Latest / The Mental Health Evolution / Ep 13: AI and Measurement-Based Care with Dr. Dylan Ross

Transcript

Ep 13: AI and Measurement-Based Care with Dr. Dylan Ross · The Mental Health Evolution

0:03
(Transcribed by TurboScribe. Go Unlimited to remove this message.) All right, everyone, welcome back to the mental health evolution podcast, where
0:10
we explore the latest trends and strategies shaping behavioral health care.
0:15
Today we are joined by Dr. Dylan Ross, head of clinical at
0:20
Blueprint AI.
0:22
Dr. Ross brings extensive expertise in value based care and measurement based
0:27
measurement based care and he has applied this expertise to Blueprint AI
0:32
to help shape tools that support clinicians in outpatient behavioral health
0:37
settings.
0:39
At Blueprint, he leads initiatives to ensure AI assisted tools like automated
0:44
session notes, treatment plan suggestions, and outcome tracking are aligned
0:49
with evidence-based practices and outcome measurement, supporting both
0:54
clinician workflow and client outcomes.
0:57
Today we are going to dive into some topics that intersect with
1:01
our guest's experience as well as current topics for the mental health
1:05
industry.
1:06
We will go where the conversation takes us, but we will look
1:10
at things like measurement based care, value based care, and potentially
1:14
touching on the ethics of AI a bit.
1:17
Before we dive into the conversation, I want to highlight a few
1:21
articles that you can dig into as you get your bearing on
1:25
the topics that Dylan and I are going to be having a
1:28
conversation about.
1:30
First is an article called, Value-Based Care, What It Is and
1:35
Why It's Needed.
1:37
I thought this was a great sort of foundational piece by Commonwealth
1:42
Fund and this article explains the core principles of value-based care,
1:47
including how it shifts reimbursement from volume to outcomes.
1:52
If you're not familiar, value-based care means that providers are paid
1:58
based on results more so than based on just their time.
2:03
That's kind of the big shift that's happening in value-based care.
2:06
There's a lot of it already happening in the healthcare industry.
2:10
Mental health is still kind of something that payers are figuring out
2:15
how to get into, how to best measure and decide how to
2:20
do value-based care.
2:21
But this article discusses the benefits for patients and providers and why
2:26
adopting it is increasingly critical in healthcare, including behavioral
2:31
health.
2:32
And then the second article is called, Getting On Board With Measurement,
2:37
Implementing Measurement Informed Mental Health Care.
2:40
And this is from the Institute for Healthcare Improvement.
2:43
And this guide provides practical strategies for implementing measurement
2:47
informed care in behavioral health settings.
2:50
It includes how clinicians can standardize assessments to look at outcomes, to
2:55
inform treatment decisions and improve client engagement.
2:59
This is a big piece of where Blueprint started and I think
3:03
you still offer this, Dylan, if I'm not mistaken.
3:05
So I'm excited to talk a little bit about that.
3:09
And lastly, I want to highlight an article that our guest wrote
3:14
titled AI in Behavioral Health Documentation, Ethical Considerations for Mental
3:20
Health Clinicians.
3:22
In this article, our guest here outlines Blueprint's approach to integrating AI
3:27
into behavioral health documentation while maintaining ethical standards,
3:32
clinical accuracy, and patient privacy.
3:35
He discusses how AI can support clinicians without replacing their judgment and
3:40
how these tools can complement measurement based care.
3:45
This article is going to be a great launching point for our
3:48
discussion as well.
3:49
So I encourage you to look at the show notes and go
3:52
ahead and click on that so that you can read more in
3:55
depth about what Dr. Ross has had to say about that.
3:59
So with all of that background, Dylan, I'd love to dive into
4:05
talking about where you are coming from.
4:08
So maybe a little bit about your journey in the mental health
4:12
industry, kind of where you've been, places you've worked, what you've seen,
4:16
and kind of your background.
4:18
You bet.
4:19
Well, thinking back to the beginning of my career, and while there's
4:23
a couple chapters that precede this, I stepped into the managed care
4:29
arena in behavioral health in the early 2000s working for Kaiser Permanente
4:34
in Northern California.
