Skip to content

What is frailty? Kate Callahan relates a clear metaphor on today’s podcast.  A frail person is like an origami boat: fine in still water, but can’t withstand a breeze, or waves.  Fundamentally, frailty is about vulnerability to stress.

In 2021 we talked with Linda Fried about phenotypic frailty.  Today we talk with Kate Callahan, Ariela Orkaby, & Dae Kim about deficit accumulation frailty.  What is the difference, you ask?  George Kushel probably explained it best in graphical terms (in JAGS), using the iconic golden gate bridge as a metaphor (Eric and I get to see the bridge daily driving or biking in to work). Phoenotypic frailty is like the main orange towers and thick orange support cables that run between towers.  Damage to those critical functions and the bridge can collapse.  Deficit accumulation frailty is like the hundreds of smaller vertical cables that connect the thick orange support cables to the bridge itself. Miss a few and you might be OK.  But miss a bunch and things fall apart.  Resilience is the ability of the bridge to withstand stress, like bridge traffic,  wind, waves, and the occasional earthquake (hey it’s California!).

Frailty research has come a long way.  We’re now at a point where frailty can be measured automatically, or electronically, as we put in the title.  Kate created an eFrailty tool that measures frailty based on the electronic health record (EHR) data.  Ariela created a VA frailty index based on the EHR of veterans.  And Dae created an index using Medicare Claims.  Today we’re beginning to discuss not just how to measure, but how to use these electronic frailty indexes to improve care of patients.

We should not get too hung up on battles over frailty.  As Kate writes in her JAGS editorial, “If geriatricians wage internecine battles over how to measure frailty, we risk squandering the opportunity to elevate frailty to the level of a vital sign. Learning from the past, a lack of consensus on metrics impeded the mainstream adoption of valuable functional assessments, including gait speed.”

To that end, modeled after ePrognosis, Dae and Ariela have launched a new tool for clinicians that includes multiple frailty measures, with guidance on how to use them and in what settings.  It’s called eFrailty, check it out now!

Did I cheat and play the guitar part for Sting’s Fragile at ⅔ speed then speed it up?  Maybe…but hey, I still only have 2 usable fingers on my left hand, give me a break!



Additional Links:

eFrailty website is: ( is fine).

Dae’s Frailty indexes
CGA-based frailty index web calculator for clinical use:

The Medicare claims-based frailty index program for research:

Ariela’s VA-FI:
Original VA frailty index:

ICD-10 version

Link to the code for investigators (included in the appendix): 

As an FYI for those in VA the code is readily available through the Centralized Interactive Phenomics Resource (CIPHER)

Recent validation against clinical measures of frailty:

Kate’s eFrailty Index

Our original eFI paper

eFI and surgery

JAGS Editorial -The future of frailty: Opportunity is knocking

JAGS Editorial -Frailty as an upstream target for intervention: A unifying approach to intervening in the trajectories of health, function, and disease in late life


***** Claim your CME credit for this episode! *****

Claim your CME credit for EP303 “Electronic Frailty Indexes”

If you have not already registered for the annual CME subscription (cost is $100 for a year’s worth of CME podcasts), you can register here

For more info on the CME credit, go to

Moderators Drs. Widera and Smith have no relationships to disclose.  Guest Kate Callahan and Ariela Orkaby have no relationships to disclose. Dae Kim discloses being a consultant for VillageMD and Alosa Health on 12/31/2022.

In support of improving patient care, UCSF Office of CME is jointly accredited by the Accreditation Council for Continuing Medical Education (ACCME), the Accreditation Council for Pharmacy Education (ACPE), and the American Nurses Credentialing Center (ANCC), to provide continuing education for the healthcare team.

University of California, San Francisco, designates this enduring material for a maximum of 0.75 AMA PRA Category 1 credit(s)™. Physicians should claim only the credit commensurate with the extent of their participation in the activity.

Successful completion of this CME activity, which includes participation in the evaluation component, enables the participant to earn up to 0.75 MOC points per podcast in the American Board of Internal Medicine’s (ABIM) Maintenance of Certification (MOC) program. It is the CME activity provider’s responsibility to submit participant completion information to ACCME for the purpose of granting ABIM MOC credit.

