Nudges for Prognosis and Comfort Care in the ICU: Kate Courtright, Scott Halpern, & Jaspal Singh

Eric 00:00

Welcome to the GeriPal Podcast. This is Eric Widera.

Alex 00:03

This is Alex Smith.

Eric 00:04

And Alex, who do we have with us today?

Alex 00:06

We are delighted to welcome back returning guest Scott Halpern, who is a pulmonary critical care physician and newly minted palliative medicine physician and researcher and director of the PEAR center at the University of Pennsylvania. PAIR stands for Palliative and Advanced Illness Research Center. Scott, welcome back to the GeriPal Podcast.

Scott 00:28

Always great to be with you guys.

Alex 00:30

And we’re delighted to welcome back Kate Courtright, who is also a pulmonary critical care and palliative care physician and researcher and core faculty at the Paris center at the University of Pennsylvania. Kate, welcome back to GeriPal.

Kate 00:42

Thank you for having me again.

Alex 00:44

And we’re delighted to welcome Jaspal Singh, who is a pulmonary critical care and sleep medicine physician who practices at 8 Atrium Health and the Wake Forest University School of Medicine. Jaspal, welcome to the GeriPal Podcast.

Jaspal 00:58

Very happy to be here. Thank you for having me.

Eric 01:01

So we got a good topic today. We’re going to be talking about a JAMA IM paper that just came out not too long ago, a nudging clinicians promote serious illness communication for critically ill patients. We’ve done a podcast already on nudging. Where was it?

Alex 01:13

The nudging podcast We’ve done a couple I think.

Eric 01:17

We had Scott and Kate talk about a nudge on default palliative care. We also had a podcast with...

Alex 01:25

One with Jenny Blumenthal Barbie and Scott about what are nudges in general.

Eric 01:29

We’ll talk a little bit about that, encourage you guys to listen to both of those. We’ll have links to that in our podcast notes. But before we dive into this topic, Jaspal, do you have a song request for Alex?

Jaspal 01:44

Well, I thought we requested Imagine by John Lennon.

Eric 01:49

Why did you choose Imagine by John Lennon?

Jaspal 01:52

Well, it’s such a nice song about compassion, kindness, and the idea of us thinking of a world where the right things happen consistently all the time. And that’s the culture and which is what we’re trying to accomplish with these types of nudges.

Eric 02:08

Great.

Alex 02:09

I love it. And we could all use a little John Lennon imagine right now. Here’s a bit.

Alex 02:22

(singing)

Eric 03:30

Thank you, Alex.

Alex 03:32

Thank you for that choice Jespal needed. Thank you.

Eric 03:35

So let’s jump into this. I feel like nudges and behavioral economics, it’s so hot right now. There have been a lot of trials. There was just a couple out on One was on polypharmacy using behavioral economics. Another one was on decreasing low value care. That was published also in, I think JAMA IM before we talk about what you did in this study, maybe Scott, I’m going to turn to you. Can you give me a reminder what a nudge is?

Scott 04:07

Sure.

Eric 04:07

I think it’s…oh! Alex just hit me! [laughter]

Alex 04:10

Maybe that was an elbow, not a nudge. Sorry. Go ahead, Scott. [laughter]

Scott 04:20

This response, you can’t hit me. So that’s good. I think it’s good to place nudges in context. So there are a lot of policy levelers at the disposal of any party that’s trying to influence behaviors or outcomes. At one end of the spectrum you can provide decision makers with information that doesn’t work very well in most contexts. At the other end of the spectrum is you can totally remove certain options. So you could think about, you know, formularies at the hospitals we all practice on. I can’t prescribe certain drugs, I’m just not allowed to. And that works really well.

But it can be seen as constraining the autonomy of the decision makers. Behavioral economics sort of operates in the middle of those two poles and can be defined as in any feature of the environment in which a choice is being made or in which or the way a choice is being framed that can foreseeably influence the probability that a specific choice is the one that’s made without removing any choices or materially changing the incentives to make one choice or another.

Eric 05:43

So if you’re like a Starbucks and you have a small and a large coffee, you want to get more people to order the large coffee, you add a super large coffee. So you haven’t restrained or constrained choice, you’ve just driven more people to buy the large coffee because it’s kind of the middle option.

Scott 06:02

That’s exactly right. And that would be what we would call like expanded choice set nudge. And in fact that’s a nudge that’s kind of near and dear to Kate’s in my heart since it was the topic of her master’s thesis many moons ago.

