Welcome back to the ICU Ed and Todd cast ICU ed like education ed and Todd, Toddcast podcast. I'm your cohost with the most Eddie and he is Todd and we are your podcast today. We're going to be talking about phenobarbital and alcohol withdrawal and reviewing transfusion thresholds and critical care as our old article. Thanks for all the listeners who message and email in. We'll get to some of those shortly on things, but before we get that, anything you wanted to say? 

By default, am I the cohost without the most? I think that's correct. Or the cohost with the least. Yeah, but least it in rhyme. So beast with the least 

Last episode, we were talking about things that make or break a critical care doctor's day. We had one listener who requested to remain unnamed that said the things that make their day is when people report out their ventilator compliance variables as part of their vitals. So that means like your peak and maybe plateau pressures, if you're on a volume targeted mode or a minute ventilation and a pressure targeted mode. I guess I never really considered that because I'm not sure I've ever heard that reported out routinely, but I think it would make me happy. 

Uh, it would be like, you know, indication as a team that we're like keeping track of the important variables for our Yeah, that would be a highly functioning ICU on bedside rounds. You know, I'm happy when they report to me the ventilator settings. I'm still not so old that I can't see the settings from, you know, wherever I'm standing either in the hallway or in the patient's room. So I often can get them even if they don't report them to me. 

We had a colleague who remarked to me once in the ICU, if there's something going on with a patient, a patient is decompensating. Everyone wants to know what's, you know, like what's wrong with the machine, what's wrong with the ventilator, the circuit for ECMO, dialysis. But you know, for like knowing, for example, that you're... vent pressures haven't changed is a pretty quick way to get to, Hey, maybe the ventilator isn't the problem here. They're decompensating for another reason. 

Maybe it's just my bias. I don't know, but I think it, for me, it's even more apparent when you do pressure control modes, because then the pressure hasn't changed. The PEEP hasn't changed. Hopefully the eye time hasn't changed, but the title volumes changed as the compliance is changing. And so when a patient's on a pressure control mode, I often will look at the title volume, some of that's probably because, because that's the way I was trained, I think in title volume. But it also I think is a subliminal way of me getting what's this patient's compliance done. 

Yeah, I feel like minute ventilation is a little bit easier there, right? Just the after the breath to breath things can change. 

Yeah, but minute ventilation is only partly related to compliance. Right. Like I can increase your minute ventilation without changing your compliance by change, by increasing your respiratory rate. Right. And oftentimes when somebody's not doing well, they are to tachypneic. So their minute ventilation may be higher and their compliance may not have changed, but their minute ventilation is higher simply because their respiratory rates are. 

Yeah. I mean, this is just, I think a conversation is just saying that our patients are complicated. They have a lot of things going on. A lot of variables going on with them. 

This right here shows you successful training all the way through residency. Now you've finished your pulm critical care fellowship and you've gotten to the point where you're like, the ICU patients are complicated. Especially the ones that are on the ventilator. I mean, they're really complicated. 

They, they have another machine. They're dinging. I have to look at the monitor, look at the vent screen. 

I mean, wait until we get you through the, the patient has both respiratory failure and renal failure and they're on the ventilator and another machine. 

No. 

Yeah. 

It's probably the laziest form of an intro topic, but you know, taking things off of Twitter or X or whatever it's called now. But this one does segue into our next topic pretty well, so c'est la vie. Dr. Moinhatton, X or Twitter or whatever handle, at E.E. Moin tweeted, xeeted, she put on social media, what's the last paper you read in critical care medicine that's not an RCT that you think was exceptionally well done? 

It's a paper we're going to talk about today. 

For me, uh, you know, my first thought was, you know, retrospective cohort studies, but it includes like this pre post analysis. We're going to talk about target trial emulations, which we've talked about before with the fludrocortisone and sepsis. 

Yeah, but that's cheating. The target trial is trying to emulate a randomized trial. Sure. 

But also like epidemiologic studies, we've talked about that with WEANSAFE from the last episode. I think the trial we're going to talk about today is probably my most recent.

Um, but we've talked about some of these, we talked about hyponatremia and Osmo, uh, osmotic demyelination syndrome, which was not a randomized trial. I think they're not as common in my practice, but that there are cohort studies, there are pre post studies that I use to somewhat inform what I'm going to do in practice. Uh, we were talking recently about you know, what should you do with a patient with atrial fibrillation that has sepsis and you think it's atrial fibrillation that's occurred in sepsis, it's not chronic atrial fibrillation. And we were talking about, should you anticoagulate these patients? And I was relaying that, you know, initially the data came out that these patients do worse than patients that never have AFib when they're in their critical illness, or even worse than patients who have chronic AFib. And with that, I was like, Oh, I should do something about this. I should anticoagulate them. And then the data came out when you looked at anticoagulation with these, that it didn't actually decrease their risk of stroke or any thrombotic type event, and they had more bleeding risks. And so I kind of quit anticoagulating them. And so there I changed practice twice once into something and then back out of something based off of non-randomized data, waiting kind of now for randomized data, which tell me, you know, what's the right thing to do here. And I think there are times where the only data you have are non-randomized data and so you have to kind of use it to try and guide practice. Or the only data you're ever going to get, so for example trials that look at harm, it turns out that IRBs in this country don't let us randomize people the things that we think are harmful. You know, hey we'd like to give Johnny and 100 of his closest friends cigarette smoke, you know, we're not going to give Bobby cigarette smoke because we think maybe it'll give cancer to Johnny. The IRB goes, you know. Not sure I like that study design. I think sometimes it's either the only data that are available or it's the only data we're going to get because you can't do a randomized trial. 

