Morgenstern, J. Liberal or restrictive transfusion in brain injury (The HEMOTION trial), First10EM, October 7, 2024. Available at:
https://doi.org/10.51684/FIRS.137642
How much blood people need in their body is an ongoing question. Physicians have moved beyond the days of removing blood from their patients, but we still aren’t totally sure when we are supposed to put blood back in. Most studies suggest we should transfuse less, but there are specific populations who, at least theoretically, could benefit from more oxygen carrying capacity. The MINT trial compared liberal and restrictive transfusion practices in the setting of MI, and may or may not have shown benefit of more liberal transfusion, depending on how you read the paper. The HEMOTION study asks the same question in the setting of traumatic brain injury, and is probably just as difficult to interpret.
The paper
The HEMOTION trial: Turgeon AF, Fergusson DA, Clayton L, et al. Liberal or Restrictive Transfusion Strategy in Patients with Traumatic Brain Injury. N Engl J Med. 2024 Jun 13. doi: 10.1056/NEJMoa2404360. Epub ahead of print. PMID: 38869931 NCT03260478
The Methods
The Hemoglobin Transfusion Threshold in Traumatic Brain Injury Optimization (HEMOTION) trial is a pragmatic, multicenter, open-label, blind end-point, randomized trial.
Patients
Adult trauma patients with an acute moderate or severe traumatic brain injury (defined as a GCS score between 3 and 12) and anemia (hemoglobin less than 100 g/L). They excluded patients with objections or contraindications to transfusion.
Intervention
Liberal transfusion (triggered by a hemoglobin less than 100 g/L).
Comparison
Restrictive transfusion (triggered by a hemoglobin less than 70 g/L).
Outcome
The primary outcome was functional outcome, measured with the Glasgow Outcome Scale–Extended (GOS-E). They used a different cutoff on this scale for different patients, depending on their initial prognosis.
The Results
They do not include a flow diagram to explain the exclusions, but out of 6188 patients screened, 742 were randomized in the trial. The majority (73%) were men, with a mean age of 49 years. Most of the injuries were severe, with a median GCS of 4. The hemoglobin at enrollment was 91 g/L in both groups.
Clinicians seemed to follow the protocol. 99% of the liberal group received a blood transfusion, as compared to only 38% of the conservative group, and the median hemoglobin was about 20 g/L different throughout most of the ICU stay.
There was not a statistical difference (but there might have been a clinically important difference) in the primary outcome of sliding-dichotomy analysis of the GOS-E score at 6 months, with 68.4% of the liberal group and 73.5% of the conservative group having an unfavorable outcome (ARR 5.4%, 95% CI -2.9 to 13.7%).
Mortality was identical (26.8% vs 26.3%) at 6 months.
The adverse events all look about the same, although ARDS does look about 2% higher in the liberal transfusion group.
My thoughts
HEMOTION is an important trial that is unfortunately difficult to interpret. The scientific interpretation is easy. This was a negative trial, but with a clear indication that a much larger follow up study is needed, to see if there is anything to this potential 5% improvement in outcomes. The clinical interpretation is more difficult. What exactly are we supposed to do while we wait for that large follow-up study?
To some extent, the issues with this trial were obvious before it started. I have previously ranted about trials setting sample sizes based on practical rather than scientific calculations. Here, they set their trial size based on the assumption that they would see a 10% absolute improvement here. That is an absurdly large benefit, that is simply unheard of in critical care trials. Anyone reviewing this protocol before the trial started would have predicted a negative trial, because almost nothing results in a 10% absolute improvement in neurologic outcomes. Instead, we are left with a negative trial, despite a very impressive 5.4% absolute improvement in outcomes. Perhaps if this trial had been appropriately sized, we would have had groundbreaking results.
Some will argue that these results are convincing enough, and that 5.4% is a big enough benefit, that we should use the liberal transfusion threshold. That is a reasonable argument. Trials are not positive or negative. There is nothing special about the arbitrary statistical thresholds we choose in medicine. We have to practice based on the best available evidence, and there is a strong hint of benefit with limited harm here, so why not transfuse literally while waiting on more data?