4:36
In that role, served as a psychotherapist working in a substance use
4:40
disorder treatment setting.
4:43
And in the lead up to the Affordable Care Act, found myself
4:46
increasingly drawn towards more system level, structural, and policy level
4:51
influences that were shaping how healthcare in America, behavioral health, was
4:55
working or not.
4:56
Ultimately went back and got a PhD as an organizational psychologist, also
5:00
duly licensed as an MFT LPC, and left direct care in 2013
5:08
and moved to a consulting role at Kaiser Permanente in their national
5:13
program offices.
5:15
And at the time, the charge was to reimagine how we might,
5:19
one, address and expand access for those with unmet behavioral health needs,
5:23
but two, looking at how we could early, in a more early
5:26
fashion, identify those unmet needs in primary care, women's health,
5:31
pediatrics, upstream.
5:33
So we were ultimately designing for the system, both collaborative and
5:37
integrated care models.
5:39
And through that work and partnerships with the Department of Defense, the
5:43
Veterans Administration, National Health Service in the UK, learned a great
5:47
deal around what models ultimately support earlier identification and
5:53
engagement.
5:54
The act with Kaiser brought me to overseeing behavioral health quality for
5:58
the health plan, as well as later addiction treatment and recovery services
6:02
as both the market operator in Colorado.
6:07
And that was the big jump into big Q quality within behavioral
6:12
health.
6:13
To fast forward the rest, left Kaiser after 12 years, joined Optum
6:18
Behavioral Health, where I oversaw a high-performing networks team, had a
6:22
portfolio of multiple initiatives, but the marquee one is really setting the
6:27
strategy for Optum around, as they refer to it, measurement-informed care,
6:31
which today sits foundational to their value-based care offerings.
6:35
Really missed care delivery.
6:37
After four years with Optum, went to Rogers Behavioral Health, led digital
6:41
strategies, the vice president of clinical innovation, through the peak of
6:46
COVID, which was no small task.
6:50
And then ultimately went off, did some consulting for a while, and
6:53
I also chair the American Psychological Association's National Advisory
6:57
Committee for measurement-based care.
6:59
And in that capacity was introduced to Danny Freed, our CEO and
7:03
founder of Blueprint.
7:04
And so today, serve as the head of clinical for Blueprint.
7:09
Blueprint started off in 2019 as a single product offering up technology
7:15
and tools.
7:16
To make the right thing easy to do.
7:18
And in this case, it was measurement-based care.
7:22
And we identified a handful of challenges in terms of selling that
7:27
technology and market.
7:28
And some of these themes I think will ring true for the
7:31
rest of our conversation.
7:32
But one, we don't have appropriate reimbursement behind the activities that are
7:36
key to measurement-based care.
7:38
Thus, we see lower adoption and use across the industry when it
7:42
comes to that.
7:42
But the second, which leads to another big innovation here at Blueprint,
7:47
was that we were competing with time or lack thereof.
7:51
And so whether it was time at the clinician level or capacity
7:53
at the organizational level, we ultimately found that the amount of
7:58
documentation and administrative tasks was ultimately getting in the way of
8:03
organizations and clinicians within those orgs, ultimately doing the things
8:06
that really matter most when it comes to advancing quality and outcomes.
8:10
And so we built out an AI scribe for behavioral health clinicians.
8:15
Today, using our technology set, we are saving anywhere between 7 to
8:19
10 hours of documentation time per week, depending on caseload.
8:23
And with this advent of AI, we also saw an important opportunity
8:27
in the market to bring in actionable insights before, during, and after
8:31
every visit using AI anchored on both transcription data and patient-reported
8:38
outcome data.
8:39
So now we're ultimately preparing clinicians to feel more prepared, more
8:43
confident, but also with those suggestions that can help them deliver the
8:47
best care possible.
8:49
Okay.
8:50
Yeah, there's a lot there.
8:52
I'd love to kind of start with measurement-based care.
8:58
How do you see that?
8:59
You mentioned that there's a struggle for adopting that largely because there's
9:04
not reimbursement for it or not adequate reimbursement for it.
9:09
What are you seeing in terms of clinician adoption of that or
9:14
the need for that or even insurance companies requiring that?
