In this episode, Andrew Wilson, Research Team Leader at the Altarum Institute’s Center for Payment Innovation, discusses the results of their study, “Debunking the argument that the Bundled Payment for Care Improvement Program (BPCI) contributed to higher procedure volumes.”
You might also like:
Mike Passanante: Hi, this is Mike Passanante. Welcome back to the Hospital Finance Podcast.
Today, I’m joined by Andrew Wilson. Andrew is the research team leader at the Altarum Institute Center for Payment Innovation, formerly the Healthcare Incentives Improvement Institute where he leads new research initiatives and provides consulting support to CPI’s core activities and products. His research interests includes risk adjustment, methods performance measurement and the design of new payment models.
In addition, he’s a doctoral candidate in the Health Services Research Program at Brandeis University’s Heller’s School for Social Policy and Management. For his dissertation, he is examining the application of network analysis in assessing integrated provider networks and the influence of these networks on healthcare costs and quality.
Andrew has joined us today to review the findings of a study he led entitled Debunking the Argument that the Bundled Payment for Care Improvement Program Contributed to Higher Procedure Volumes.
Andrew, welcome to the show.
Andrew Wilson: Thanks for having me!
Mike: So, before we begin, can you just explain to our audience a little bit about what Altarum does?
Andrew: Altarum is a research and consulting organization. Much of their work in the past has been in supporting the Department of Defense in TriCare and the Veterans Affairs Administration and some of their healthcare issues and implementations.
CPI is actually relatively new to Altarum. We were formerly the Healthcare Incentives Improvement Institute. We merged with Altarum at the beginning of the year.
Most of our work is around the development of the Prometheus Analytics software. I’m not sure, but perhaps some of your listeners are familiar with that. It’s kind of an episode grouper software that groups claims into episodes. And we use that for performance measurement and bundled payments.
So we work with payers as well as some providers on thinking about bundled payments and implementing bundled payments and providing technical assistance when they need it.
Mike: Excellent! Thanks for that explanation.
As I’ve mentioned, the study we’ll be discussing today is entitled Debunking the Argument that the Bundled Payment for Care Improvement Program Contributed to Higher Procedure Volumes.
Andrew, can you explain why this study was initiated and what the goals were?
Andrew: Sure! So, there was a study that was published in the Journal of the American Medical Association back over the fall by some folks at the Lewin Group and CMS and a couple of other organizations. And it was the first major evaluation of a BPCI program and they looked in particular at major joint procedures (so the BPCI program as it relates to hip and knee replacements).
The primary goal was to look at the impact on cost and quality. And what they found was very promising for BPCI and what everyone was hoping, and that was that relative to non-participating hospitals, BPCI participants were able to reduce their cost even further than what was observed overall. And mostly, that was driven by lower use of post-acute care services.
Importantly, they also found that there was no decrease to quality. Lower cost, no change in quality. And they measured quality, by the way, in terms of re-admissions, ED visits or mortality. So, none of those went up which is a good thing.
So, that was a promising thing for BPCI.
Now, there’s a secondary analysis in there. They addressed this issue of whether bundled payments lead to higher volumes. That’s been an ongoing academic discussion because it’s been mostly theoretical. But they did want to take a look and see whether that bore out or not during the first few months of BPCI.
They looked at that, and they looked at procedure rates in a per quarter basis. And when they compared the participants to non-participants for the post-intervention period, they found—it suggested there was maybe a slight increase, a bump in procedure rates among the participants, but their findings are actually statistically non-significant. So, in the research world, we would say that there was actually no effect on volume.
Now, the issue came in for us with an accompanying editorial that was done by a prominent health services researcher in relation to this study. And what he did in the paper was to sort of debunk the findings a little bit. And the way he did this was to actually take their findings around volume and almost accept them as facts.
So, even though the authors themselves didn’t find a true significant difference in volumes between participants and non-participants, he actually kind of suggested that that was true, and then went on to show how if you take that finding and kind of multiple the numbers out, take it to total cost, and you compare those total changes over time, he actually found that the participants decreased spending less than the non-participants.
To us, that was a rather irresponsible application of the results a) because we can’t actually conclude from the findings of the study that volumes were actually not higher among participants; and second of all, the others themselves stated in the conclusions that, well, they did the analysis, but because of the methods and the fact that they didn’t have the complete universe of procedures, lower joint procedures that were completed paid for by Medicare during that period, they really can’t say a whole lot or be definitive about the impact on volume.
So, we really had a problem with the assertion that was made in the editorial. So, our goal was to actually go back to the data, see if we could get the full universe of the lower joint procedures, re-do the analysis, support the findings of the original study, that there was no increase in volume, and then see if we could do a secondary analysis to further refute the claims and show that there were likely other factors that would explain the increase in volume over time.
So, that’s where we get into this regional issue in the paper, but we’ll get into that.
Mike: Andrew, can you briefly go over your methods and the data you reviewed in your study?
Andrew: Yeah. So, we worked with a group that we have a relationship with called CareSet. And they have access to 100% of Medicare claims over several years. They provided us with the full universe of hip-and-knee replacement procedures performed at all hospitals in the United States over the period of 2011 to 2015.
Given that data, we then went and supplemented it with publicly available data from CMS indicating what hospitals were participants in BPCI and those that weren’t.