Are there other situations in which that you would view non-randomized data as kind of like the best that you're going to get other than kind of harm? 

Yeah. I think harm is one really, really distant outcomes. You know, if you have to randomize somebody, then follow them for 20 years to get their outcomes, that's hard in a randomized trial and you're much better probably doing it with a trial that starts with the outcomes and then tries to go back 20 years instead of following people for 20 years. 

Your case control. 

Yeah, exactly. And the other thing that I think are hard to do in a randomized trial are rare, really rare events, you know, to try and understand differences in two treatment arms between rare events. You need tons and tons and tons of patients. And. the signal is pretty small. And so you're not getting a lot of events, you know, that's why they're rare, but those are also really hard to do in a randomized trial fashion. But there are, I mean, and this is what we talk about all the time, but we've never explicitly said that, you know, the benefits of the randomization, the benefits of the perspective nature of the studies, eliminating multiple forms of bias, selection bias for the randomization, uh, some circumstances allow you to blind, which decreases observer bias. So, you know, question. Anything else that you're thinking about when you're talking about that? No, I mean, you know this and many people who are listening know this. One of the hats that I wear is a clinical trialist. And as a clinical trialist, I believe that a randomized controlled trial is the... best evidence that can be obtained for the benefits and probably the safety of interventions, treatments, you know, strategies, that sort of stuff. I'm not an epidemiologist and so I don't do a ton of my own work in epidemiology, case control studies, cohort studies and that sort of stuff. But I think it's important to recognize that there are times where that's all you're going to get and you have to be able to accept that's all you're going to get and interpret those data and then translate them as best you can into what you think is best. clinical practice and best for your patients. 

Yeah. So that's a little bit of a manufactured segue into our first article for the day. And we're going to talk about some of those things. The title of the article is phenobarbital based protocol for alcohol withdrawal syndrome in a medical ICU, a pre post implementation study. It was published in critical care explorations, April of 2023 by Awakeel et al. There's no acronym here right off the bat in a pre post study. what kind of biases or things should we be looking out for? I think. 

Do you have a bias on this study because it doesn't have an acronym? 

A little bit. 

Do you immediately kind of knock it down a tier because it didn't get an acronym? 

I mean, I was pushing hard to not do this article when we were talking about planning for the podcast specifically because it didn't have an acronym. That you could grade. 

Absolutely. Yeah. I mean, we can just spit our way into one. Pfft. Yeah, that's lots of Ps. Phenobarbital protocol, pre-post, it would just be... 

Yeah. Or you can skip that base and do phenobarbital protocol, alcohol withdrawal syndrome, and you got P-PAWS. 

Let's just go back to where I was heading with that. So in a pre-post study, we are looking at time, I think, is the most obvious bias. Seasonality is one of them and our patients different in different seasons. And then did anything else change at the same time? That could be a confounder, like another protocol or a concurrent quality metric.

 Yeah. We're talking about P PAW-MICU, right? Yeah. I mean, the pre post design is set up to have some biases and there's things that you can do to try and offset those biases. So for example, if your rate of whatever you're studying is changing before you implement something right, and it continues to change. And all you do is take a central tendency of before and after, it's going to look like it's lower after than it is before. And offset that you could say, okay, instead of just comparing the mean or the median before and after, I'm going to compare the rate of change before it to the rate of change after and see if, you know, the rate of change changed between before and after. So there's some sophisticated, more sophisticated, statistical techniques that you can use to try and get through those. But ultimately you don't have a concomitant data that you're getting from both arms of the trial. And so there are lots of potential confounders, as you said, different seasons for the before and the after, or, and this happens more than you know. And sometimes off of your radar screen, something changes in the practice. It doesn't have to be exactly between before and after, but even, you know, one month into your after something changes and it's going to potentially affect all of the data that you have in the after. 

Yeah, and these things can be completely outside of your control, right? 

Right. 

We'll talk about this later, about COVID, what about pharmacy shortages, right? 

So there's so many different things. Yeah, and those things are unplanned, but there's even planned things like, hey, somebody does a QI project and they figure out that maybe we should do X for our patients, and they implement X because they think it's the right thing to do for the patients in the hospital in the middle of you trying to do a pre-post study and that can dramatically affect the results that you're going to see. 

Yeah. And so this is not a criticism though, you know, I'm aware it might come off as such. Truly thankful for that effort that went into executing this change, gathering the data, reporting it. But for you Todd, your other hat is one of your other hats is medical director. So if you're approached as a medical director of ICU about this change, would you have thought, hey, but this would be better off as a prospective randomized trial. You know, a consideration is both your priors on the topic. Do you think the existing evidence is enough that randomization would be harmful? And then also the effort and time that it takes to do that. 

