CRASH2, MATTERS, MATTERS2, non-trauma bleeding, thromboembolism. What are we to do with TXA? Read on to see where the literature is up to.
When the CRASH 2 Trial came out a few years back I was impressed but skeptical. On the positive side it was a massive trial, without any big-pharma powerhouse driving it, looking at a very cheap and simple intervention for a common problem and finding a seemingly significant benefit. However, I wondered how it could be that a drug that stopped you from bleeding didn’t seem to cause you to clot; in fact it seemed to cause less clotting AND less bleeding. That was seeming a lot like a herbal tonic or some sort of homeopathic jiggery-pokery. Moreover, it seemed questionable whether a trial conducted largely in India, Columbia and other countries where rapidly responsive and highly integrated trauma systems like we have in Victoria (or Maryland or London) just don’t exist could be really applicable to the management of trauma patients where I work.
Now, we are a regional trauma service situated about 150km from the big city so we see a bit of trauma but not heaps. And it’s Australia: people don’t routinely shoot each other here, they tend to wear seat belts, the roads aren’t particularly crowded. So we tossed around the TXA question amongst our group, had a peep at what they were doing down the road, wrote a blog post about it. and went back to managing ED flow and basic EM education and arguing about stroke thrombolysis and all the other things that emergency physicians busy themselves with.
Then a couple of weeks back someone pointed out the MATTERS trial to me (only 2 years post publication- what was that about rapid knowledge translation?). It was still sitting in my pile of things to read when I walked into resus check on a complex blunt trauma patients one of my registrars was managing and there, hanging from the IV hook, was a 100mL bag labeled “Tranexamic Acid, 1000mg”. Now this particular registrar is a part-time soldier so it seemed kind of fitting that the paper I had been meaning to read was a battlefield study. Since then I’ve trawled the literature since Crash2 in an attempt to provide a synthesis of where things are at with this intervention and here is the result.
The MATTERS trial, or Military Application of Tranexamic Acid in Trauma Emergency Resuscitation study, was published in 2011 in Archives of Surgery (now JAMA Surg). It was a retrospective analysis of 896 patients treated over 2 years at a single battlefield surgical facility in Afganistan. The were a mixture of NATO troops and locals and represented a wide variety of injury types. What they had in common was that they were really smashed up. All needed to receive a unit of bood in the first 24 hours to be enrolled; about a third had a massive transfusion and half got an acute operation. This is in comparison to CRASH2 where enrollment occured when thte treating doctor thought about giving a transfusion, which seems to equate to “if you’re sending a group and hold, randomise the patient”. The groups were well balanced for NATO troops and locals and for blood component ratios. The TXA group were more severely injured on the basis of ISS and RTS scores, got more blood products and had more extremity injuries. (As a civiian doc I read “extremity injury” and think broken ankles and dislocated elbows and think that that represents a head start for TXA; then I think of all the legless and armless war veterans around the world and it occurs to me that “extremity injury” means different things in different places). The benefit associated with TXA was bigger than in CRASH2 with a 6% ARR in in hospital mortality, blowing out to a 14% ARR in the massive transfusion subset. There was a small excess of thromboembolic disease in the TXA group but htis may well have been a function of a greater injury burden in this group or a survival effect (with more patients living long enough to get a clot).
So what is there to be cautious about in interpreting MATTERS? Well, like all retrospective registry reviews it is subject to various biases. For the first year of the data collection administration of TXA was on the basis of clinician prefernece and for the second year, as part of a protocolised system. The people treating the patients and collecting the data clearly had decided that TXA was a valuable intervention already and this may have introduced unconscious bias. The patients were incredibly complex and high severity trauma cases in comparison to the trauma casemix of a lot of civilian settings.
