- 2 years ago
Dr. Rajaie Batniji, Co-founder and CEO, Waymark, Dr. Sachin Jain, Chief Executive Officer, SCAN Group; SCAN Health Plan, Brent Nicholson, Co-founder and Chief Partner Officer, Carrum Health Moderator: Chrissy Farr, Second Opinion Media
Category
🤖
TechTranscript
00:00 Value-based care.
00:01 So maybe five years ago at a similar conference
00:05 and another time we would have been the hot panel,
00:08 but now it's the AI panels that are the hot panels.
00:11 So I think I'm gonna start with Sachin,
00:13 who's written a lot about this topic of value-based care
00:17 and how much it's real and whether the implementations
00:20 are actually doing what they say.
00:22 Would love to get your sense with where we're at now
00:25 just on value-based care in general.
00:27 Has it met the promise?
00:29 - I think it's easy to say no, it has not met the promise.
00:32 And I think one of the reasons is that I think everybody
00:34 still has got a split personality about it.
00:37 And when you actually look at what's happening
00:40 under the hood in most organizations
00:42 that are so-called committed to value-based care,
00:45 the commitment is only an inch deep and a mile wide.
00:49 And I think a lot of organizations haven't done the hard
00:52 work to really ask themselves the existential questions
00:56 about how they're gonna actually evolve
00:58 to a new business model because they haven't necessarily
01:00 had to do that.
01:01 So I think it's largely been an incomplete experiment
01:06 so far from where I sit.
01:07 I wouldn't say that that's uniformly true.
01:10 I think the future is here in many places
01:12 where value-based care has been implemented at scale.
01:15 But I would say nationally, we're still in the early innings
01:18 of the value-based care story.
01:20 - Rajay, do you agree?
01:22 Do you have a more optimistic view?
01:23 - I mostly actually agree and actually it's a pleasure
01:26 to agree with Sachin.
01:28 I would say that the biggest issue,
01:30 and Waymark, we work in Medicaid,
01:31 the biggest issue we see in value-based care
01:34 is that there's so much fake value-based care.
01:37 I wrote a piece on this for Health Affairs
01:39 called Value Veneers where I think what we're actually
01:42 seeing is there are a lot of programs that sort of like
01:45 check the box on value-based care.
01:47 They technically meet the requirements of it
01:50 so that a health plan and a provider could say,
01:53 yes, CMS, we're doing value-based care,
01:57 but they actually failed to do what value-based care
01:59 was intended to do and value-based care was intended
02:02 to actually change the nature of care delivery.
02:05 And so I think my rubric for looking at a value-based
02:07 program is not is there incentive payment for X, Y, or Z,
02:12 but has this program changed the nature of care delivery?
02:15 And for us, that means have we created a new healthcare
02:19 workforce that can reach patients where they are?
02:22 Are we able to move responsibility and accountability
02:25 for care outside of the traditional brick and mortar
02:27 healthcare system and directly into the communities
02:30 that we're serving where those interventions
02:33 can be most impactful?
02:34 And so I think unfortunately, so much value-based care
02:37 is kind of operating almost like a loyalty rewards program
02:40 rather than truly something that can change
02:43 what's paid for in healthcare.
02:45 And that's actually what we have a true shortage of
02:48 is real value-based care.
02:50 We have an excess of veneers.
02:51 - So one of the things that I think Satchin
02:55 has also been writing about is just that value-based care
02:58 can focus on a few things like you look at hospitalizations
03:02 and making sure that people are getting discharged quickly.
03:04 That can be one way of thinking about value-based care.
03:07 And there are others that we tend to focus on
03:09 a lot of that based on how the billing works
03:11 and how we record outcomes.
03:13 Brent, for you, as you look at what is value-based care
03:16 and what isn't and through this context of thinking about
03:19 fake value-based care and veneers,
03:21 what do you think are the true measures or outcomes
03:23 that we should be looking at
03:24 so that we don't just get fixated on a few things
03:27 that may or may not be right for that individual patient?