9:19
Yeah, maybe just to set the stage, putting a definition against what
9:24
this thing called measurement-based care is could be helpful.
9:27
So, and I'm going to use the acronym MBC for ease of
9:31
the conversation, measurement-based care.
9:35
MBC is the routine and standardized use of patient-reported outcome
9:39
instruments.
9:40
These are the PHQ-9s and GAD-7s of the world to
9:43
help ultimately inform and prepare clinicians to act on the data that
9:50
oftentimes is focused on symptomology, such as PHQ-9 for depression or
9:55
the GAD-7 for anxiety.
9:57
Functional measures to look at global functioning across key domains,
10:02
occupational, social, and beyond, as well as looking at some of the
10:06
process-oriented dynamics of therapy, such as the patient's perception of the
10:13
quality of the working relationship between the care provider and they
10:16
themselves.
10:18
And so, by using measurement-based care as an evidence-based practice,
10:23
which there's over 25 years of research supporting this, bringing in data
10:28
from patient-reported outcome measures, this helps serve as an additional
10:35
signal to help prepare the clinician to take action sooner across a
10:41
treatment episode.
10:43
And I do think it's important to note that it never replaces
10:47
clinical judgment.
10:48
Again, all data being good data.
10:50
This is very helpful for clinicians to act on this signal earlier.
10:55
And so, when we look at the research around measurement-based care,
10:59
not only do we see the outcomes improve by bringing in the
11:05
patient voice through patient-reported outcome measures, we see the therapeutic
11:10
alliance, the quality of that working relationship improve between the client
11:15
and the therapist.
11:16
We also see something that's notable, is the length of treatment episodes
11:23
also decreases as goals of treatment are able to be obtained earlier
11:29
in that treatment episode.
11:30
And the why behind that is that by having the standardized and
11:34
routine use of the collection of patient-reported outcome data, these data,
11:40
again, serve as that signal that can be picked up on in
11:43
a more reliable fashion earlier in the treatment episode.
11:47
And thus, as we see treatment episodes ultimately reduce in terms of
11:50
length, goal attainment for that specific case also improve.
11:56
We also see cost savings.
11:57
So, from an affordability standpoint, whether you're a payer or anyone else,
12:03
if you see the length of treatment episodes reduce, there's also that
12:07
financial upside.
12:08
So, that's kind of a working definition of measurement-based care.
12:12
But high level, if we think about this, in the same way
12:16
on the physical health side of healthcare, every chronic condition has a
12:21
vital sign, whether hypertension has blood pressure, diabetes has hemoglobin
12:27
A1c, whether we're looking at obesity, maybe it's BMI and weight, we
12:32
don't have those same objective vital signs in the mental health arena.
12:39
And while, yes, we are starting to find different biological underpinnings of
12:44
certain conditions, and I do anticipate that as we move forward, technology
12:49
will really help us bring additional objective data into the care process.
12:55
In lieu of these vital signs, we look to the patient-reported
13:00
outcome measures, and there's hundreds of these validated tools that are out
13:03
there to serve that purpose, which I think is really important.
13:09
I see the value of it, definitely.
13:12
And I think in actual practice, sometimes it gets a little bit
13:17
difficult.
13:17
For example, having patients that are willing to fill out this measure
13:23
on a regular basis, what do you see in regard to that?
13:29
It really starts with the provider.
13:33
And if a clinician is asking their patient, their client, to complete
13:40
these survey-based tools over and over again, yet the value isn't
13:46
necessarily linked to the request of taking time out of their busy
13:51
lives to complete them.
13:52
If the data isn't ultimately reviewed and becoming just a core part
13:57
of what's discussed in a psychotherapy visit, for example, you're going to
14:01
see drop-off in terms of buy-in, ultimately, on the client's
14:05
side to complete these.
14:08
Contrasting that with, let's say, one done well, this becomes an ongoing
14:13
topic of conversation.
14:15
Looking at total scores in terms of symptomology or overall global functioning,
14:20
using item-level scores to help inform and really focus the conversation
14:25
around sleep, attention, whatever it might be in terms of the target
14:31
of treatment.
14:32
It's really important that this becomes just a part of the conversation.
