Dr. Joan Teno Exposes the Hidden Problems with Hospice Ratings and Quality Scores | Part One

Are hospice quality ratings really measuring what matters most?

In this powerful episode of TCNtalks / Anatomy of Leadership, Chris Comeaux and Cordt Kassner sit down with nationally recognized hospice researcher Dr. Joan Teno to uncover the hidden problems behind hospice ratings, quality scores, and public reporting systems.

Drawing on more than 30 years of experience shaping hospice and palliative care policy, Dr. Teno explains why many publicly reported quality measures may fail to capture the true experiences of patients and families. She discusses the challenges of transparency, the unintended consequences of current reporting systems, concerns about fraud and oversight, and how artificial intelligence could help transform the future of hospice quality measurement.

Whether you’re a healthcare executive, hospice leader, clinician, policymaker, or nonprofit professional, this conversation offers valuable insights into the complex realities behind quality scores and what must change to better serve patients and families at the end of life.

In this episode, you’ll learn:
✔ Why hospice ratings may not tell the full story
✔ The biggest blind spots in today’s quality measurement systems
✔ How some providers avoid meaningful public reporting
✔ The role of AI and data analytics in improving oversight
✔ What meaningful transparency should look like for consumers

Chapters:

0:00     Leadership Sets The Tone
0:22.    Monthly Hospice News And Framework
3:38.    Why Oversight Beats Payment Redesign
8:28.    Transparency Limits And Small Hospice Loopholes
13:10.  Blind Spots In Quality Reporting
18:56   Using AI To Pick A Hospice
23:45.  Moratorium And Rebuilding Trust
27:30   Clinicians Needed To Catch What Models Miss
29:18   Part Two Teaser


Guest: Dr. Joan Teno
Adjunct Professor, Brown School of Public Health | Hospice & Palliative Care Researcher | 2023 AAHPM Lifetime Achievement Award Recipient

Host: Chris Comeaux, President / CEO of TELEIOS and author of The Anatomy of Leadership

C0-Host: Cordt Kassner, 

👉 Subscribe for more conversations on leadership, healthcare innovation, hospice care, and mission-driven organizations.

Teleios Collaborative Network   /   https://www.teleioscn.org/tcntalkspodcast

Transcript

Melody King: 00:00
Everything rises and falls on leadership. The ability to lead well is fueled by living your cause and purpose. This podcast will equip you with the tools to do just that. Live and lead with cause and purpose. And now, author of the book, The Anatomy of Leadership, and our host, Chris Comeaux.
 
 
Chris Comeaux: 00:23
Hello and welcome. I'm so excited. It is my favorite time of the month where we do the top news stories of the month. Cordt Kassner, welcome.
 
 Cordt Kassner: 00:30
Thank you so much.
 
Chris Comeaux: 00:32
Yeah, this is really fun. So as you and I adjusted this year, Cordt, um, we are always going to keep in the show notes because if any listeners like, when do you guys do the top news stories of the month? On the top of the quarter, Cordt and I will actually go past the over the past quarter together, but every month in the show notes. But what we committed this year is that we were going to use Dr. Ira Byock's framework. And if you guys will remember, zero tolerance for waste fraud and abuse, clinical and programmatic standards, making meaningful data readily available, and driving competition based upon quality. So we've been using that framework. And so we bring incredible guests. And this is a guest that we're super excited to introduce to you. Sure, I really feel like she needs no introduction, but Dr. Joan Tino, welcome. It's so good to have you.
 
Dr Joan Teno: 01:15
It's I'm just delighted to be here. And I want to give a shout out to all the hard work you guys are doing. Um, you know, I I just tell everybody they have to listen to your podcast.
 
