Skip to content
Headshot of Doug White

This week’s guest is Doug White, Professor of Critical Care Medicine at the University of Pittsburgh and lead author of a randomized controlled study of a nurse-led intervention to provide emotional support to families of seriously ill patients in the ICU and improve the quality of communication, published in the NEJM.

Nurses trained in standard palliative care communication techniques (think NURSE statements) provided the intervention, which included regular family meetings with family members and daily check ins for emotional support.

The primary outcome, difference between the intervention and control groups in anxiety or depressive symptoms at 6 months, was negative. There was just nothing going on. Zilch. So was it a bust?

Well… multiple secondary outcomes were positive, including improved quality of communication, greater perception of patient-centered care, and 28% costs in the intervention group compared to control. Well, that sounds great! So was it a breakthrough?

Doug discusses his surprise about the mixed findings. Here’s a teaser:

I think we have pretty clear data that the intervention substantially improved the surrogates experience in the ICU. The ratings of the quality of communication were much higher. When you look at the individual items on the Quality of Communications Scale it’s things like, “I felt that the doctors were giving me the information that I needed to know. I felt like they were involving me in the ways I wanted to be involved. I felt like they were giving me emotional support and caring about my loved one as a person.” So all of these things that we think are really great, important things. And I think many of us, myself included, thought that if we only could improve this aspect of care that that might substantially lessen the trauma of the ICU experience in such a way that would modify their long-term psychological outcomes. And I think that assumption was just wrong. You know? I think we really did modify their experience in the ICU and the sense of feeling supported and that didn’t change their long-term psychological outcomes.

This study is an intervention that improves family/surrogate experience in the ICU, at lower costs, features nurses, AND establishes that we need other pathways to modify long term psychological outcomes. More work to be done!

Should we be widely adopting the intervention? Shmaybe? More on that below.

And hey… I got to sing Gillian Welch, one of my favorite Folk musicians!


by: Alex Smith, @AlexSmithMD


Eric: Welcome to the GeriPal Podcast! This Eric Widera.

Alex: This is Alex Smith.

Eric: And Alex we have a guest in our studio audience today.

Alex: We have a guest in the studio, in Eric’s office. Sometime we’re going to have to tweet out a picture of this. It’s quite comical. But today, we have Brian Block who is a Pulmonary Critical Care Fellow here at UCSF. Welcome to the GeriPal Podcast, Brian.

Brian: Thanks for having me join you guys.

Alex: And, we have a guest who’s joining us from the University of Pittsburgh and it’s Doug White who is Professor of Critical Care Medicine at the University of Pittsburgh and just published a study in the New England Journal of Medicine that we’re gonna talk about.

Eric: I heard about that journal!

Alex: You heard of that one?

Eric: Yeah, it’s a good one.

Alex: Journal this small medical society and it’s some North Eastern State. Minor Journal. Welcome to the GeriPal Podcast, Doug.

Doug: Thank you very much for having me.

Eric: So before we dive into the topic, this New England Journal paper, we always ask our guests to give a song request for Alex. Doug do you have a song request?

Doug: I do, yeah. Gillian Welch, Look at Miss Ohio which I think is a beautiful song and also just a little bit subversive in the roles we sometimes feel pressured to playing in life.

Eric: Great.

Alex: Good? Alright, I’m gonna play it. [Singing].

Eric: Awesome!

Alex: Right, so I’d like to see Gillian Welch in concert. She’s such a great musician. Appalachian, Folk, sorta Rocky thing going.

Doug: Absolutely, Alex that was fantastic! Nice work.

Alex: Thank you, Doug!

Eric: Before we jump into this New England Journal of Medicine paper, “A Randomized Trial of Family-Support Intervention in the ICU” that just came out this last week, maybe you can take us back a little bit and just tell us how you got interested in this subject.

Doug: Uh, yeah well, you know … So I practice in Intensive Care Units at the University of Pittsburgh, and, uh, part of the time. And the rest of the time I’m doing research and this, the issue of seeing families overwhelmed and really struggling with making decisions for critically ill patients has always been something that I have thought we could do better. And we, you know, we haven’t really brought to it like a critical lens, of … Let’s think about a careful system that’s theoretically grounded and let’s give it a whirl. Let’s see what we can do. It really came out of my clinical experience.