First, let me say the medical director hat that I have doesn't fit very well. 

Too big or too small. 

Yes. The answer to a bunch of these questions is this would be way better done as a... I shouldn't say way better done. That's not fair to the authors. This would be a higher level of evidence if it was done as a randomized trial. There are logistics and doing it as a randomized trial that make it probably harder to do as a randomized trial. And so this gives us data, I think data from what the logistics allow. and some data that you can at least look at and say, you know, how do I think that might be implemented in my practice? What might that do for outcomes in my practice? And should I, should I be working to try and do that? Yeah. I mean, part of this is right. Like someone had to be excited enough about this, that they wanted to implement this as a policy, right? By the time it reaches your desk as the, by the medical director, like I think that ship has just sailed a little bit as far as like the momentum. Yeah. I mean, the other, the other, and I don't know this for this are sometimes the only way you can get data on something is if there's not equipoise, then if everybody in this... hospital, this environment where this trial was done, thought, you know, we have to go to a phenobarb based protocol because what we're doing with our, with our DTs is, uh, our alcohol withdrawal is just not working and we need to move to it. Right. You can't really do a randomized trial because the people that are in the non phenobarb arm are, you know, going to get different treatments and they're going to get crossed over and contaminated and a lot of that stuff. And so in those situations, you may end up with the best we can do as a, as a pre post because there's no equipoise. we're moving to this, but you know, to do a pre-post, you still have to set it up to set to collect data, get the data, analyze the data, you know, all those things. 

Yeah. I mean, that's good foreshadowing a little bit for when we're talking about TRICC. But so this is a pre-post study of phenobarb based protocol compared to benzodiazepine based protocol. So for a long time, we've been using benzodiazepines for alcohol withdrawal. I'm going to oversimplify this, but... The alcohol is an indirect GABA agonist in the central nervous system. So when you remove the alcohol withdrawal, you have a relative GABA deficiency when your body needs to create its own and build up that period causing the activating symptoms of alcohol withdrawal. Benzodiazepines and barbiturates both impact that GABA pathway to ease that transition. Barbiturates are older, they're older drug and they are a more longer acting drug and had given way to benzodiazepines. I think this is mostly driven by safety in the outpatient setting. different impression of that or why is, why are we interested in switching now? 

I think it's safety. Barbiturates, especially phenobarbital has been around for such a long time that it's cheap. There's not much cost for phenobarbital. So you could say this is also a cheaper protocol and it's saving the hospital money or saving the patient money or, you know, whatever, but it's not like non-phenobarbital approaches to alcohol withdrawal, largely benzodiazepines are ridiculously expensive and we're spending a million dollars on a patient for, for benzodiazepines. I think it comes from the fact that we're looking at trying to improve outcomes in these patients and most of the improvement in outcomes as you said come from avoiding safety problems. So avoiding things like having to intubate the patient because I gave them enough benzos to make their alcohol withdrawal symptoms better but so much that they're not very responsive and you know I now need to intubate them. 

Yeah and there's also this concept that I'm not super familiar with but talking about like benzodiazepine resistant withdrawal and thinking that phenobarb and again I didn't get into details hitting a different part of that pathway as a direct receptor agonist could be beneficial. 

It makes sense. The other I guess potential reason that people may be looking at this is it easier to administer? 

Yeah, a longer duration of action. 

Correct. And does that make it easier to control patient symptoms and does it make it easier to administer you know less intense for the bedside nurse less intense for the provider, et cetera. 

Less peaks and troughs. So they're not getting that super obtunded sedation from the boluses of benzodiazepines more frequently. 

Look at Eddie bringing in pharmacokinetics into the podcast. 

Well, it was the last time we talked about pharmacokinetics. I got so embarrassed. I went back and reread that chapter in my book. So, so who are these patients? Patients were admitted to a tertiary academic medical center from January 2019 to February, 2022 with a primary diagnosis of alcohol withdrawal. And they split those at baseline into. complicated versus severe, where severe was DTs, delirium tremens, which was basically just defined as any delirium, or withdrawal seizures. These patients were found via an ICD code search and the use of alcohol withdrawal order set and then manually reviewed for exclusions. They have a flow diagram, which I guess technically is not a consort, but same idea, where it looked like 353 patients were identified and 251 were excluded, mostly because if you're collapsing categories, alcohol withdrawal wasn't the primary reason for their ICU admission. 

I will say, as you kind of outline this, that it brings up another potential risk for bias in these studies, which is that are the pre and the post population the same? Are you comparing apples and apples? And oftentimes, the way we, especially if we go back and look at these in a retrospective nature, the way that we identify patients has to be different in the two arms. But because the only way to identify patients is different in the two arms, you may end up with slightly different patient populations or dramatically different patient populations. And then you have a harder time knowing is the difference in the outcomes that I'm seeing between these patient populations, the difference in the treatments that they got or a difference in... some of the baseline differences that they had in the apples and crab apples comparison. 