Eric 06:19

Oh yeah.

Alex 06:20

Can I what’s the healthcare equivalent to the Starbucks coffee?

Eric 06:24

Yeah.

Kate 06:26

I invited patients with end stage renal disease to complete a short advanced directive, a medium length advanced directive with some more options and the really full, lengthy living will advanced directive with all the.

Alex 06:40

Legalese, the Ayn Rand version and more.

Kate 06:44

Patients were willing to complete advanced directive and picked the middle one.

Alex 06:49

That’s good.

Scott 06:50

Yeah. And these multiple choice set nudges, and some of the nudges we’re going to talk about today are they’re nudges that can often be very useful, but they’re not kind of like the granddaddy nudge, if you will. The granddaddy nudge is the one we talked about on our last podcast with you where you change the default option. The events are conditions that’ll be set into place unless some active choice to the contrary is chosen. So that’s by far the best studied. It works in almost every context in which it’s been studied, et cetera. What we’re gonna be talking about today are more nuanced nudges for which the evidence base can be considered to still be evolving.

Eric 07:35

Yeah, so I always remember back to your advanced directive nudge study where you, you changed what the default is for the advanced directive, Right. So it was already X’ ed out, like comfort care or dnr.

Alex 07:50

Or you got one that was selected.

Eric 07:52

It was selected.

Alex 07:53

Comfort Care was selected.

Eric 07:54

Comfort Care, you either got one that was selected and if you didn’t want it, you had to cross it out, or like full treatment, full code was selected and if you didn’t want it, you cross it out. And what I remember from that, correct me if I’m wrong, Scott, is that if you got that comfort focused one, like 80% of people wanted the comfort care focused one, but if you got the other one, it was like much like it depended which one your preferences, your end of life preferences, depended on which advanced directive you got. Did that sound right? Am I summarizing that correct?

Scott 08:24

Yeah, you’re summarizing it perfectly. It was actually two separate randomized trials that had the same general conclusion that you could meaningfully change the proportion of people who made choices about their future care in comfort oriented or life extending oriented directions by just flipping the way the choices were framed. And those effects were true, even though we told people that we were doing this at the end and they had.

Eric 08:54

A chance to change it and nobody changed it.

Alex 08:56

Yeah, not nobody, but fewer. Almost no one.

Scott 08:59

And so the key insight there was, I think, which is relevant here too, is sometimes people are making choices about things that they don’t have, like pre existing, really well ordered preferences about that. Like they might have preferences in sort of a gestalt sense, but they haven’t given Them a lot of thought, and those are areas that are kind of ripe for nudging.

Alex 09:23

Yeah. And some of this might just be like. It takes effort to overcome that default. Right. And so this morning, I came in and there was coffee grounds that were hot, but there was no coffee left. And so what I did is I made some more coffee grounds. I added them on top of the existing grounds, and then I poured more water in the coffee maker. And then I told Eric what I did. And so it was my laziness that did that. And then Eric came back with a norming statement.

Eric 09:53

Dear Jerry Palliser, I’m searching for a new co host for the J. PAL podcast.

Alex 09:57

This is worse than anything he’s ever heard.

Eric 10:00

Heard of.

Alex 10:00

He was, like, dying inside because I had reused the coffee grounds. So my question is, is norming probably the most common nudge in our society? Inside or outside of healthcare? Would you guess?

Scott 10:16

I mean, I don’t even know what to call that, Alex. That’s just absurd.

Eric 10:26

This is what I should have done when Alex came in. I should have used the nudge. I could have said, most people who make coffee do not reuse coffee ground. So I’m nudging Alex using social norms to not do something that he should never do. Yeah. I’m not constraining choice. Right. I am just nudging him.

Scott 10:44

Does your wife know about this, Alex?

Alex 10:47

She will when she hears this, but yeah.

Eric 10:51

I just want to highlight the last time we had Kate and Scott on. I highly encourage you. We did another nudging trial where the deep default was for default palliative care consultations for high risk patients, and it significantly increased individuals who were receiving palliative care. And again, we can go into kind of what they found on the topic. Highly encourage you to jump in and listen, but I gotta ask. So you’re doing these nudging trials.

Alex 11:17

Wait, can I ask about nudging before?

Eric 11:18

Go ahead.

Alex 11:19

Okay. Because my wife yesterday, she said, what are you doing? I said, I’m doing a podcast about nudges in the ICU for critically ill patients. And she said, is that the word? The accepted word? Now, do we say nudges? Is that. Okay.