03:30 - Yeah, sure.
03:31 I think what we focus on is actually engaging
03:35 in arrangements with providers
03:37 that have both upside and downside risk.
03:40 That's in our version of true value-based care.
03:43 It's engaging with the provider
03:45 to ensure that they have a shared accountability
03:47 and actually are invested in the outcomes of the patient.
03:49 And so what we're looking at is condition
03:54 or episodic-based care in the specialty care area
03:58 and engaging with providers to have more ownership
04:01 outside, say, the four walls of an operating room
04:04 to extend their accountability to not just,
04:08 did we technically perform this procedure correctly,
04:11 but did it actually get the patient better?
04:13 Did it actually address what they went into the facility
04:17 to get the operation for in the first place?
04:20 - And maybe I'll just kind of jump off of that
04:22 and throw the question out more broadly,
04:24 but does that kind of work require actually thinking
04:27 about which physicians are good at what they do
04:29 and not good at what they do?
04:31 And I know this is a very uncomfortable topic
04:33 for a lot of people when we start thinking
04:35 about how do we rate and rank physicians,
04:38 but that is essentially part of what you're saying, right?
04:41 - Absolutely.
04:42 I mean, we are not believers in the general hospital model
04:46 that you can have a building that is good at healthcare
04:49 across the spectrum.
04:51 And so when we're looking at,
04:52 we're zeroing into specific episodic conditions
04:55 and looking for expertise and top performance
04:59 down to a very granular level.
05:01 And we use the power of massive data sets
05:05 to help us gain insights into which providers
05:10 do meet that very high bar of top 10% performance.
05:15 And we're looking to engage them
05:17 in these upside-downside risk arrangements
05:20 so that they can really do what they do well
05:24 and don't have to be saddled with the administrative burdens
05:27 of figuring out how do we then figure out
05:29 how to submit claims and how do we then figure out
05:32 how to do the authorizations
05:33 and how do we figure out then haggling with the payers
05:36 to make sure that we get the money in the door.
05:38 We wanna find the best providers,
05:39 allow them to do what they do well,
05:41 and then make sure that they are rewarded for that.
05:44 - So I wanna just say I love this conversation
05:46 because I think a lot of value-based care conversation
05:50 is actually not about what you're talking about.
05:52 It's not about getting patients to the best clinicians.
05:55 It's actually about labor arbitrage.
05:57 Is how do we take the work that specialist MDs
06:01 were doing to generalist MDs
06:02 and then take work that generalist MDs were doing
06:05 and take them to nurse practitioners
06:06 and then take work that nurse practitioners were doing
06:08 and take them to RNs and take work that RNs were doing
06:11 and then take them to community health workers
06:12 and then take the work that community health workers
06:14 were doing and then give them to people's family members.
06:16 And I think we have to talk about labor arbitrage
06:19 and we have to talk about actually
06:21 what is an excellent healthcare system actually look like?
06:24 And I think most of us intuitively get
06:26 that you have to actually go to a great clinician
06:29 who really understands the condition well
06:30 and knows how to treat it
06:32 and knows the boundary conditions of that condition
06:35 to be able to effectively get the kind of treatment
06:38 that we need, but that's not the conversation we're having
06:40 about value-based care for the most part.
06:42 - I totally agree on the labor arbitrage piece.
06:45 And then to flip that question, push it further, Rajay,
06:48 how do you see the role of the patient here?
06:51 Because if they feel like they're just getting pushed
06:53 into a lower level of care with less of a specialist,
06:56 wouldn't they as a patient say,
06:58 "I wanna be in the other system.
07:00 "I wanna be in the system where I can see the specialists
07:02 "that I wanna see."
07:03 - I mean, I think to me, we center our approach
07:08 literally around the patient.