Primary outcome here was ICU length of stay. Secondary outcomes included intubation, hospital length of stay, other medications for alcohol withdrawal, restraints, ICU readmission. They included their power calculation too and looking for effectively a one-day shortening in ICU length of stay. You know, I'm not as familiar with pre-post studies. I appreciated them including their power calculation. But that aside, is ICU length of stay the right outcome here? I think so. It's not really a disease state in which I think mortality is the right barometer. 

No, I don't think it's mortality. You know, some of the pre, there've been a few previous studies of phenobarbital in this area, not randomized trials. And then I think maybe one randomized trial, but at least one of those used intubation rates as a, you know, did you get intubated as the outcome? I think I see length of stay as probably a reasonable outcome. Honestly, it scares me a little bit when I always think of it for my own studies because it's- really, really variable. And oftentimes the ICU discharge is not entirely related to does the patient actually still need a critical care and related to bed availability and social stuff and that sort of stuff. And so that, that makes me a little leery of trying to use it as an outcome. But ultimately, if you look at resource utilization in a, in an institution, whether or not that ICU bed is free for the next critically ill patient for me to take care of is a huge deal. 

What was their intervention? So. Well, September, 2020, about 20 months into the observation period, this institution made a switch from a benzodiazepine driven protocol to phenobarbital protocol. The benzodiazepine protocol was summarized as lorazepam. It was one milligram every two hours PRN for a CWA score of six to eight, two milligrams every two hours PRN for a CWA score of nine to 19, and then two milligrams. every hour as needed for a CWAS score greater than 20. The phenobarbital protocol kicked in for a CWAS score greater than 10 and was effectively a 260 milligram bolus IV followed by subsequent 130 milligram doses every 15 to 30 minutes until you got an adequate response, which wasn't defined, got to 15 milligrams per kilogram of ideal body weight or four totaled subsequent doses as the max dose. There weren't any subsequent doses beyond that loading dose, but the protocol suggests that Haloperidol, dexmedetomidine, melatonin, or quetiapine are other agents you can use. Any thoughts on that protocol? It's pretty spot on with up to date dosing recommendations for phenobarbital. for the initial dose, but you and I, and I think as well as up to date, add a taper to this dose. 

The other thing I'll say is this, is that because the protocols are slightly different with when they dose for CWAS scores and that sort of stuff, you're really comparing here to strategies for the treatment of alcohol withdrawal syndromes. As opposed to the medications. Right, it's not really phenobarb versus benzos. It's sort of a phenobarb-based, and they say that in the title, phenobarb-based protocol versus Benzodaisapine based protocol. Yeah. And there's no way to do that to make that comparison better necessarily, unless you wanted to use drugs with different half lives at the same durations, which would probably be not right. Yeah. I mean, you could, I can imagine that you and I could sit here for two hours and come up with a blinded way to do it with placebo given at one hour in the Benzo group, sorry, in the phenobarb group, because you don't need it yet. And Benzo's given at one hour and then. you know, eight hours later, a placebo benzo and a phenobarb thing, but it would be, it would be highly complicated. 

I think that's fair. So we'll get into that when we talk about their table one baseline demographics. It's a retrospective analysis. So P values are included. The median age here was 56, a 75% male, mostly white. There was a difference in BMI by two points, which is close to statistically significant, but I'm not sure it's probably super clinically important. About half the patients had delirium, which took them out of the uncomplicated quote category. getting a little bit into their separation between groups and what we were talking about before, the median dose of lorazepam and the benzodiazepine group was 21 milligrams and the median dose of phenobarbital in that group was 520 milligrams. So doing some rough math, the time it took to get the median amount was about one hour in the phenobarbital group, right? So 260 milligrams initial dose, then two 130 milligram doses for a total of 520. That was compared to somewhere in the 20 to 30 hour range for a one to two milligrams every one to two hours PRN to get up to 21 milligrams. And this was doses given in ICU, not additional doses given on the wards. So there was a difference in time in just the amount of drug that they were given. 

Yeah, I mean, the drugs kind of lend themselves to that. So I'm not entirely surprised. 

But yeah, I'm just saying like, if your outcome is ICU length of stay, kind of stacking the deck a little bit, right? 

Yeah. 

One protocol is over a day before the other. 

Yeah. 

That's the type of analysis we love to hear from you, Todd. So their results, their unadjusted outcome, certainly seemed to favor phenobarbital. The primary outcome of ICU length of stay was 1.5 days compared to 2.3. And that change was, again, reflected in their hospital length of stay of 3 versus 6 days. The highest CWA score was also lower, 16 compared to 21. There was less use of restraints, 37% vs 57%, and less patients who received mechanical ventilation, 2% vs 20%. It also appeared that the use of all adjunctive medications, any adjunctive medication at all, favored the phenobarbital group. Their adjusted model for their primary outcome of ICU length of stay was adjusted for age, sex, BMI, comorbidities, history of alcohol withdrawal, last drink, baseline COS score, and severity of withdrawal at baseline. From that, it looked like the phenobarb vs. benzo, group delineation, sex, and severity of alcohol withdrawal at baseline impacted the length of stay. So, I mean, the results all look pretty promising for this phenobarbital protocol. That you had stacked the deck in favor of. 