Alex: So, let’s start talking about the study. Let’s dive right in here. Tell us, you know, sort of briefly, high level overview of what you did here.

Doug: Yeah, so. This was what’s called a stepped wedge randomized trial.

Alex: What?

Eric: Stepped wedge, is that a podiatry, like some type of shoe?

Doug: It is clunky, but it’s not a shoe, it’s essentially a method where you randomize ICU’s to the time at which they get the intervention. So every ICU starts in the usual care phase and then they, at random time points, they transition to the intervention phase.

And the reason we chose this method is that the, as you’ll hear, the family support intervention that we tested is really an ICU level intervention. And if we had tried to randomize individual patients the care that the control arm got would’ve been contaminated.

So, stepped wedge trial. We randomized fourteen hundred and twenty patients and their families, to either the usual care arm or what we call the partner family support intervention. And we followed them both through the hospitalization and then we followed surrogates to six months out. And the primary outcome of the trial was the surrogates psychological distress measured using the HAD scale, the Hospital Anxiety and Depression scale, and then we looked at several prespecified secondary outcome measures including the Surrogates Perceptions of the Patient Centeredness of Care, the Surrogates Ratings of the Quality of Communication, and then a variety of measures of healthcare utilization like costs and ICU and hospital length of stay.

Alex: Great. That’s a terrific overview. Let’s dive in a little bit deeper into who exactly you studied, because these weren’t just any patients in the ICU. You had specific criteria for who could get into the study.

Doug: That’s right, so these were, you know, for lack of a better term, kind of the sickest of the sick. We asked the clinicians to identify patients who they said had either, at least a 40% chance of death in the hospital or a 40% chance of severe long term functional impairment. Or the third criteria was that they had been on mechanical ventilation for at least ninety-six hours continuously.

Alex: Alright, so this was a pretty sick population as you say, and the mortality rate was fairly high.

Doug: Yes, very sick population. 35% of the patients died during the primary hospitalization and then by six months 55% had died. And the other, I think, important thing about the population is that almost no one at six months was living independently at home. It was, I think less than two percent in both arms.

Brian: And Doug, this is Brian here. How about the people carrying out the intervention? Which members of the team or were you bringing in other people external to the standard ICU team?

Doug: Yeah, let me tell you a little bit about what the partner intervention was. And, essentially what we set out to do with this intervention was take whatever resources that are in the ICU and use only those resources to deploy the intervention. And so basically we developed this interprofessional family support intervention as a way to sort of flip on it’s head how we typically support families. Which, in my experience, in a variety of ICU’s around the country is often somewhat ad hoc. The individual attending physician has his or her own personality and that determines a lot of the family support that is delivered. And instead we thought, let’s make something more consistent that it doesn’t matter who the attending is that week, let’s develop a family support intervention that happens relatively regularly regardless of who’s there.

And the second thing that’s, I think, somewhat novel about it is a lot of the work of family support and the intervention is delivered by existing ICU nurses who undergo advanced training. So it’s four to six nurses per ICU, go through twelve hours of communication skills training for the kind of things that we want them to be able to do in the intervention and then they’re there and they’re the ones who are interacting with the family on a daily basis. Delivering what I think of as sort of both emotional support and practical support. About you know, what’s happening in the ICU and how the patient is doing. And then the second thing these nurses do is they’re responsible for making sure that family meetings happen at set intervals for the patients. So no later than within forty-eight hours of being enrolled and then every five to seven days there after. These partner nurses are arranging interdisciplinary family meetings. The attendings run.

Brian: So are these partner nurses the patients existing bedside nurse? Or is it an additional nurse beyond the bedside nurse?

Doug: They are, it varies. So it’s not always the bedside nurses. At UPMC there are charge nurses who are there everyday, who usually don’t have heavy bedside responsibilities and there are nurses called resource nurses who are there to help with education and ensure protocols are being followed for ICU care. And so it’s these individuals. And you know sometimes it actually is that a partner nurse has a bedside assignment with another patient and he or she makes the time during the day to go meet with the family that, for example, is going to have a meeting that day. So, it takes a little bit of organization.