Eric 11:31

Behavioral interventions or.

Alex 11:33

I mean, there are all these terms?

Scott 11:35

Yeah, I think so. I mean, Cass Sunstein and Dick Thauer wrote a book entitled nudge. Right. Best New York Times bestseller for many moons. To reuse the statement I used before. I also think it’s useful in the sense that it is differentiated from sludge. So you guys referred to sort of the work Involved to overcome the default or whatever. Right. That. That is what we call kind of like sludge. Like, it’s. If you make it hard to make a specific choice, that’s sludge. That’s not a nudge. And so the fact that they end in the same sort of.

Eric 12:18

Okay, like a default makes it a little bit harder. Right. You have to expand a little bit more energy to. To make that change. Why is that not sludge?

Scott 12:30

Because a default has to be set in one. It oftentimes has to be set in one direction or the other. So you’re. It’s an apples to apples comparison of the effort. There’s always some effort involved. By contrast, you could make it so that, like the fact that a default is set is totally hidden and you’ve got to click through 27 different prompts in the EHR to overcome it. That would be smudge.

Alex 12:57

Okay, all right, this is good.

Eric 12:59

All right, hold on to that, because I got questions once we talk about the nudges that you did in this trial. But, Kate, I’m going to turn to you. Why did you decide to do this trial at all?

Kate 13:09

This particular trial, the Ponder ICU trial, we were, I’d say the overarching objective was to improve adherence to really best practice guideline, recommended communication with seriously ill patients in the icu. By that I mean those at high risk of death in the ICU or in the hospital or severe functional morbidity thereafter. And it’s been pretty well written about for a long time that that’s the right thing to do, but we don’t systematically or routinely do it. And so how do we, as Scott said, sort of use our levers to. We don’t think clinicians don’t do it because they don’t want to, or they don’t think it’s the right thing to do, but perhaps they just need it reminded, put in their face this is the right context and sort of give them that support and facilitation.

Eric 13:56

So what, what did you do in this study?

Kate 13:59

What did we do? We tested two different nudges individually and then we combined them. So there were four study arms, one in which we had no nudges, and then one for each individual nudge and one combined. And the nudges were focused on two areas of communication, one prognostication, in which we asked the clinicians. This was based on something called a focusing effect, whereby, you know, ICU clinicians in particular have a lot of data coming at them and maybe tend to focus on the blood pressure and the heart rate. And we simply wanted to focus them on prognostication and specifically death and functional morbidity, as I mentioned.

And so we simply asked them to do that, prognosticate six months for mortality and if alive, what they thought the functional status would be. We didn’t make them do anything with that information, just document it. We didn’t make them communicate it, but we focused them on it, which makes them more likely to then utilize that information. And then the second nudge was based on multiple recommendations from professional societies that for these types of patients, we should offer the option of care focused solely on comfort on the same footing as care focused on longevity. So continuing life support and comfort focused care are very reasonable options for patients with these potential outcomes.

But we don’t tend to offer care focused on comfort until we think there’s literally nothing at all left to do to extend this person’s life. And we sort of don’t talk to people about them in the same way. And so multiple professional societies have recommended this. So we leveraged a nudge that is called accountable. Justification has been used in multiple other contexts. But the idea here is simply that clinicians are trained to make decisions backed by reason. And so we simply asked clinicians, have you offered this patient and, or their family the option of care focused on comfort? If not, why not? And that’s key. That justification is a key part of it in that it’s a normal part of their decision making to have reasoning. And we’re simply asking them to be accountable for said reasoning, just like they would document why they did or didn’t give an antibiotic for a pneumonia.

Eric 16:16

So I can imagine for the accountable part, somebody would actually have to see it. Like it just doesn’t go into the ether if you put that reasoning.

Kate 16:25

Yeah, right. These were EHR documents in the patient’s chart built within the Atrium Health’s electronic health record, where they document other things.

Eric 16:36

Jaspa, you practice in Atrium Health, Right?

Jaspal 16:39

Right.

Eric 16:39

What, what is that? What did this look like actually for the ICU physician? Was this like a, a researcher coming in and asking these questions? It sounds like. No, it was built into the ehr.

