Ok, so it was pretty cool to see an NEJM issue basically dedicated to septic shock management, I must admit. But let’s dig a little deeper, shall we?
So here is where they are: http://www.nejm.org, and fully available for now.
I won’t go through all the details and numbers, after all they are in the papers, so let’s just analyze them from two principles:
a. the N=1 principle – how was therapy individualized?
b. was there any integrated monitoring of the therapeutic goals?
…and we’ll conclude by looking at the potential practice-changing potential of each of these studies.
So first of all,
High vs Low BP Target in Septic Shock, by Asfar et al.
So basically a negative study except for two findings, the increased incidence of afib in the high target group and the decreased need for renal replacement therapy among chronic hypertensives in the high target group.
so N=1 is not really revealed:
“Refractoriness to fluid resuscitation was defined as a lack of response to the administration of 30 ml of normal saline per kilogram of body weight or of colloids or was determined according to a clinician’s assessment of inadequate hemodynamic results on the basis of values obtained during right-heart catheterization, pulse-pressure measurement, stroke-volume measurement, or echocardiography (although study investigators did not record the values for these variables).”
So lets just hope that the variability evens itself out between the groups, since we don’t really know. The numbers don’t really tell the tale, because the average fluids received (10 liters over 5 days) could mean one patient got 15 and one got 5 – although let’s trust they followed the French Fluid Resus protocol…
So the atrial fibrillation makes total sense – more B agonism should result in that, and the decreased renal failure also does.
As the authors note, the actual BP averages were higher than planned. For those of us practicing critical care, we know most nurses titrating prefer having a little bit of extra BP – even when I prescribe MAP 65, I usually see the 70 or so unless I make a point to tell them. Understandable. They also note the underpowered-ness of their own study, but I think it is still worth looking at their results.
So…bottom line? I think it’s a great study for a couple of reasons.
The first is to remind us to pay a little more N=1 attention to the chronic hypertensives, and that it is probably worth aiming for slightly higher MAPs.
The second, debunking the myth of “levophed, leave’em dead” (which I heard throughout residency at McGill), and the concept of doing everything (ie juicing patient into a michelin man) in order to avoid the “dreaded and dangerous” vasopressors. So really I think an alternative way to conclude this study is that it isn’t harmful to have higher doses of vasopressors. I think this is actually a really good study on which to base assessment of more aggressive vasopressor support vs fluid resuscitation, in the right patients.
It would have been interesting to have echo data on those who developed a fib – were they patients who had normal to hyperdynamic LVs who in truth did not need B agonism at all and would have been fine with phenylephrine? Perhaps…
Cool. I like it.
Albumin Replacement in Patients with Severe Sepsis or Septic Shock, by Caironi et al. The ALBIOS study (a Gattinoni crew)
So basically showed no difference, so pretty much a solid italian remake of the SAFE study in a sense, confirming that albumin is indeed safe overall, and may be better in those with shock. As the authors note, mortality was low, organ failure was low, so study power a little low as well. Note the mean lactates in the 2’s at baseline. The albumin levels of the crytalloid only gorup were also not that low, low to mid 20’s, whereas I often see 15-20 range in my patients, especially if I inherit them after a few days, as I do use albumin myself a fair bit. They also used a target albumin level, not albumin as a resuscitation fluid purely.
In my mind the benefit of albumin would be greatest in those with significant capillary leak, particularly those with intra-abdominal and pulmonary pathology. It would have been nice to see a subgroup analysis where extravascular lung water was looked at (especially coming from a Gattinoni crew!).
Another interesting thing would have been to know the infusion time of the albumin, since animal data tells us that a 3hr infusion decreases extravasation and improves vascular filling vs shorter infusion times. I routinely insist on 3hr infusion per unit, which sometimes results in 9-12hr infusions, almost albumin drips!
I like it. Reinforces that albumin is safe, so makes me even more comfortable in using it in the patients where my N=1 analysis tells me to be wary of third-spacing. Also the fact that they used 20% – in Canada we have 100cc bottles of 25% for the most part – is nice, since the SAFE data used 4%.
