Podcast 112 – A Response to the Marik Sepsis Fluids Lecture


Last week I posted a lecture an incredible by Paul Marik on Fluid Management in severe sepsis. The lecture is the equivalent of a bucket of ice water poured over your head. Now let’s give you a towel and discuss.

Want to add a journal club to this flipped classroom?

Then read these pieces in Critical Care:

The Low-Fluid Volume Early Pressor Experiment has Already Been Tried

It was called standard care 15-20 years ago–patients did not do all that well.

Should we Increase DO2?

We know from that shooting for supranormal DO2 is actually harmful. The original goal-directed therapy trials did not pan out and this may be why.

There is definitely a group of Severe Sepsis patients that are receiving inadequate oxygen delivery. I have treated these patients; I have documented ScvO2s in the 50’s and 60’s on initial check in a EGDT-type algorithm.

Far more commonly, patients are in pure vasodilatory shock (Jones’ trial patients). The latter fact doesn’t disprove the former. Using the studies demonstrating the deleterious effects of shooting for above normal delivery doesn’t say anything about normalizing patients with low delivery.

Marik’s Goals

  • Achieve adequate perfusion pressure
  • Improve microcirculatory flow
  • Limit Tissue Edema

All right on point; how do we get there is the question.

Rivers Trial as a Waterfall?

This is not actually what that trial showed. And the way CVP was used was not actually debunked by the 7 mares article (Note: I’m not advocating you use CVP, I’m just pointing this out! We have better ways to accomplish assessing fluid responsiveness so CVP should be sent to the junk bin)

I will make the utterly blasphemous statement that Dr. Rivers and Dr. Marik are actually in near-complete agreement if you followed both of their protocols explicitly in the ED.

MAP and Association to Survival

Not sure what this is proving: both groups (the flow-optimizers and the desert-inducers) believe in shooting for MAP goals. Patients in whom it is impossible to get the MAP up will die more frequently.

But is it Flow or Pressure that Matters?

I must say, I am still in the tissue flow camp

On to the Glycocalyx

Now this is where stuff gets really interesting. Every day, there is increasing research and more publications on the fundamental role of the vascular glycocalyx. But how do we integrate this clinically?

Chris Nickson tweeted this amazing lecture from Rob Wise. It will explain the Glycocalyx in 5 minutes. It lives on Life in the Fast Lane.


Now as to its relation to fluids in sepsis, we are being told hypervolemia is bad. But if the fluids are leaking from the vascular space are we ever seeing hypervolemia in the vasculature or is the problem whole body volume overload–not if we believe BNP/ANP are the root of the problem, they only respond to vasculature fluids. If we are doing a fluid responsiveness strategy, we should not be seeing the vascular overload.

Do balanced vs. unbalanced fluids have a different effect on the glycocalyx in Sepsis? I have no answers to any of these questions.

ANP/BNP damage the Glycocalyx

But are we causing Myocardial Wall Stress to generate these peptides. Well here is where things get interesting. If you really start thinking this through, CVP EGDT style would make more sense–its use solely to measure RA/RV overdistension rather than fluid responsiveness. Not advocating this in any way, just something to think about.

And is a majority of the BNP coming from fluid resuscitation or from sepsis-induced cardiomyopathy? Is the latter why higher BNP levels are associated with worse outcomes?

Do Crystalloids Stick Around?

I’m not sure, but the Bark Study quoted was a study in septic rats. (23318490). This study, if it has any practice-changing value, would be to make one consider albumin. Clinically this lacks all face validity. Start doing IVC exams on your patients, fill until there is no resp collapse. Come back in a few hours, they will have stayed fuller than when you started by a significant amount. Some of this fluid is sticking around.

Let’s Talk about Marshmallow People and the Inflammatory State

I’m not sure anything we do makes a damn bit of difference when it comes to patients holding on to water if you are using a reasoned fluid responsiveness strategy, but I am open to new data telling me I am wrong.

Excess Fluid Increases Mortality

This is when the lecture really diverges from clean cause/effect progression. Associations are being taken for causation, so let’s spend a bit of time on these:

I’m going to ignore the experimental trials and move to the clinical.

  • Fluid Balance doesn’t equal fluid administration
  • Fluid balance associations certainly do not equate to fluid administration causation (SOAP, etc.)
  • CVP associations with fluid administration–Is Dr. Marik, the ultimate CVP debunker, actually equating these two?

Optimal survival occured with a 3-liter positive fluid balance at the 12 hour mark?


Dr. Marik’s statements on this one need to be looked at carefully. The FEAST trial was an amazing accomplishment with results contradictory to what we all expected. Giving severely ill kiddies fluid increased mortality. But what doesn’t get discussed at all is this little article:

subgroup analysis (BMC Medicine 2013, 11:68 ) of why pts actually died revealed CV collapse as opposed to the expected volume overload.

Norepi to tense the Tank

  • Critical Care 2007, 11(Suppl 2):P37
  • Crit Care Med 2011;39(4):689-94
  • Critical Care 2010, 14:R142
  • Crit Care Med. 2012;40(12):3146-3153
  • Crit Care Med. 2013 Jan;41(1):143-50

Then go read this editorial to get some perspective on the issues involved. (23269148)

Marik-Phillip Curves

Would indicate that fluid accumulation in the lungs occurs only on the flat portion of the Starling Curve

Should we be using Fluid Responsiveness on all Severe Sepsis Patients?

I think we should. I think we should stop blind fluid loading; it will take a while for us to get there in many EDs.

My modification of Marik’s Algorithm:

EMCrit Fluids in Sepsis Modification


Where are the Endpoints and Is Lactate Clearance is Flawed?

I have already addressed this in a wee when Marik’s article was published. See that link if you want all the nitty-gritty details, but here is the essence. Anyone who has talked about lactate clearance for the past decade has already known and perpetrated that the lactate elevation in severe sepsis is not solely from tissue hypoperfusion. We know this. This fact in no way changes that elevated lactates mirror the inflammatory response, and strategies that cause it to clear (probably to near-normal, rather than just 10%) are associated with improvement and decreased mortality.