Chris Comeaux: 01:25
Well, thank you. That's I'm so honored that you would say that. And you told me in the green room not to call you Dr. Joan Teno, and I did say that would be like telling Mother Teresa telling me you can't call me Mother Teresa, just Teresa. But I will honor your request and we'll call you Joan. But hopefully you take it as the compliment that is meant. And so I don't think you really need an introduction, but let me just go ahead and do it here. So Dr. Teno is an adjunct professor of health services, policy and practice, Brown School of Public Health. She's part of the adjunct staff of RAN Corporation. She's a retired physician, health services researcher. That's where she's done so much amazing, good for our field, whose work has helped shape hospice and palliative care quality measurement in the United States. That's why Cordt and I have been wanting to get her on the show for quite a while for this year, especially under Dr. Byock's framework. She's led research initiatives funded by the NIH, worked on the centers for Medicare and Medicaid Service contracts, including work on the hospice cap survey. Her research focuses on improving care for serious ill older adults, people with advanced dementia, and she has authored more than 340 publications. 340 publications, just put an exclamation point there, and hospice and quality care, end of life. Dr. Tino received a well-deserved 2023 AHPM Lifetime Achievement Award. She was appointed a fellow of King's College in London in 2025. How flipping cool is that? And every Saturday, she was telling Cordt and I about this in the green room. Her and her dog Bean volunteers a pet therapy team in a hospice inpatient unit. And Dr. Teno was sharing, Joan was sharing that. They just know my dog. They don't have a clue that the lady walking around with the dog has literally shaped a lot of this hospice and powder care field and has continued to do so. So again, Joan, it's so good to have you. Thank you for being here.
 
Dr Joan Teno: 03:14
Well, thank you for having me. I'm just delighted to be here.
 
Chris Comeaux: 03:17
Yeah. Cordt and I have been chomping at the bit, so we kind of divvied up our questions back and forth. And so really looking forward because quite often in so many shows last year, not just the top news stories of the month, but we just keep bumping into some of the challenges and the way we measure quality today, the number of hospices that don't even participate in the CAP surveys, et cetera. So, we're gonna let you unpack all of that. And again, to connect the dots to Dr. Byock's framework, our focus today was really about making meaningful data readily available and driving competition based upon quality. So, Cordt, you want to go first?
 
Cordt Kassner: 03:49
You bet. And Joan, I'll just echo uh all the accolades Chris has been mentioning about you. I I will say for me personally, uh I have shared in presentations, I share through Hospice and Palliative Care today, uh a very common quote, which is if Joan writes it, I read it. And I just really appreciate not only the academic work that you've done, the way that you've shaped CMS hospice policy, but I've really been enjoying your Substack because that feels like a maybe a more personal uh communication, like a letter to me. I just wanted to let Cordt know, hey, here's what I think about hospice in Rhode Island or whatever. It's so uh so interesting. So I really want to encourage all of our listeners to pay attention and you know take a look back at the uh articles, the book chapters that you've written in the past, the Substack that you're writing now, because it really is uh incredible uh insight and helps shape where hospice is driving.
 
Dr Joan Teno: 04:54
I view the Substack as sorting uh writing a letter on things I wish I could have done better and my hopes for the future. Um, you know, I've been so lucky and with a lot of gratitude that uh I've worked in hospice for almost 30 years. And uh, you know, what makes hospice great, and you know, well, I know we're having a lot of discussion about fraud, but what makes hospice great is the staff. And I think the majority of hospices out there are just really excellent. Um, there's a lot of fake hospices in LA County, which we need to deal with now.
 
Cordt Kassner: 05:30
Absolutely. And that kind of launches into the first question that that we have. You've
 
Cordt Kassner: 05:35
 written extensively about how ownership structure influences hospice behavior, especially the growth of private equity-backed and highly fragmented for-profit providers. At this point, do you believe the current Medicare hospice payment model itself is structurally incentivizing undesirable behavior? Or is the larger problem inadequate oversight and transparency, or maybe something different?
 