ABIM MOC credit will be offered to subscribers in November, 2024.  Subscribers will claim MOC credit by completing an evaluation with self-reflection questions. For any MOC questions, please email


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

Alex: This is Alex Smith.

Eric: And Alex, we’re going to be revisiting frailty today.

Who do we have on our podcast today?

Alex: We are delighted to welcome Kate Callahan, who’s a Geriatrician Researcher and Associate Professor at Wake Forest University School of Medicine and Advocate Health, and Creator of the eFrailty Index. Kate, I’ve know you well for a long time since residency at the Brigham. Welcome to GeriPal.

Kate: Delighted to be here, Alex and Eric. Thank you so much for having me.

Alex: And we’re delighted to welcome Dae Kim, who’s a Geriatrician Researcher and Associate Professor at Hebrew SeniorLife and Beth Israel Deaconess Medical Center in Harvard Medical School, and Creator of a claims-based frailty index colloquially called Medicare FI for Frailty Index. Dae, welcome to GeriPal.

Dae: Honored to be here.

Alex: And we’re delighted to welcome Ariela Orkaby, who is a Geriatrician and has training in preventive cardiology. And she’s a Researcher, an Assistant Professor at Brigham and Women’s Hospital and Harvard Medical School in Boston VA, and the New England GRECC, and Creator of a VA-based claims-based eFrailty Index, and with Dae, is launching an eFrailty website, We’ll talk about that today. Ariela, welcome to GeriPal.

Ariela: Glad to be here. Thank you.

Eric: Back in 2021, we had Linda Fried talk about frailty, so we’re going to revisit frailty in particular to talk about eFrailty as well. What is it? How should we be thinking about it? But we got a lot to cover, including, I got to remind myself what frailty is. But before we get into that, Kate, do you have a song request for Alex?

Kate: I do have a song request for Alex. I would love to hear Sting’s Fragile.

Eric: I was going to ask, why’d you pick this song? But it seems fairly clear.

Alex: Clear enough. Yeah. Here’s a short version for YouTube. You get the longer version for those of you listening. (singing)

Eric: Wow, very apropos of the topic. So, let’s talk about it. So, can we jump into the question of what is frailty? Because I remember, even though we did an entire podcast with Linda Fried, I’m still a little bit confused. I always go back to-

Oh, is it Justice Oliver Wendell Holmes’ definition? “You know it when you see it.”

Kate: That is true.

Eric: How much should we be thinking about that versus something else? I’m going to start off with you, Ariela. Can you tell me what frailty is?

Ariela: Great question. You’re right, we think we know it when we see it. But actually, we don’t always know it when we see it. So, I like to think of frailty as giving me some insight into a patient who is vulnerable in the event that they were to experience some kind of negative event, whether that’s getting sick or undergoing a surgery. And then, how do we then define it? How do we measure it clinically is, I think, really the question that we all struggle with.

And so, there are different ways of thinking about how do we actually measure or define frailty? Because we think we know it when we see it, but we don’t always. And so, whether that’s somebody who’s walking solely, whether that’s somebody who’s not managing well with a lot of different morbidities and that’s impairing their functional status, or whether it’s their overall assessment once we do a comprehensive geriatric assessment and we say, “There are a lot of different parts here that are not working well.”

So, I’ve actually just given us multiple different ways of defining frailty, which may not be fully answering the question, but getting at the complexity of what is frailty.

Eric: So, it’s super complex. And I agree, we’re probably under seeing it if we were just looking at somebody and, “Oh, yeah, they look frail.” So, the know it when you see it. Kate, how do you think about it?

Kate: I just read that about… Yeah. I was just going to say that studies show that up to 40% of the time, we’re wrong when we say we know it when we see it. So, there’s the Potter Stewart rule. The end of the Battagram is not super reliable. And I think it’s important that we know the direction in which it’s not reliable, which is that most often, when we’re eyeballing somebody and we think that they’re frail, or I’m sorry, when we’re eyeballing somebody, the problem is that we think they’re okay and they’re not.

We think that this person is going to sail through whatever stressor is in front of them, if that’s an infection, if that’s an injury, a surgery, whatever it is. But in fact, this person is, like Ariela said, vulnerable. They don’t have that much gas left in the tank. And so, that lack of reserve means that when they get hit with another wave, that’s a problem.

