FlameProof Preconference Course at Essentials of EM
Mentioned in this episode: Rob Orman and I are putting on the Flameproof: Shift KickAssery workshop on May 29, 2023, at the Cosmopolitan Hotel, Las Vegas. 9a-1p. Limited space (intentionally).
This is a PreCon for Essentials of Emergency Medicine.
The CLOVERS Trial
The Bottom Line Review
Multicenter, randomized, unblinded superiority trial.
Inclusion
- Patients over 18 years old with either a suspected or confirmed infection and sepsis-induced hypotension (SBP<100 after 1 liter fluid)
Pertinent Exclusion
- More than 3L of IVF prior to enrollment (including EMS fluids)
- Fluid Overload
- Severe volume depletion from non-sepsis causes
Intervention and Comparison
Both groups received 1-3 L of fluid prior to randomization
3 Big Sepsis Trials (PROMISE, ARISE, PROCESS) got between 2-3 liters of fluid
Ran the Protocol for 24 hours
Outcome
1563 patients at 60 hospitals. ~ 780 in each group.
Average age < 60 y/o
Primary outcome was in-hospital mortality by 90 days.
14% in the restricted fluid group and 14.9% in the liberal fluid group (non sig. -0.9 favoring Restrictive CI -4.4 to 2.6)
Stopped for futility due to lack of difference between primary and secondary outcomes (I view this as an enormous misstep, though certainly not on the part of the researchers, but on the part of the safety board)
End-stage Renal after already excluding volume overloaded patients
27.3 % mortality in restrictive vs. 47.5 % in liberal (CI -41.9-1.5)
Peripheral Vasopressors
500 patients on peripheral vasopressors. Amongst those 500 patients, they only had three extravs all of which resolved without complication or intervention
Justin Morgenstern's response to the ?, “What does this say about the CMS Measure Urging 30 ml/kg?”
I am not a statistician, so I might be missing something, but I really hate the way we present scientific data, and talk about things as if a non-inferiority trial is somewhat physically different than a superiority trial. The methods of the trials are identical, its just what stats you decide to run that differentiates them.
As a side note, I really hate non-inferiority trials. I think they have tremendous problems – I actually have a long and nerdy blog post about it that I haven't published yet. What you would really want here is an equivalence trial, but that distinction is irrelevant.
The big problem with either a non-inferiority or a superiority trial is that you make a massive assumption in which arm you consider the standard. That assumption biases the entire results towards the side you choose. An example is the RCT looking at conservative management of pneumothorax. Their technical answer is that ‘conservative therapy is not non-inferior to invasive.” But running the exact same stats in reverse, assuming that conservative therapy should be considered standard, they would have reached that conclusion that ‘invasive therapy is not non-inferior to conservative.” Its crazy that a methodologic assumption has such a huge impact on how we interpret trials.1) Absolutely you can run a non-inferiority calculation on this. Many non-inferiority trials specifically say in their trial ‘we are going to start as a superiority trial, but if that fails, we will run a non-inferiority calculation”. The only problem with doing it after the fact is setting your non-inferiority margins is sketchy when you already know the outcomes.
2) One of the major problems with non-inferiority trials is that their non-inferiority margin is often laughably large. They seem to get away with it in trials, but if you try to run that calculation after the fact, people likely wouldn't accept it.
3) You don't really need to do that. The way trials like this should be presented is not “positive” or “negative”, or even “non-inferior”. We should just present the actual range of possibility. So, excluding the influence of potential bias, the appropriate conclusion for this trial is that:
We have a 95% certainty that restrictive fluids, as compare to liberal, result in as much as a 4.4% decrease in mortality to as much as a 2.6% increase in mortalityI agree with you that this is an equivalent result
I agree with you that this could be presented as RCT data that the 30 mL / kg is not superior to less fluids
The only problem you might run into is that this trial is still relatively small, and you are stuck with that 95% CI that reaches 2.6% on the side of restrictive fluids increasing mortality. I think that would be a silly place to focus, seeing as the is an equally likely change that they decrease mortality by 4.4%. I am not sure whether the people who write your core measures will get caught up on that fact.
Luckily, we don't have to worry about that, so the sane areas of the world will just use clinical judgement to determine how much fluid each individual patient needs.
Justin's take on CLOVERS
Additional New Information
Retrospective, multi-center study shows no benefit of 30ml/kg vs. lower doses of fluid
More on EMCrit
Additional Resources
Thanks for the review, Scott. I know I am a frequenter on these posts, but I have some comments regarding your podcast I would like to share. As you mentioned, you were hoping to have Nate Shapiro on your podcast and perhaps this could serve as some points of discussion. First, to your overall points on the study: 1) I agree with you that completing the enrollment would have been best. These studies are expensive, but they also require the proverbial movement of mountains to get up off the ground. Given there was no harm, I think completing the remainder would… Read more »
Mike, It is always a pleasure to read your comments!! – It was not the width of the CIs I was talking about, it was where the CI actually sat. This is the true definition of a trend. – I did not say that this study supported 20ml over 30. My reading of this study is that it fully supports the safety and equivalency of a conservative approach to fluids based on clinician judgment. I do not want to see SSC go back to 20mls/kg (though that would be better than 30), I want them to eliminate dictating any specific… Read more »
Thanks for the reply, Scott. I appreciate the time you put into your responses. I will keep my eyes and ears open for your discussion with Nate.
Mike
Kia ora and thanks Scott for hosting Nate as I am hoping he can answer a question regarding Peripheral Vasopressors.
I could not find the concentration used in the CLOVERS manuscript or supplementary material or additional references. I’m in Aotearoa/NZ and we use syringe drivers with 4mg Norad in 60 mls (0.067 mg/ml) as opposed to 4 mg/250 mls (0.016 mcg/ml). Should I be concerned about higher concentration of pressor if peripheral IV extravasates even though delivery of actual drug may be same for equivalent rates.
Best, Michael in Whangarei (Fung-a-ray)