Dr Joan Teno: 06:04
No, I think you  hit it right with your last comments. We need better oversight and better transparency. Um there's a uh sort of trust me part of Medicare Advantage plans. There's a trust me uh with an alternative payment model, which really the Medicare hospice benefit is. You know, you get paid a daily rate. Um, there's no set really minimum requirements on how often you have to visit a person. So there's a really trust me aspect, and there's unfortunately uh huge profit margins that can be made by enrolling patients who are known to have a long length of stay. You know, I don't know how these various CEOs and CFO balance the fact that you have a benefit where maybe a third of your patients are not profitable because of their very short length of stay, because all your costs are upfront. But to really get at your question, where I think we need to go is really data-driven oversight. We have the data now, let's use it. Uh, I am getting more and more impressed with what AI can do to use analytics. So, if I had to say there was one thing that if I could go back 10 years and say we could have done, was to really look at geography and um the Medicare certification number. Um, right now, Medicare certification number really relates only to um a state, but uh it's very clear when you look in hindsight, and hindsight's always 2020, that there were these clusters in LA County that spread like the measles out to Harris County, to Maricopa County, and you saw this sudden rapid growth of hospices. I was so impressed with the deputy director when she said there was 4.5 Starbucks or 4.5 hospices than the times of Starbucks. Well, I have another data for you. In LA County, there's one hospice for every 38 deaths. Um, so obviously, you know, in in retrospect, we didn't deal with geography and the clusters. Um, and I think there are some very appropriate tools in terms of, you know, uh ADRs, um, the RACs, TPEs, and then UPEX to really start dealing with this. Um, I think transparency, we really need to have a debate about transparency. Um and uh part of that debate should be uh how much information do we give consumers? So right now, uh, I'm gonna go a little statistical on you and I'll try to go easy on statistics, but right now, through the process, you have to have something that's called sufficient inner unit reliability, which means that you're getting more signal than there is noise. So, you can say hospice A is better than hospice B. Now, the problem that we have run into with this inner unit reliability is small hospice programs. And the sort of suspicious side of me thinks there's a lot of programs are purposely flying under the radar by staying small. And in that staying small, they're avoiding public reporting. Uh, and they're also uh avoiding, you know, um, you know, hopefully not being subject to any of these TPEs or these UPICs because, you know, they're not going to go after me because I'm a small provider. We've created a huge problem with growth of CCNs and flying under the radar that hopefully will be corrected under this current thing, current administration with the current deputy director. Um but we still have this problem of how much information to give consumers. Uh, I'll give you an example. Um look at the hospice care index. Um hospice care index basically is topped out, and a lot of it reasons it's topped out is to achieve this sufficient inner unit variability. And, you know, um you take a measure like the burdensome transition measure, or they call it BT1, which is a transition from hospice discharge, hospitalization, and uh hospice readmission. Um for the most part, that's fairly low. Um, but I would argue that it's almost equivalent to a sentinel event. You know, it's something you want to avoid and it also can be a suspicious behavior. Now, I don't expect it to be zero, but the issue is they can't achieve sufficient inner unit reliability unless there's fairly large numbers. Yet that information as packaged within the HCI is not really providing information to consumers. So, you know, maybe one potential solution is to say, well, let's report it as a sentinel event and you know, and let's educate consumers how to interpret the data. Maybe I have more trust in consumers, but I don't know about you, but every day I go to Amazon, I go to Yelp, I go to Google, and I know how to deal with the sample size and know how to make decisions. So, you know, maybe we should give consumers more information and uh, you know, not worry as much about this interunit reliability, but that's beyond my pay grade. Um, but I would I think what's exciting to me about the AI revolution is maybe it's going to empower consumers to make better choices.
 
Chris Comeaux: 11:54
That's really good. Thank you for what you just said, Dr. Teno. You educated me. And Cordt and I again keep bumping into this quite often, just over the past two years together. Um, I did share it with Kim Brandt in the podcast. I can't remember if it was on air or off air. I think it was a bit of both.
 