I’m mixing my metaphors here, but I think my favorite metaphor for frailty is one that I’ve heard Andrew Clyde use, which is that, “A frail, older adult can be like a beautiful origami boat on a pond.” If you’ve ever made those paper boats with your kids or whatever, they can sail around for a while, looking great, and as long as there’s no chop on the water, as long as there’s no breeze. But once that comes, it’s very difficult to recover from that event.

Alex: Wow, that’s a great analogy.

Eric: Well, I’m trying to think about that analogy. Because I feel like the chop, the wind, the rain, those are the stressors, and frailty is something intrinsic to not the stressors, but intrinsic to the origami. So, I always think back to George Kuchel’s… God, I don’t even remember. What year was that, Alex? George’s article?

Alex: I don’t know. It’s like six or seven years ago.

Eric: We’ll have a link to it and we’ll include the picture, which is the Golden Gate Bridge. Me and Alex, San Francisco.

Alex: Yeah, we love it for some reason. We’re biased to the San Francisco area.

Eric: I think the three and him, what I remember is that we have phenotypic frailty with the Golden Gate Bridge, where it’s the structural things going wrong with the Golden Gate Bridge. So, think about phenotypical frailty is one. The second one being…

Alex: I think the phenotypic frailty were the iconic pylons off of which the cables run, and then the frailty index were all of the little struts that go up and down the cables that take the weight. And there are so many more of those than there are the major pylons. Does that sound right?

Eric: And it’s the cumulation of those deficits in that that matter. And then, he also threw in resiliency is the ability for the bridge to bounce back from stressors, and defining that as that resiliency component. How do you think about that? Does that analogy work?

Dae: Yeah, I think that analogy is very clever. Because when we think about it, when we look at the biology of frailty and there’s a lot of interest in understanding geroscience and what really happens with aging at the cellular or subcellular level, so all these changes, the hallmarks of aging that affect the cellular function that leads to dysfunction or impairment in physiologic system and organ system. And then, that give rise to clinical manifestation of frailty that we, as a clinician, can measure by physical exam, by doing a blood test and other observations.

So, when I think about these two main theories of frailty, like physical frailty phenotype and deficit accumulation frailty, these are just too correlated, but still, different manifestation of underlying altered physiology that results from cellular or subcellular changes with aging. So, if we happen to measure that five factors that go into the physical frailty phenotype, we are really capturing phenotypic frailty. If we’re looking at multiple abnormalities that occur, different organ system, we are really capturing the total amount of abnormal physiology that’s going on in an aging person.

Eric: Okay. So, let me get this straight, and correct me if I’m wrong. So, frailty, you have this aging-related syndrome. You have this physiologic decline and we know that’s associated with a bunch of bad outcomes. So, the very first one, physical frailty or phenotypical frailty, this is the Linda Fried Index? Linda Fried Frailty Tool?

Ariela: The Fried Physical Phenotype.

Kate: Yeah.

Eric: Well, five things. Weight loss, exhaustion, weakness, slow walking-speed, and decreased physical activity. Those are the five core features of phenotip frailty. We won’t spend a whole lot of time on that. If you’re interested in learning more about that, take a listen to the Fried podcast that we did. So, when we think about the deficit accumulation or index frailty, can you, one more time, explain that to me?

Dae: So, because the aging process happens to the entire body, so why not just measuring multiple things from multiple organ system? So, the more things abnormal with a person, the more frail that person is.

Ariela: The other thing I would add is that the physical phenotype only considers physical contributions to fit to frailty. And so, it doesn’t account for things like cognition and mental health nutrition, well, a little bit nutrition, but not really, socioeconomic and environmental situations.

And what the cumulative deficit approach does is it takes a wider lens approach of all the different inputs that can be contributing to that vulnerability in an older adult. And actually, frailty, although associated with age, is not necessarily only seen in older adults, which I think is an important thing that we’ve learned, especially with the accumulation of deficit approach.

Kate: Absolutely.

Eric: Yeah.

Kate: I think that philosophically, these are different constructs. We’re getting at this thing that we all see, which is, I think if you talk to people on the street and say, “What is this?” there’s some sense that there is a vulnerability that comes with age, whether you see it or not. But I think two ways of thinking about it. One is the phenotypic frailty is a very biologically-based conceptualization, whereas one of the cool things about the deficit accumulation is that it’s bringing in concepts from engineering.

