“But doctor, he’s vomiting blood!!!” – The NEJM GI Bleed article by Villanueva: Yup, time to reassess transfusion in GI bleed! @FOAMed, @FOAMcc

Screen Shot 2014-01-08 at 3.13.37 PM

Last january a highly anticipated paper came out in the NEJM (http://www.nejm.org/doi/pdf/10.1056/NEJMoa1211801), which should be a game changer, given a few provisos.  Villanueva et al reported on their large (almost 1,000 patients) randomized study on liberal (<90 mg/dl) vs restrictive (<70 mg/dl) strategy.  Interestingly but no longer surprisingly, the patients in the restrictive strategy did better.

Hmmm…sound familiar?  By now everyone  accepts the TRICC trial threshold of 70 in ICU patients, but when it first came out, there were a fair bit of disbelievers and concerned health care workers.  At the time, they excluded GI bleeds and acute coronary syndromes, understandably,

So what do the numbers say?  First lets see if there was any difference in the actual treatment. Definitely. In the restrictive group, 51% of patients required transfusion, vs 86% in the liberal group.  Sizeable difference. Now in terms of outcomes:

a. rebleeding decreased: 10% vs 16%

b. 6 week survival was better: 95% vs 91%

c. less adverse events: 40% vs 48%

But you know what, even if we throw out all the above, the absence of any difference would result in the same conclusion: there is no advantage to a higher hemoglobin goal in GI bleed.

…except for this important proviso: in their institution, UGI endoscopy was performed within the first 6 hours.  Now I’m not necessarily suggesting that the results would have been any different with a real-life, “random” timing of endoscopy, but in the spirit of scientific rigour, it is important to note.

Now there are a few important points to make that are highly relevant in the whole transfusion issue.

Why is transfusion bad? For those who may not be aware, here are a couple of reasons why transfusion looks good on paper but not necessarily so good for the patients: (a) due to stored red cell loss of deformability, the capillary bed can suffer micro-occlusion, and (b) there seems to be some immune suppression related to receiving transfusion. And I don’t (yet) have a clue what it might do to the glycocalyx…

So this is the sort of situation (not uncommon even in other clinical scenarios) where the physicians and nurses breathe a sigh of relief when seeing a nice number on a paper or a screen following transfusion, but the patient’s microcirculation is actually worse off than it was.

How valid is a hemoglobin value in GI bleed?  The answer to this is why we have MD’s after our names, and hopefully live up to that. Let me give you a quick anecdote: sometime in the past year I was called on to help a sick patient in the OR, a young woman with some post-partum bleeding who remained unstable. So the anaesthetists are concerned, the OB’s are trying to fix things, and she’s pretty unstable. A quick CUSE (critical ultrasound examination) shows she has a tiny IVC (almost virtual) and a hyperdynamic LV/RV with a near-100% EFs.  They were concerned because they had already given her fluid, etc…might she have suffered an amniotic embolism? So we were reassured that her RV was fine but clearly she needed more fluid of some sort (SBP 60’s, HR 140’s).  Just at that moment a call comes thru and the OB resident declares with a smile “its ok, her hemoglobin is 105!” and for a split second, there seems to be a sense of relief in the room, which worries me even more. I remind the resident that if he bleeds fast enough, the very last drop of his blood will have a very similar hemoglobin to the one he had while quietly sipping coffee at starbucks… So although we tend to use hemoglobin as an answer to “how much blood does my patient have,” what it really says is what the concentration of red cells is in the blood that my patient has – hence a clinical assessment (preferably with ultrasound) is required to at least determine in our minds whether our patient is empty or full, and only then can a hemoglobin value be properly interpreted. I doubt I would treat the same was someone with only 2 liters of blood going around with a hemoglobin of 75 versus someone with 4 liters of that same blood going around.  In that exact moment their hemoglobins are the same, but in all likelihood, in the ensuing hours, the first will continue to drop as the intravascular volume seeks to replete itself from the interstitial space and from possible GI intake or IV fluids.  So consideration has to be given to what fluids the patient has been or is receiving.  This is why we have MDs. If it was just about using a number, the clerk could manage our patient off the computer census and consult GI. Yep, its the N=1 concept again.

GI bleed = Trauma?   Conceptually, GI bleed is not that different from trauma.  You have a breach in vascular integrity. What are the determinants of rate of bleeding? The size of the hole and the pressure gradient.  Our body will try to shrink the hole with a fibrin mesh and a platelet plug and some vasoconstriction, and hypotension will decrease the pressure gradient.  That’s why in major trauma we don’t want to “stay and play” but want to “scoop and run” to an OR while just maintaining life until someone can plug the hole surgically (just look up some of Karim Brohi’s great #FOAMed lectures), because all that medical resuscitation (big crystalloid boluses) will do is blow those fibrin meshes away and increase the pressure gradient!  So aggressive fluid resuscitation may not be the best idea in GI bleed either. And yes, I’ve used tranexamic acid in bad GI bleeders while waiting for intervention.

Philosophical rambling… 

I think most of us forget that phrase…what was it again? Primum…primum non nocere?  Oh yes, we all said it, know it, but do we really remember to apply it to everything we do, every day? And why not? Because, in our desire to help, we more often than not feel like we have to “do something.”  And in truth, so does staff, families and the patients themselves.