07:09 And so if you think about what's broken
07:11 in the healthcare journey,
07:13 imagine you're covered by Medicaid,
07:15 which all of our patients are
07:16 in all the states that we're in.
07:18 People have a hard time accessing a physician.
07:22 They have a hard time navigating the healthcare system.
07:25 They need to use social resources
07:27 that no one has helped them get to.
07:30 And so to me, the point of value-based care
07:32 is let's actually get the physicians to see them.
07:35 Let's make availability happen.
07:38 Let's wrap those physicians with the expertise
07:41 that they need to appropriately navigate this member
07:43 and literally guide them the whole way through their journey.
07:46 And so to me, it's about complementing the existing system
07:49 by creating net new resources.
07:52 I mean, if you look at the data,
07:53 and we always start with data.
07:55 If you look at the data about what works
07:57 in healthcare delivery for a certain population,
07:58 and I think you need to do this population by population,
08:01 in Medicaid, for example, it's totally clear.
08:04 Study after study proves that these same models work.
08:07 The challenge is that those models have been funded
08:09 either by a one-off grant.
08:12 They've been funded out of a special budget
08:14 from a health plan or health system.
08:16 And to actually do these things and do them for a long time
08:18 so that communities can know that they can rely
08:20 on these services to exist in the future,
08:23 you can do them under a value-based arrangement
08:25 so that you can have that kind of permanent source
08:28 of funding where actually accountability for care
08:31 can actually move towards these new care delivery groups.
08:35 - So just a reminder to the audience
08:37 that we are taking questions.
08:39 I'll flip to all of you next.
08:41 I'm gonna ask a question about AI,
08:43 'cause that has been a big topic of this conference.
08:46 We've talked a lot about actually using AI in diagnosis,
08:49 not just kind of back office,
08:51 which is where it tends to be more kind of discussed today.
08:54 Sachin, do you think in a value-based world,
08:58 a true value-based world,
08:59 there is more room for AI in that context
09:03 because it becomes less about who's gonna pay for this?
09:06 - Yeah, I think yes.
09:09 A lot of kind of value-based care
09:12 is in like full delegation of risk to a downstream entity.
09:15 I think the real challenge of freeing up enough resources
09:19 to actually invest in AI is then how that downstream entity
09:24 that's taking full risk actually allocates those dollars
09:28 to sub-entities, how they pay specialists,
09:30 how they pay primary care doctors,
09:32 how they pay ancillary care providers.
09:34 And that is still happening very much
09:36 in a fee-for-service way.
09:38 And so then the question becomes like,
09:39 where does the AI get introduced?
09:41 Does it get introduced at the risk-bearing entity level
09:43 or does it get introduced at the level of
09:47 the primary care doctor or the specialist?
09:49 And is there enough margin
09:50 in what is actually being delegated,
09:52 subcapitated to the primary care doctor or the specialist
09:55 to actually integrate AI in ways that are gonna reduce costs
09:59 and improve quality of care?
10:00 And that's a long way of saying,
10:02 I think we're still figuring it out.
10:03 - Yeah, I would add to that.
10:05 I mean, I think so much of the way we talk about AI
10:07 in healthcare is about, well, in the future,
10:10 humans will be talking to machines,
10:11 machines will be doing the diagnosis.
10:13 I think that's actually totally wrong.
10:15 And our approach to AI development is really,
10:18 let's automate the routine tasks
10:22 so that we can allow our caregivers
10:24 to spend even more time with patients.
10:27 And second, let's automate the identification
10:31 of which patients are gonna benefit
10:33 from certain services so that we can actually spend
10:37 our time targeting the appropriate patients.
10:39 And we've been publishing actually quite a bit
10:42 on our development in this realm,
10:44 because I think that is where you have step change
10:48 improvement in improving access to care,
10:51 is you're gonna create and enable a net new workforce
10:54 through AI and actually give them the pattern matching
10:57 of somebody with years more experience.