14:37
And there's lots of research that ultimately points to the intervention, as
14:41
it were, in terms of sharing and graphing change over time with
14:46
clients, that this is empowering.
14:49
To be able to see the mile markers of both success or
14:52
even challenge in terms of progress across a treatment episode, it really
14:58
helps with engaging the client in care, but also makes what can
15:02
at times feel very opaque in a change process much more clear.
15:07
And so ultimately, lots of different models out there.
15:11
Some take a very treat-to-target focus, really using these types
15:14
of tools to target the symptomology or the underlying presentation of whatever
15:19
the issue might be.
15:20
And this becomes one of the north stars to gauge success by.
15:24
But ultimately, it shouldn't be the only ruler that is used to
15:29
gauge success.
15:31
So why do we not see it?
15:33
And just to put some numbers to this, outside of large group
15:39
-based practices, oftentimes these kind of aggregators, the headways, rulas,
15:45
almas, the growth therapies of the world, we see in the wild,
15:50
as it were, 19% of psychiatrists, 11% of psychologists, and
15:55
anywhere between 3% to 5% of master's-level clinicians endorsing
16:00
routine use of patient-reported outcome measures.
16:02
If we look to those large groups, again, let's just choose headway
16:05
as an example, 40,000 therapists in this network, we see a
16:11
much larger adoption and use of these, given the kind of payer
16:15
requirements around value-based care, which I know we're going to talk
16:18
about.
16:19
So even if we're being generous, less than 20% of the
16:24
field are using this measurement-based care as an evidence-based practice.
16:29
And we've got our ways to go to ultimately take steps forward
16:33
on that.
16:34
Yeah, yeah, definitely.
16:37
There's a lot of dynamics there, but I do want to go
16:40
into value-based care because that is more and more going to
16:45
be a requirement.
16:47
I know payers that I've worked with have said, this is coming,
16:50
get ready, all of these things.
16:53
What are you seeing specifically in behavioral health care for the patterns
16:57
for value-based care?
16:59
And I think we've defined it a little bit at the beginning,
17:02
but it is basically using some sort of measurement or outcome to
17:07
determine payment.
17:09
I know I was just actually at a conference earlier this week,
17:13
and it was a payers conference, and they were talking about value
17:17
-based care as sort of a shared risk kind of piece with
17:21
providers where sometimes payers feel like they are shouldering all the risk
17:27
because they are paying for the service.
17:29
And this is a way that providers have a shared risk or
17:33
some accountability, if you will, in the outcome.
17:36
But I'm curious to know just some of your views of value
17:40
-based care.
17:42
Yeah.
17:42
So let's take a moment and imagine that we're all working for
17:47
a large payer.
17:50
And just looking, let's say, at Optum.
17:53
Optum has over 500,000 behavioral health providers in their network.
17:58
One of the teams that I led while at Optum was the
18:01
high-performing networks team.
18:03
And the charge of this team was to try to tier the
18:06
network and try to understand through data who within this exceptionally large
18:11
network was delivering differentiated and exceptional outcomes.
18:16
The challenge the payers face is really a lack of actionable data.
18:22
Claims-based data is highly blunt.
18:25
You may have a location code in terms of where the care
18:30
is happening, a diagnosis.
18:33
You'll have CPT codes and other types of codes that map back
18:36
to reimbursement and billing.
18:38
But it's really hard for a payer to ultimately identify and answer
18:44
a seemingly simple question.
18:46
Did our member get better?
18:48
And if so, by how much?
18:50
So that's a member-client level view.
18:53
At the provider level, whether it's an independent clinician in private
18:57
practice or a very large behavioral health organization, that same question is
19:02
equally as difficult to answer.
19:04
To what degree does provider A versus provider B move the needle
19:10
where it matters most in terms of clinical effectiveness as well as
19:14
clinical efficiency?
19:16
And embedded within that, there's obviously a cost-based question.
19:20
And if we look at where the important interest holders within the
19:26
payer arenas kind of sit, employers in particular are really critical.
19:32
And it's interesting to think back in the arc of my 20
19:36
-plus years, 25 years in the field, where we've gone from the
19:41
beginning in terms of sophistication and how informed purchasers were, and in
19:47
this case, we'll talk about employers, compared to where we are today.
