We care about trauma.
All of us are practising clinicians in healthcare and all of us see trauma in our clinical jobs. Some of my most memorable cases are the saves that the team has made, but usually they are the traumas that went really well, but where the outcome was not the desired one. We train for it, reflect on it, blog and tweet about it, and probably carry the after effects around with us more than we should do.
Trauma care has come a long way since an orthopaedic surgeon crashed his plane, killing his wife and critically injuring 3 of his 4 children. To say he was unimpressed by the care provided at the rural hospital – initially closed on his arrival -is probably an understatement. Hence the ATLS beast was born. The rights and the wrongs of ATLS are not for debate here, as much as we love both dogmalysis and iconoclasm. Needless to say I’ve probably not given 2 litres of warmed crystalloid to a critically injured patient in well over a decade.
Whilst ATLS is an international beast, the UK has suffered from it’s own trauma reforms. In 1988, 40 years after the NHS was formed, the Royal College of Surgeons of England proposed important reforms that speedily took off to improve trauma care. In 2000 the RCS and the British Orthopaedic Association produced Better Care for the Severely Injured, which is well worth a read to understand how we “do trauma” in the UK. This included amongst other recommendations the setting up of a National Trauma Audit and Research Network (NTARN) to “collect data from all hospital Trusts that receive severely injured patients. The development, improvement and monitoring of the standards of care for severely injured patients would thus be assured“. NCEPOD (for whom we at St Emlyn’s have a lot of time due to the excellent quality of their reports) produced the most catalytic report “Trauma, who cares?” in 2007. This has resulted in a system of regional trauma networks, with predetermined ambulance bypass to Major Trauma Centres (MTCs) and Trauma Units (TUs). There are hospitals that are neither. There are further recommendations in the report that now seem axiomatic, such as consultant-led multidisciplinary trauma teams all day, every day, increased consultant presence for trauma laparotomy, and the use of pan-scanning.
A wise man once said that the road to hell was paved with good intentions. Concentrating patients in one location may lead to better outcomes for those that arrive there, but colleagues in non-MTCs complain of feeling deskilled and underprepared to receive the occasional critically injured patient. There is also the very real effect of over-triage, unnecessary transport and, perhaps, prioritisation of the critically injured over other, as deserving cases. Part of the battle is the additional money that comes with trauma – funding of the NHS is a zero sum equation, so a bigger slice for some means a smaller slice for others.
However, an unintended benefit is the national adoption of submission of data to TARN, now shorn, but deserving of its National status, which has resulted in a huge database for interrogation. The paper under discussion is a review of the first decade of TARN submissions. The authors, and the Lancet, should be congratulated for making this paper open access (FOAM).
So what did they do?
This is an observational study that looked at the year on year submissions to TARN between 2008 and 2017. For each year they collected data on outcomes and also processes. Using their own methods they extrapolated for lost data and tested these assumptions. The clever thing about TARN is that they scrupulously collect data not only on the injury but also on comorbidities that may affect outcome (Ed – they did not always do this, but they do now and I hear that it’s going to get even better in 2019). The injury is classified by the injury severity score, about which you can read more here. The combination of ISS and comorbidities, along with demographic data and GCS can be plumbed into the TARN probability of survival calculator. With this you can spend a happy hour working out what probabilty of survival a 74 year old woman with a minor TBI and a fractured hip would have if they had metastatic cancer and high blood pressure (92%, if you’re interested).
From the entire dataset they compared data from 35 constantly submitting hospitals over the time period with the remainder who joined piecemeal over the decade or didn’t submit for some time periods. This provides a control group of sorts in that the constant submitters contained 43% MTCs (15/35) which was much higher than the proportion of all hospitals (27/169; 16%).
The salient outcome figure is that there is an Odds Ratio (OR) of survival in constant submitters of 1.19 and an OR for all hospitals of 1.21. That gives a number needed to treat (NNT) of around 5 which is something to be proud of, for the trauma team, the trusts, and the wider NHS. Certainly, the NHS Department of Self-Promotion picked up this headline figure along with reporting an “extra 1 600 lives saved” which is nowhere I can find in the original article. This was then reported in the Times and the Telegraph and I suspect that’s a trifecta that the authors were no doubt justifiably chuffed about.
So was it the right thing to do?
Undoubtedly yes. It’s a huge database of nigh on a quarter million patients. They further refined this to an ISS > 9 – postulating that more minor injuries would be unaffected by the changes – and further excluded proximal femoral and isolated pubic ramus fractures. The rights and wrongs of excluding common fractures such as these (especially those – PFF – that have an ISS of 9) is again outwith the scope of this discussion. There are problems with ISS though – it can only be calculated after all the injuries are known and it is hard to predict which patients may have an ISS >= 9 from limited information in the prehospital environment. This does lead to a degree of under- and over-triage; we see (and get excited enough to mobilise the trauma team for) minor penetrating torso trauma, whilst the elder with the PFF and minor head injury (eg contusion) gets the PGY2 alone at an hour down the line.
Huge registries will have problems with missing data; TARN is no different – there was an attempt using previously documented methods to account for these missing data and performed sensitivity analyses with and without the cases with missing data.
So what did they find?