 Chris Comeaux: 12:10
 That, um, and I do think it's intentional. I'll just go out on the limb, that you know, they've got these big brands, but in the reality, if you start putting it under a microscope, it's literally a bunch of little 10, 15, 20, 30-day patient-day hospices and multiple provider numbers. And then we wonder why some of these quality programs are getting the audits, it's because there's one provider number. It's, you know, when my granddaddy taught me like that fishing hole's got a lot of fish in it. That fishing hole doesn't have many. Well, those little hospices, why are they not going to go audit them? I don't think there are a lot of fish in that fishing hole. But the reality is it's an aggregated as a, if you look at it as a whole, all of these little 10, 20, 30 patient-day hospices under the one brand, it's almost a strategy to fly under the radar screen. I feel like it is a bit intentional. And it does seem like we need to do something to affect that, which maybe I'll I know you'll probably want to make a comment to that, but let me go ahead and give you my second question, which is a good lead-in from Cordt's. So you've spent decades of your life just helping to build the hospice quality measurement. And I love what you say about life, hindsight is 2020, right? In your recent Substack series, which again, we're going to create a link for, you suggest that the current public reporting system may still miss major warning signs. What are the biggest blind spots in today's hospice quality framework? And how vulnerable is the system to gaming and outright fraud and why?
 
Dr Joan Teno: 13:38
Well, I think the outright fraud and why is a really testament when you hear the deputy director of CMS say, uh, maybe 800 to 1,000 hospice programs are going to, you know, be suspended and go on to be closed. Um, you know, when you start losing a thousand out of seven thousand hospices, you know, that's gonna affect rates. Um, it's gonna affect that. You know, I I've always been interested, um, MedPAC and uh CMS, they count a hospice if there's one bill. Um and, you know, I think we all know uh Judi Lund Person always told me that, you know, the way CCNs were assigned were sometimes every branch office got a CCN because that was easier for them to do the data collection. And then you're the opposite extreme, you have a CCN for the entire state. Um, you know, one provider uh has a CCN for the entire state of Florida. And, you know, that can hide a lot of variation within all the various locations um in that thing. So I think we need to, you know, if you want to make me the CMS, you know, advisor for the day, which is probably never gonna happen. Yes, I do. But that's the point of this podcast. I would really love them to fix the sort of geography thing um and think and link the CCN more clearly to um a service area. Um, you know, even if they don't want to create a hospice service area, start with counties. You know, you have deaths reported by counties. You could easily, you know, track a rapid growth of hospice programs. Um so one more point, hopefully you'll let me make and bear with me. Um PPO, PPEO, I love all the acronyms of CMS. I I really think I'm gonna write a a substack note just listing all the acronyms they have because they don't have to. That's a great idea. That's a great idea. But PPO only goes after new hospice programs. Okay. But the problem is not new hospice programs. I was like, you know, I started playing with the data. Um in, you know, in 2019, there was 481 hospices in LA. And out of that 481 hospices, back in 2019, 208 did not have a sufficient number of visits in the last, you know, the visits in the last three days of life to report the visit measure. Um, so visit in the last day, out of that 208, 130% or one third uh were listed to the same building address. Um so you know, PPEO going after new providers is not going to pick this up. And, you know, the data I'm re reporting back on those 208 out of those 481 who um didn't, you know, the who didn't report their visit data is in 2025. So I really think, you know, CMS has the data to do it. Um the other thing is CMS starts needs to start looking at longitudinal data instead of looking at just like a snapshot, how do things change over the snapshots? Um I think the AI revolution is here. Uh I really want CMS to embrace that AI rev revolution and really start thinking about how you use all these measures together to look for data inconsistencies. Um, my biggest concern when I start looking at some of this data is um, you know, it's not audited. Okay. So what's stopping a hospice from getting 90% of visits in the last three days of life and bumping up their SA payment, um, their service intensity add-on payment, just filling out a piece of paperwork. But if that same hospice has a low cap score on an item that reports about help when needed, uh there's a data inconsistency that should be really data-driven down. We have some of the data now, let's make use of it and maybe, you know, uh do it with ADRs or do it with TPE. Um, but I think it's time to really start cleaning this up because there's a lot of concerns going with these small hospice programs.
 