What about Microcirculatory Endpoints?

Anyone have any good ones? I want one, I want one…

Sepsis Mortality goes Down if you Care!

Paul actually sent me this study. (24201173) Control arm mortality has dropped consistently since 1991. If you care and do a modicum of resuscitation, I think you are seeing a vast majority of the benefit. Our Sepsis Collab data has shown the same thing.

Read the Comments of the Marik Post

Some amazing stuff buried in the comment section including an inotropy first pathway and the USCOM Device

My Thanks

to Chris Nickson, Cliff Reid, and Minh Le Cong for pre-publication peer review and many helpful comments

Now on to the Podcast…


You finished the 'cast,
Now get CME credit

Already an EMCrit CME Subscriber?
Click Here to Get CME Credit for the Episode

Not a subcriber yet? Why the heck not?
By subscribing, you can...

  • Get CME hours
  • Support the show
  • Write it off on your taxes or get reimbursed by your department

Sign Up Today!


Subscribe Now

If you enjoyed this post, you will almost certainly enjoy our others. Subscribe to our email list to keep informed on all of the ED Critical Care goodness.

This Post was by , MD, published 1 year ago. We never spam; we hate spammers! Spammers probably work for the Joint Commission.


  1. Craig Rosebrock says

    I am so eager to listen to the podcast….gonna be my treat later. However one thing I want to throw out that is on topic. Methylene blue. I would love to hear some thoughts on please. Have you guys here ever used methylene blue as a pressor in sepsis?

    • Erik Kistler says

      Methylene blue putatively scavenges nitric oxide, which is overproduced by the iNOS isoform (as opposed to eNOS which is inhibited in sepsis). The iNOS form is probably most active in the larger vessels. Thus giving methylene blue should increase your MAP. This may make you feel better but you are not increasing microvascular perfusion and studies tend to bear this out. In contrast, in true vasoplegia like after on-pump CABG, where there is generalized vasodilation but apparently functioning microcirculatory mechanisms, methylene blue will not impair tissue perfusion and works quite nicely. But then one could also use high(er) dose pressors like norepi. Take your pick – the blue/green urine is a bonus :)

      • Craig Rosebrock says


        Interesting…but my understanding of the literature is that their are some contradictory findings regarding tissue perfusion and role of enos and inos. Mont studies seem to be in knock out mice and rodent models. I suspect you are correct that it is increasing Bp due to large vessel affects and maybe so what, however it seems less toxic than our true pressors we use. Thanks for the reply…it is really a treat to talk bout this stuff.

        • Erik Kistler says

          Sorry to get off topic a little – the nitric oxide story has been the stuff dreams are made of – can’t see it, difficult to measure, yet it is seemingly everywhere and does everything. Back in the early 90s there was a lot of hope that this was “the” mediator for sepsis, but even in pre-clinical trials the data was equivocal and non-specific inhibitors led to higher mortality. The thought was that iNOS was ‘bad’ and eNOS was ‘good’ and if somehow we could switch off the ‘bad’ and switch on the ‘good’ things would be ok. Which never really panned out. NO does a lot of stuff but whether it’s ‘good’ or ‘bad’ depends in part on the tissue studied, the condition, the temporal aspect of the condition, concentration, etc. Lots of good reviews on the topic in sepsis if you’re interested. There still may be a clinical role some place for its modulation (besides inhaled NO for PHTN) in sepsis.

  2. Mike Jasumback says

    This is why FOAMed works! The Hegelian Dialectic at it’s best.

    I think the most profound portion’s of each of your discussions is the need for measuring our success in volume resuscitation. Dr. Marik expounds on giving fluid ounce by ounce and giving enough but not an ounce more, and you comment on the fact that our management of sepsis pts has resulted in better outcomes, largely through close monitoring and CARE of the patient. We cannot lose sight of that fact. Caring for the patient may be as powerful as what we give and how we give it.

    So my final thought is all of this discussion is basically to reiterate that what we are doing is not throwing out EGDT, but changing how we view and manage the Goals!

    Thanks again for an excellent summary.

    Mike J.

  3. Kevin Klassen says

    Hi Scott
    You and Paul both agree a supranormal DO2 is bad.So how do we really know when enough fluid is enough, and when someone has reached a cardiac output that is giving normal levels of oxygen delivery? Is it when someone’s no longer fluid responsive (based off whatever measurement we want to use), or is it possibly well before that? I’m not sure if Paul’s flowsheet targeting a MAP>65 is answering that, and maybe targeting fluid responsiveness isn’t either. Does that come down to us finding better measurements of microcirculation, or can we use Scvo2, or something else?

    • says

      DO2 can be discovered either through surrogates or directly calculated. If you don’t have a cardiac output monitor, then you need to use surrogates:
      Is Hb>7
      Is SpO2>90%
      Is HR reasonable
      Then give a reasonable shot towards optimizing preload
      Then look at the heart on echo, is it pumping reasonably?
      Do you see a decent pulse ox wave form?
      Have the pt’s extremities stayed warm?

      • Sean Marshall says

        So in your view, how important is DO2 in critical care and how do we use it? Intuitively oxygenating the blood and delivering it to the tissues is one of our prime goals. Supranormal is bad, very low is very bad… If minimally invasive CO monitoring catches on,DO2 can be trended like a vital sign in sick ED/ICU pts. Given the limitation that DO2 takes supply but not demand into account, is this perhaps our macro-perfusion resus target?

      • Kevin Klassen says

        So we have ways to measure DO2, but what I think we’re missing is a way to measure when we have obtained *adequate* DO2 where further fluid boluses are only going to lead to supranormal DO2 levels (where DO2:VO2 is >>1). What I wonder, and maybe I’m wrong on this (let me know), is whether that ratio is possibly met far before patients stop being fluid responsive.

        • says

          Yes that is the million dollar ?. When their MAP is low it is an easier call. Once the MAP is good, much tougher. I don’t have that answer.