Jaspal 16:50

It’s just depending on when you access that ehr. So basically the prognostication is happening in the background and like once a day or so this thing would prop up this little, this little button would pop up and sort of ask you to prompt through and just ask you a couple of simple questions and you had to respond and the ehr. And basically that was the extent of the intervention, it wasn’t sludge, I think like the way Scott described it. Right. It was fairly straightforward and it wasn’t very complex. And that’s where the beauty of this intervention was.

Eric 17:25

I gotta ask another question though, because I work in a healthcare system where when I open up the chart, a box comes up, I have no idea what it says. I just click, ok.

Alex 17:36

I’ve never read that box. I have no idea what it says.

Eric 17:41

No idea what it says.

Alex 17:43

So is there a sludge effect from having like one more pop up? You know what I mean?

Eric 17:47

Or fatigue pop up? Fatigue. Like there’s so many ops.

Jaspal 17:52

No, absolutely. And especially that requires a lot of education. But I think the fundamental premise here, and something I think is sometimes for your audience is probably innate. But for me, who’s practicing in North Carolina and not just at the main resource hospital, but also at the smaller hospitals, the community hospital, we look at this across the geography. A lot of our clinicians want to have these conversations, but they sometimes feel they don’t have permission, they don’t have the space, they don’t have the time, they don’t have the expertise.

And so you end up with this really complex phenomenon on a very complex problem of like people want to do well, they don’t know how to approach it and so they end up not doing any of it. Right. And so it kind of brings it up to up into your face to have these discussions. And yes, we did a lot of prep work because as you know, alert fatigue, coding billing fatigue, error. I mean there’s all these sort of pop ups that come up and we did think carefully about that, which took a lot more complexity to think through this. But you know, we managed to work it through and people got used to it and throughout the duration of the trial.

But what’s really cool is that we got insight not just from the main hubs of centers, the tertiary ICUs, like where I work, tertiary quaternary centers, but also the community centers that are, you know, 30, 40, 50 miles away from the main centers. And I think that’s where I think it’s got really interesting.

Scott 19:13

I just add one more thing on that score because I think this often comes up. I don’t want any more alerts. Get rid of all words. I already get too many. Agreed, we all get too many. But the solution to that problem isn’t to not test hypothesis driven novel alerts that are foreseeably designed to improve patient outcomes. The answer to that problem is to not allow prompts or alerts that haven’t.

Eric 19:44

Undergone such testing like the one me and Alex just clicked, but we have.

Alex 19:49

No idea what it says.

Kate 19:50

Yeah, we commonly at the start of these EHR trials have conversations with operations and informatics folks who tell us they won’t work because they have an average of 10% response rate to their alerts. And we are like, oh, they’ll work because we’re going to design them differently with science in mind. Definite. And we have much higher adherence and response rates. In fact, in this trial, you know, I was very proud of our adherence rate. As Jess Paul said, you’ve got two interventions in the middle of ICU rounds maximum or four questions if you had both combined.

But either way, you know, you’re interrupted, you’re doing something and your patients are sick, very busy or attending to multiple things at once. And we had 75% overall adherence, which is phenomenal in an implementation sense. But if you look at those who were just the way our eligibility criterion worked, if you were eligible for at least 24 hours, because if you were extubated or left the ICU shortly after you became eligible, the prompts would stop. Of course, we had almost over 90% adherence. So people did attend to this. It was not required.

Eric 21:07

And adherence is you, you completed, responded, responded to the prompt. So you put in your prognosis statement or if you got the alternative, yes, I did it or no, here’s my reason why I didn’t do it.

Kate 21:18

Correct. And I, I would agree with Scott. I don’t think there’s an example of greater than 90% or 75% adherence in the typical alert that kind of gets just turned on by a health system for, you know, all sorts of reasons.

Alex 21:33

So it seems like you had buy in. Yeah. Great.

Eric 21:38

And who are, who are the patients included in this study?

Kate 21:41

These were patients with a pre existing at least in the last 12, I think a 12 month look back of a chronic serious illness. Pretty much every organ system. We really tried to be very broad and were receiving mechanical ventilation for at least 48 hours continuously. And adults, the pretty broad criteria, but also very, very sick patients had in our preliminary estimation, we saw about a 35% hospital mortality rate among a historical cohort that met those criteria. And indeed that’s what we saw in.

Eric 22:21

The trial as well, 35%. So about a third, a little over.

Kate 22:24

A third died in the hospital and more than half died by six months.