Cordt Kassner: 18:04
Yeah, I think that's a really interesting point. And I I I agree completely about uh not only tying the different data points together, but then looking at it longitudinally. Um, I uh when I've had opportunity to to meet with the CMS hospice team in Baltimore in the past, uh that's usually the kinds of information that I'm bringing to the table. And I it's like, this is your data, this is your claims, this is your quality information. And it's interesting that they would invite me in now and then to reflect on what an outsider sees in their data, but it's right along the lines that you're talking about. We're gonna kind of hit some of these, some of these questions, I hope, are not repetitive. We're trying to hit at these quality issues from different angles. One theme running through your recent work is that some publicly reported hospice quality measures may actually fail to detect the absence of meaningful care. Uh I think some of this is that quandary between something being statistically significant but clinically insignificant. If you could redesign hospice quality measurement from scratch today, be the advisor for the day, if you will. Are there a few measures? That'd be so much fun. Wouldn't it? Are there a few measures that you would prioritize because they most closely reflect the lived experience of patients and families?
 
Dr Joan Teno: 19:37
Yeah, so so let me go into
 
Dr Joan Teno: 19:40
what my past week was like, and uh that'll lead to uh a priority. So um I had two conversations this week. Um both involved daughters whose fathers ended up dying on hospice services. Um the first daughter was totally overwhelmed with the compare website. Um, and then the second daughter was even more overwhelmed because she lived in a region where there was more than 90 hospices that served her zip code. And um, you know, so what I did for, you know, just trying to deal with 90 hospices is ridiculous, you know, in and you know, trying to get that information. So here's my Claude AI story. I fed into Claude uh three X three files, which were CSB files, and I wrote out instructions to pull out the measures I think that are most important and to create me a dashboard and uh let me sort the hospices out for this daughter so she can make a choice. And here's what I did. Um I just, you know, cap scores I think are important. And I know later we can talk about the fact that, you know, there's um the this figure going around that says that only 30% of the hospices uh have cap scores. We'll talk about that later. But I chose CAP scores, star ratings, star ratings, I'm sorry. Um, star ratings. Um the second measure I chose was um the visits in the last days of life. And that's mainly based on its reliability and its contractability again with CAP scores. Um the third measure I chose was a I asked Claude to create a gaming uh uh a uh he to use its informatics to create a gaming measure based on pr providing services that were focused on one diagnosis, focused on being in a nursing home, focused on a sniff, and create a score. And then um, yeah, I think those were the three things I keyed on. And oh yeah. The other thing was because in the second case, they um had not eaten in five days, was very close to end of life, was actively dying, was not having shortness of breath, uh treated appropriately. Um, I thought IPU. And so I was, you know, basically told that, you know, if you have more than, you know, X percent, flag this as an IPU. And it came up with a list on a dashboard that I could sort. And I was able to give three or four hospices and then go back and check them against the compare data. So why can't we end up with this type of product that I want to go one more step forward? Here's what I want CMS to do. I want to be able to write a prompt that says, I live in 02806 and I want a hospice that's four or five stars, that has a good score on HLBI, has an inpatient unit, and doesn't look like it's gaming the system. I think that should be doable. Um and with AI, I I think that's the direction we should go. But that means you need to take all these data sources together and start looking at them. So we have utilization-based measures, um, we have demographic data, we have CAPS data and start looking at how to put all of them together. I do have a bias here, which is in an urban area, if there's a choice of hospice programs, as there is in an urban area, and you have a cap score, I'd rather go with a hospice that has star ratings as evidence of its quality care. Difficult thing is what to do with rural areas. Um, you know, I I think that's always a a problem. Um and I I don't know if I have a good solution for them, to be honest with you.
 
Chris Comeaux: 23:35
Yeah. That's such a great point. Well, this kind of took this was kind of takes my next question, so I'm gonna actually interject about the moratorium since that was kind of a late breaking.
 
 Chris Comeaux: 23:45
 And so what are your thoughts about the moratorium, Joan?
 