Yeah, I mean. No, I agree with you. You know, I mean, I think based off of, as you've said, some potential biases and some methodological potential biases that overall... I don't know, phenobarb looks pretty good here. 

Yeah, and you're making fun of me for the stacking the deck reference, but... That's not a bad thing. If that's a reality of using these medications and it does decrease length of stay, which we're saying is important from certain perspectives, that's great. 

Yeah. It opens, it opens the manuscript up for a critique where somebody will say, well, I don't, that's not how I use my benzodiazepines. I don't, I don't have that extra day. And so when you compare it to mine, maybe it won't be any different, right? Obviously there's, there's no perfect study and there are things that people can critique it on all day long. If you're, if your usual care group is. not the usual care in the study, then you might not see a difference. Yeah. And that's true for randomized trials too, right? If you want to come, if you want to, if you want to critique the critical care group in the randomized trial, then you know, you may not think that the outcomes, even of a randomized trial are applicable to your practice. I think your point is well taken in that from a pragmatic standpoint, even if the protocol as set up stacks the deck a little bit, the fact is, is that if you can use a protocol that stacks the deck a little bit, and gets the patients out of the ICU a day or more sooner, then maybe I should be using that protocol, even if that protocol does stack the deck a little bit. It's sort of like, if I can come up with a protocol that gets a patient extubated sooner because it allows me to check to see if the patient should be extubated sooner, you could say, well, but you checked in that group sooner and that's why they got extubated sooner. And I would say, great. Yeah, they got extubated sooner. let's do it, let's use a protocol that checks sooner. 

Yeah, so the question here has to be, right, like is there a downside to this, right? So are there more, ICU readmissions because you're getting them out too quickly. There's, you know, you're talking about phenobarbital might be a more dangerous drug with this long half-life, so is there anything with sedation or respiratory depression? And it didn't seem to have any of those things. 

No, I agree. It's a small study. There's only 100 plus, you know, over 100 patients. So safety signals in smaller studies are a little harder to find because the rare safety signal doesn't show up in 100 patients or 50 in each arm. 

But even so it's still like it did show up and it seemed to favor phenobarbital. So at least it doesn't appear to be like trending in that direction or an event that you just have no data on. 

Yeah, completely agree. These data in this study said, suggests to me this phenobarbital protocol safe to use and is at least as safe as the benzodiazepine protocol and is something that if I think it would be beneficial for my patients I shouldn't necessarily worry about the yeah but what about the safety side of it. 

Right so one more one thing before I get to that because we referenced this. And I think soon we'll get to the point where we don't have to think about this, but they also called this out in their discussion. So just that this happened during COVID. And there was a note that cases were redirected to other facility based off of medical ICU. bed availability. There's also the consideration that what if they needed to make this change because of COVID, like a less nursing intense protocol and a time where donning and doffing into every single ICU room was not an insignificant amount of time. And I'm on a roll here, so let's go a little bit deeper. And there are reports that- 

Like butter, Eddie, like butter. Yes. There's reports- You are on a roll. 

That alcohol sales increased during the pandemic. So- Maybe the populations are even different at a baseline. Like the person who's getting alcohol withdrawal before the pandemic and during the pandemic may be different. 

Yeah, maybe. Does this factor at all for you when you're thinking about this? I mean, you always have to keep it in the back of your mind, sure. Based off of the way this trial is designed, then you have to take the next step and say, okay. Do I think that phenobarb is like a great treatment for these patients with alcohol withdrawal, but it's not very good for, you know, a different demographic of patients with alcohol withdrawal. I mean, it's a syndrome where they need some, they're withdrawn from some GABA stimulation and this provides GABA stimulation. So I don't know if you would say, you know, Hey, the patients in COVID that were doing alcohol withdrawal were younger and maybe therefore, you know, phenobarb is better for the younger patients. I don't know. Is that really true? 

The medians didn't change and I don't think anyone's... 

The median age. 

Yeah. I don't think anyone's coined this term, but we can be the first, you know, COVID alcohol withdrawal. 

CAWS. That might be your cause. Um, I, but I think in general, while I have that in the back of my mind, it's not something that I think completely invalidates the results of the study. Yeah, I agree. I mean, they were, I think rightfully so mentioned in the discussion. Uh, but I don't. I think it makes too much of a difference here. 

We've referenced already that you and I, we kind of already use phenobarbital for alcohol withdrawal in our practice. I do, although let me be explicit, and you can state whether this is true for you or not, but let me be explicit in that I don't necessarily use this protocol. 

No, you mean you're talking about that single dose and then the rapid up-tytration? 

Well, even the phenobarb without any benzos, because at my institution, patients get benzos, and they often get benzos before they get to me in the ICU, and I say, phenobarb, but it's often sort of a phenobarb added to some benzos and not necessarily a phenobarb first and phenobarb excluding benzos protocol. Yeah. For what it's worth. 

So you're saying our protocol here is still to use benzodiazepines with phenobarb is the adjunctive measure. 

Yeah, absolutely. 

