSW has therefore been lost and attention should now focus on the best means to deliver physician care to critically injured patients. This study suggests that case identification systems in Sydney need to be reviewed, and that times for response and treatment between different physician response models may not be equivalent and should be taken into account during tasking decisions.