This is a system, and I’m not an engineer, but this is a system in which multiple things are going wrong. And so, that instability in the system means that adding anything more is going to make the whole organism behave differently. And, in fact, this is a stochastic random process, that we are not programmed like a simple computer, that we go from A to B. We have multiple inputs. There’s lots of pathways to frailty.

I sometimes describe it, the deficit accumulation spits out a number between zero and one, and that can feel a little reductive. But at the same time, we deal with a lot of numbers. It’s really helpful. I think of it like a creatinine of aging. Your creatinine doesn’t actually tell you what your kidney problem is from. It tells you that there is a problem, and it also tells you some things that you probably shouldn’t do. So, if you have a creatinine of two-and-a-half, you should not be getting cat scan dye injected on a regular basis.

If you have a high frailty index, there’s things that we should do to protect in that setting and consider before we move down that space. It will take more digging to say, “All right, well, why might this be happening? What is the driver here?” There will always be a role for more explicit focus from, say, a comprehensive geriatric assessment or other processes. But having a place to start, where you can say, “Hey, at a global picture, this is new trouble ahead,” is helpful.

Alex: Yeah. Can I ask you a question, while I have the experts of frailty here on the podcast? One of the issues that I had, and I know one of my mentors had, I don’t know if he wants to be named so I won’t name him, with the phenotypic frailty I think, maybe also the frailty index, is that it excludes people who have disability and who are in nursing homes. Because that’s an outcome of frailty, not frailty.

But when I think of people in nursing homes, I think of them as very frail. And I think of frailty in the simplest sense as vulnerability to stress. And is that too simplistic? And is there a problem with my thinking about those people in nursing homes as being very frail, just as somebody who’s walking across the street on a crosswalk very slowly is also frail?

Dae: Yeah, I think that’s really the core concept of frailty, which is the vulnerability to a stressor. I think the distinguishing frailty from multi-morbidity and disability as a physical frailty phenotype does, if you focus on the prevention of somebody developing disability or being in a nursing home, then it is important to detect a pre-disability state, which is still vulnerable. They have not developed disability yet.

So, in that sense, not including people with existing disability already in the nursing home makes sense, because that’s an opportunity to prevent those bad outcome in the future. But when we are seeing these patients in the clinic, we need to manage these patients, including patients in the nursing home, patients who already have disability.

So, we still need to have a concept that is broad enough to put these people on a spectrum of fit to really advance frailty, and decide clinical decisions based on their vulnerability, and then goals and preferences of the patients.

Ariela: And I would say, not every nursing home patient is frail. There’s certainly a much higher prevalence of frailty in a nursing home, but it wouldn’t be true to say that every nursing home patient is frail.

Eric: Yeah. And I just-

Kate: Well, I don’t know what you all think, but do you think… Oh, I’m sorry. Go ahead.

Eric: No, go ahead, Kate.

Kate: I was just going to say, I don’t know that it’s important that we exclude those with disability. I think that one can be frail and have a disability, one can be frail and not have a disability. You go back to the landmark-free paper where there’s the Venn diagram overlapping. You could have 50% of the people who have multi-morbidity also had frailty, and something around, I want to say, 25 or 30% also had a disability. And you guys correct my numbers if I’m wrong. But you know what? What you’re looking at in that situation is that they may be vulnerable from another perspective.

Does someone have a disability and then they’re vulnerable to infection? What is the outcome? And I think that there’s, unfortunately, myriad adverse outcomes that are associated with aging that are more common in those with frailty. And when you’re thinking about it, what you’re thinking about is, how can we prevent bad things from happening, and worse outcomes that negatively affect quality of life, and negatively affect the experiences that an older adult is having? And that we’re not delivering a one-size-fits-all approach. Personalization is what frailty makes possible.

Eric: Can I ask you? Just because you brought up not all disability is frailty. I get that from the Fried Frailty Tool. Because you have these five things; weight loss, exhaustion, weakness, slow-walking, decreased physical activity. There may be some association, but not all diseases people would… You can have CHF and you cannot be frail by this definition. COPD, you can have disability and still not be frail by this definition.

And I’d love to talk more about the index frailty or disease accumulation frailty, because that feels like we could be calling somebody with diseases or multiple diseases, each one of those being a deficit accumulation. Potentially, is that the same as being frail, now that you have three different diseases?