A really important methodological point is that there were three primary outcome measures. That means that a positive result in any one of them is only one third as significant as a positive result where there is only one primary outcome measure because you are getting three throws of the dice. If all three were positive that would be fine but not all three were. The trial looked at 24 hour, 48 hour and “in hospital” mortality which was 30 day mortality really and found no differnce at 24 hours and a growing difference from 48hrs to 30 days. While this raises interesting pharacological questions about the potiential role of TXA as a modulator of the interface between the clotting and inflammatory pathways it is hard to understand in relation to a drug that is supposed to work by stopping haemmorhage. There were a number of patients lost to follow up also and others lost at 48 hours and found again by 30 days. These were rtoughly spread across the TXA and no-TXA groups but if we assumed the worst (all the lost TXA cases did badly and all the lost no-TXA cases did well) then the results are more than reversed.
These objections aside, the MATTERS trial does sit very nicely with CRASH2 as a next puzzle peice. CRASH2 seemed to suggest that getting the drug in early made the biggest difference and that it made the biggest differnece in the sickest patients. These patients certainly got treated quickly and efficiently and they were nothing if not sick. Essensitally, the hypotheses that were supported by the CRASH2 trial received further support from the findings of MATTERS.
Since CRASH2 and MATTERS one might have expeceted a flood of further research on this promising topic but there is little to be found. The British NHS has swallowed the Kool-Aid altogether and made TXA part of standard care for trauma patients. Articles have been written in the medical and lay press implying a moral imperative to use it and groups of cardiganed British doctors have recorded youtube music videos on the topic (this little animation is cute too and a bit less cringeworthy).
An excellent review article in Shock early this year and another from the Journal of Trauma and Acute Care Surgery summarise the evidence from CRASH2, MATTERS and the non-trauma literature on TXA. More importantly they lay out the current knowledge gaps and priorities for further research and the second one suggests a rational approach to use of TXA in trauma It points out that experience in hip and knee replacements is largely positive with lower trandfusion requirements and no increase in thromboembolic disease (an anaesthetic registrar told me when I was an intern that I must always send a group and hold for orthopedic surgical cases becasue the only form of haemostasis known to orthopedics was the sucker). The cardiac surgical literature reports an increased freqency of post-op seizures with TXA and the applicability of this to trauma is unclear.
Finally, MATTERS2 was published in JAMASurg in March 2013, again retrospective battlefield work, looking at TXA with or without cryoprecipitate and reporting that TXA and Cryo were about equal in benefit and were almost 100% additive to one another in benefit. EMCrit has had a look at it in all in this podcast. Expect to see more on this sort of thing.
So who am I going to use it in and what am I going to agitate to have included in the next revision of our hospital policy on major haemorrhage?
- First of all, if it is going to be given it must be given early. If you are able to give it within an hour of injury you probably have the most potential to do good, on the basis of the information we have now. If 3 hours have elapsed since injury, don’t bother.
- Next, save it for the sickest patients. It does not appear that there is a lot to be gained by making TXA part of the routine management of all trauma patients in an industrialised nation with an established trauma system.
- Consider the kind of bleeding you are trying to deal with. A hole in a vessel that you can put your finger on does not need TXA, it needs a finger. If the trauma is focal and surgically correctable the value of TXA seems to me to be questionable, based on mechanism of action. On the other hand if you are watching blood drip off the sides of the trauma trolley and onto the floor then TXA might be just the thing, Here’s were the link with the severe extremity injuries in MATTERS comes in. Some messy extremity injuries can just bleed and bleed. Civillian cases that come to mind from over the years are a 10 year old who slipped off his father’s lap on the ride on mower and had his lower limbs chopped to bids with multiple lacerations and multiple fractures; or the pillion passenger on a lane-splitting motorcycle who’s knee clipped a parked car and arrived with the rest of the leg in an esky being carried by a pale faced policeman while the rest of him arrived on ambulance trolley. The mushy pelvic fracture that needs to be transferred to the city from a regional trauma service for interventional radiology and who is going to ooze and ooze in the meantime is another good candidate.
And finally, if you are giving it but not as part of an ongoing trial, it will be important to report seizures and thromboembolic complications to your adverse drug reaction system, as data on this remains incomplete.
Please let me know what you are doing with TXA where you are.