Eric 22:29

Okay, so very high risk group and pretty low bar intervention like this intervention is, sounds like it’s scalable, easy to plug into a healthcare system, your primary outcome was similar to the primary outcome the last time you joined us on the primary on the palliative care the GeriPal podcast, which was on default pal of care length of stay.

Scott 22:54

I thought we agreed, not before.

Kate 22:55

Yeah, I thought we were going to refer people to the past podcast.

Eric 23:03

If you want. You should have nudged me before we started not to talk about this.

Alex 23:09

Wait, can we talk about it quickly though? Because it’s complicated and we don’t want to over complicate prayer listeners. You want to get the results.

Eric 23:17

I just would. Why did you choose length of stay as your primary outcome here?

Alex 23:21

Hospital length of stay.

Eric 23:22

Hospital length of stay.

Kate 23:23

Hospital length of stay.

Scott 23:25

This is a palliative care study. Ultimately, it’s not a study of palliative care clinicians per se, but it’s a study of the construct of providing high quality palliative care. And at the time we launched the study, the prevailing business case for palliative care had to do with the premise that it provision in a high quality way could reduce length of stay. And as such, it would be important to conceivably all stakeholders, like patients, want to get out of the hospital sooner. Their family members want them to get out of the hospital sooner. The hospital CEO wants them to get out of the hospital sooner so they can admit a new patient.

The clinicians want them to get out of the hospital sooner for myriad reasons. So it aligns in the sense that it’s important to all stakeholders. It had a prevailing conceptual foundation in the sense that whether or not you believe the data that existed at the time that it might reduce length of stay, that was part of the way it was sold. And so that’s why we chose it. And as I think Kate and I both articulated in all the ways we possibly could, we have sworn on our graves that we will never ever design a palliative care study with language, a primary outcome ever again.

Alex 24:57

And why?

Scott 24:59

There’s just too many competing factors, right? Like the ability of someone to leave the hospital for social needs reasons. The ability that all the different things that are completely outside the domains of palliative care that influence length of stay and hence really degrade the signal to noise ratio of that outcome measure. And then finally, that there’s actually no agreed upon, straightforward method for analyzing length of stay data that doesn’t rely on certain untestable assumptions.

Alex 25:41

Yeah. And just briefly, to ward off those of you who’ve read the article and may have concerns that death was included as one of the longest lengths of stay in the primary analysis. And you might think, oh boy, that’s like punishing death. Like, it’s as if they had a huge long length of stay. But then our authors today did a billion different sensitivity analyses that basically showed no matter how you slice it, the results were the same where they you counted death as like a really short length of stairware.

Eric 26:11

And we dived deep into that on our last podcast.

Alex 26:14

I don’t feel like we got to the other podcast. Yeah.

Scott 26:17

And only 900 million of those billion sensitivity analyses were requested by the handling editor.

Eric 26:27

Okay, let’s jump into what did you find? Did it work?

Kate 26:32

Oh, well, are we talking about the primary outcome or.

Eric 26:36

Yeah. What the primary outcome, length of stay.

Kate 26:39

Alex just stole the thunder. No matter how we analyze length of stay, these nudges combined or individually did not change hospital length of stay. However, you know, we saw increase significant increases in hospice enrollments with the accountable justification nudge. The have you offered care focused on comfort nudge both unto itself and when combined with prognostication. And we saw earlier comfort care orders. So those without an increase in the proportion who rehab comfort care orders.

And I think that’s important and without an increase in hospital mortality because what it really says to us are that these nudges identified for clinicians, patients for whom who were likely to die in hospital and likely to ultimately receive comfort focused care, at some point, they just received it earlier, arguing potentially, you know, less time receiving life supportive therapies that were non beneficial and me be not wanted as identified perhaps during a conversation. So those were the main takeaways.

The prognostication nudge alone was not effective. It had no signal in any primary or secondary outcomes. And then when combined with the treatment alternative, the idea of there’s another way to do this, there was some signal in those outcomes. So I think that was probably the active ingredient in the combined set of nudges as well was the accountable justification.

Eric 28:15

So let me get this straight. So just having people think about or focus on prognosis, prognostication didn’t really change any of these outcomes, primary or secondary. 18 secondary outcomes. Right.

Kate 28:31

You looked at it did not. Let me just add that the premise for that particular intervention, interestingly, was a simulation study study done by colleagues at Johns Hopkins, I think, led by Alison Turnbull, where she showed a signal for willingness to of clinicians who were told to focus on prognosis in a hypothetical study were then more willing to offer the alternative comfort focused care just by thinking about prognosis.