Well, some of it isn’t very shocking. There has been an increase in the number of cases seen in MTCs; a rise of 53% to 72% with MTC as the initial destination and 73% to 82% with MTC as the final destination, which suggests that pre-hospital triage gets it right more often than not. There has been a 44% rise in CT scanning (50% to 72%) and an increase in the number of cases seen by a trauma team led by a consultant Emergency Physician (29% to 63%). We now give tranexamic acid, which we never used to do before 2007, as CRASH-2 wasn’t out until 2010!
Some of it is a validation of what we already know – our trauma patients are apparently getting older (median age increased over the decade from 45 to 59) and the percentage over 65 increased from 22% to 42%. Or are they? This might reflect an ageing population but the rate of change of population demographics cannot explain this. It’s more likely that we are now including patients into TARN who were previously excluded or missed. Whole Body CT use may also have an impact here as our threshold for WBCT is quite low in the elderly, thus we pick up more injuries that might well have been previously missed in years past. Those extra injuries found may well bring them into the TARN spotlight.
These patients are also getting sicker (or we are getting better at recording comorbidities as it changes the mortality statistics?) with 64% having a recorded Charlson comorbidity. Despite this, we are intubating fewer patients in the ED, admitting less to critical care (31-24%) and those that do get to critical care are being discharged sooner – the median length of stay has dropped by a day to 3 days.
Another thing that’s nice to have spelled out by the data is the massive increase in work that we do. There is an increase of 260% of patients with ISS>8 presenting to the constant submitter hospitals. Over the study period, England’s ED attendances increased from 19.6 million to 23.4 million, so trauma is making up more of our workload – or we’re recognising it better.
If you end up at a constant submitter hospital you are more likely to be sicker. This is based on that there were more deaths, probably reflecting higher numbers due to the concentration of services and increases in ISS.
The authors also looked at Ws scores over time. These are based on more than just the ISS (a measure of anatomical injury. Ws scores incorporate physiological derangement, age, comorbidity and anatomical injury to provide a probability of survival for individual patients that can then be combined into an overall score for a hospital that indicates whether they are getting more or fewer expected deaths. You can read more about the probability of survival model here. Ws has the advantage of adjusting for case mix, but there are problems. For example I’ve never really understood why centres of excellence often have excess death indicators on TARN data……., it doesn’t really make sense. For example here is the data from the Royal London Hospital. Do we really think they have excess deaths, or is the model struggling to reflect local circumstance, populations and prehospital care?
Let’s be cautious though. Ws may have its problems, but it is a useful tool to look at trends and in comparisons and in this study there was a significant change in Ws trajectories across the country with an overall improvement in Ws scores. That at least suggests that change happened at the point when trauma centres were created.
So what’s the problem?
At the risk of coming over all “reviewer 2”, this is neither the paper that I would write, nor the paper that needs to be written to draw the conclusions that MTCs save lives. What is up for debate is the assertion that MTCs are better, get better outcomes, and therefore deserve the extra funding that has come to MTCs following the readjustment. The statistical jiggery-pokery that TARN do isn’t detailed here and it needs to be taken on trust that the outcomes are better for MTCs. I also cannot find any data to support the assertion that 1 600 lives are saved, though that may be the case if we were able to crunch the numbers ourselves.
The premise in papers like this is that we try to compare outcomes from one year to the next. If that’s the case then we would hope that the comparison cohorts are similar. However, that really does not look to be the case here. Even taking into account the changes in submitting hospitals we look at these findings and it really makes us thinks that we are comparing apples and oranges.
- There were 5338 patients analysed in 08/09 as compared to 19,197 in 16/17. The world has not become 3-4 times more dangerous and so these are clearly very different populations.
- Average age has gone from 45-59. This is not a life expectance/demographic change. It’s because we are looking at different populations.
- Low falls have gone from 35.8% of the database to 47.6% of the database. This is clearly a very different cohort.
- In the adjusted analysis of constant submitting hospitals there is a significant change in numbers of patients seen.
So it clearly looks as though there are very significant differences between patients in the early years on the database and in those more recently analysed.
So what can we take from this?
Well, our workload that we now identify as ‘trauma’ has more than doubled, (Ed – where was it before I wonder?), and as it’s supposed to be a consultant-led and -delivered service, this does provide some ammunition to expand consultant numbers in the ED and for the other services that deal with trauma.
The other major thing to take is that the “Silver Tsunami”, a rather trite name for the increase visibility of traumatically injured elders, is very real. We’re also less good at seeing it as trauma from the off, largely for the problems with ISS detailed above. Certainly these make up a large proportion of the missed TARN +ve cases that we see (or rather don’t, with the trauma team) in my ED.
So, file under interesting, but not practice changing. And like all things FOAM, the system has changed immeasurably in the decade in question. As well as comparing apples and pears in terms of patients and locations, the system is constantly evolving and changing faster than the journal publication time.
On a final positive note we must recognise that this is probably the best possible study that could have been done with the data available. Sure, it has bias and uncertainty, but there is a signal in here. We’re not convinced that the effect size quoted in the press is really justified, but it might be. We also want to reassure colleagues, patients and the health economy that we do believe that MTCs have made a difference and that we’re not keen to go back to the pre-MTC era.