Dr Joan Teno: 23:49
Yeah, um, so I think the big question is um, you know, is it six months or is it gonna be a repetitive episode of six months? Um, you know, from a initial policy standpoint, I think we need the moratorium. Okay. And my substack this week, I sort of outlined, you know, I think we need to restore trust and integrity. I suspect they have a huge backload. Um, you know, even if they just go into looking at the data from Harris County in Texas, LA County, Maricopa, and um in Nebraska, or yeah, not Nebraska, but um Nevada. Nevada. Yeah, um, Clark County, um, there's probably a lot of clearing of um suspensions they're gonna need to do. Um, and there's a lot of suspicious data there. Um the issue is, you know, um rightly so, someone's pointed out that they're just relying on, and I don't I have no knowledge of this, so I could be totally wrong, but there was a a hospice news article written that said basically they're relying on discharge, live discharges. And you want to set a threshold where you're not penalizing good providers, okay? Um, and they should have the data analytics and learn very quickly on how they can get a positive predictive value that's good enough. I think it's somewhat reassuring, although I hate to use Fox News as a source of information, but Fox News reported that 90% of the people who hospices that were suspended didn't appeal so far. So that may be, you know, on track. But the issue is you want to be careful not to um to harm a good provider. So they need to figure out how to balance that in implementation. And the proof is gonna be how well um Kim Brunt can pull together these various silos of CMS and get them to work together to solve this problem. Um, and I don't know, I'm a little bit excited. Uh I I she might be the person to do this. And uh if she does, my hat's off to her, you know. Um, I I'll I'll be a fangirl of her work if she pulls this off.
 
Chris Comeaux: 26:16
Yeah, we're already a fan. And you know, the interesting thing too, Joan, is so we met her at an NPHI meeting. We brought her actually to our TCN network, and I said, I please, I would love to do a podcast with you. So it was kind of months in the making. And but when we brought her to our Teleios network, she started talking about this chili cook-off idea of how they were bringing all these tech vendors. And so that in itself gave me huge help. And like just listening to everything you're talking about, you know, there were statistical analysis tools that was maybe only for a a select few, but now this is putting powerful tools in many people's hands. And the fact that they're trying to be on the cutting edge, the whole point of the cook-off was to try to take like a myriad of 200-something vendors and get down to the best of the best. Well, obviously they're trying to apply it to something. So yeah, I like you was super encouraged. And the other interesting thing, I I don't, I think it was NPR, but it was a um press conference with JD Vance. I find myself listening now since this whole initiative under fraud. I don't know if he let it slip, but he said it in a way that would lead you to believe that the this pause, or timeout is the way he framed the six months. But maybe they have to extend it because you know the work is going to extend beyond the six months.
 
Dr Joan Teno: 27:30
If I had one piece of advice to give to the team who's working on this, um a lot of this ends up being really data driven and having health economists do this. And uh there's sort of an almost an over-reliance on statistical methods. I think it's really important to have one or two clinicians sitting at the table who can do the sort of statistics talk, but also put the data in context. Okay. And my perfect example is the burdensome transition one. Think about that measure. That measure is uh discharge, hospitalization, readmission. Well, there's one thing that these fake hospices being run out of a burrito stand can't do, which is fake a hospitalization. So, their zero score, you know, um is suspect. You know, there could be a zero score that they're doing a really good job, but there also could be a zero score because they have fake patients or they're enrolling people who are not terminally ill. So, you know, I think it's really important to have clinicians um look at the data and try to understand what's in the black box. I still think we're sort of at a faced validity looking at some of these measures and understanding, you know, making sure you understand the sort of Medicare hospice benefit, how hospice works. I find it's fair and very invaluable that I was a medical director for 20 plus years, but I also looked at the data and brought those two worlds together. So, I think it's really important for CMS to continue not just hire not just focusing on health economists who do these beautiful models, but to make sure there's a clinician to sniff out problems.
 
Jeff Haffner: 29:18
Don't miss part two of this episode coming this Friday.