She followed that up with another study to test that in a more real environment and it was null. That obviously was published well into our trial. So we sort of wonder if perhaps the premise of the focusing effect and prognostication in the end, when put to the real test in a, in a another study was not, not quite as powerful as we had hoped.

Eric 29:25

Yeah, it feels like it has face validity though, like, like when I’m in these pre family meetings where we’re sitting down, actually having people think about prognosis kind of helps them think about like thoughts and their recommendations, whether it be like oncologists, primary team members.

Alex 29:41

Right. Though we should note that’s a different setting. That’s in a, before a family meeting where you’re probably going to have a goals of care discussion, not in daily rounds.

Eric 29:48

Yeah, good point.

Scott 29:49

But it’s important that what we’re not able to say, Alex and Eric, is it’s possible that these interventions, while not exerting effects on things that are easily measured in an ehr, which is what we do in practice, pragmatic trials, could have foreseeably nonetheless changed the types of communication offered in family meetings and even more kind of rote day to day conversations on rounds with family members who are in attendance. So we don’t have any way of getting at that. But it is a plausible premise.

Eric 30:31

So I gotta ask about the, so the positive findings you had around the alternative type of care, the comfort focused care, it was two out of the 18 different secondary analyses. I think this came up in the editorial. How much weight should I put into that? Given that statistically if you look at 20 different outcomes, maybe one of them.

Alex 30:55

Will be positive by chance alone.

Eric 30:58

By chance alone, yeah.

Kate 31:00

I think it’s a fair criticism. You know, my two sort of off the cuff thoughts are pretty plausible mechanism for how we got there. It’s not like we have to work that hard to put together how A to B led to C for those two particular outcomes for these interventions. I think that matters a lot here. The other one is simply that, you know, these, as Scott just said, these pragmatic trials are trying to tell a whole story. And you know, there is a struggle between sort of the traditional really limiting what you’re looking at and being very, very narrow and these efficacy trials and this idea of I’m trying to understand an ecosystem but reliant on existing data and trying to be able to explain the findings the best we can.

And that if you look into sort of how to think about pragmatic trials, an answer to many, many different stakeholders which they are intended to do. You really do need to look at a number of different things. Is there a world in which we can call some of these tertiary? Maybe. Maybe we just start leveling them and we don’t get so upset about how many secondaries. But I think it’s a less interesting study if we were to just pick three secondary outcomes. And there’s this whole other slew of things we’re all interested in understanding. And we don’t do it because, you know, we’re worried about that kind of criticism. But it’s a fair criticism.

Eric 32:26

Just.

Kate 32:26

Paul Scott, if you have anything, you know, statistically or otherwise to sort of respond to that kind of, you know, thinking.

Scott 32:33

I do, but I don’t know if. Jaspal, you wanted to weigh in on this.

Jaspal 32:37

Yeah, I mean, I think you’re. The criticism is fair. And were we disappointed that length of stay didn’t, didn’t change? Well, maybe. But the bottom line is, I think for such a complex topic, to distill it to a behavioral intervention, I think to understand the environment of critical care in the area in which I work and the different hospitals and different centers, I think actually the idea of comfort directed care improving in terms of the uptick in that, I think that’s pretty powerful, especially in this region of the United States. Culturally, it’s a little bit different than perhaps other areas where palliative care is not ubiquitous.

There’s a shortage. There’s a massive shortage. In fact, many of the centers that we studied this intervention in are places where you don’t have intensivists, you have hospitalists, you have NPs and PAs rounding in the ICU for them. They may not be as comfortable seeking permission to engage with family. These aren’t things that we necessarily could do with a large trial at that. On the outset, when we designed this study, when we implemented this study, should say on my end, but looking backwards actually at it, actually, I thought it made a huge cultural difference.

And if these markers are secondary, sure, there’s, it’s great that we analyzed all these, what Kate said, but I think there’s probably a powerful signal in there that actually is much more complex than complex and nuanced, especially when we think about social justice areas that are under resourced particularly, I think we have to think about those populations a little bit. And I think, yes, the science gets diluted out in terms of the signal that we want to see, but I still think it’s pretty powerful.

Eric 34:20

So do you believe it sounds like you do believe that get the nudge of offering that physicians should offer comfort focused treatment alternatives had this effect on both hospice utilization and comfort care orders.