        • Brendan Smith says

          I think there is no doubt that we can achieve more than adequate levels of DO2 long before we reach the limit of fluid responsiveness provided that somebody has reasonable cardiac function. We are all sitting here reading this nowhere near our limit of fluid responsiveness but with more than adequate DO2. The problem comes in the patient with low inotropy where they show minimal response to fluid challenge, exactly as Starling’s law predicts. In this situation it is quite possible to more than exceed their limit of fluid responsiveness without providing an adequate DO2.
          In most cases of adult sepsis however we find that their DO2 is way above normal even when we first see them, which is typically after they have had a litre or two fluid in the emergency Department or the Ward. A DO2I of 450 ml/m2/min is enough for most people, but septic patients frequently have levels more than double this. The very high cardiac output that results from very low vascular impedance usually guarantees the DO2 is not a problem. It only seems to become a problem after they’ve had too much fluid which is compromising their cardiac and the pulmonary function.
          Now if we take a central venous blood sample, which is close enough to a true mixed venous sample for our purposes, we can easily calculate the oxygen content. If we have measured the cardiac output, then we can calculate VO2 very easily. Typically this is about 150-175ml/m2/min, so the ratio of about 3:1 is adequate. Most of us sit at around 4:1 in normal health.

  4. Jeff says

    Scott, great podcast. I am not a physician but a pre-hospital provider (Paramedic). I was very interested in the part of the podcast discussing Norepinephrine and its use is Septic shock. When I was going through my training in the mid 1980’s this vasopresser was thought poorly of for use in any kind of shock. It is now being widely used in most hospitals. What has changed?

    • says

      Jeff, the levophed leaves them dead paradigm was based on the fact that norepi was only given as a hail mary when the pt’s fate had been sealed. The evidence has lead to many of us to stop using dopamine.

  5. Jeff says

    Any future for Isolyte as a resuscitative fluid in sepsis? pH 7.4 and electrolytes balanced to more normal physiologic levels seems appealing when compared to NS.

  6. Craig Rosebrock says

    Wow the podcast was awesome, thanks.

    Few things….

    Please correct me if my logic seems off or makes no biological sense.. !

    We have spent a good amount of space talking about fluid administration, monitors to estimate adequate hemodynamics, and mortality markers.

    My brain seems to tell me the following. I think this way and this is how I see the whole thing. Being an organic chemist prior to medical school, I am used to the physical inorganic chemist (Jedi vs darth’s … Had to throw in the Star Wars humor) folks giving me a hard time because us organic folks get into weird somewhat vague explanations of the world….but we still make good chemicals. My point is this. We have made evidence based practice the gold standard…evidence based data are population biased by nature. I have yet to find an icu patient I can randomize. Apache scores may be the same but the pathophysiology is different. You can’t find 1000 left upper lobe lobectomy patients with liver failure, S/P bowel resection for ischemic gut, in the setting of an MI three weeks ago on plavix bleeding with a sat of 78 percent who happens to be septic still and oh yes has a pacemaker for afib. What you can find is 1000 patients 50-80 years old with “septic shock” getting admitted or transferred to the icu. My point is that all the invasive and noninvasive stuff goes out the window when people are really sick. This is my clinical opinion, but I have only been out of fellowship for 5 years. I have always found some frustration when using the invasive things. The lidco for example is wonderful unless you are in a fib. Nobody can really know what swan numbers mean unless maybe you are worried about tamponade. That being said, I have seen swan dx tamponade when the heart was small on film and echo was taking too long to get. At the end of the day, I get lost as to what gadget is really going to change the way I practice. I have been to 17 location in five years as a locums….I love it and I find the clinical variance is quite high. The patients still do pretty well regardless of what most of my colleagues do. The choice or pressors, how much fluid, using cvp, using lidco….all of it is variable. I have noticed that folks just keep giving fluids to see what happens. Pressors get added by pulm/crit folks a little later than the anesthesia/cc or the ed docs do. I see a lot of early dopamine or levo in the er after a liter of saline. LR vs saline battle still rage between surgery and medicine. All of this to say there seems to really be no followed standard of care out there. Folks know the rivers stuff and some of the other articles, but do non academic folks follow it. My observation over five years is no! However patient maps come up, they start to pee, lactates come down, they get better. Is this because we just happen to follow markers that get better but really mean nothing? I really don’t know the answer to that.

    Here is my major point which is what I would like you guys to chime in on. I recall learning in medical school that most of our fluid was usually not in the vasculature. If that is the case then does it not make sense that we would mobilize that fluid to the vessels in time of perceived low filling pressure ie low returning volume. I think the most important concept is CO and the stroke volume component in sepsis. If we believe that patients get vasodilated plus or minus leaky in the caps., then we must argue that preload drops. No preload, no good. Yes this is a simple point but seems important to me. If we become hypotensive from vasodilation, then seems to me the body would off set this by allowing fluid from the extra vascular space to re-enter the vascular space. I tend to think of what would the cave man do if he was septic and was away from a water source….seems that this might have happened a lot. The strong likely figured a way though it. (Evolved to increase preload somehow —- since we know tone is not going to magically get better — if that was the case then we would not need all the pressors we throw into the mix). What simple evolutionary thing allows those kids in Africa we just read about to do well with no fluid. Hmmm.

    So I tend to think that when I hammer folks with fluid that I am somehow filling some space other than vascular initially. Once that bucket is full, then I start to get some vessel retention of fluid….boom now I start to see map increases. Yes this is kindergarten thinking academically and my prof. Emeritus back home will hunt me down for such a layman way to think. However the reason this seems clinically plausible is that we really do not know how dry somebody is…we really do not know how much fluid is where. It is important to experiment to advance understanding but we sometimes tend to over rate markers for and against mortality benefit. Just read the journals….we basically question everything every few years. Steroids yes, steroids no, psa good to measure, psa bad to measure, no need to screen this person, oops we should screen this person….it is really hard to keep up with things i knew and then had to forget just to be reminded to know them again.

    For me each patient, as for most of you too, is a unique challenge. The bedside assessment, good clinical judgement, some labs, and eagerness to do the right thing are what keep me on track.

    Sorry to take up so much space, but would love some feedback.