Yeah, I think a lot of this also occurs outside the ICU too. I think giving a dose of phenobarbital, giving kind of like that, I don't know, like that basal, like to draw a comparison to insulin, giving that kind of basal GABA allows people to have CUS scores that allow them to not come to the ICU in the first place. 

Yeah, everybody needs some basal phenobarbital. Just put it in the water, right, like statins or GLP-1s now. That's how I'm actually gonna get rich. a GLP-1 and phenobarbol on the same pill. You'd probably the... Take one every single day. The most popular person. The paper here does highlight that the prior literature even using ICU length of stay is mixed. For you as individual provider and then also from like a hospital policy perspective, what are your thoughts and preferences here? 

I mean I think this to me says that it's a reasonable option to use. It doesn't say to me that it's the only option to use. we should fairly strong handedly move our protocol to a phenobarb based protocol. Not to get rid of benzos. But yeah. And, you know, the scientist side of me says this needs a large RCT. Now, having said that, talking to some people who've tried to do an RCT, I think the RCT was of dexmedetomidine versus benzos and not phenobarb. Consent in the RCT is actually a difficult thing. And it turns out this is all a social situation. It turns out that by the time many alcoholics have hit the hospital with alcohol withdrawal syndrome, they've burned the bridge of many of their next of kin. And you know, the patient has delirium tremens, so they're probably not of the mind that they're able to give their own informed consent. And you're trying to get a surrogate from a next of kin. And a lot of the time, the next of the kin is just kind of over it, right? And they don't want to be involved in John's life anymore. And John's, you know, they've tried to help John 57 times. John's burn that bridge every single time. So I think there are some logistics issues in trying to do a consented randomized controlled trial in this population. 

And it's not like you're telling this family member that this is going to help them long-term with their abstinence from alcohol, right? This is just in the acute setting. 

Yeah, exactly. Let me bring us to this. I think that in a lot of practices, benzos remain the primary treatment for patients with alcohol withdrawal syndrome. I think that they work. I think there are patients and you called patients, you know, alcohol resistant or alcohol refractory, sorry, not alcohol refractory, benzo resistant or benzo refractory alcohol withdrawal. And I think it just points out that they don't work perfect in everybody. And I think having something like a protocol where you could use phenobarb as an adjunct or by itself in patients that either you were worried benzos weren't going to work or benzos weren't working in, is a reasonable thing to do in your practice. 

We're heading back into our time machine and going to be talking about TRICC or a Multi-Center Randomized Control Trial for Transfusion Requirements in Critical Care. It was published in the New England Journal of Medicine by Herbert et al. in February 1999. 

What were you doing in February of 1999, Eddie? 

I was turning nine. 

Uh, is that third grade, second grade? 

Third or fourth, right? 

Second grade for you though? 

Yeah. 

The second year of second grade. 

Second year of second grade. 

That's it. Perfect. 

In our typical old second article fashion, we're going to hit the high points and give away to discussion. But first, this is an A plus acronym. TRICC, T-R-I-C-C, transfusion requirements in critical care. We've really gone downhill since Trick Todd. 

It doesn't bother you that they misspelled trick? A little bit. I was, I was going to say that I had it out here in my outline and I took it out saying it would be nice if it was T R I C K, but it's so nice. Every day, like you use the preposition in T R I C C. Man, I love it. That's great.

 Why, I mean, you know, go into maybe a song that people have heard on the radio, why can't we do K A R E care for kids? 

Is it, it's C A R E care for kids? 

Well, no, the song is K A R car for kids. 

No, it's C A R cars for kids. 

Oh, I thought it was K. Nevermind then. 

So this is one of those landmark trials in critical care that we don't typically think about since we've all adjusted our practice to this. 

Wait, I got another one to try. What about like E K G? Where the K is German for cardio. Can we do German for intensive care? 

Unless the German word for care starts with a K. I don't know. especially from these Canadian trialists, right? 

Yeah, Canadian with a C, right? Not a K. 

No comment there. 

So I don't even think Todd was practicing by the time this trial came out. I was actually a resident. 

Yeah, I was a resident. February of 99 I was in PGY2. 

Dictating all the care for your critical ill patients in the hospital. 

No, we still used handwritten notes at that time. 

And cutting and pasting EKG strips and pasting them in the chart? 

Of course. That's where cut and paste came from, right? 

Yeah, cut with a K. 

We'll not bury the lead here. We all know the fallout from this trial. We transfused her a hemoglobin threshold of seven grams per deciliter compared to the transfusion threshold of a hemoglobin of 10 grams per deciliter, which was studied here. You were practicing here, Todd. Do you, are you aware of the rationale for this? Why did we ever think that 10 was better? 

I was, I was a PGY 2. I was like trying to keep my head above water.

 You were cutting and pacing with a capital K. 

Yeah. Trying to keep my head above water, which I think I did moderately well. So ultimately the direct answer is no. idea what the rationale was. At the time, I had no idea what the rationale for the trial was. 

At least in the introduction, they have a couple of references to... what seemed like retrospective studies, one of them from this author group that suggests that critically ill patients with lower hemoglobins do worse. Now I didn't pull up that article to see what it was, but it kind of strikes me as, you know, sicker patients are sicker, right? Sicker patients have low hemoglobins and sicker patients do worse, kind of like as the confounder. 

