If you’ve read St.Emlyn’s before you will know that we adore any papers that challenge dogma and so it is that this week I came across this paper in the Journal of Trauma and Acute Care Surgery. St.Emlyn would have loved a bit of #dogmalysis.
So the dogma increasingly seems to be is that judicious prehospital fluid resuscitation is a bad thing. This paper has the title – and it’s important that we note the title here – of….
Now that’s a pretty unequivocal title. It pretty much says that trauma patients who get prehospital fluids survive more often. A title is just a summary though, so we should really look at the abstract as well.
So, the abstract again tells us that prehospital fluids are associated with a decreased mortality with a statistically significant reduction in mortality (Hazard ratio 0.84 (95% CI 0.72-0.98).
The conclusion states that IV fluids decrease in-hospital mortality. On that basis we may indeed be up for a bit of #dogmalysis here. Should we be returning to prehospital fluid boluses for trauma patients?
Let’s look a little more closely before we decide…..
[learn_more caption=”What kind of trial is this?”] This is a prospective observational trial as a part of the ‘PRospective Observational Multicenter Massive Transfusion (PROMMT) study. This is a prospective study across 10 level 1 trauma centres in the US. The focus of the trial is to look at the in-hospital management of major transfusion. A series of 14 papers are published this month as a supplement to the Journal of Trauma and Acute Care Surgery. It’s worth a read. There are 1245 patients in the study. The researchers did not direct therapies (this is observational data), but rather observed what happens. The intention of the PROMMT study is to develop future interventional studies and hypotheses to test. That’s entirely reasonable, but it does mean that there is the possibility (probability?) of bias entering into the findings.[/learn_more]
[learn_more caption=”Tell me about the patients”]OK, so there were 1245 patients and in this sub-analysis they are looking at mortality outcomes by comparing outcome between patients who did or did not receive pre-hospital fluids. 64.5 % of the patients were blunt and 35.5% were penetrating trauma. More penetrating than in the UK (even in Virchester), but a fair mix none the less.
There is a good analysis of baseline data between the two groups and they are broadly similar. SBP was slightly lower in the fluids group (100 vs 110mmHg), and those in the fluids group had a non-statistical rise in SBP of 5mmHg before EDThis is observational so the numbers of patients in the groups is not matched. In fact 1009 (84%) did receive fluids and 191 (16%) did not. This suggests that the standard operating practice in these trauma systems is to give fluids. I don’t know this, but it certainly suggests this. If that’s the case then perhaps those that didn’t were ‘different’ in some way to those that did.[/learn_more][learn_more caption=”So how much fluid was given?”] In the 84% who received fluid an median of 700ml fluid, though this varied greatly both between patients and institutions. One study site had a median of 1500ml fluid given.
I may have missed it but I cannot see a breakdown of fluid type, though the documented increase in chloride amongst fluid patients suggests that saline must have been a common choice.[/learn_more]
[learn_more caption=”So do patients who get pre-hospital fluids survive more?”]Well, if you look at the title and the abstract you would think so, but let’s look at the overall mortality figures first.
In the fluids group the overall mortality was 21%
In the non-fluids group the overall mortality was 23%
This was not statistically significant
These figures are not in the abstract.
Rather, the abstract and title appear to be based on a Cox regression analysis that adjusts for age, sex, mechanism of injury, ISS, ED, and GCS score. In that model the headline figure is a reduction in mortality as described as a hazard ratio reduction of 0.84; (95% confidence interval 0.72-0.98; p = 0.03).
I’m a bit confused at this point as a Cox Regression analysis is described as a time to event test. I am unclear why time to event is a significant outcome here. What exactly does this figure represent in this study? How do we interpret a statistical test aimed at comparing time to event? The details do not really appear to be here and a conversation with colleagues asked the question of why did they not focus on the main outcome (death) or if they really wanted to they could have used a logistic regression model (thanks Rick).
So we only see a statistically significant difference after controlling for the factors above and by performing a time to event test. Overall mortality is not statistically different and we are dealing with observational data that is likely prone to selection bias. Nor can we conclude that the overall 2% difference does not exist – this is just too small a study to determine that (likelihood of a type 2 error).
On the basis of these concerns I am not convinced that the data clearly shows a clear benefit to fluids and that the headline/abstract misses the important data around actual mortality rates.
[/learn_more][learn_more caption=”So what’s the major lesson here?”] The lesson to me is that critical appraisal is important and that if an abstract looks enticing, that’s all it is. If you want to know more you’re going to have to read the full paper.
Now, maybe we’ve missed something here. Perhaps this is a really fantastic way of looking at the data…., let us know if you think the approach is the right one.
Simon Carley MB ChB, PGDip, DipIMC (RCS Ed), FRCS (Ed)(1998), FHEA, FAcadMed, FRCEM, MPhil, MD, PhD is Creator, Webmaster, owner and Editor in Chief of the St Emlyn’s blog and podcast. He is visiting Professor at Manchester Metropolitan University and a Consultant in adult and paediatric Emergency Medicine at Manchester Foundation Trust. He is co-founder of BestBets, St.Emlyns and the MSc in emergency medicine at Manchester Metropolitan University. He is an Education Associate with the General Medical Council and is an Associate Editor for the Emergency Medicine Journal. His research interests include diagnostics, MedEd, Major incidents & Evidence based Emergency Medicine. He is verified on twitter as @EMManchester