    • says

      great comment my friend. Yes, at the end of the day your clinical judgement is prob. just as good as all of this other crap. One of the pathologies of severe sepsis is they lose the ability to mobilize back into the vasculature a bunch of the fluid lost into the interstitium. At least that was always my understanding, Paul will probably tell me I’m wrong.

      • Erik Kistler says

        Back in the 70s there was a lot of interest in mechanisms of fluid transport in the body and there was the idea that the ‘shocked’ patient (depending on etiology) was fluid down – including the cellular compartment, which can swell and shrink. The idea, conceptually, which I still adhere to, was that one would try to resuscitate the cellular component of the body with crystalloid first (and yes I use NS. Or LR. or whatever’s closest) and once replete the idea was to shift to a colloid to avoid flooding the extravascular (or 3d space) space because it made little sense to try and use oncotic forces if the extra vascular milieu was not yet resuscitated. Perhaps a little simplistic but I find it useful

      • John Hinds says

        I’ve often wondered why ICU docs don’t pursue this aspect more vigorously.

        We are getting good at aggressively resuscitating, but a lot of units are happy to just potter along in the days and weeks post sepsis without pursuing aggressive DE-resuscitation.
        Why not?
        Aiming for deresusciation endpoints is at least as key as the “sexy” bit at the start with all the lines, tubes and pressers.

        We’ve been doing this for years in our unit, and our length of stay, mortality and ARDS incidence outcomes are great as a result.

        Aiming for a negative balance at day 3, tight electrolyte control (using CVC hydrochloric acid, and CVC free water, for example), rigorous anitmicrobial stewardship and deescalation strategy is just as key to the patients transit.

        Sepsis resuscitation is referred to as the “Ebb and Flow” of fluids. I think we’re keen to pour them in at times, but overly keen to sit back and let the patients wallow in them needlessly in the aftermath.

        It wouldn’t be uncommon to see repatriated patients to our unit have a 10-15L negative balance within the first three days of their stay


          • Erik Kistler says

            You are correct, Scott :) I agree with the above gentleman. As our immediate mortality from sepsis has declined (due to EGDT? Surviving Sepsis Bundles? Recognition and early antibiotics? (probably)), long term outcomes haven’t appreciably changed. I don’t have the figures but I think most of our septic deaths now are at hospital day 10-30. As far as I can tell, we have no theory about what to do with patients here hemodynamically or otherwise. Yeah, we try to get them fluid negative, have flirted with beta blockers, steroids, etc, and we put all sorts of monitors on them, but we’re really fishing here. To me, this is the new frontier for sepsis research – the “After goal-directed therapy” time

  7. Gavin denton says

    In the UK we have about 1/7th of the ICU capacity of the use per-capita. As a result, septic patients on wards often won’t get near ICU until they received 5+L, and probably abnormal saline to boot. In the UK these are probably the high risk patients that need vassopressor far earlier, but without the resources to provide it early enough. This could be mitigated if we were able to assess fluid responsiveness out side ICU with systems such as uSCom or NICOM. We don’t keep patients in ED for this sort of care (4 hour max stay in ED nationally).

    • Brendan Smith says

      We have the same 4 hour limit in ED in Australia – in theory anyway. But that’s plenty of time. 80% of our septic patients in ED are haemodynamically optimised in 1 hour and 95% within 90 minutes. By the time they reach ICU a couple of hours after admission I will usually have gone home! Maybe that is the real answer to why our SS mortality is so low at 6% – we don’t let them wallow in hypotensive hell for hours…

  8. Nikolay Yusupov says

    So as the case with most things middle ground seems to be the way to go. Striving to achieve physiologic levels as opposed to supra/infra physiologic levels in resuscitation.

    Scott would you still resus according to your ACEP talk, given cited literature? Or would you adjust this scheme in anyway?

    Fix the Sat
    Dial the Rate
    Fill the Tank
    Squeeze the Pipes
    Flog the Pump
    Bolster the Crit
    Calm the Beast

    • says

      everything is the same, and I mention the early use of pressors to squeeze the venous side in that talk. The only thing is that bronze medal is prob. ok in the ED, we are probably moving away from empiric fluids after those 1st few hours as being acceptable.

  9. Graeme says

    The benefit of using a non invasive CO monitor is it makes things so much easier than flying by the seat of your pants. You titrate noradren to SVR (rather than risking giving too much or too little). You titrate inotropes to inotropy. You accurately assess preload. Etc. it saves you time and worry.
    I would no longer suggest just cracking-on in a really sick patient with an airway issue simply because most times it would work out OK. This site has pulled us into a better place than that. There is a better place also with haemodynamics in sepsis.

    • says

      Agree monitors would be great, unfortunately they currently all have flaws. As to SVR, be wary–it is a calculation, not a measurement and it is predicated on and changes with CO.

      • John Hinds says


        “Cardiac Output” monitors can measure pressure, estimate or infer flow, and attempt to derive SVR. It is, at best, a second level derivation.

        Even if we could magically measure it, we’ve no idea what best suits the patient.

        This is what you can often seem to radically change the SVR by giving fluid with a CO monitor in situ. The pressure or the inferred CO may change; and the derived SVR will follow regardless of what the “real” SVR may or may not be.

        CO monitors are best thought of as a little box of trend measuring voodoo!


        • says

          yep, if i had the ability to customize these things, I would love to see 1 screen with just SVi and perhaps a small display of SVV (only useful in a small subset of pts). This is assuming the SVi is actually accurate, which I am unsure of with most of the devices out there.

      • Brendan Smith says

        There are a number of points I feel I must pick up on here. Low fluid volume-early pressor therapy was certainly tried in the past, the big problem of course is that its effects were not adequately monitored. Simply using a vasopressor does not cut the mustard in septic shock, because we have to deal with the inotropy issues as well. What tended to happen was that patients would be drastically over-squeezed resulting in a very high afterload in a patient with low inotropy. The left ventricle would just dilate and fail. “Levophed – leave ‘em for dead!” was a common saying. It’s hardly surprising that they didn’t do well. We can do so much better these days.

