Universal Outcome Measurement for Older Adults

Join Dr. Michael Steinman, from the University of California, San Francisco and Dr. Greg Ouellet, Assistant Professor at the Yale School of Medicine and the VA Connecticut in West Haven, as they define universal outcomes in contrast with disease-specific outcomes and explain why the former are particularly helpful when studying older adults with multiple chronic conditions. They also delineate between composite and domain-specific outcome measures and discuss how to choose between them.

To view a transcript click here then select the transcript tab. 

Transcript

Michael Steinman, MD: Hello, I'm Mike Steinman. I'm a geriatrician and a professor of medicine at the University of California, San Francisco and the San Francisco VA and I'm here today with Dr. Greg Ouellet, who's a geriatrician and an assistant professor in the Division of Geriatrics at Yale. And today we'll be discussing the key points from the Data Measures and Measurements Module that was authored by Dr. Ouellet and Natalia Festa on the topic of Universal Outcome Measures. 

Greg, nice to have you here, and can you take a moment to introduce yourself?

Gregory Oullet, MD, MHS: Yeah, absolutely. Thanks for having me, Mike. So, yeah, I'm Greg Oulett. I'm a geriatrician at Yale in the West Haven VA, and I've been engaged in both educational and research activities related to the care of older adults with multiple chronic conditions and so universal outcome measurement is near and dear to my heart. 

Michael Steinman, MD: Wonderful. So, Greg, to start us off, could you describe what are universal outcomes and how do they stand in contrast with disease specific outcomes? 

Gregory Oullet, MD, MHS: What [00:01:00] really stands out about universal health outcomes is their ability to bridge different disease states. One of the things that brought me to geriatrics was a focus on the why: why we treat particular conditions, and what is the ultimate goal that impacts the patient's life. I can remember actually being a second year resident and having Mary Tinetti ask me, why are we treating to this A1C? What is the outcome that actually matters to each individual patient?

And I really had to think about it. Okay. We're thinking about, okay, preventing a variety of complications, and what do those complications have an effect on the life of the patient, whether it be longevity, whether it be function, whether it be quality of life related outcomes? You know, thinking about mortality, which is often something that's measured, but also things like function, like quality of life, like feeling like there's a social purpose. All of those things are things that matter to patients above and beyond the things that we often [00:02:00] think about as doctors that might be a particular measure for a particular condition, like for instance, hemoglobin A1C and diabetes, or a target blood pressure for patients with hypertension.

Michael Steinman, MD: What's the difference between composite universal outcome measures and domain specific ones? Why might you choose, uh, either composite measure or domain specific universal outcome measure? 

Gregory Oullet, MD, MHS: Yeah, so that's a really great question. I think big picture, if we take a step back to thinking back 20 years ago or so when folks were studying particular conditions, like I was saying, we have lots of process related outcomes and then fortunately, many of the studies would look at, would look at mortality as well, so something that kind of cut across different outcomes. I think one of the main things that we were hoping with introducing universal health outcomes into the study of older adults with multiple chronic conditions over the last 20 years was to really capture things other [00:03:00] than just mortality in these disease specifics.

So there've been a variety of outcome measures that are composites that basically take a variety of different domains of things that could impact the life of patients, like pain, physical function, sleep, mental health, and putting them into, into indices that are validated in a population like the one that you're going to study.

And these are sort of a, a really great way of, having parsimony, if you will, right. So they're having one measure that that's looking at a variety of patient-friendly outcomes in addition to something maybe that you're interested in that is disease specific or just mortality. The trouble with the composite outcomes is that they're putting together a lot of different domains that may be important to patients and sort of weighing them the same for every individual patient. 

So it doesn't mean that they're useless. In fact, they're very useful because [00:04:00] they're probably a lot briefer than trying to do a lot of different domain specific measurements. But if you have a situation, for instance, where you know that the main trade off involved in a particular situation might be between longevity and function, for instance, and that people might have very strong disagreements about whether they value function more or longevity more, having a composite measure might not be able to really pick up on on those differences individually between people.

So I think it has to with if the study that you're trying to do is really in the clinical decision making space, having some more specific domains might be of value. But if you're just trying to say, okay, I'd like to add some patient specific outcomes related to quality of life and overall function, and all those domains together, a domain specific is absolutely a great way to start.

Michael Steinman, MD: One thing that might be helpful for people listening to this podcast is just some [00:05:00] examples of like, you know, to take these sort of principles you've been talking about and how they're actually operationalized in this specific measure or study. So are there some specific universal outcome measures that are good exemplars that we might share about, like, you know, what they are, what the name of those things are, and how they're constructed and how they're measured.

Gregory Oullet, MD, MHS: Yeah, absolutely. So my colleague, Dr. Festa, who worked on this with me, presented these in the accompanying webinar, but basically about a decade ago, um, uh, a panel convened by folks at the NIA led by Marcel Salive, tried to put together what are some validated measures that pull together lots of these domains.

And so three that they recommended in the, in the article that came out of that process back in 2012 were the SF 8 and the SF 36, the SF 8 being a briefer questionnaire compared to the SF 36, and then also the Promise 29 and to sort of go back to what I talked about a little bit earlier, is that that these composite [00:06:00] measurements put together a variety of domains. So the SF 8 and SF 36 include general health, pain, fatigue, physical function, sleep, mental health, social purpose, and role and then the, the Promise touches on very similar domains. It doesn't include the general health domain. And so in that article with other colleagues that sort of gave expert recommendations around this, they recommended for if you're using the Promises to, to sort of supplement that with a, with a measure of general health.