Eric: But no additional staffing requirements?

Doug: No additional staffing.

Eric: This is all under the ICU FTE.

Doug: That is right. So that was sort of the thing that we said, “Let’s not bring in external people. Let’s not add new people to the ICU team, let’s see what we can do with the existing resources that we just reengineer the care processes around them”.

Alex: So would you call this a combined effectiveness and implementation trial in some sense?

Doug: You know, this is, substantially an implementation trial. We framed this as an intervention that is delivering recommended practices in terms of intensive emotional support and frequent family meetings. And the only thing that is different is that we wrapped it in a care delivery package that is more consistent than usual care is.

So we both wanted to see, can we deploy the intervention and then also what are the effects of the intervention once deployed.

Eric: So how did it deploy and what were the effects?

Doug: Alright so, we trained these nurses, as I said, twelve hours of communications skills training with standardized patients and some pretty good communication skills training coaches here at UPMC. And then we had an implementation specialist which many hospitals have, you know people who help deploy quality improvement interventions, helped each ICU develop care practices and protocols to make sure that the partner intervention was being deployed as prescribed or as directed.

And so after a week of pretty intensive support, each ICU is kind of released to do the intervention. With weekly site visits by the implementation specialists to see how things were going. And what we found was that the intervention on the primary outcome measure, on Surrogates Longterm Psychological Distress had no impact. It really didn’t change any markers on their symptoms of depression, anxiety or PTSD.

Eric: So it’s a negative study. So what’s next for you Doug? You put all this hard work into this ..

Alex: And that’s the ned of the podcast! And onto the song again. No? There’s more.

Doug: So there is a little more, I would say there’s a lot more, but I also think that a negative trial is super important. I think we learned a ton about this kind of intervention not moving the needle on surrogate psychological distress, and maybe we can come back to sort of unpacking why.

But what was positive about the trial was every other outcome measure. So the measures of the Patient Centeredness of Care were much better in the intervention arm compared to the control arm. The measures of the quality of communication were much better, the length of stay was shorter in the intervention arm. Substantially caused by a decrease in the length of stay before death among the patients who died.

And then the other thing that was substantially different were costs across arms. The partner intervention was associated with about a six thousand dollar decrease in costs per patient in total hospitalization costs.

Eric: And then was that largely just driven by length of stay? Or other factors?

Doug: Uh, it was substantially by length of stay. Less use of things like blood transfusions, antibiotics, vasopressors, and then the costs of being in the ICU.

Eric: And it sounds like by not including additional staff beyond those normally in the ICU you didn’t incur extra costs for each patient.

Doug: Correct. So we did a, I should say we did a pretty careful costing of how much it costs to deploy the intervention, so the … We paid for the nurses time to go through the training. We included the costs of the training. The meals, the rental of the facility, the actors. And we also included in our cost estimates the, all of the time and the travel costs of this quality improvement specialist to go to the ICU’s.

With all of that we found it costs about one hundred and seventy dollars per patient to deploy the intervention and got us about a six thousand dollar savings per patient.

Eric: Wow.

Alex: Now just sticking on the costs for a moment, I noticed that the, or the press officer for UPMC reached out to us about this study. I noticed the press release is titled, “Team Approach to Support Families Improves ICU Patient Center Care and Lowers Costs.” It really leads with this cost finding in the first paragraph. And then there seems to be a contrast between that story here and the way it was framed in the New England Journal. I wonder if there’s a story there about how the cost finding was emphasized in the press release and not as much in the scientific publication.

Doug: Uh, there … Yeah. There’s a bit of the story. I think I mean you’ll see in the New England piece the cost data are not included with the exception of the, what it cost to deploy the intervention. So the cost savings were not in there. And that was an editorial decision on the part of the New England Journal. I think they really wanted to focus on the patient centered aspects of this and not complicate the message with even more quantitative data about healthcare utilization.

Brian: So as a contrast you mentioned that this was a study that used just resources in the ICU and comparing to a couple of years ago a study by Randy Curtis’ group in the Blue Journal that looked at communications facilitators who were external to the normal ICU team. They actually did include some cost data with some similar findings in terms of shortening length of stay. I was wondering if you could comment on why that made it into that article but maybe not into yours?