        DO2 only has to be adequate – not high, not low, just enough. VO2 must also be adequate for normal tissue metabolism as Paul Marik pointed out. The problem of course is how do we ensure that this happens? We could use surrogates allied to some educated guesswork, or we could just measure them. The problem is that we need to know what the cardiac output is in each case. Now maybe I’m just an old country doctor, but to my way of thinking cardiac output is the amount of blood that comes through the aortic valve in one minute. Stroke volume is the amount that comes through the aortic valve in one beat. It is not a pressure measured at the wrist and multiplied by some weird and unproven formula, it’s not an electrical current, or a resistance, or an impedance, or a change in temperature in the pulmonary artery, or something measured by complex mathematics based on highly dubious assumptions. The only cardiac output I accept is that which is measured at the aortic valve, which means either using echocardiography or its derivative, the USCOM. Simple.

        Now as both of these methods can measure cardiac output with an accuracy of 95% or more, and if we can measure mean arterial pressure with a similar accuracy, then the calculation of SVR is accurate. Of course it changes with cardiac output, that’s how it’s calculated. What is perhaps a more relevant question though is what does SVR actually mean?

        SVR is a pretty poor indicator of vascular impedance, we all know this, but what do we use instead? We know that an SVR of 300 indicates a widely vasodilated patient, whilst an SVR of 2000 is a vasoconstricted patient. And guess what? If your SVR is about 900 you’re about normal. The problem however is that we are deriving the value for SVR as if cardiac output was constant throughout one minute, but it isn’t. What we need is a measure of dynamic impedance during ventricular ejection. A far better value to use is PKR, which we published along with our inotropy index. The full text is available here- http://bja.oxfordjournals.org/content/early/2013/05/02/bja.aet118.full.pdf

        A normal PKR is around 30. We have found that the PKR is the very first thing to change in septic patients, and a value of 3 or 4 is very typical for established septic shock. This should come as no surprise when we look at the pathophysiology of sepsis in adults, where the arterial tree progressively dilates and the cardiac output attempts to compensate by increasing dramatically. Ultimately, when this process fails because the cardiac output can increase no further, the blood pressure drops and we have decompensated septic shock. The drop in vascular impedance is what was leading the charge and therefore a sensitive measure of vascular impedance such as PKR is extremely valuable.

        So-called cold shock, predominantly seen in children, is associated with a low cardiac output and a high PKR, typically over 60 and often much higher, and low inotropy, indicating that this is really cardiac failure rather than peripheral vascular failure. And while we’re talking about kids, just to give my spin on the FEAST trial, if you’re severely anaemic and your cardiac output is maximal already, then diluting the remaining haemoglobin really isn’t a good idea if you want to maintain tissue oxygenation…

        In the many years I have been involved in intensive care, I have noticed that patients don’t usually die with normal numbers! Getting the numbers back to something like normal is exactly what we do in every branch of medicine, be it a blood sugar level, an electrolyte level or whatever. Why then is it such a bizarre idea to say “let’s get the cardiac output, the vascular impedance, the inotropy, the preload and the arterial pressure back to normal”? It certainly seems to me to be the obvious way to go. Or we can just kid ourselves that we actually know what we’re doing and use a one size fits all approach. Flashing our ultrasound over the IVC to show that we can get a rough estimate of the CVP and then throwing fluid into the patient on the basis that they can respond to it, doesn’t seem logical on any level, especially as the evidence strongly suggests that the fluid doesn’t stay in the vascular compartment anyway. We can’t even claim that the CVP or IVC distensibility index gives us an accurate indication of over-distension of the ventricles, because we don’t know the ventricular compliance.

        Finally, I would ask the question “just because the patient is fluid responsive does it mean that they need fluid?” As I sit here writing this, I am sure that I’m fluid responsive, but I do not need fluid right now. (Although a couple of beers later would go down very nicely!) We really have to start thinking in terms of what is adequate on each of our parameters if we are to optimise the care of the patient. We certainly don’t want to maximise the haemodynamic parameters, but conversely, we don’t want to compromise them either. Optimise? Maximise? Compromise? The only way I know of doing this properly is to measure all the haemodynamic parameters involved, and correct them accordingly. As to what’s normal, we have studied over 3500 normal subjects from the cradle to the grave so that we know exactly what is normal at any age. If anybody wants a copy of the tables just send me an email and I’ll forward the tables to you.

          • says

            Now an offer like that is hard to refuse! The tables are winging their way through the ether to you on that worldwide google interface twitterbook thing.

            I would have thought smaccGOLD was a real possibility. I could even work out a few comments for your debate like “real airway docs of a certain age can’t even remember where their checklist is!” (For everybody else, see the smaccGOLD program for day 3 – Friday at their website).

            • says

              I would love to look at them as well, you have my email.

              As to the comment, you very well turn out to be right about your strategy. The thing is you need to actually do the study. Your before and after is quite nice and is just the thing to establish that an irb should accept your research proposal–it is not, unfortunately the kind of proof you need to establish this is any better than all the other stuff we do based on physiology and insightful theory.

              If there is one thing we have learned from the >15,000 patients in the NYC Sepsis Collaborative, it is that incredible absolute risk reductions in mortality come simply from having a protocol (any protocol) and being aware of these patients.

  10. Susana Planck says

    Scott, I very much agree with you that most of Marik’s arguments do commit the same logical fallacy again and again. He points to end point, eg. positive net fluid balance, and takes for granted a causal relationship with fluid administration which appears unjustified. However, I do think his strongest and most interesting evidence comes from the FEAST trial. My understanding of your rebuttal of this trial is that the patients didn’t die from fluid overload so Marik’s theory of the particular mechanism of iatrogenicity is incorrect. Sure, but does the ‘why’ really matter? The evidence seems to state that in the one and only RCT fluid administration proved to be deleterious. This seems tremendously important to me. Whether it was from increased acidosis or not the take home message for me is that these kids did worse with fluids. To be fair, we need to be cognizant that this could be a function of the particular kind of crystalloid (NS). Also, I just did a quick lit search and couldn’t find it but I thought I remembered a study done in the prehospital literature that found that no fluid administration proved superior to even judicious fluid administration in hemorrhagic shock patients. Does anyone else remember this?