Indeed, one might argue that liberal transfusion is the current standard or usual care. If this trial had been set up as a non-inferiority trial, it almost certainly would have failed to demonstrate that conservative therapy was non-inferior to liberal. (This highlights how many assumptions go into our trials, and why large replications are necessary to get us closer to the truth.)
However, statistics are not our only problem. We also have to consider bias. This was an open label trial using a subjective primary outcome, which is the combination most prone to bias. (Outcome assessors were blinded, but if patients or their families knew the group assignment, bias is still easily introduced.) Not only is this statistically negative, but it is also data at high risk of being biased.
When you add that to the fact that we know that we have a replication problem in medicine, and we know that trials have a tendency to overestimate benefits while underestimating harms, there are plenty of reasons to be cautious here.
This trial also showed benefits from a restrictive transfusion strategy. They clearly used less blood, and we know there are harms associated with blood. There was also probably less ARDS in the conservative group.
Ultimately, I don’t think the clinical implications are as clear as many will make them out to be. I think we are going to have to use clinical judgement, carefully considering the potential harms of transfusion, as well as indicators of benefit. (Is this patient acutely or chronically anemic? Are there signs of hypoperfusion?) I think it is fair to err on the side of liberal transfusion, but I don’t think we should make it the standard of care.
Two other minor points:
There were some baseline imbalances between the two groups which could be important, such as 6% more hypotension and 6% more patients with a GCS motor score of 1 in the restrictive group. This is a large enough trial that randomization should balance the groups, but obviously still small enough not to be able to demonstrate a 5% absolute benefit, and so these baseline imbalances could still be important in a relatively small trial.
I don’t know that it affects my overall interpretation of this study, but I have never seen a study adjust the primary outcome threshold based on your initial prognosis. I am not sure that defining a bad outcome as a good outcome just because you started sicker makes a ton of sense. There is a threshold on this scale where patients want to end up, and so the most patient oriented outcome is probably to use the same functional outcome for everyone. Pain studies do something like this, so that a change from 10/10 pain to 7/10 pain can be seen as equivalent to the change from 4/10 to 1/10. However, this feels different, because when we are talking about functional outcomes, I am not sure the person is going to be all that happy to be at 7/10 pain, no matter where they started.
Bottom line
HEMOTION is a large multicenter RCT comparing liberal and restrictive transfusion triggers in patients with traumatic brain injury was statistically negative, but may actually support more liberal transfusion.
Other FOAMed
The MINT trial: liberal or restrictive transfusion in MI
Evidence based medicine is easy
Evidence based medicine resources
References
Carson JL, Brooks MM, Hébert PC, et al; MINT Investigators. Restrictive or Liberal Transfusion Strategy in Myocardial Infarction and Anemia. N Engl J Med. 2023 Nov 11. doi: 10.1056/NEJMoa2307983. Epub ahead of print. PMID: 37952133
HEMOTION: Turgeon AF, Fergusson DA, Clayton L, Patton MP, Neveu X, Walsh TS, Docherty A, Malbouisson LM, Pili-Floury S, English SW, Zarychanski R, Moore L, Bonaventure PL, Laroche V, Verret M, Scales DC, Adhikari NKJ, Greenbaum J, Kramer A, Rey VG, Ball I, Khwaja K, Wise M, Harvey D, Lamontagne F, Chabanne R, Algird A, Krueper S, Pottecher J, Zeiler F, Rhodes J, Rigamonti A, Burns KEA, Marshall J, Griesdale DE, Sisconetto LS, Kutsogiannis DJ, Roger C, Green R, Boyd JG, Wright J, Charbonney E, Nair P, Astles T, Sy E, Hébert PC, Chassé M, Gomez A, Ramsay T, Taljaard M, Fox-Robichaud A, Tinmouth A, St-Onge M, Costerousse O, Lauzier F; HEMOTION Trial Investigators on behalf of the Canadian Critical Care Trials Group, the Canadian Perioperative Anesthesia Clinical Trials Group, and the Canadian Traumatic Brain Injury Research Consortium. Liberal or Restrictive Transfusion Strategy in Patients with Traumatic Brain Injury. N Engl J Med. 2024 Jun 13. doi: 10.1056/NEJMoa2404360. Epub ahead of print. PMID: 38869931