Jaspal 34:36

I do believe that. And so as the site based PI that was responsible for education in this space. Right. It’s amazing how many conversations I had to have with people like what is this? Oh, you mean I can sort of act on this like it was shocking to me at how many people hadn’t. Hadn’t occurred to them to actually, you know, when the, when the patient has clearly a prognostic challenge that basically that clinicians were uncomfortable and all of a sudden became like, oh, okay, I get it now what you’re trying to do.

And throughout the duration of the trial, which we didn’t talk about the trial being interrupted with COVID and all the dynamics of all of that. Right. That all of a sudden that end of life discussions all of a sudden became somewhat normative in areas that were not. And I think that’s an important aspect of the study which I think was hard to capture because of all the factors in the study design.

Eric 35:26

Scott, any additional thoughts on that?

Scott 35:29

My first thought is just to thank Jaspal for that. That really like lucid and insightful explanation from, from, you know, eye level view on the ground. Yeah, I think it really adds a lot of color. The other, only, the only other thing I was going to add sort of in response to like this issue of the criticism, which I agree is, is valid from a statistical sense, but I think we need to think about like where the burden of proof is or how high the burden of proof is here. The burden of proof would be really, really high if we’re talking about what’s often studied, which is hiring extra fte, whether it’s nurse led navigators in the ICU or extra social workers, extra palliative care clinicians.

All these things that cost a lot of money and have real opportunity costs associated with them. Or if we were talking about an intervention, no matter how much it cost, that had real potential to harm. Neither of those things are true here. Right. So we’re talking about dirt cheap, readily scalable interventions that you’d have to be really creative to think about how they might harm someone. And in fact for sure we didn’t see any evidence of harm. It’s in fact very important to point out that mortality didn’t differ across the arms length of stay wasn’t shortened, but it also wasn’t lengthened like etc.

So sure. Would, would the scientists in us want to see a replication study that showed the same Results in a different setting, 100%. But if we don’t get that study in the near future, health systems still have a choice to make and, you know, a little evidence in the right direction for conceptually compelling outcomes that have a clear mechanistic basis to them. I might be in favor.

Alex 37:38

Yeah, I really like that sort of public health angle here, both in terms of what you were saying. With this low resource, it’s pretty easy to implement, as opposed to going back to Eric’s analogy, having a trained palliative care physician going in and asking them what they think about prognosis and making eye contact with them. You’re a little bit different from an electronic prompt and a lot more expensive. And I also appreciate it from what Jaspal was talking about, about reaching community hospitals and critical care facilities.

Like, one of the beauties of this study is just the population and the hospitals that were included and the potential generalizability of these findings outside of the sort of ivory tower academic centers which are so often the subjects of the palliative care trials that we have people on and talk about. Are you going to ask the what’s next or if you could do it again?

Eric 38:32

I just want to know what do we do with this information? Yeah, you have this trial. What do you do with it? Kate, should we be implementing this? How should we be thinking about.

Kate 38:41

Yeah, if I were running my own study at my own health system, I would be asking my ICU docs if we could implement treatment alter. Maybe it’s not a prompt. You know, maybe we don’t want an alert. Maybe there’s a smarter way to get it targeted to the ICU docs. But given the conversations like to Jess point some of the operations and sort of the larger culture. We’re talking about this stuff all the time in our ICU faculty meetings is we’re underutilizing hospice. There aren’t enough communication documentations. Where are these advanced care planning notes for these patients? And, you know, like, here you go.

You could, you could just simply remind us and. And we would do it more often. It seems so. But, you know, that would mess up my ongoing trial. So that is one health system taker. I, you know, I do think to Scott’s point, it is, it is off the shelf, easily replicable to institute. It is a low lift and a low lift for clinicians, too. And it’s the right thing to do clinically for patients and families. Like, it doesn’t. Doesn’t have a lot of knocks.

Eric 39:49

Jaspal did. You did. Was it deleted from Matrix Health or is it still being used the prompt?

Jaspal 39:55

Well, it got deleted for the. After the study was finished. But I think we learned a lot about decision support, about decision analysis, about the idea of nudging and how to design interventions. I think it was just complex, right, to kind of continue the moving forward. I think that if you ask what’s next, I think it provided tremendous insight as to how we go about sort of approaching very complex interventions at scale. When I mean scale, I don’t mean just one hospital, one facility, but as a large organization. You know, what is. Back to our initial song. What is, you know, ICU rounding, right. ICU rounding right now is so heavily nuanced. Certain people, it depends on the attending, depends on the environment, depends on the hospital, depends on the. I mean, at some point it doesn’t make any sense, right?