  11. says

    Thanks for this post Scott. Paul’s talk was great value for stimulating thinking on the topic and shocking people into taking the decision to prescribe fluids seriously however it really needed to be taken apart in a critical fashion like this.
    The common ground you both seem to share is you give some fluid and take a look and then give more. I sometimes reckon we are overthinking fluid responsiveness assessment. It’s fun to play with leg lifts and carotid flow times but when I have primary responsibility for a crashing patient I assess fluid responsiveness by giving some fluid and looking at my patient again. Any strategy that doesn’t include frequent reassessments of a sick patient is pretty much destined for failure.
    That’s why I really like where this talk went. Do the basics properly, watch your sick ones closely, titrated treatment to response and never “fire and forget”.

  12. John says

    In the last 5 years or so, we have had a better understanding of capillary fluid dynamics, particularly in conjunction with an appreciation of the glycocalyx. We now know that the glycocalyx normally ‘traps’ about a litre and half of plasma water in it (due to its hydrophilic chemical composition!) and that normally in the capillaries, there is a central moving layer of plasma and a relatively immobile layer closer to the endothelium….the bit that is bound to the glycocalyx. This explains the differences in measured capillary and venous hematocrit values, and also why Crystalloid : Colloid equivalence is 1.3 : 1 rather than 4: 1 as previously thought.

    We have also acquired a better understanding of the mechanisms of edema formation in critical illness and more importantly, the magical phenomenon of improved diuresis that we have all marvelled at, during the recovery phase.

    In short, we have kinda debunked the original Starling theory of fluid dynamics in the capillary.

    We now know that the colloid osmotic pressure in the intravascular space will only oppose the outward movement of water, and increasing the colloid osmotic pressure by synthetic colloids will not reverse the flow and draw fluid from the interstitial to the intravascular space. ( Multiple trials , starting with the SAFE trial have proved the futility of using synthetic colloids !) What they end up doing is, probably drawing water from the glycocalyx in the intravascular space itself and dehydrating and then disintegrating this vital layer. As a result you will find a transient improvement in blood pressures, but afterwards, a lot of this fluid will track into the extravascular space. Any hyperosmolar solution can do this including Soda Bicarb….we have all seen the very transient increase in blood pressure after bicarb which has always been incorrectly attributed to ‘reversal of acidosis’…bah!!

    Extravasation of fluid from the capillaries is predominantly dependant on capillary hydrostatic pressure and not on decreased intravascular colloid osmotic pressure— because we have realised that interstitial and intravascular colloid osmotic pressures are very close to each other.

    The way to prevent overloading and thus extravasation would be to minimise rapid increases in capillary hydrostatic pressure. How can we do that? – By small volume crystalloid boluses and early use of alpha1 agonists—the latter work by afferent arteriolar constriction and thus minimising huge increases in capillary hydrostaic pressures. This is where Marik’s argument takes a strong foothold.

    Albumin is needed for the integrity of glycocalyx, — explaining why albumin is making a comeback into our fluid armamentarium.

    The lymphatics have assumed a pivotal role in the normal mechanisms that prevent edema formation. We have realised that they are a very active conduit to return of interstitial fluid to the central circulation, and they they have contractile collecting ducts and passages that are calcium dependant. They are inhibited by the terrible twins ANP and BNP—therefore shutting down in active sepsis, where the twins tend to dominate. (This also partly explains the peripheral edema commonly seen with Ca channel blockers when they are used as antihypertensives). Once the sepsis resolves, ANP and BNP levels drop and the lymphatics recover their contractile elements. All the interstitial fluid can now be returned to the central circulation causing an improved diuresis.

    In any case, fluids should only be used as any other drug should be— only if needed. We need to realise that fluid requirement and fluid responsiveness are two completely different things and focus on appropriate fluid balance rather than branding it as either restrictive or liberal.

    • says

      Fascinating and a more erudite description of the situation than any I have seen thus far. Which John are you? And what is your affiliation?

      • John says

        Hey Scott,
        I have always been an avid consumer of the fantastic pearls of wisdom scattered throughout your site….You are doing great work !.

        I am an intensivist from India, presently retired from hospital practice, and concentrating on teaching post graduate trainees in Emergency medicine and Critical care.

        Keep up the good work, mate..


        • Paul MArik says

          John: Brilliant I agree 100%, one limits tissue edema “By small volume crystalloid boluses and early use of alpha1-agonists” and then closely monitoring the patients response. The days of giving a 30mlk/kg bolus are gone; this leads to iatrogenic salt water drowning. We also need to revise the starling concept; fluid does not move back into the post capillary vascular system by osmotic forces; lymph flow is the primary mechanism that fluid returns to the vascular compartment; so giving colloids to suck fluid into the vascular compartment is doomed. So in the end it is the capillary hydrostatic forces together with damage to the glycocalyx and endothelial junctions that cause tissue edema. Rapid infusion of large volumes of fluid will increase capillary hydrostatic pressures and increase ANP/BNP which damages the glycocolyx and inhibits lymphatic flow.

          A lot of folks have gone bla bla bla about the fact that the greater fluid balance may just be a marker of disease severity since the studies were observational and not RCT’s (like the FEAST study). I stupidly omitted two very important points from my talk, 1) There are over 30 randomized controlled trials (RCT’S) in patients undergoing elective non-cardiac surgery that demonstrate that a conservative fluid strategy vs standard care based on resuscitating the non-existent “third space” are associated with a significantly reduced incidence of postoperative complications and death. 2) A meta-analysis just published in Crit Care Med (epub) demonstrated that in both RCT’s and cohort studies that a conservative fluid strategy (less fluid and NOT using the ATLS protocol) is associated with a lower risk of DEATH. Interestingly, an ER study showed that trauma patients who received more than 1.7 l of fluid in the ER had an increased risk of death. The Rivers paper is the only study that I am aware of that demonstrates aggressive fluid resuscitation improves outcome. Many folks cite the KP data to support EGDT. An analysis of this data (soon to be published) has shown that this approach(EGDT) actually increases the risk of DEATH.