Ariela: So, it’s a really good point. And if I may clarify, frailty index is not a disease index. And so, it’s an accumulation of deficits across multiple different systems and different types of variables that we think about associated with age. So, if we were just counting up diseases, we’ve got great tools that do that. That wouldn’t be frailty.

If we were just measuring disability markers or function, we’ve got tools that do that. That would not be frailty. It’s putting these all together, where we’re combining function and chronic diseases and their management, and polypharmacy and falls and all of the different components together. You can have three or four chronic conditions, but you’re not frail, right?

Kate: Mm-hmm.

Eric: Yeah.

So, let’s talk about that. What are the tools? I know you all have some tools. You don’t have to talk about all of them, but what are the things that you look at? I’m going to start with you, Kate.

Kate: Sure.

Eric: Your tool is called what?

Kate: Well, we call it the eFrailty Index, the Electronic Frailty Index. And here, at Wake Forest, it’s been embedded in our electronic health record since 2019. So, that’s dot eFI. You type it into our local version of Epic. And that’s been a great tool.

To Ariela’s point, it’s not only diagnoses. And to further that point, the diagnoses that are drawn in are not just any random diseases. It’s important that they are associated with aging. So, we pull in diagnosis codes, yes, but we also pull in laboratory values, biometrics, vital signs, health behaviors. And we’re available, and this is where we’re fortunate. Because we have a large presence of the Accountable Care Organization, Next-Gen Accountable Care Organization.

So, we have a fair amount of completion of the annual wellness visits and therefore, we are also able to capture function. And those elements come into it as well. So, ours has a total of 54 deficits. Like I said, it’s a simple proportion of what somebody has over the total number measured. That gives us a number between zero and one. And typically speaking, there’s some variability, but we use a cutoff of 0.21. That if a value is above that, then you’re categorized in the frailty, in the frailty scope.

Alex: And just to drill down on this a little bit more, your eFrailty Index then, to distinguish it from others, it’s not a claims-based index. It’s an electronic-health-record-based index? Would that be fair to say?

Kate: Correct. So, what we do is we extract and derive data that comes from routine clinical care. So, our process downloads that data every week, and Epic has its Saturday-night bash and goes offline and updates everything. And REFI gets updated at the same time. And then, everything is calculated externally to the EHR and then ported back into it.

And we’ve been able to map this not only to Epic, but also to Cerner. And so, we’re able to provide to Cerner as well. But yes, it’s an EHR-based frailty index. So, there’s lots of pros and lots of cons when it comes to electronic health record data. I’ll be the first to tell you the limitations.

Alex: List the top two of each. Top two pros, top two cons.

Kate: Well, I think the top two pros are, there’s a wonderful depth and breadth of information. Being able to say, “All right, well, I know what their hemoglobin was, or their albumin,” or whether or not they had had a fall recently. And it’s fresh data.

One of the cons that goes along with that is, diagnoses in the electronic health record can be like barnacles on the side of a ship, and people don’t always go through and scrape everything off again. So, that can be a downside.

Eric: Yep, it’s only as good as the information people are plugging in.

Kate: It is. It is. Another good thing about it is that because it’s being drawn from the data that our health system is using to make decisions. And so, our population health team that is tasked with this huge dilemma, which is, how do we look at a huge population of older adults and figure out who needs what?

You need something that is scalable. You need something that is quick and that you don’t have to necessarily leave the health system infrastructure in order to calculate. So, we’ve been able to apply the eFI for care navigation, for prioritization for COVID vaccination to make sure that the frail folks were the first to get the COVID vaccine, and other decisions.

Eric: Also, those are system levels. Individual providers, do they do something different when they see this number, you think?

Kate: Some of them.

Eric: Or do you? What does Kate Callahan do?

Kate: It’s a push-pull. I do track my frailty index. The residents make fun of me. They say that they always know when they have a patient of mine, because their eFI is in the first sentence. But I think that what it tells me is when my eyeball test is wrong. I’m a geriatrician. I look at gait speed, I look at all of these things. But sometimes I’ll have someone who just moves great and they’re doing great and their cognition is great, and I’m thinking, “All right, we’re cooking with gas.”