So, how will this change my practice?  It will make me completely comfortable in allowing my hemodynamically stable and nearly euvolemic (IVC can be seen and LV/RV not just a tachycardic blur) GI bleeder to have a hemoglobin of 70. Will I insist on an endoscopy within 6 hours? Not necessarily, but I will be monitoring (as I usually do) for any deterioration to get on the phone and push a little or a lot.

What’s next…?   I’m thinking down the line we might just see small volume resuscitation (100 cc at a timewith a glycocalyx-friendly albumin), a bolus and infusion of tranexamic acid and a quick endoscopic intervention will be the way to go, while keeping a hemoglobin no greater than 70…cuz who knows, maybe 60 or 50 might be better…

love to hear what anyone has to say!


Check out Ken and I discussing this on the SGEMs Podcast:


9 thoughts on ““But doctor, he’s vomiting blood!!!” – The NEJM GI Bleed article by Villanueva: Yup, time to reassess transfusion in GI bleed! @FOAMed, @FOAMcc

  1. Hi Phillippe
    I wrote about this article last January here: http://broomedocs.com/2013/01/blood-n-guts-hold-the-blood-save-the-patient/
    I agree – the best option seems to be getting a scope down and making a diagnosis / intervening sooner rather than later.
    We have seen the evidence basis for PPIs showing not much benefit prior to scope for all comers – and I think this paper steers us in the same direction. Get a scope, plug / inject / stop bleeding and then they don’t need as many potentially harmful interventions as “stop gap measures” – e.g.. PPI or transfusion.

  2. Phillippe,

    As a fairly newish visitor to this place, I am totally impressed by the good work you do …. You have some really great stuff on this site.

    I thought I might add some quirky ideas to your discussion.

    We are now getting familiar with the concept of endothelial cells covered by a surface glycocalyx layer, that forms part of the barrier and mechano-sensing functions of the blood-tissue interface. We have discussed in some detail, the role of the glycocalyx in preserving endothelial integrity. I am gonna try and add a bit more spice into the whole transfusion drama.

    In recent times, we have started talking a lot about a bioactive phospholipid called sphingosine-1-phosphate (S1P), as a crucial element in preserving vascular barrier integrity by ‘protecting’ the Glycolcalyx. (Most geeky papers on TRALI and other transfusion related complications do mention it).

    Because albumin is one of the primary carriers of sphingosine-1-phosphate (S1P), it is possible that S1P, acting via S1P1 receptors, plays the primary role in stabilizing the endothelial glycocalyx. Infact, antagonism of S1P1 receptors have been shown to cause widespread shedding of the glycocalyx, as evidenced by increased serum concentrations of Heparan sulphate and Chondroitin sulphate. (This might probably be one of the mechanisms how albumin is glycocalyx friendly).

    RBC transfusions are a double edged sword…..especially in situations of acute anemia as in post hemorrhagic situations ( major GI bleed or trauma.)….I totally agree with you in that the two are conceptually very similar.

    Erythrocytes have been identified as an important buffer for sphingosine-1-phosphate . In mice, depletion of plasma S1P by genetic inactivation of S1P synthesizing enzymes (sphingosine kinases 1 and 2) elicits profound pulmonary vascular leak, which can be reversed by restoring circulating S1P via RBC transfusion.

    In humans, hematocrit (Hct) predicts plasma S1P levels. There also seems to be a dynamic equlibrium between SIP levels of the plasma, and the circulating RBCs. It has been demonstrated that in anemic individuals, plasma S1P levels are not uniformly restored by RBC transfusion. Rather, the age of the RBC unit at the time of transfusion tended to negatively correlate with the ability of RBC transfusion to replenish plasma S1P. During storage, the S1P content of human RBC markedly declines, likely due to enzymatic degradation. Because erythrocytes serve as a buffer for circulating S1P, aged RBC with low S1P content may be incapable of restoring plasma S1P levels and may actually remove S1P from plasma, which in turn could contribute to increased endothelial permeability, capillary leak, and infiltration of inflammatory cells.

    I hope this partly answers your question as to how the glycocalyx may be impacted by inappropriate and irresponsible transfusion triggers. I agree that these are all very novel ideas and as such, exist in the realm of experimental clinical physiology, but my gut tells me that the delicate Glycocalyx may hold the clue to a lot of answers to questions that have plagued us for a long long time!

    John from India…

    • Thank you so much for your kind comments and for entering the fray. I think your voice is one of the few that can help us seemingly bring the bench to the bedside, which is really key, since most clinicians (such as myself) are not really scientists and tend to wait for things to distill through – knowledge translation usually delays this a decade or so…

      Keep shining the light!

  3. Really enjoyed this blog entry. I covered this paper for BEEM conference. Can you email me to discuss something? TheSGEM@gmail.com

  4. […] Dr. Rola came to my attention because of the great blog post he wrote on the paper we are reviewing today. It was called “But doctor, he’s vomiting blood!!!”. […]

Leave a Reply to “But doctor, he’s vomiting blood!!!” – The NEJM GI Bleed article by Villanueva: Yup, time to reassess transfusion in GI bleed! @FOAMed, @FOAMcc – the SilverBird9 Cancel reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.