10:58 - Yeah, I would offer a slightly alternate perspective
11:01 from SCAN, which is, our view is that AI
11:04 is going to enable humans to become superhumans.
11:06 And so, when you call your health plan
11:08 member service representative,
11:10 or you're talking to a nurse
11:11 from a care coordination department,
11:13 they're gonna be able to do more and know more about you
11:16 and be able to leverage everything within our system
11:20 through advanced decision support
11:22 to be able to make those interactions
11:23 more valuable and more meaningful.
11:24 That's how we're thinking about it.
11:27 - I guess two other angles, at least from our point of view
11:30 at Carim is, agree with Rajay, what you said about
11:34 using the power of AI and specifically machine learning
11:37 to help identify the right patients,
11:39 get them to the right programs and services
11:41 at the right time.
11:42 Because there's just a sea of information out there,
11:47 and it can be very hard to understand and navigate
11:49 for a common patient who just has too many options
11:52 available to them, too many mouths
11:54 trying to get their attention.
11:56 And so, being able to do that and say,
11:59 identifying only a small percentage of the population,
12:03 in our instance, that we're able to then say,
12:05 we can correctly identify, say, up to 95% of the surgeries
12:09 that are likely to occur over the coming years.
12:12 And then being able to serve up the appropriate program
12:16 to meet the needs of that patient.
12:18 And then from the provider's point of view,
12:20 being able to help these providers perform better
12:22 in value-based care arrangements.
12:25 We've seen some tools out there that help to
12:28 create these personalized care plans that
12:31 harvest all of the information from
12:34 millions of different patients,
12:36 rather than just the handful of patients
12:37 that one doctor has seen,
12:39 to be able to do things like optimize variables around
12:42 the length of stay, the discharge destination,
12:46 the post-operative care protocols,
12:48 so that they can zero in on a likelihood of readmission,
12:52 and then be able to adjust and tailor their follow-up
12:55 and check-in protocols accordingly
12:58 to then result in the best outcomes,
13:00 and then perform better under these value-based arrangements.
13:04 - So I'm gonna flip for a question from the audience,
13:06 who's got a burning question on value-based care.
13:09 - Hi, Eric J. Daza, Stats of One,
13:21 inference for the individual.
13:22 So I love what I'm hearing about
13:24 these patient care journeys
13:25 and how you're really tailoring them.
13:27 I'd love to know what the biggest challenge is
13:30 to incorporating AI or anything like that
13:33 into that workflow.
13:35 - Brent?
13:39 - Well, I'd say for us,
13:42 one of the considerations is to try and
13:45 shield the patient from all of the complexities
13:48 of what we're talking about here up on stage,
13:50 is really to make it so that the patient doesn't have to
13:55 understand or navigate,
13:57 well, am I in a value-based program, am I not?
13:59 And so one of the things that we work on is
14:04 ensuring that we are working very tightly with our
14:08 providers on the patient identification
14:10 and the patient attribution.
14:12 And so that's one of the things where
14:15 we have a very specific implementation process,
14:18 we have a very specific pre-registration
14:20 identification process that ensures that
14:23 when the providers that we're working with
14:25 are seeing a care inpatient,
14:27 they can understand that and
14:30 they can follow the program protocols accordingly.
14:34 - I think we have a similar challenge.
14:36 For us, the biggest challenge is like,
14:37 just imagine, right, we serve a new market,
14:39 we're serving 50,000 patients covered by Medicaid there,
14:42 you know, where do we start, right?
14:44 Like, how do you know which patients to reach out to?
14:46 And so, you know, our first AI use case
14:48 was we built this product that
14:50 allowed us to bring in all of the historical medical claims,
14:54 the local zip level social determinants data,
14:56 information from food banks and housing registries,
14:59 and actually build this composite picture of a patient.
