Blood ‘n Guts: hold the blood, save the patient?

One of the trends of recent trauma literature has been the use of “minimal volume” or “permissive hypotension” strategies in the face of bleeding.

Now there is a new study from the New England Journal that compared 2 transfusion strategies in acute upper GI bleeding. And guess what?  It looks like less is more in he upper GI bleed  as well – if you but the data.

The reference is:

Transfusion Strategies for Acute Upper Gastrointestinal Bleeding by Càndid Villanueva, M.D., Alan Colomo, M.D., Alba Bosch, M.D. et al

NEJM, Jan 3 2013


IN Summary:

  • It is an RCT of 921 patients with acute, severe upper GI bleed presenting to ED (this was done in Spain)
  • The strategies being compared for transfusion of blood were:
    • Liberal = transfusion trigger at 90 g/L, with a target of 110  vs.
    • Restrictive = transfusion trigger of 70, with a target of 90 g/L

[In Australia (at least when I was in a big hospital) the magic number for transfusion was 80g/L – which is painfully in the middle of these study groups.]

  • In he restrictive group only about 50% got any blood, compared with 85% in the “liberal” arm.
  • Not many patients bled to death in either group – mortality was largely due to the underlying disease eg. cirrhosis
  • There was a significant mortality improvement in the “restrictive” arm.
    • This was more apparent in the patients with milder underlying liver disease
    • Those with severe hepatic disease all did about the same regardless of strategy
    • There was no mention of INR or other coagulopathy parameters that I could see
    • In peptic ulcer bleeds the trend was there but did not reach significance

So how does this translate into our ED practice?  Should we use more restrictive transfusion triggers?

Can we start generalizing? Can we now say that blood products are probably a bit poisonous if infused into sick patients?

Should blood be considered an intervention we utilise only when all else has failed, or if the cardiac output is crashing?  My guess is that a lot of the upper GI bleeders are still perfusing their brains and hearts with Hbs of 70 – 90.

We covered PPIs in acute GI bleeding a while back. They also seem to be not so great in the acute, pre-endoscopy period of resuscitation.

Like trauma, the best option seems to be to get in and plug the hole, stop the bleeding – rather than transfuse or medicate until the Hb looks pretty.

Love to hear your point of view and critique of the trial above.

Hit me on the comments or join the twitter conversation @broomedocs



  1. Casey,
    Thanks for sharing this post and interesting article. Great timing for me, as I just a cirrhotic variceal bleeder on my last shift. Melena stools for a week followed by bright red blood per rectum and hematemesis for several hours. Hemoglobin (Hb) 82 (right in the “grey zone”). INR 1.8.

    I tried to access the article but alas my institution login has failed me. I did however go over the abstract and compare the findings/conclusions to recent guidelines.

    The hypotensive resus approach seems to make sense in theory. But when we narrow down to a strict number, as suggested by using Hb in this article, I think generalizing becomes dangerous. Other factors to consider in the transfusion decision would be the following:
    1) How much crystalloid did the patient get (or should the patient get)?
    2) What is the patient’s baseline Hb (ie how much has been lost)?
    3) Serial Hb measurements can indicate how brisk a bleed is, better than a single lab value.
    4) Is the patient coagulopathic?

    Certainly a provocative article. The fallout will be interesting. I gave blood and wasn’t interested in waiting to see a second value in the 60’s. Gave octreotide and ceftriaxone but held the PPI, which was subsequently given by my consultant, of course.

    Really like your blog/site, by the way.


    • Hi Elisha

      Agree – hard to pull a number out of this study and say that it is a good transfusion trigger.
      Would like to see the comparison of the patients who were transfused vs. the ~ 65% who did not get any blood at all – then compare apples to apples, rather than strategies.

      The general principle seems to be supported – blood transfusion if not required ultimately does do some harm to patients


  2. Hi Casey, great post, Even in my medical culture the threshold for Hb was 80; however I think that 70 is very fair as transfusion threshold and not so low….I read case reports about high survival rate even in patient with starting Hb of 30…..Indeed i think that restrictive transfusion triggers are the future in bleedings management (except of course for Brain Injuries)…..Going back to GI bleedings i think that more studies on this topic are needed taking into consideration the coagulation status of cirrotic patients because as you said in this study there is no mention about coagulation datas of patients…

Speak Your Mind