          This has been a fascinating dialogue and I am indebted to the our wonderfully smart and politically correct leader …the one and only .. Scott.


    • says


      Absolutely fascinating. I’d vaguely heard about the glycocalyx and after hearing Paul’s talk, started to dig a little, but as Scott says, fascinating and highly sensical description. Thank you for lifting some of the fog, I will pursue this with even greater interest.

      I’d love to discuss this live sometime!


  13. Tom W. says

    Very nice summary, John in India. Just want to caution that the glycocalyx seems to function well at very low [alb] and we dont know if bottled albumin is as effective as your own. Not enough evidence yet to crack open the 4% at the first sign of sepsis.

    • John says

      Thank you Tom W, and I completely agree…..I would be extremely hesitant to reach for that bottle of Albumin too. I work in a resource poor part of the world where stuff like Albumin can only be found in the deep pockets of the privileged elite, anyways!! Again on the topic of Albumin, it is good that you mentioned 4%, as opposed to the 20% that Marik mentions in his slide set… I believe that if you are going to use it at all, you should stick with iso-oncotic Albumin and not hyper-oncotic because using the latter is not going to ‘protect’ the already fragile glycocalyx.

      Personally, I find the concept of glycocalyx and the endothelial surface layer a very attractive one, as it helps to answer a lot of the questions which had plagued me all the way from Med- School. I know we are only just about beginning to understand the details of it….. but I believe that in the future, a lot of our therapies will be aimed at trying to preserve this biological layer which seems to be vital and extremely fragile at the same time! It has already helped in targeting therapy in other ‘colder’ areas of medicine like Atherosclerosis, Diabetic nephropathy etc.

      I think we are now in an era, where minimising harm has to be the thrust of any critical care intervention. So often do we see our patients suffer from ‘status iatrogenicus’. I think 5 years of undergraduate and several more years of post graduate training, fills us with an urge – to ‘do something – do anything’ when probably the best course of action would be to leave well alone! We are now looking at all of our previous gospels of ‘goals of resuscitation’ with skepticism and sometimes, outright disdain.

      This is precisely the charm of Intensive care medicine as it keeps dishing out exciting and humbling experiences in rather equal measures!

      • Tom W. says

        Right again about 20%. When thinking through RSE&GM (Woodcock & Woodcock BJA 2012) I thought I had fluffed it when I found Margarson & Soni had demonstrated that 20% HAS increases red cell dilution by more than the infused volume, which seems to prove plasma pi can reverse filtration, disproving Michel/ Weinbaum/ Curry/ Levick/ Adamson. Eureka moment when I realised hyperoncotic 20% will suck water back from the glycocalyx and so deflate it before any reabsorption can occur! I told Mike Margarson about this at a conference, he said he agrees with my interpretation. RSE&GM predicts hyperoncotic solutions are very bad for you. Someone should tell Marik…

        • Paul MArik says

          TOM.. That’s surely a EUREKA thought. My only reservation is that Prof Gattinoni used a 20% solution in his ALBIOS study which showed improved outcome in septic shock. 20% solution does increase intravascular volume by 4 times the volume infused; so the question becomes does this fluid come from the endothelial surface layer.. .. would be an interseting study to do.

          • Paul Marik says

            So it would appear that the only way a 20% solution can increase intravascular volume when given as a bolus is to draw fluid out of the ESL (glycocalyx). However I was suggesting giving the 20% solution as a continuous infusion (not as a bolus)… this is currently what we do in our ICU at a rate of 10mls/hr; this should negate the dehydrating effect. It would appear to be a bad thing to give a 20% solution as a bolus. It also appears that if a 5% albumin solution is given as a slow bolus rather than a rapid bolus less leaks into the interstitial space. So frankly I think there may be a role for albumin to “repair” the glycocalyx, but how much and how to give is a mystery to me. We also need to wait for the publication of the ALBIOS study.

  14. says

    I don’t know why the Rivers protocol always gets held up as an example of the successful use of CVP in guiding fluid resuscitation. Both the treatment and control groups in the Rivers study used CVP (as well as MAP and UOP) in their protocols. The difference in protocol between the groups was in the use of ScvO2 as an endpoint. This translated into higher numbers of PRBC transfusions and more inotropic support in the treatment group. I realize there was also more fluids in the EGDT group, but how much of this was PRBCs? I am anxious for the results of the Process study to hopefully iron this question out a bit more. But I have always had a physiological problem with the concept of giving crystalloids to increase cardiac output and thus oxygen delivery. Yes, you may increase CO (to a point) with fluids. But with every drop of crystalloid you give you decrease the hematocrit an equal amount, thus having no effect on DO2. Or at least that’s the way it would seem to me. What ultimately matters is the number of hemoglobin molecules trickling through the capillaries per unit time, not the volume of fluids going through them per unit time. You might make your blood pressure numbers look better when you give a fluid bolus, and allow your sphincter to relax a bit, but I don’t know if you’re really helping the patient. And the lecture by Marik regarding the glycocalyx is really making me think hard about it.

    • says

      I agree. The question must be asked though why did the patients need so many PRBC transfusions if there is this massive capillary leak syndrome in sepsis? Their [Hb] would be high wouldn’t it? They only became anaemic due to all the fluid that was given, and crystalloids don’t carry oxygen, although they may affect the rheology of the peripheral circulation leading to (hopefully) better tissue perfusion, but only up to a point. After that, the reduction in Hb offsets the increase in cardiac output and there is no net gain, or even a loss.

      Now Paul Marik always tells us that transfused stored blood doesn’t improve oxygen dynamics in the tissues because it doesn’t offload oxygen to the tissues. Now if he’s right, and I think he is, then this means ScvO2 goes up because the blood is coming back to the heart with most of its oxygen still attached. ScvO2 looks better, but did tissue O2 flux improve? I don’t think it does, at least, not to any great degree. The danger in this of course is that you have just ruined the one parameter, ScvO2, that you are relying upon to act as a marker of cardiac output. Is the ScvO2 increase due to improved haemodynamics or just the transfused blood hanging on to its haemoglobin?