There are certain things, to Scott’s earlier point, about decisions that just need to be made. You know, we’re having the same conversations and parallel types of intervention the ICU call the ICU liberation or ventilator weaning or other aspects where some decisions just have to be taken away or have to be at least limited. Some decisions need to be potentially trimmed down in terms of options, in terms of what clinicians choose. And it can’t just wait for the super specialist, in this case the palliative care specialist to come in because that’s a Monday through Friday at the secondary tertiary center, but not at a primary center of ICU care. That’s not an option. Even then there’s workforce issues, there’s subjectivity, there’s biases that are all throughout the color, that interaction.

So then you step back and how you design an ICU and how do you deliver care effectively across an entire geography. At the minimum standard, right. Has to somehow be, I think nudging has to almost be engineered into the decision making process, which is again, if I have were to imagine how I see rounding, I would probably design it very differently, right? And I would actually start putting things like nudges on this issue, nudges on mechanical ventilation, decisions for liberation, for example. There are certain things that could potentially relieve the clinician from all those decisions that they’re making and move towards what we call a standard that we define and that society’s working on defining. That’s how I look at it, is I think, yeah, we didn’t continue this intervention, but to Kate’s point, this may not be the perfect intervention, maybe other things to think through this space, but I think this triggered a lot in me to think about how we can do better, especially since we carried this to the COVID pandemic. You can imagine the complexities at a rapid fighter pace, end of life discussions, mechanical ventilations, ECMO questions, huge survivability implications, having to make clinicians put them in very uncomfortable situations at a rapid pace, at a massive onslaught of acuity. It had to really think through. If we had to redo things, should we prepare for that situation?

Eric 42:56

And I guess the last question for you, Scott, is so I love what Josh Paul is saying is like introducing these nudges and but it also feels like the system is currently designed with these nudges. They’re just nudges to do very aggressive things to people at the end of life.

Alex 43:12

The default pathway.

Eric 43:13

Default pathway is to keep people in the icu is to keep them full code. It’s the default nudges are built into the system. And is it to think about? Are there alternative nudges that may also improve quality of care? Thoughts on that?

Scott 43:30

I’m not sure my thoughts will reveal any more than your smart question revealed, which is that we ought to be, in a normative sense, defining nudges mindfully rather than our tradition of implementing nudges unknowingly or haphazardly. There’s no option to not nudge in many, many contexts. And as you correctly stated, a lot of the prevailing nudges are in one direction. So well thought out, well designed nudges that might move in an opposite direction could be value added.

And then, you know, it’s hard to get through a podcast on these at the intersection of palliative care and ICU medicine without thinking about Randy Curtis. And one of the many, many wise things that Randy taught, so many, was that oftentimes you learn more from the interventions that quote, unquote, don’t work than the ones that do. And we’re running lots of trials now that I think will be more effective in this general space.

Eric 44:49

So should we be expecting more behavioral economics studies? Kate, sounds like you’re working on one.

Kate 44:56

Yeah, you know it. I agree. I think we’re really these aren’t just one and done. We’re really building and learning and getting smarter about implementation and honestly getting smarter about understanding the culture and context in a larger sense, as Jaspal described, and trying to get in to be more eyes and boots on the ground during the studies in the centers, learning as it’s happening instead of only relying on our hypotheses. And I think that’ll be Also a really big difference in future studies.

Eric 45:30

I love that I can imagine a world where we’re thinking about behavioral nudges differently than our current defaults. Which takes us to the song Alex.

Alex 45:54

(singing)

Eric 46:54

Kate, Jaspal, Scott, thank you for joining us on this GeriPal podcast.

Kate 46:59

Thank you for having me.

Scott 47:00

Yeah. And, Alex, when I’m in San Francisco for ATS this year, I’m taking you to Blue Bottle to show you what some real coffee tastes like.

Eric 47:08

Oh, Scott, I got some better coffee places to try. Andy, down here. Oh, you got to talk to me.

Alex 47:13

Yeah.

Eric 47:14

And thank you to all of our listeners for your continued support. Most of our listeners have rated us 5 out of 5 on the favorite podcast app.

Alex 47:22

There’s the nudge [laughing].

Eric 47:27

Please consider rating the GeriPal podcast on your favorite podcasting app. Thank you everyone!

This episode is not CME eligible.