Just so you know, sicker has both a C and a K in it. 

Thanks, Todd. 

I do, I will say that at the time, and I think still are teachings, that a hemoglobin of 10 is sort of the optimal minimum threshold for oxygen carrying capacity for a hemoglobin. And if you need to increase oxygen carrying capacity and the patient is anemic with a hemoglobin less than 10, giving them a transfusion and moving it to 10 is one of, if not the most efficient ways of increasing oxygen carrying capacity. Part of me wonders if there's some rationale in it also of get to a point where our oxygen carrying capacity is plateaued a little bit and quote unquote optimized. 

There's also risks of transfusions, right? So that's the kind of the competing risk, I guess, where you have decreased oxygen carrying capacity, but then the risk of the transfusion, including overload, including lung injury amongst other complications. And I think it was interesting, this is also around the time when there was concern about the safety of blood transfusions from like a transmissible disease perspective. Now as far as like things that I don't really know how to explain maybe you can why seven? Why did we use seven as opposed to six or eight? 

Well, it's a nice clean number. It only has you know kind of two Lines one angle in it.

 What why was six afraid of seven exactly?

 Because seven eight nine. 

There you go. 

Yeah, you are starting the dad jokes really early for having a one-year-old 

I mean, I've been doing dad jokes since my baby was like negative five years old. 

So since before I was a dad, yeah. Awesome. No, I don't, I don't know why, why seven was chosen here. It's a good question, but not one that, that I'm old enough to answer. 

Yeah. If anyone knows the answer to that question or it just has a better supported guest, let us know at, you know, ICUEdandToddcast@gmail.com or @ICU_cast on social media at X Twitter. Anyway, the primary outcome was a 30 day mortality and it was not different though. It trended towards a. benefit with the restrictive strategy 18.7 vs 23.3% with a p-value of 0.11. From a complications perspective, the only one that reached a p-value of less than 0.05 was less pulmonary edema and the restrictive group 5.3 vs 10.7%. This study spawned a number of other studies in different populations which to the best of my knowledge are all similar in that no study TRISSin septic shock and TRICS 3 in cardiac surgery. 

More acronyms you like? 

Less good than TRICC, right? 

Well, the sequels are always not as good as the original movie. 

For me, the story here is pretty straightforward, which was TRICC compared 7 versus 10, but excluded patients who were actively bleeding and who had quote unquote active coronary disease, although I don't think they ever defined that, but active coronary disease. as our threshold except in patients who are actively bleeding, in patients who were worried about cardiac stuff. And then subsequent studies, TRIX, an actual study in patients that have upper GI bleeds, demonstrated that seven was at least as good as 10, if not better than 10 in those populations also. And so then I think it got expanded. The one population that I have seen data, and it's not a randomized trial, it's actually a cohort study that said maybe a higher transfusion threshold, sorry, a lower transfusion threshold, so a higher hemoglobin. target may be better in was in oncology patients. And again, it's not a randomized trial. It's actually not something I've incorporated in my practice because I don't entirely believe that I know the mechanism behind it. But I think it's one population that if you said, is there a population that you still might have a question about? Do we know the optimal transfusion threshold? Maybe patients with solid malignancies may be that population. Yeah. But I'd like to see a trial before I move my practice that way. 

Right. We were just talking about like sicker patients are sicker. It wouldn't, it wouldn't be shocking to me if you said the patient who has a solid organ tumor with a lower hemoglobin is sicker than the patient with a higher hemoglobin, right? For sure. 

This trial is also interesting. I'll admit that I've never read this article before prepping for this podcast. Their power calculation was for 2300 patients and they stopped early at 838 for quote, a variety of medical administrative reasons. that aren't expanded on. That's like 35% of their intended sample size. That would be crucified if that were today. It's not diminishing the achievement, but just more commentary on the advancement and evolution of evidence-based medicine. 

Yeah, I think a couple thoughts on that is one is I think you're exactly right. If you went back to February of 1999 and you were reading this for the first time, you and you were in today's critique level, you would say, wow, I don't like that at all. You know, they're more likely to have found a false positive and you just have stopped and you know, that sort of stuff. I will say that now that we have a bunch of data post this that have kind of been very, very supportive of this and have found very similar findings. It bothers me less that they stopped. Yeah. 

So the next time I'm reviewing a paper, I'm just going to like look into the future and see what the other data show. You hop in your time machine. You go to the future, you see what the other data show, and then you go back. 

Obviously. 

Yeah, it's pretty simple. 

The other interesting point that I thought was here, that they had more patients or providers refuse consent than were actually enrolled. So 838 patients were enrolled, 598 patients, the physician refused, and another 603 patients, the patient or family refused consent. Maybe that's the admin reason for stopping? 

Yeah, if you're gonna enroll another 65% of the patients 65% of the people you approach say no, you know, now you're going to have to enroll. I don't know what that math is, but it's going to be, you know, 200% more patients. They had their concert diagram. 

They'd already screened like 6,000 patients and they only were at 800. 