19:50
And what we've seen is an evolution in terms of priority.
19:55
Early days, access mattered more than anything else.
19:58
And because of that, payers ultimately built out exceptionally large networks.
20:05
They wanted to make sure that employers and other purchasers could get
20:10
their constituents, consumers into care quickly, because we have a long history
20:16
in behavioral health of delays in terms of those that raise a
20:19
hand and ask for help versus those that can get in quickly
20:22
to get the care they need.
20:24
But then fast forward, what we're now seeing is what is being
20:28
asked of from payers by the purchasers, employers in this case, is
20:34
yes, table stakes are access, we expect it.
20:38
But now they're asking for demonstrated outcomes.
20:41
What are we getting for every dollar that we spend in terms
20:45
of partnering with you, payer, when it comes to the quality of
20:48
outcomes?
20:50
And so this has put an incentive behind payers to ultimately, in
20:55
turn, swivel their head towards their provider networks and challenge them to
20:59
ultimately demonstrate effectiveness.
21:02
Did this member get better?
21:04
Did a population improve in terms of the care that you're providing?
21:07
And it's very hard to answer these questions unless we have some
21:12
sort of objectionable data.
21:14
And that's where measurement-based care comes in.
21:17
Again, these are the vital signs for behavioral health care.
21:20
If we're not collecting outcomes, we can't answer these seemingly simple
21:23
questions.
21:25
Within this, we've also seen a significant increase in terms of outpatient
21:30
utilization across the payer landscape.
21:33
We're seeing on average a 20% year-over-year increase in
21:36
terms of behavioral health utilization.
21:39
And what that looks like to the bottom line of payers is
21:43
they see it as runaway outpatient behavioral health spend.
21:47
Benefit expense continues to rise in this segment across their whole book
21:52
of business.
21:54
And so what we're seeing is a real evolution and kind of
21:58
a stepping up when it comes to what providers are being asked
22:02
to do, if and when they're going to partner with payers and
22:06
be direct in network.
22:07
And where we're seeing a lot of this is in the provider
22:10
aggregator space, so managed service organizations, IPAs, the headways of the
22:15
world, for example, where payers are basically saying, look, if you want
22:20
to continue to grow your network, you're going to need to agree
22:22
to these certain terms.
22:24
One, you've enjoyed premium rates of reimbursement, especially coming out of
22:29
COVID when virtual care access was one of the North Star priorities.
22:34
But now what they're seeing is anywhere between 20% to 30%.
22:38
Higher rates of reimbursement have historically been aligned to these large
22:42
aggregator groups.
22:43
And what the internal evaluation in terms of both cost and quality
22:50
from the payer standpoint, what they've ultimately come to realize is the
22:55
quality of care happening in many of these groups is not that
22:58
much better than the quality of care happening outside of those groups.
23:02
And as they look to those that are, again, just direct and
23:04
network small, midsize group practices, even independent clinicians in private
23:10
practice, they're asking themselves, why would we want to pay a 20
23:14
% to 30% premium for the same type of outcomes?
23:18
And so it's a complex space.
23:21
Measurement's going to be key.
23:23
And so one of the strategies in terms of frameworks that payers
23:26
are embedding within their networks is this alternative payment models.
23:33
So value-based care lives on a continuum.
23:37
It goes from fee for service, far left, all the way to
23:40
full capitated, full shared risk, far right.
23:43
But lots of different steps in between.
23:46
What we see most in terms of behavioral health, in terms of
23:48
value-based care, is ultimately a pay for performance type model, where
23:53
if a provider organization successfully obtains certain thresholds across key
24:01
metrics, and many of these are process-oriented, very few seem to
24:07
have ultimately arrived at true outcomes, then there's an additional upside
24:11
that can be paid out.
24:12
And so why we haven't seen such a large shift towards value
24:17
-based care compared to, you know, joint replacement or other types of
24:23
physical health examples of value-based care that have been around for
24:27
many years, is really this linchpin of measurement.
24:31
And so back to Blueprint, we see that there is a critical
24:36
need within the provider community, within the practice community, to make
24:41
bringing in this evidence-based practice of using standardized outcome
24:45
measurement.
