Providence Population Health Management is leading the Medicare Shared Savings Program (MSSP) initiative for a multistate health system that is the largest ACO in the nation, covering more than 150,000 members. The program has generated more than $230 million in cost savings for CMS between 2017 and 2022 and is expected to generate more than $200 million in savings for CMS over the next two years. Deepak Sadagopan, chief operating officer for Providence Population Health Management, recently said: Healthcare Innovation We talk about what's key to successful value-based care.
HCI: Can you talk a little bit about the keys to success in value-based care? Did Providence have to make any course corrections to get to those results?
Sadagopan: We have been with the MSSP since 2014 and have supported the program consistently. The period from 2014 to 2017 was a period of poor performance for us, not only with the MSSP but also with the other risk programs we were a part of. As an organization, we were still learning how total cost of care risk worked. The healthcare delivery system does not have the built-in capabilities, processes, and systems to actually manage risk. On the provider side, we have to learn to think like a payer so that we can manage risk, which doesn't come naturally. We're not wired to think that way.
MSSP was one of the programs that we used as a vehicle to learn how this works in a broader managed care environment. The main reason was that MSSP had a single standard model of contracting with CMS that covered many of our operating territories, so it provided us with a great way to learn about value-based care. That learning came in a variety of ways. One was how we built partnerships in each of our territories with groups that were working on enhancing primary care, providing primary care access to these patients or beneficiaries, assessing chronic disease, and providing ways to manage this within the primary care environment before it gets worse.
HCI: What kind of partnerships are you talking about?
Sadagopan: At that point, we needed to figure out, within Providence itself, all of the primary care groups, which medical groups were the most committed. We needed to identify the right medical groups, and then identify referral partners in each area outside of Providence. I think part of it is understanding the network and interconnectedness of care.
The second part is really about integrating all the information and data and associated analytics around total cost of care, recognizing that the vast majority of health care services take place within the walls of Providence. Improving visibility through a unified data and analytics framework has been a key part of this.
HCI: Does this include working with post-acute care providers?
Sadagopan: Yes, we also need to build a post-acute network. In a traditional hospital operation, you discharge a patient and don't know what happens to them after that. The biggest change we made was reaching out to each of our skilled nursing facilities and building data connections with each of them. This dramatically changed our ability to manage the full scope of care. We can now say, “This patient has been in the SNF for over 37 days, that's crazy. We need to step in now and figure out how to get this patient home.”
We can now ask these questions and make and initiate interventions outside the walls of the system, which has been critical for us to administer these types of programs.
HCI: Are there other cultural changes at play?
Sadagopan: “When it comes to integrating these value-based care programs into our health care delivery system, I would say there's an incredible capacity gap, not just at Providence but across the industry. Right now, 99% of all the claims and information that we track goes through the revenue cycle, through billing. The billing system is wired only to reimburse through claims. The shared savings payments that we receive for keeping our claims low in exchange for accountability don't even flow through the regular billing system, they get set aside.”
We had to take a step back and say, whoa. Our health care economics have changed dramatically. It's no longer just about the P&L or the financial performance of a hospital or a medical group. We have to learn the concept of risk pooling. What is a risk pool? It's all the patients that are assigned to our care, the amount of resources that are allocated by the payer or Medicare to fund the risk pool, how much we get paid for claims, and even if we're not getting paid for that claim, how is that risk pool performing? That's part of thinking like a payer, and we're not doing that. So one of the biggest changes that we've made is we've really looked at integrating value-based care into our mainstream delivery system.
This impacts every part of what we do: If we need to allocate more resources to hire community health workers, or increase behavioral health resources to care for our residents, or increase mobile care assets to provide care in remote areas, all of that funding will come from recognition of this Integrated Performance Pool, not just funding that comes from claims.
HCI: Currently, approximately 35% of all patients served by Providence are covered under alternative payment models. Is there a goal to increase this percentage significantly? Also, is this 35% number still a “one foot in both canoes” problem? Are you starting to build muscle memory of working in a value-based care regime?
Sadagopan: The key to changing that muscle memory is what I just explained, which is to actually recognize the economics of this risk pool and put it at the forefront of how health care delivery system leaders actually think about the economics of care. The performance of that risk pool is just as important as the financial sustainability of each of the core lines of business in the hospital system.
The challenge isn't to stand and balance the two canoes, I think the challenge is how to get the two canoes together and connected so that they don't lose balance in the first place.
HCI: You mentioned at the beginning that it was important to figure out which primary care groups were ready to do this value-based care work. What about those who weren't ready initially? What will you do to get them on board?
Sadagopan: Take this MSSP program for example. There are 30-32 TINs. [taxpayer identification numbers] These are companies participating in our MSSP program, which is one of the largest in the country.
We take a data-driven approach, asking, “What is the total cost of care? What is the industry benchmark?” We have refined our techniques to assess the benchmark at an individual level and compare it to the total cost of care experience and quality outcomes.
For those groups that weren't adhering to the benchmarks and quality scores, we kept them as part of the organizational structure of the network. Although we didn't physically include them in the MSSP program, we did all the same things we do for all other members and made sure they were consistently included in all reporting and outreach activities, which allowed us to learn alongside the other groups.
Another thing we have established is an annual evaluation process, where we re-evaluate every year and all the organizations that actually meet the criteria rejoin the network. As a result, most of the organizations that were out of the network for a few years were back in the network within three years. If you look at the membership graph of the MSSP program, it was declining a little bit, but now it is increasing again.
HCI: From a policy perspective, is there anything you would like to see CMS change regarding the MSSP or ACO REACH programs?
Sadagopan: From a policy perspective, one of the things we’re very excited about is CMMI is experimenting with something they call “shadow bundles.” A big part of any value-based care effort is how you engage experts, and we’ve been thinking a lot about that.
HCI: There aren't many APMs for specialists…
Sadagopan: Outside of bundling. Yes, that's right. Bundling is a great tool, but the challenge with bundling is that it's not sustainable unless you hit a certain volume. So you go back to a pseudo fee-for-service mode. But bundling has a really useful property of reducing episode-to-episode variability. You don't see spikes where in episode A it's $50,000 for a knee replacement and in episode B it's only $15,000. It evens out over a given time period.
When you build that episode predictability into a structure like the MSSP, you create less variability across all acute care episodes and make it easier to hit your total cost of care goals, which in turn leads to better quality because there are fewer readmissions. I think if you create a shadow bundle within the total cost of care program and engage specialists within that and provide incentives to participate in the total cost of care program, that would be a great engagement tool for specialists.
CMS and CMMI are looking at ways to reduce the variability across these programs from a measurement standpoint — how the data from these programs is used.
HCI: Regarding multi-payer collaboration with private insurance companies…
Sadagopan: Yes. At Providence, for example, of the 1.7 million patients who have some form of value-based care contract, we have over 130 unique value-based care contracts, and each contract has multiple cohorts. Very few of the 130 contracts are exactly like each other. They're all different. This creates a challenge of administrative complexity, because we've moved from revenue cycle complexity to value-based care management complexity. This is antithetical to our goal of simplifying healthcare, not complicating it. So I think having a standardized set of quality standards and structures for how these programs are administered would dramatically simplify administration within the broader context.