Right. So, I mean, as you, as you alluded to earlier, that could be a suggestion that providers didn't have equipoise in this area. It's also kind of a time when. clinical trials in the ICU were less commonly done. And it could be providers just weren't comfortable with it and or patients just weren't comfortable with, my loved ones really, we hear this a lot, my loved ones really, really sick. I'm not sure I should put them in a trial, but you're right. It does suggest from an administrative standpoint that to get your next 1,400 patients, you're gonna have to look at 13,000 patients to get there. 

They had already enrolled for three years to get to that 800. So that's nine years total for enrollment. That's a long time. I don't blame them. I'm just that I just said it was interesting 

So with all those critiques, are you telling me you don't believe this study? You want to do this study as a follow-up study for this as your, your. 

No, if I went in, if I went back in time with my time machine to 1999 was a reviewer for this study, I would just hop back in my time machine to 2023, look at the evidence and say, yeah, I know this is a great study. 

It's not really use in asking you how you'd apply this data, but I have a question here. Why not six? Why not five? Are you interested in RCT that looks at a lower target, a more restrictive target, quote unquote. 

If somebody was doing it, I'd be intrigued. to know the results of it. Obviously, there is a J shape curve here, there has to be a J shape curve here, right? Right. In the fact that, you know, one, zero is gonna be bad. Yeah, but take zero out of there, you know, even one is gonna be bad, right? You know, and so where is where is that probably gonna be bad? Where is that? bottom of the J or you or whatever letter you would like to make it, flip it upside down and make it an N if you'd like, depending on if you're doing mortality, survival, good outcomes, bad outcomes. Right. It's either a hill or it's a valley. 

Isn't there, isn't like two hills, right? Is it M? 

Could be. Yeah. And bimodal. That's what we call that in medicine, Eddie. Bimodal.

You're already using letters Todd. Now you're going to get on me. 

I think the thought process probably is, is that the bang for the buck between six and seven isn't big enough that, you know, have to put in a ton of patience to see if there's a difference there and it just hasn't been something that's been important enough for somebody to do. 

But this is just a strategy, right? So like you're saying, I think one of the arguments I heard of there's like you can't do six versus seven is because you wouldn't have like a good separation between groups. But if your strategy of targeting six doesn't give you a large separation between groups as your hemoglobin level from a strategy targeting seven, then why not target six, right? And then if you end up in the same place. 

Your point being that if you're targeting seven, not every patient's gonna have seven. And in fact, when you look at mean hemoglobin concentration, it's eight in that group. And there's a percentage of patients that I can't find in the manuscript, but there's a percentage of patients who never get to seven and never get a transfusion. And no matter which group they get randomized to, those patients aren't getting a transfusion in either seven or six. So there's going to be a group of patients that. there won't be any difference in treatment, whether you targeted a seven or a six as the hemoglobin. And that I think goes to the logistics of doing that trial mean you're going to have to enroll a lot of a lot of patients and looking for a signal that is relatively small and is going to be non-existent in some patients because they're not even going to get differential treatment effects. And the question is just, is the bang worth the buck there? Is the squeeze worth the juice? You know, those phrases. 

Juice worth the squeeze. I am NOT advocating for that. I'm just you know providing the devil's advocate there But anyway, that's all we have for episode 16 of the ICU Ed and Todd cast you have any questions Yeah, 16 you ever think we get here? Not really Yeah Well if you want to tell Todd that he got it all wrong because he just got that wrong Anything else you want us to talk about in the future you can hit us up at ICU Ed and Todd cast at gmail.com You can also hit us up on the social at ICU cast at Ed Chien at todrise underscore ICU. Thank you Todd again for your insights. Thank you again to the study teams for all their hard work. Congratulations on another completion of a trial. Thank you Mike Gannon for the intro outro music. Thank you everyone listening and we will see you next time. Let's go save some lives. 

Let's go give some phenobarb. 

Eddie, I think you're too young for this, but there used to be a charity, and actually I think it's still around, where the goal of the charity was to get you to donate your car. And the charity was for Kids and the benefits the proceeds went to kids cars spelled with a C time, but not in the charity It isn't C So they had a nice little jingle and they made it a K So that you would remember it in the jingle and even old guys like me still remember this and so the jingle would go 1-8-7-7 cars for kids K a R s cars for kids 1-8-7-7 cars for kids as 

Even old guys like me remember this, right? 

I can't remember the end part of it. Give your car today or something like that. 

You know Todd, even young people can remember things with Google as you're using right now. Shoot, I remember things from like the 80s because I can Google it. 

Donate your car today. There we go. 

15 seconds later. 

So the song went, 1-8-7-7 cars for kids. 

I think we got it, Todd. 

K-A-R-S cars for kids. 

I'm going to cut all this out. 

1-8-7-7 cars for kids. 

It's fading. 

Donate your car today. 

This podcast is made for educational purposes. The content provided in this podcast and in any length of material is not intended and should not be considered as medical advice and should not be used to diagnose or treat any medical condition. It's inevitable, but we try to stay away from opinions, but all opinions represent our own and not of any entity that we work at. Please keep this in mind as you enjoy the podcast.