HEMS doesn’t work – at least not for the Dutch
We all love helicopters, don’t we We love the idea of swooping from the sky to save a patient at the moment of near death before whisking them off to the Trauma Centre for more life saving surgery, not to mention the tea and medals that will inevitably come.
Surely then it is pretty obvious that helicopters save lives, isn’t it? Or is it? The theoretical benefits of having an experienced and skilled clinician able to attend the severely injured patient almost immediately appear obvious. They can assess the patient, instigate immediate treatment and liaise with other medical staff to ensure the appropriate services are ready and waiting at the hospital: this should all combine to improve the outcome for the patients in a way that should be apparent and measurable. Well in theory it should, but to be sure we should really look at some data to see if there is any evidence for the role of helicopters in improving outcome from trauma.
So it was with this in mind that we selected a paper entitled ‘The effect of Helicopter Emergency Medical Services on trauma patient mortality in the Netherlands’, published in Injury, for this week’s Journal Club at St.Emlyn’s. You will probably know that Virchester has an ambulance helicopter but that it is relatively new to the scene and has only recently started carrying doctors. The city is therefore on a bit of a learning curve on how it will impact on trauma care.
Now, to this interesting paper asking a pretty relevant question: ‘Does a helicopter emergency medical service (HEMS) improve patient outcome in trauma?’
In this retrospective study from the Netherlands, the authors looked at patients who had suffered trauma (Ed – good that’s the group of patients we are interested in), and to see how the helicopter affected mortality the authors matched patients who had been seen by the HEMS with patients who had not been seen by HEMS. Patients were matched according to ISS, age and the presence or absence of severe traumatic brain injury.
It is important to note that the HEMS service in the Netherlands is different to that provided in other countries. The service is all about bringing the hospital (or at least a surgeon or an anaesthetist) to the critically ill patient rather than bringing the patient to the hospital. In other countries with different geographic conditions the priority is often about transporting the patient by helicopter in order to reduce the pre-hospital time. In this study, the majority of patients who were treated by the HEMS team were still transported to hospital by road ambulance. So this is about moving the doctors as much as it is about moving the patient. An interesting model and quite different to what we try and do in the UK.
So what did they find?
The authors looked at the Day 1 mortality rates and the total in-patient mortality rates for each of the two groups of patients. They also sub-divided them into traumatic brain injury (TBI) and non-TBI patients. They found a slight increase in in-hospital mortality in the HEMS group for patients with TBI and a slight reduction in mortality for non-TBI patients. Neither of the results were significant with very large confidence intervals each way.
Mortality is a pretty hard outcome so we are not overly concerned about the lack of blinding.
They did find that patients where HEMS were involved spent significantly longer at scene, took longer to get to hospital and were much more likely to be intubated. These findings are not surprising but the question remains as to whether this was beneficial or detrimental to the patient outcome.
This is the next issue, were the two populations comparable at baseline?
But matching for ISS should sort this shouldn’t it?
Using ISS is one way of matching patients in a study of this kind as we could argue that having patients with similar levels of anatomical injury is comparable……or is it? As resuscitationists, we know that the anatomical injury is but one part of the problem. It is the effect on the victims physiology that holds our focus in the resus room. However, the authors did not take this into account in the matching process.
One scoring system that may reflect how physiologically sick the patients were, is the revised trauma score (RTS). This includes physiological parameters to show any clinical decompensation in response to the injury. This was recorded but unfortunately the authors were not able to match the patients for this due to ‘practical reasons’. It may be that it was not possible to match this due to the limited size of the patient population available. This was adjusted for in the multivariate logistic regression but was significantly different at baseline in the two groups.
One stage further would have been to combine physiological, age and anatomical data into a survival prediction tool. TRISS is probably one of those that you are most familiar with, but do follow this link to the TARN site for an online survival calculator for trauma based on European data. Something like this could have been used to match patients against probability of survival rather than just anatomical injury.
So what do we know?
There was no attempt at a sample size calculation but given the non-significant results one suspects that a larger study may have been more conclusive if there is truly a difference to find. This study suggests that if HEMS does make a difference it is not massive enough to be reliable demonstrated with the numbers here (though to be honest the effect would have to be very large with this size of cohort). The geeky ones in Journal Club spotted some unusual methods in the statistics (like using t-tests for comparing RTS data – almost certainly not the right test), but our fundamental concerns were really about the overall method and patient selection rather than the analysis. No amount of clever (or not) analysis can compensate for fundamental methodological flaws.
RCTs may be difficult to achieve though as randomising patients to not get the very expensive, all singing, all dancing shiny helicopter will always be tricky and the ethical questions were highlighted in the online journal club discussions.
So what about other methods? One model would be a before and after study such as this one from Scandinavia published this year.
An alternative approach would be to look at very large databases with access to patient data and outcomes as this article in JAMA did last year.
Both of these articles suggest a survival benefit although one could still criticise the methodology for not being of a perfect design.
Perhaps then we need to think a little more broadly on this subject and consider whether we are really looking at helicopters or whether we are looking at trauma systems. You can’t have one without the other after all, but the focus of many studies revolves around the aircraft and not the system as a whole. This is perhaps not the place to consider the risks/benefits of HEMS, rather it’s a really good paper to stop and think carefully about what we are reading before coming to any firm decision on whether we can use this data for the benefit of our own patients.
Sadly the absence of really well executed trials means that HEMS is seen by some as an expensive service to provide with very little evidence of benefit despite the obvious theoretical ‘fact’ that it should work.
As EBM practitioners, we need to be careful how we interpret the data and methods used within the description of the trauma systems being analysed. This is a good example of how a retrospective, small study with inherent biases may not quite answer the question asked.