So the nice thing about these measures is that they have been really well validated that they've been used in older adults with multiple chronic conditions and they have really robust predictive power for future health states, for future function, as well as future mortality, as well as healthcare utilization.

These are the recommended instruments to use if we're going to use a composite scale, and they're very well used, they're very [00:07:00] popular, and you're gonna be able to compare them across different studies. 

Michael Steinman, MD: One challenge that I've always sort of struggled with as I think about the use of universal outcome measures is that it's hard enough to change any one thing with a health intervention. It's really hard to, say decrease falls or, improve quality of life, or help people sleep better. And here with universal outcome measures, we're sort of in a way sort of trying to measure, we want like the entire person to be better, sort of head to toe in all domains: physical, mental, social, et cetera. And how do you think about the role of universal outcome measures given the inherent challenges that we face with sort of, you know, making a meaningful dent across the totality of a person's life? 

Gregory Oullet, MD, MHS: To start with, I, I think knowing what, what you're hoping that your study is trying to get at, like if you really are in the particular [00:08:00] space of wanting to inform particular patients, you want your study to give results that you're gonna tell a patient, "Hey, I think this intervention is going to be able to do X for you, and this might be the thing that it'll help and this might be the thing that might not help at all." I think sort of identifying that upfront can help you to be pretty judicious about what other things that you want to measure in addition to, you know, I'll give the example of, say a, a person with heart failure, right? And so traditionally they may say, okay, recurrence of another heart failure, hospitalization, and mortality as outcome. And in this particular space, you may say, okay, well in addition to that, patients might really value their symptom control as well as as their function. And so maybe those are the two things that I'd like to add because I'd really like my study to inform that.[00:09:00] 

On the other hand, if for instance, you're doing a very big clinical trial and you're not really in the space of, of giving all of the, the, the very specific information to make a decision guide, and you wanna say, okay, I, I wanna just see is there any signal that, that this intervention is having some, some impact on patient related outcomes. Go for a brief composite outcome so that you can at least get a sense of that, get a signal for that without adding too much burden in, in terms of what we do. And also, knowing that we may not get a signal and may have to dig deeper later, but it's probably, you know, it's probably unreasonable with a composite outcome to think that we're necessarily going to find a signal in every study, every study that we do.

But I think it's a start when we start looking at it, when we weren't looking at it before.

Michael Steinman, MD: There's a really nice example of this slide deck where you spoke about, you know, gave an example where someone might be conducting an [00:10:00] observational study of guideline support on medications and older adults with multiple chronic conditions, and you have these people with MI and people with chronic kidney disease and people with COPD exacerbation and you're sort of mushing all these people together across these multimorbidities. You're trying to kinda look at outcomes and can you sort of, in that example, provide an example of your thought process about what outcome you might be looking at, and then what specific measures you might choose to do to help to measure those outcomes? 

Gregory Oullet, MD, MHS: That's a great question. 

So I, I think, uh, uh, you know, if we're doing a, a study of lots and lots of different people and here we're, we're looking at that, they have, let's see, so we said they're at risk for a lot of different outcomes that you mentioned before, so like in MI, chronic kidney disease, death, as well as stroke. I think in this particular case, measuring each of these individual outcomes that are disease specific probably gets [00:11:00] very, very unwieldy. Probably measuring mortality is going to come natural to most people. And ho- honestly, it does matter to patients most of the time whether or not there's a, a mortality benefit. I would, would venture to say, depending on whether or not you have some of this data already. So if you're, if they're using, if this is secondary data and there's already data on, for instance, patients' function and quality of life, I'd absolutely go with that. That's sort of a consideration, like the ease of what kind of information you have.

I would also venture to say like, depends on if you're, if you're talking about, um, one of these particular conditions that you know is particularly disabling, I might opt for function as a related outcome. But I would venture to say if you're, if you're doing a large one where you're doing primary data collection and you're looking at this many potential conditions, I [00:12:00] probably would opt for a composite at least initially, just because of how complicated it can be and how much each of these conditions could contribute differently to a variety of different domain specific outcomes. 

Now, if you had a narrower population and you have, you know, for instance, like you're, you're really interested in MI, right? You're looking at MI, but the bulk in the population, most of the people have chronic conditions, then I think it's probably easier to zone in on, on what are your particular domain specific outcomes that may be a trade off with the mortality benefit.

Michael Steinman, MD: Anything else that we haven't talked about that you think would be useful for people to know? 

Gregory Oullet, MD, MHS: I think one thing to really hammer home is just the importance of measuring other outcomes other than disease specific outcomes and mortality alone. And yes, there are considerations to picking which instrument or domain specific or not, [00:13:00] but I would encourage folks to use something. Because honestly, measuring something that is relevant to the patients is better than not measuring anything at all. And we've gotta, we have to start somewhere. And I think we have been, and I think that that's making all the difference in the world at trying to make a dent in the way that we approach medicine and trying to make it more relevant to patient's lives.

Michael Steinman, MD: Well, couldn't ask for a better concluding statement than that. Thanks so much, Greg, for joining today. 

Gregory Oullet, MD, MHS: Thank you so much.