And so, using that data, we then went and replicated the analysis, the original study, looking at changes over time between participants and non-participants, and the number of procedures that they performed. And then, we compared those changes to see if there was any difference between them. So, if the change was greater among participants and non-participants. It was suggested there was higher volume.
Of course, we looked at the raw numbers and the trends over time as well. And then, as a secondary analysis, to kind of get at this idea that to buttress the findings or what we hope to be the findings that there was no chain in volume that could potentially explain the higher volumes over time overall by some regional factors and show that it was these regional demographic and market trends that was actually driving these increases in volume over time and not the BPCI program itself.
So, a) revisiting the analysis of the general paper, and b) then trying to buttress our analysis with some further data.
Mike: Andrew, can you tell us what you found the effect of BPCI to be on hospital level volumes?
Andrew: In our primary analysis, our replication of the original JAMA study, we essentially confirmed the results. We found that there were no differences in the change in volume before implementation of the BPCI program versus after. So, the participants did not perform more procedures than non-participants because of the BCPI program itself.
And I think that’s the main takeaway from our study, which is that BCPI, we don’t think it’s associated with any increases in volume despite the assertions made by that editorial.
Interestingly, when we look at the raw data—and this is a secondary piece to this—if you look at just the raw changes over time between participants and non-participants, the increases among the non-participants were actually greater on a percentage change basis than the actual participants.
And even when you go down further in a year by year level, we actually saw in the final years of our data that procedure volumes actually decreased a little bit among participants.
So, clearly and definitively, we cannot attribute any higher volume or volume increases to the BCPI program. That just seems to not have happened.
Mike: So, BPCI alone didn’t seem to increase the volume. Were there any other regional factors or other factors associated with volume increases that you did see?
Andrew: Yeah! And this was the secondary piece to our analysis because we wanted to illustrate that, probably, what we’re seeing are broader demographic and market trends.
So, when we looked at the association of some of these factors with changes and volume over time—and by the way, we looked at this at the regional level, not at the hospital level—we did find some interesting results.
We found that, as you might expect, increases in the number of Medicare beneficiaries, particularly fee per service beneficiaries in the area, contributed to volume. We expect that because of higher demand.
We found that greater competition is associated with more procedure volumes or greater procedure rates. The number of facilities in the HR was associated with higher procedure rates.
And interestingly, the Medicare Advantage penetration in the region was associated with lower procedure rates. And then, finally, we found that greater minority populations in an area was associated with a lower procedure rate or change in the procedure rates over time.
Mike: Let me drill in on something you just said in your previous response. You found that hospital referral regions with high competition actually have a larger number of lower joint procedures. Can you give us some insights into why that may be?
Andrew: I have two theories that I think about. This is actually a question we’ve been asked a lot. It’s one of the more slightly interesting aspects of the study and our findings.
The first piece of this I think is this idea of competitive pressures in a market. And what has been termed in the past maybe as the medical arms race.
So, as hospitals compete aggressively with each other in an area, there’s been a tendency to be the go-to hospital or to offer the services that is in demand and kind of makes you that go-to place.
So, hospitals adopt technologies and offer service lines to compete with one another to draw in patients. But once they get those service lines and those technologies in place, then there’s got to be some effort to justify having those and provide revenues for devoting resources to those service lines.
So, I think those competitive pressures to both offer the service lines, but then have to justify providing them leads to some aggressive activities to boost procedure rates and volumes over time.
And we’ve seen this in a lot of aspects. As listeners of your shows should know, greater competition leads to greater adoption of technologies—for example, robotics surgeries. We know that in areas with greater competition, there’s both greater adoption of robotic surgery, and then as a result, greater use of robotics surgery. That conforms with all sorts of previous literature.
The second piece to this I think is this idea of supply-sensitive care. It’s very related to what I just mentioned. So, the more supply of services in an area, the greater use of those services.
This has been shown repeatedly over time. Areas with more specialists tend to use much more specialty care. The same applies in this case likely with joint replacements. The more hospitals provide joint replacement services, the more they’re going to get used in those areas.
So, that’s just the way. That’s something we’ve witnessed over time in a lot of different areas. And I think joint replacements are just as much as supply-sensitive a service as anything, as plenty of other things. It follows along with that line of reasoning.
Mike: Andrew, are there any other final thoughts you’d like to mention about the study?
Andrew: Yeah, one of the things we haven’t talked much about is that I think our findings show that BPCI itself hasn’t contributed to volume. So it addresses this lingering question, this ongoing question we’ve had about the impact of bundled payments on volume.
But I don’t want this to be taken as the final word. Clearly, BCPI has an effect on volume. But we’re still kind of early in the episode in bundled payment space, and it’s just something we should continue to keep our eye on.
I’ve always been a little bit skeptical of this volume question and the ability of bundled payments to lead to higher volumes. That’s a really tough thing to do, to manufacture things from a provider’s perspective. But it’s something we should continue to look at.
So, it shouldn’t be taken as a final word. It’s possible that things could’ve happened in the past, so let’s look at it. But let’s not get too crazy about worrying about this.
Mike: Andrew, that was great information. Thanks for taking some time to explain your study to us and helping us understand more about the effect of BPCI on procedure volumes.
Andrew: Thank you, Mike. It’s a lot of fun!