Doug: Well I would be speculating. You know I think, to be clear, the length of stay data are included in the New England Journal paper. The only utilization data that weren’t are the dollars. You know, as I said, I think editors have a lot of discretion about the shape an article takes, in terms of what’s the right amount of data to put in it. What creates a compelling story and then in contrast, what becomes too much to tell in one paper and I think that that was the decision that they made and the piece of data they thought would best be dropped.

Eric: So, I’m going to take a step back. So you, it was a tantalizing statement you said about how this is a negative primary outcome study that made it into the New England Journal that you learned a lot from it. About surrogates, care givers, family members and the emotional burdens of the ICU. You wanna take a little bit of a deeper dive on that?

Doug: Yeah, uh. So, I think we have pretty clear data that the intervention substantially improved the surrogates experience in the ICU. The ratings of the quality of communication were much higher and these are, when you look at the individual items on the Quality of Communications Scale it’s things like, “I felt that the doctors were giving me the information that I needed to know. I felt like they were involving me in the ways I wanted to be involved. I felt like they were giving me emotional support and caring about my loved one as a person.” So all of these things that we think are really great, important things. And I think many of us, myself included, thought that if we only could improve this aspect of care that that might substantially lessen the trauma of the ICU experience in such a way that would modify their long-term psychological outcomes. And I think that assumption was just wrong. You know? I think we really did modify their experience in the ICU and the sense of feeling supported and that didn’t change their long-term psychological outcomes.

Alex: Yeah, that also brings up the issue of why were the long term psychological outcomes the primary endpoint in a study that seems to be designed initially to, as far as my read, to improve communication and improve emotional support for the caregivers of surrogates in the ICU? Patients in the ICU?

Doug: Yeah, great point. We in the design phase of the study we thought a lot about should we even designate a primary outcome? And then because of sort of existing norms and the need to register one on we picked the has as the primary outcomes.

But what I want to say is the way we conceptualized the study and the outcome assessment from the start is that they’re three important domains of outcomes. There are the surrogates psychological outcomes, that are certainly important in and of themselves. But separate from that, the experience and the patient centeredness of care we viewed as it’s own separate domain of outcomes. And then the third outcome of importance is healthcare utilization. Both when we conceptualized the paper and the project and when we wrote it, we said there are three primary domains of interest. And because of the psychometric properties of the hads, because they’re the best of all of the outcomes, we specified this as the primary outcome. Not because it was in our view the most important outcome to change.

Brian: So Doug I had the opportunity to see you present this paper last week at the American Thoracic Society Conference, and one thing you emphasized there was that you did not use rational actor theory for your decision theory of the study, but instead used a cognitive emotional theory. Could you talk a little bit about that?

Doug: Yes, thank you. This is one of the things that I really loved about this intervention is that we, in contrast to the support intervention that was really about lets make sure we give people the right numbers and the right information and hopefully good decisions will logically flow from that …

Eric: That’s the rational actor?

Doug: Yeah, in decision sciences that’s expected utility theory or the rational actor theory that with good decision, with good information, good decisions will follow. And there’s been a great deal of work in decision sciences over the last couple of decades that suggest that there are many things that get in the way of this theory that good information will lead to good decisions. And in our context one of the most important considerations is the role of strong emotions. Like fear and anxiety. And really good decisions, scientists and experimental psychologists have documented how fear and anxiety change how people both are able to process information and also the decisions they make with that information. What appear to be problematic ways, tends to make people make, really sort of snap judgements that are grounded in less deliberation than you might hope for for such weighty decisions. So, we felt that not only do we need to make sure that surrogates are getting good information, we also need to deploy an intervention that substantially targets providing emotional support with the goal of moving these surrogates from what I think of as a hot state, you know emotionally overwhelmed, to a cooler state, where their emotions are there and I think it’s inevitable that they will be there, but they are not overriding their ability to think clearly and to think about trade-offs.

Brian: So what would you recommend for those of us practicing in San Francisco, where it’s a rather temperate state?