15:02 And the result of that, which we published earlier this year
15:04 in Nature as a scientific report,
15:06 was that we had 90% predictive power,
15:09 positive predictive value,
15:11 in anticipating who's going to use acute care
15:13 in the coming days and weeks.
15:15 That means ER visits, that means inpatient admissions.
15:19 Compared to the Next Best Model,
15:20 that was performing at like 30%, right?
15:22 So that means that now one out of 10 of our outreaches
15:25 to patients are to a patient that we can't serve,
15:27 versus seven out of 10, if we're using a prior model.
15:30 And so I think you have to get really efficient
15:33 at doing that modeling so that you can actually
15:35 direct your resources as effectively as possible.
15:38 - Satyen, I'm gonna call on you for a last question.
15:41 We have a few minutes left.
15:42 So one of my, you write a lot of pieces,
15:47 you've written a lot for Forbes.
15:49 And I loved what you said about just the churn problem,
15:52 and having a patient for a single year,
15:55 because obviously we do enrollment annually,
15:58 how problematic that is.
16:00 Can you just end us on a note of talking about your idea
16:03 with a three year experiment?
16:04 - Yeah, absolutely.
16:05 So I think one of our biggest challenges in our industry,
16:07 when we actually think about interventions
16:10 that will actually move health outcomes,
16:12 is that most people are in their health plan for a year.
16:16 That's the most that we can actually be certain
16:17 that someone's gonna be in their health plan for a year.
16:19 - What can you do in a year?
16:20 - So very little, right?
16:22 And the most you can do in a year
16:23 is maybe you can do an annual wellness visit,
16:25 maybe you can get them appropriately coded.
16:27 But as far as actually better managing their chronic disease
16:30 and then seeing the return for it,
16:32 it's just not gonna happen.
16:33 And so I believe, at least in the Medicare space,
16:36 we have to get out of single year enrollments.
16:38 We have to be thinking about three to five year enrollments,
16:42 which would then give us the confidence
16:44 that our investments in year one
16:46 would have some payback in year two, three, four, and five.
16:48 I believe that's a foundation of value-based care
16:51 that has been missing in most of the
16:53 so-called value-based care models,
16:54 because without the reliability
16:56 that a member is going to be on your books
16:58 in three or five years,
17:00 you're unlikely to wanna necessarily
17:02 make the investment in year one
17:03 to actually create some health outcome benefit
17:07 in year three or year five.
17:09 Or you may want to, but it doesn't necessarily pencil
17:12 from a financial perspective.
17:13 And I just think it's these types of structural changes
17:16 that are gonna actually lead us to true value-based care
17:19 as opposed to the value veneers
17:21 that Rajay has written so eloquently about.
17:25 - How optimistic are you that it's gonna happen?
17:27 - So I think the number one objection people have is like,
17:30 what if somebody finds their way into a plan
17:31 that is really bad for them
17:33 or they're being treated badly by that plan in year one
17:35 and they really need to get out?
17:37 I think we have to get comfortable with the trade-offs.
17:39 So I think it'll take a while for us to get comfortable
17:42 with those kinds of trade-offs.
17:44 But I believe that the Centers for Medicare
17:45 and Medicaid Innovation,
17:46 in addition to all the science projects
17:48 that they're running right now,
17:49 should add one more science project,
17:52 which is to really look at what a three or five year
17:54 enrollment in a plan would do and have it be voluntary.
17:57 People should wanna enroll into it.
17:59 I think people will wanna enroll into it
18:00 because it'll be intuitive that a health plan
18:03 that owns your life for three or five years
18:05 is going to view your care very differently
18:08 than a plan that owns it for a single year.
18:09 - The time horizon for risk
18:11 and the time horizon for change need to line up.
18:13 - Exactly.
18:14 - Well, you had it here.
18:15 We love it.
18:16 Thanks everybody for joining us.
18:17 - Thanks so much.
18:18 (audience applauding)
18:21 [BLANK_AUDIO]
Comments