And then, I see their eFI and I realize, “Oh, man, I didn’t realize that they were at 0.27, 0.3.” I have to remember that physiologically, this is somebody who is not squaring. And that may be because they have had outstanding health behaviors, they have had the support of a good income and a good education and the time to take care of themselves. But that, even so, when that individual gets sick, they still have the loss of reserve. It is not-

Eric: And maybe doing even some, for surgery, prehab before the surgery for those individuals.

Kate: Sure, prehabilitation. Absolutely. And we’ve tested a few different… We have a few different things in tests to say, how might we consider this? So, for instance, de-prescribing for high-risk medications in diabetes, your prehabilitation for surgery. We’re testing looking at targeting unmet social needs in folks with frailty. So, yeah.

Eric: Do you know if it’s improving outcomes yet, or is that a little farther down the line?

Kate: Not yet. Further down the line.

Eric: Well, I want to hear from you, too-

Kate: I’ll come back in a couple of years.

Eric: Ariela, I’m going to turn to you. You also have… Is it in the VA?

Ariela: Yeah. So, ours is what we’ve named the VA Frailty Index, and we’re not quite at point of care the way that Kate has been able to move her frailty index. So, what we did, we started back in 2016 or ’17 or so. We saw what was going on in the UK with Andy Clegg’s work, who had applied a claims-based frailty index to primary care across the US. And we said VA, it’s an incredible integrated healthcare system, the largest one in the United States. Something like 10 million active veterans who come through the system regularly.

And so, we said, “We’ve got all this rich data that goes back for decades, at least going back to about 2001, 2002.” And so, we developed based on a comprehensive geriatric assessment with a few geriatricians sitting around a table in 2017. We said, “If we could go into the health record and look at the claims that are being coded from routine care, what would we pull out that would fit with a frailty index based on accumulation of deficits model?”

Eric: Okay. So, I got to ask the question, because Kate was talking about performance tests, physical function. I’d imagine that would be very hard to do on routine. What’s in your frailty-

Ariela: Excellent question. So, we settled on 31 different variables that map to multiple different domains. And in order to capture function, one of the things that we did, which I think is quite novel, is capturing durable medical equipment use.

And so, rather than just saying, “This is somebody who has a code for durability,” let’s say, we could also pull, “Are they needing a wheelchair, walker?” And because of the VA and the way that veterans have access to care, all of these things are being prescribed and paid for by VA. So, the capture of this information for veterans who are coming in for regular use of VA care across the nation is actually remarkable.

We also were able to pull in codes for slow gait speed, which is used to help refer people to physical therapy, and multiple different functional-related codes like that, which is different than just saying fall, or associated fracture, let’s say.

Eric: Okay.

Alex: Can I ask how reliable you think those durable medical equipment codes are? Are there people who have medical equipment who don’t have a code for it in the VA, or people who have a claim for it, but don’t use it, and don’t need it? I don’t know.

Ariela: That’s a great question. Whether somebody’s using it or not, we certainly can’t say that. But we tried to be thoughtful as we picked out which durable medical equipment we were using in looking back at the literature and what others had done to try to capture mobility impairment in claims data. And so, when you have a physical therapist who’s prescribing a walker, there’s a pretty good reason why somebody needs that.

And also, we tried to not just look back at forever. “Were you ever prescribed something?” But rather, in the last year or all the way back to three years for people who are only coming… For the healthy, well, you only need to come in once every three years to maintain your VA primary care. So, we don’t want to lose people who are healthy and well.

Eric: I also wonder, as you think about this, as you put this together, I imagine there’s a sensitivity and specificity discussion. In part, it feels like you want to use these tools, like a broad scale. Kate, that sounds like what you’re doing. Broad scale, look at a lot of people, figure out who may be frail. You may not figure exactly who is frail, but it may be the question of who may be frail and who needs more assessment. Rather than high specificity, this person is absolutely frail with losing a lot of folks in that because of poor sensitivity. Is that how you think about this?

Ariela: That’s not quite how we approached this. We really approached the VA Frailty Index based on the comprehensive geriatric assessment frailty index approach, where we said, “We want to capture domains related to cognition, related to mental health, related to medical conditions, related to function,” and so forth. And so, that’s how we settled on those 31 variables.

We’ve actually compared it to an in-person frailty assessment, both gait speed, functional assessments, a frailty index from comprehensive geriatric assessment and a few other things. It correlates reasonably well. It’s not perfect and that’s because we’re relying on claims, just as Kate was saying.