Doug: I think cool is the right place to be, and I don’t mean cold. I mean I really don’t. I think it would be a mistaken view of the intervention if someone walked away and said, “well they were just trying to do away with emotions and actually get people to a place where they could be purely rational.” And that’s just not what we were trying to do and I don’t think that’s even a realistic goal. We were really trying to bring emotions and rationality together in ways that they can each bring out the best in the other. So a lot of the work of emotions when we make decisions is to alert us to the salient features of a decision. Alright the thing that sort of tugs at our heart is sometimes the thing that we need to be thinking about.

Alex: That’s well put. I wanted to ask you about the mortality finding. It seems that the mortality in the intervention group was 6.7 days and 7.4 days, versus 7.4 days in the control group and that the difference was statistically significant and yet mortality at six months was the same between both groups so that, you know, if I could put this in other words, you might say that we weren’t prolonging the dying for patients who were in the ICU who were going to die anyway. Does that sound like it’s in line with what you would describe?

Doug: Yes, that is how I interpret the data also. You know I think that it’s really, really important that we collected data on families’ perceptions of the patient centeredness of care because if we only had the finding, the intervention led to more deaths or speedier deaths in the intervention arm, we would be left with this question of, “Gosh is that good or bad?” And the data that show the intervention yielded more patient centered care, suggests to us that these are decisions that the family felt reflected the patients values and preferences to not be kept on life support in the face of a poor prognosis.

Eric: So with all this data and this study coming out, what’s next with this? Is this something that all ICU’s should be adopting or is there a next phase? Partner version 2.0…?

Doug: Well, so very good question. I’ll answer it in two ways. One I will tell you what’s happening at UPMC and then two I will talk about what I think needs to happen from a research standpoint.

So, UPMC is a twenty-five-ish hospital system and they have made the decision to roll out the partner intervention across all the ICU’s in the health system and that has been ongoing for the last few months. So UPMC has decided that this intervention makes good sense and improves the care of patients and families and we should go forward with it.

I think that that is reasonable. One other interesting thing about the editorial process in the New England Journal is they, the editors were very clear with us that we needed to include a statement in the discussion that said, the results need to be confirmed in a larger pragmatic trial that involves multiple health systems in multiple regions of the country.

So I guess that is sort of two ways to illustrate how two different groups viewed the implications of the trial.

Eric: I got another question, what was palliative cares involvement in this? Do they no longer need to see ICU patients cause you got these trained nurses there?

Doug: That’s not how I view it. So, you know I think in an ideal world this intervention would be sort of what we think of and call generalist palliative care or primary palliative care, delivered by an ICU team that is just a little bit more skilled and a little bit more coordinated than they were before the intervention. And I think that that will most definitely not do away with our need for our palliative care colleagues to help us both in terms of managing higher level palliative care issues that arise, such as conflict between the clinicians and family, or conflict within a family. Or even attending to families who are highly emotionally distraught and the best basic interventions of the care team are just not moving it in the ways that we would’ve hoped to.

So you know when I give talks on this I liken it to a cardiology consult. So as an intensivist I am comfortable managing a variety of cardiac issues, you know, straight forward a fib, a flutter, v tach, v fib. But there are some things that get complicated and where I really am not sure what the next thing to do is and that’s when I involve my palliative care colleagues or my cardiology consults in this circumstance and that’s how I hope people will view how to integrate primary palliative care and specialist palliative care.

Eric: So let me ask you another follow up question to that. So a couple years ago you wrote and editorial on a JAMA paper, about proactive family meetings for chronically critically ill patients. Some dubbed this proactive palliative care consult, although that’s a whole other podcast that we could talk about. How would you, like how do you think about these two studies? Your study and the last study around like communications in the ICU, the JAMA study on chronically critically ill patients.

Doug: Yeah, Uh. So I think it’s really important to point out that as the distinction you made is the right one, that was not a palliative care, a full palliative care consult, was prognostic information from a mortality prediction model, delivered by palliative care clinicians to the family, presumably in a very sensitive way. But it’s still, that’s still a pretty cognitive intervention. Right? That’s the one or two meetings where the interventionists in the sit trial, their role was to make sure they talked to the family about the patients prognosis and then to think about the implications for goals of care decisions. And you know, on its face, that’s a wonderful thing to do, right?