24:46
This is really key to not only improving care and outcomes for
24:50
every client and patient served, but ultimately helping to unlock value across
24:55
all level of the health system, including at the payer level, but
24:59
also at a population level, policy level, and beyond.
25:04
Yeah.
25:05
So what do you think is the value add for patients in
25:09
this value-based model?
25:12
Well, if I were looking for a therapist for somebody that was
25:18
in my family or even myself, and if we wanted to draw
25:22
an analog to another health condition, imagine, and unfortunately, let's say in
25:28
this case, you know, a loved one was diagnosed with one of
25:33
300 different cancer types.
25:35
My guess is that you would want to make sure that the
25:38
oncologist that you're connecting your loved one with not only is expert
25:42
in treating that subspecialty focus area in terms of that type of
25:48
cancer, but that the outcomes in terms of mortality, morbidity, but also
25:53
recovery rates, that those were stacked up.
25:56
That's why, from a consumer standpoint, you'd want to make sure that
26:01
the right provider is getting connected to the right patient at the
26:04
right time for the right condition, and they have a track record
26:07
of success in terms of moving on outcomes.
26:12
Similarly, in behavioral health, while yes, in the mild to moderate arenas,
26:16
and given prevalence rates of depression and anxiety, there's lots of providers
26:21
that are well-positioned to address these things.
26:24
But as we look towards perhaps serious and persistent mental illness, eating
26:28
disorders, OCD, substance use disorders, it really requires a certain training
26:33
and specialty type.
26:35
Back to it, it's really important from just across the landscape that
26:41
we have data that can help answer, again, these seemingly simple questions,
26:46
but those that we've been kicking the can down the road as
26:49
a system.
26:51
I love it.
26:52
Yeah, I think it's important to see the benefit for all the
26:56
players in the landscape for measurement and base care.
27:00
I know we don't have a whole lot of time left, but
27:04
I want to kind of ask you about Blueprint specifically and some
27:09
of your thoughts moving into AI.
27:12
How does AI relate to this conversation of value-based care and
27:16
measurement-based care?
27:17
Yeah, so as I mentioned moments ago, the thesis that we had,
27:23
the hypothesis ultimately, was that if we could offload some of the
27:29
administrative demands from clinicians and within an organization and free up
27:36
capacity, some breathing room, for clinicians to really do what they're best
27:40
at, which is providing care, that we would see a few things.
27:45
One, our hypothesis was that if we introduced an ambient scribing technology,
27:52
such as our documentation automation solutions, our AI scribe, that we would
27:57
save time, and again, in tune between 7 to 10 hours per
28:00
week, and that with that time, clinicians would do the things that
28:05
they historically have had less time to do, such as measurement-based
28:10
care.
28:11
And what we've learned as a system, that when we do introduce
28:15
an AI scribe, that we're seeing close to a 60% increase
28:19
in adoption and use of measurement-based care on our platform.
28:22
That's huge.
28:24
That is huge.
28:24
And so that's important.
28:27
Second, if we zoom out, we have a workforce challenge when it
28:32
comes to the practice community, and by any respect, it's somewhere around
28:39
320 individual Americans in our nation who have an unmet behavioral health
28:47
need, could benefit from treatment.
28:49
We see, depending on the data source, anywhere between one in four
28:54
or one in five adults and adolescents will experience some sort of
28:59
mental health issue within a given year.
29:02
And so if we look at 320 individuals that could benefit from
29:05
care for every one provider that's ultimately licensed and capable of providing
29:10
that care, this is a real kind of supply-demand mismatch.
29:15
And so we need to be more efficient in terms of how
29:19
we deliver care, and we also need to ultimately create sustainable work
29:23
models.
29:24
So we're not seeing rates of burnout and ultimately therapists leaving the
29:30
profession, which is top of mind for many organizations today.
29:36
I have the privilege of spending an ordinate amount of time with
29:40
clinical leaders across hundreds and hundreds of different organizations, and
29:44
what we hear loud and clear is that there's a competition for
29:49
clinical talent, that once you get a quote-unquote good therapist or
29:53
good provider in the organization, keeping them engaged, keeping them happy,
29:57
and providing the right support so that they can