But as we’re moving towards making this tool available for clinicians at the bedside, where we’re starting is actually at the population health level, so that the frailty index can be available to all primary-care clinic teams across the nation.

Eric: All right, Dae, I’m going to turn to you. I actually want to have time to talk about the eFrailty website that you’re doing. But real briefly, you’re also doing something with Medicare claims, is that right?

Dae: Yes. So, actually, I do have two versions. One is a clinical version that we use in our EHR, and also there’s a web-based calculator. That’s a CGA or comprehensive geriatric assessment-based frailty index calculator, which is a 50-item frailty index using the information that we collect from comprehensive geriatric assessments. So, this is not based on the claims, but it is actually based on the assessment. So, the clinician has to do assessment of their functional status, physical performance, cognitive function, but it makes the calculation of FI and the visualization of the results more appealing.

Eric: FI? Frailty index?

Dae: Yes. Yes.

Eric: All right.

Dae: So, you don’t have to count while you’re actually seeing a patient like, “Oh, how many I counted? The thing’s abnormal.” And so on. So, that’s the clinical part. That’s embedded in our EHR at Beth Israel Deaconess Medical Center. And then, the Medicare part is what I use for research, using Medicare. So, it’s pharmacoepidemiology research and health services research, looking at outcomes of drug therapy, surgical procedure, and health services, using claims data.

In that setting, we developed a Medicare claims-based frailty index that uses 93 variables defined using diagnosis code, procedure code, and durable medical equipment code, that Ariela just explained, in the past year to capture someone’s deficit accumulation frailty index. And we found that the cut point, it ranges from zero to one, just like a deficit accumulation frailty index. But about 0.25 is the cut point that we use.

And that achieves about 70 to 80% sensitivity and specificity in detecting frailty phenotype, as well as the frailty based on the clinical deficit accumulation frailty index.

Eric: All right. Now, my head is about to explode because we’ve talked about three. I’m guessing there’s a lot more frailty index.

Alex: Yeah.

Eric: Is there an easy button I can press to help me figure out which tool should I be using?

Dae: Yeah, that’s the most frequently-asked questions to us. “There are so many validated tools. Which one should I use?” I always tell, “Depends on what you are measuring frailty for, what the purpose is.” So, we, as a team, Ariela and many others at Harvard and the affiliated hospitals here, we came up with an algorithm and then we also developed the web-based interactive calculators for 15 different validated frailty tools. 10 for general purpose and then five for specific populations, such as surgery, TAVR, cancer, and so on.

So, we developed a web-based calculator. You plug in numbers and it will calculate the results and give you the results. And then, we have an algorithm that asks, what’s the reason for assessing frailty? I just want brief risk assessment. I want more comprehensive care planning. I want somebody’s risk stratification before a stressful treatment, such as chemotherapy or surgical procedure. And answering that question will give you options that have been tested in that specific setting and that you can use to measure someone’s frailty level.

Eric: And that’s website?

Dae: Yes.

Alex:, launched. We are recording this in advance, prelaunch, but we won’t publish it until it’s launched. So, at the time that you hear this, dear listeners, just go to and check it out.

Eric: Yeah.

Dae: I just want to add that I was inspired so much by ePrognosis and how much I used and then I enjoyed using it. And then, I was thinking, “I should do something like this for frailty.” There are so many frailty tools and people just have to reach them.”

Eric: I love it.

I guess a question for Dae, and a little pushback too for all of you is, so ePrognosis, we have indices, mortality indices on ePrognosis. They include function, they include diseases for some of them, some lab values. Are they just mortality indices, or are they frailty indices, too? Because they’re predicting bad outcomes.

Ariela: They’re certainly related, but what’s been fascinating as we’ve learned more about frailty and how distinct it is from multi-morbidity or disability and so forth, is that it can help reset life expectancy estimates at every age, and it can help identify overall risk beyond just mortality. So many of our patients, and my patients will say to me, “Death is part of life. That’s going to happen, eventually. What I really worry about is, how many more years can I spend at home? Am I going to fall? Am I going to be okay through the surgery?”