And if the rational actor theory was true, that should’ve been enough, but it wasn’t. And I think that study also, compelled a lot of us to think a little more critically about what is it that needs to happen to help families make good decisions? And I think many of us are coming around to this idea that it’s a lot about aligning with families and about establishing trusting relationships and helping them move through tragic and really hard experiences. Which typically isn’t just done by laying out the bad news and then hoping that we’ll do all the work. And I do want to say, the sit investigators are fantastic investigators, I think they did a really nice job of answering a hugely important scientific question, in ways that really advance the field’s thinking about what’s gonna be needed to move forward.

Eric: I agree. I love this study because it shows you the impact that a negative study can actually have. It teaches us a lot more than I think it would’ve taught us if it was a positive study.

Doug: Absolutely. I think if it had been a positive study we all would’ve said, “well of course it’s a positive study that intervention makes perfect sense.” Because at the time it did make perfect sense. Saying let’s bring in the heavy hitters, really skilled communicators and just have them deliver this news and talk about goals of care and see how that goes. And well I think we got a good answer to the question of is that enough? I think the answer is that it’s probably not enough.

Alex: Any sense … This article has been out about a week now, is that right?

Doug: Mm-hmm (affirmative).

Alex: And so, any sense of the reaction from the community? I’m particularly interested in the reaction from the nursing community. I mean, this is a nurse led intervention and to my mind, although it’s primary outcome was negative, wasn’t the outcome I actually cared about most, it was very positive in some ways. And so I’m curious whether there’s been uptake in the nursing community, enthusiasm, and also in the physician community. How are they responding? And perhaps what kind of questions were people asking at the American Thoracic Society last week when you presented these at the time the article was published.

Doug: So, yes there was a lot of enthusiasm about really fully leveraging the interprofessional team and giving nurses the kind of responsibility that fits with their professional competence. You know I think the nurses in ICUs are incredibly skilled and in many circumstances not being used to their fullest and that’s both kind of unsatisfying for many nurses and also leaves them on the sidelines witnessing things that are quite distressing to them. So from the stand point of the nursing community the feedback I’ve gotten has been really quite enthusiastic about you know “we want this role and we think that this is the direction the field should go.”

The other interesting comments were people kind of coming, some of the emails were people saying “It’s really great to see that the Kahneman Tversky model is the model that we should be focusing on and thinking about rather than the rational actor model. So you know it’s really interesting that people are sort of embracing this more modern view of what goes into good decisions and recognizing the hugely important role of affect in that.

Brian: So Doug, you described this, parts in the manuscript as a quality improvement study, where you’re using guideline recommended interventions. What does it say about guidelines that some of the outcomes were positive and some were negative?

Doug: Well, these guidelines are the kind of guidelines that are from expert consensus, largely. They’re not based on high quality randomized control trials and I think part of the goal of a trial like this, which like you said is a hybrid implementation effectiveness design, is to actually provide some empirical evidence about whether our guidelines are the right ones.

You know I think it’s both appropriate to have framed this as a quality improvement project because we were basing it on the best available professional society statements about how family should be supported and you know I think it does raise interesting questions about what should be the next steps.

Eric: Well Doug, I really want to thank you for your time with us and thanks for going over your article, and really great work. I love this study.

Alex: This was terrific Doug, you should be so proud of what you’ve done here and to publish in the New England Journal, just gives it this enormous mouthpiece. We will try to add to that in our own tiny way with our GeriPal Podcast.

Eric: It’s almost as much of an honor as being on the GeriPal Podcast.

Doug: This has been great fun for me, thank you guys. Your questions were fantastic, it was really fun to look into it. Can I just say also that I think the trial was the culmination of lots of peoples efforts. And so I am glad to have gotten to be the one to be in the first author position and write it up, but it really took lots of nurses and docs in these ICU’s agreeing to do this work and experimenting with how can we improve things for families, so I just want to acknowledge everyone who was involved in it too.

Eric: Great, and great work again to everybody. How about before we end, Alex, do you want to give us a little bit more of that song?

Alex: A little bit more Gillian Welch here? [Singing].

Eric: Wonderful! I want to thank you, Doug again and thank you to all the Jerry Powell listeners. We look forward to talking to you next week. Bye.

Back To Top