A Randomized Trial of Protocol-Based Care for Early Septic Shock – The ProCESS Trial.
So right off that bat my allergy to protocols flares up, so I’ll try to remain impartial. It just goes against the N=1 principle. The absolutely awesome thing about protocols is that it primes the team/system to react – so clearly protocols are better than no-protocol-at-all, but strict adherence would clearly not fit everyone, so that some built-in flexibility should be present.
This being said, the ProCESS study is really interesting, for a number of reasons. They have three groups, and compare basically (1) Rivers’ EGDT to (2) their own protocol (see the S2 appendix online) which gives a little more flexibility and (3) “usual care”. Net result is that all are pretty equal, no change in mortality. As the authors note, their mortality was low, so again may not have been able to detect a difference.
So, what does this mean. To me it’s a little worrisome because I doubt that the “usual care” represents the true usual care found in EDs/ICUs all over the world, so I am concerned that many docs will use this as a reason to justify not changing their practice, similarly to many I’ve heard say they don’t need to cool anymore after the TTM trial. Human nature for some I guess.
Bottom line? You don’t have to follow EGDT if you’re conscientious and reassessing your patient frequently and have done all the other good things (abx, source control, etc). I think that’s really important because giving blood (see my post about S1P) to those with hb > 70 and giving dobutamine to patients with potentially normal or hyper dynamic LVs never made physiological sense to me, and the problem with a multi intervention study such as EGDT is that you can’t tease out the good from the bad or the neutral. Again, studies such as EGDT are pivotal in changing practice and raising awareness, so this is not a knock against a necessary study, just to highlight the point that each study is a step along the way of refining our resuscitation, and the important thing is to move on. In fact, the reason that this is a negative study is probably due to the improvement in “usual care” that EGDT brought along.
Conclusion: No new ground broken, but these studies do make me feel more confident and validated in continuing to not do certain things (strict EGDT) and doing others (albumin and earlier use of vasopressors).
Kudos to all investigators.
let me know what you think!
“As the authors note, mortality was low” — wasn’t the mortality at 28d ~32% and at 90d ~41-43%, compared to the the ARISE 28d mortality ~15%, 90d mortality of ~18%??? and the PROCESS 60d ~18-21% and 90d 30-34% (actually interesting in my mind that there was a 10% increase?)… my point is the mortality in this study was HIGH compared to other septic shock trials (outside of the esmolol in septic shock trial 49.4% vs 80.5% — interestingly both these studies were done in Italy). Interesting that the lactates were ~2 but had a mortality of 32% and 41-43%? So the argument that the mortality is too low to find an difference doesn’t hold true.Moreover, look at page 29 of the supplement (which has been talked about as an editorial to the authors). Why do they make the conclusion based on ‘unbalances’ versus ‘clinically relevant variables’ — because one gets them a positive outcome and the other a negative–this to me is unethical publishing given they never made note of this issue.
Other issues: isn’t ~32L at 7d between the 2 groups seem like A LOT of fluid given that restrictive strategies are showing promise for improvement in outcome. ~96% of the fluid in the albumin group was crystalloid at 7d–why were these patients requiring so much other fluid (ie seems like they should have been limiting/concentrating drips, etc). Hb generally greater than 10 in an ICU population — how much blood were these patients getting or were they really that sick– I would have to go with not that sick given there was minimal blood products given if their supplement is accurate? I hardly find a net fluid balance of -1618 to -858 all that impressive at 7d — another interesting point, look at 1d fluid balance 1229 vs 1504 — I have a hard time getting this to the ~3L mark which has shown to be associated with the best mortality and they got to ~1.5L (another indication that these patient’s weren’t that sick).
Overall interesting that they had a SOFA score of ~4-5 (est mortality < 10% and the lactates ~2) but had a mortality at 28d of ~32% and at 90d ~41-43%… something is wrong here??? I get the idea of using albumin but I don't see any evidence that it is worth the 32x cost at my hospital in my severe sepsis or septic shock patients. Love to hear your thoughts.
U of Wisconsin CCM fellow