And so, I know that my group has worked on looking at time to long-term care needs using the frailty indices. So, how interchangeable all of these different scores are, I think that remains to be seen. But when we think about frailty, we’re not just looking at mortality and endpoint. We’re looking at so much more. And Kate, I wonder if you want to add to that.

Kate: Yeah. So, I agree, people can live with frailty for quite a while. And I think that particularly the frail folks, there’s so much that we do that worsens outcomes for the frail that if we were to change how we approach them in the way we deliver care in the health system would help everybody. It’s the curb-cut effect of like, you cut the curb so that you can have people who use wheelchairs in order to move more readily. But then that helps people are making deliveries and have strollers and so on. If we make the healthcare system better for those who are more vulnerable with frailty, that helps everyone.

I also wanted to do a quick pitch about using frailty as an outcome. The question that George Kuchel asked a while, “Are we ready for this?” Recognizing that yes, frailty is associated with all of these negative sequelae, but in and of itself, thinking about what we can do upstream, today’s point about the geroscience hypothesis and how are we integrating this? And Ariela will hopefully have some great news about some work that she’s leading and I get to tag along with about looking at how we can prevent frailty.

Because I think there’s going to be elements, yes, at the individual level, but also at the social and structural level in the way that we deliver care, that may be able to prevent or delay the onset of frailty. And thus, really, the gap between health span and lifespan that we’re all trying to get to be closer.

Eric: Yeah. Go ahead, Dae.

Dae: Let me actually add one comment quickly, to Eric’s question. Both the ePrognosis mortality prediction models and then many frailty models use similar information, such as function, lab value, and so on. But the mortality prediction model, the weight of that model coefficient are really optimized to predict mortality versus frailty. Mostly, it’s a deficit accumulation, some sort of other process-defining weight.

But when we compare deficit accumulation frailty index and, for example, lead index, the mortality prediction is similar. But when we look at some of the geriatric outcomes, such as worsening disability, falls, and sniff stay, frailty index does much better than mortality prediction index. So, that’s the difference between the two types of models.

Ariela: I would also add that most of the mortality scores are heavily weighted by age, and almost all of the frailty definitions don’t include age. Because we’re thinking about, what else is going on beyond that chronologic number?.

Eric: Well, that’s fabulous. I think, Alex, we’re going to need a revisiting revisited frailty episode, because we have a lot more to talk about.

Alex: Yeah, there’s so much more we can talk about. We didn’t get enough into what outcomes. What can we do about it? There’s a lot more to cover. Well, we have to end with-

Eric: Well, it sounds like it’s really complicated when it comes to what we can do about it, because what I’m also hearing is that it really depends on what you’re using it for.

Alex: Yeah, that’s right.

Eric: For the presurgical, we’ve talked about prehab for surgery patients. In your general medicine clinic, polypharmacy. So, it sounds like what we can do about it isn’t as simple as just, “You should do more exercise,” or, “Eat better.”

Kate: Can I make a quick outro, though? Which is that the work that Dae and Ariela have done, and that I’ve tried to do as well, gets us so much closer to being able to answer that question. A major limitation up until this point has been that if you didn’t go to a geriatrician, or if you weren’t in a research trial, there’s no way that someone would have their frailty level assessed.

Are these tools perfect? No, but the scalability of the means that more and more folks are going to be able to have a number, that gets us closer to understanding this. Which means that we now can move the field forward into doing tests that can say either, what can we do to optimally manage the people who are frail and deserve personalized care plans now, and what can we do to try to prevent or delay frailty? Because we can actually measure what’s happening. So, it’s incredible work and so inspiring.

Ariela: I would say we’re finally moving away from defining frailty to being able to do something about it.

Eric: Because that was the biggest question a couple of years ago, is defining frailty. So, it’s great that we’re moving towards first, then figuring out how do we actually catch people who are frail or pre-frail? And then, what do we do about it? With that said, Alex, you want to end this off with a little bit more Sting?

Alex: Yes, a little bit more.

Eric: Because we had such a beautiful outro from Kate. She had to take over by a roll.

Alex: Yeah. Here’s a little bit of what I can do with a broken hand. You get the more full version with chimes and synthesizer on the audio podcast. (singing)

Eric: Dae, Ariela, Kate, thanks for joining us on this podcast.

Kate: Thank you.

Dae: Thank you so much.

Eric: And to all of our listeners, thank you for your continued support.

Back To Top