      • says

        Agree and I have made the same arguments of ScvO2. What I have heard in terms of hidden back story is that many patients in the trial got blood outside of study protocol, i.e treating docs felt pt would benefit from blood. This would explain why so many people got it.

        Old blood does regain its ability to release a few hours after transfusion. Heaton A, Keegan T, Holme S. In vivo regeneration of red cell 2,3-diphosphoglycerate following transfusion of DPG-depleted AS-1, AS-3 and CPDA-1 red cells. Br J Haematol. 1989;71:131–6.

        That being said, I can count on one hand the number of patients I have had to transfuse (using Hb<7) as a result of sepsis resuscitation. Not sure if the Rivers’ pts already had a good degree of chronic anemia prior to their sepsis course. These were predominantly patients with surgical sepsis and comorbidities.

      • says

        I think Marik might be making too general of a statement if he says that stored blood doesn’t offload oxygen to the tissues. Sure, it might not offload as well, but to say it doesn’t offload O2 at all is a bit of an oversimplification.

        Granted, there are other concerns with PRBC transfusions including immunosuppression, and nitric oxide scavenging leading to endothelial dysfunction, etc. Plus, what Hct do you target? 30 is pretty arbitrary. And assuming your SaO2 is good then ScvO2 just tells you what is going on with your brain and arms. I dont entirely agree with the idea that it trends with SvO2 in the setting of severe sepsis where your normal vascular regulatory mechanisms, designed to maintain perfusion to the brain at all costs, go completely out of whack in a completely unpredictable manner.

        Whatever the case may be, there was a dramatic decrease in mortality in the Rivers study, so you really have to take a hard look at what was done differently between the treatment and control groups. And clearly that wasn’t through CVP monitoring, or targeting a MAP>65 via pressors, because that was done in both groups. Maybe it’s true that the reason ScvO2 went up was because there was less extraction from transfused Hb molecules, but obviously something went right overall because more patients were surviving.

        Maybe the thrust of our efforts should be in identifying how to make PRBCs more effective with regards to O2 unloading, membrane integrity (in terms of nitric oxide scavenging), deformability, and safety, etc?

  15. says

    This is great stuff, and an awesome debate.

    I’m out of my league in this comment thread, but I wanted to bring something up:

    This was an interesting article looking at fluid balance based on diastolic function and it’s relationship to mortality. Basically, they found that in patients with preceding diastolic dysfunction, those with a lower fluid balance had a higher mortality. Those with a “volume overload” (as predicted by a pseudonormalization of diastolic function (presumed PCWP >15 mmHg) actually did better. They also got more fluid!!!!!!
    This is interesting to me because this is the very group we worry about fluid overloading- the ones with heart failure. Does early presser venoconstriction have the same effect in this group of patients by increasing preload without damaging the glycocalyx? I guess we won’t know until we figure out how to ultrasound the glycocalyx and settle this thing for real……. Working on the protocol now.

    My impression would be that every patient doesn’t start with the same fluid balance or the same capillary leak. No protocol that describes a specific amount of fluid administration will be right all of the time. For that reason dynamic measures of fluid responsiveness still make more sense after an initial fluid bolus.
    More food for thought.

      • says

        That is a beautiful explanation of diastolic dysfunction and it’s progression and relationship to filling pressures. It wasn’t long ago that ECHO reports would conclude “normal LV function; diastolic heart failure cannot be excluded” and now ECHO is offering a truly sophisticated explaination of the pathophysioligy- fantastic!

        I’m not sure about drawing any conclusions though about diastolic dysfunction’s relationship to mortality in sepsis (except for the conclusion that starting out with a normal heart is better than starting out with a sick one). The statistics used to hypothesise about differences between grades of diastolic dysfunction and mortality have the feel of bullseyes drawn around a dart and the numbers in each group are very small. It’s an attractive idea that grade 2 did better because their increased filling pressures (due to resus or to increased renal water and salt conservation as part of the heart failure compensation) gave them an edge: this would justify our practice of giving people fluid) but it looks to me like more subjects are needed.

  16. says

    Scott, thanks for the fantastic Marik lecture and your excellent critical response. One comment re your response. You state the FEAST study “needs to be thoroughly debunked” but I don’t believe you have done this nor does the article by the authors you linked to. If anything that article firmly debunks most of the criticisms of the FEAST study by showing findings such as no difference in harm from boluses in kids with and without anaemia/malaria.
    While I agree your acidosis theory has merit, you failed to mention the primary theory put forward by the authors in the article you referenced and by Myburgh & Finfer http://www.ncbi.nlm.nih.gov/pubmed/23497460, that “although there was evidence of improved short-term hemodynamic effects with boluses in these patients, bolus administration may have resulted in a rapid reduction in sympathetically mediated compensatory mechanisms, leading to cardiovascular dysfunction and death”.
    I find this more plausible as an explanation as to why the main Terminal Clinical Events that increased due to boluses were cardiovascular collapse rather than fluid overload.
    We should not ignore this landmark trial because it makes us all somewhat uncomfortable about our current practice.

    • says

      Anand, the thing that needs to be debunked is not the FEAST trial, it was Paul’s assertions based on it in the context of the debate. FEAST was amazing as we discussed. But this trial has little applicability to our patient populations. I will clarify that in the shownotes.

      Now as to fluid somehow being a sustained sympatholytic, I can’t see a shred of physiology or literature as to why that would be the case.

  17. Joshua Bigler says

    What about Nitroglycerine infusion in septic shock. I know this has been studied before and it is very backwards in traditional thinking. This would unmask volume status to allow for fluid resuscitation. This would maximize 2 of Marik’s initial goals ( 2- improved microcirculation and 3- limit tissue edema) If we prevent the dysoxia to the mitochondria we can maybe prevent the glycocalyx injury, reperfusion injury, improved flow. Again I am sure I am way over my head. Would love to hear from what others thinks, form this knowledgeable group.


Speak Your Mind (Along with your name, job, and affiliation)