Multilevel Interventions in Health Care Conference: Introduction by Stephen Taplin, MD, MPH

>>>DR. STEPHEN TAPLIN: I'm Stephen Taplin from the
National Cancer Institute. I was a family doc for twenty
years in the Northwest before coming here. And I can't multi
task - I can't introduce myself and - there we go.
Alright, so thank you, thank you all for coming,
thank you Arnie for setting the stage so well and thank you for
your wisdom during the development of this conference.
Working with Dr. Kaluzny was an incredible privilege for us and
it's one of the great things about being at the National
Cancer Institute is that we have a chance to convene great
people and talk with them and to bring them together and
support this kind of work. So thank you for being part of
this, Arnie. Also I wanted to acknowledge that we have some
others that were able to bring in: Dr. Richard Scott,
who will be talking later today; another luminary,
Otis Brawley, tomorrow from the ACS who is known all over
the country for the work and for his ability to
say things the way they are. And so we are very excited
about having all of these people here and the chance
to interact. Also wanted to thank the people here you see
listed on this slide who are all part of the thinking behind
the conference and the thinking behind this paper and hopefully
the thinking in the follow-up, as well as taking care of many
of the details that made it all possible.
Lots of things happen. If the conference goes well,
it is because these people have worked really hard to make
sure all the details are there, so thank you to all of them.
But let's start. You've heard the plan.
You've heard some of the problems.
Let's elaborate and let's get to the discussion.
A little more background: U.S. healthcare overspends.
If you look across the nineteen developed countries in the
world, the U.S. is far and away the most expensive
place to get care. But it's not, of course, that simple;
it's worse than that. It's also that the U.S. healthcare
under achieves. If you look at what we get for our dollar,
55 percent of people with chronic illness get the
recommended care - 55 percent get what is recommended in care
when they come into that expensive healthcare system.
We are 19th among 19 developed countries in the reduction of
avoidable mortality. It's a measure of looking at - if you
look at conditions in which you can make a difference,
how quickly are we reducing deaths due to those conditions.
The U.S. is 19th out of 19 developed countries.
We are 13th in infant mortality.
We have a ways to go, we have things we need to do and the
U.S., we know one thing is absolutely certain: things are
going to get more expensive. So we have, we are
underachieving and we know we are facing higher costs.
We simply cannot keep doing what we are doing,
thinking the way we have been thinking.
124, 125 billion - I'll just round it up - 125 billion
dollars in 2010, 27 percent increase anticipated for 2020
by my colleague Martin Brown and his team at the National
Cancer Institute. So we need to do better and in fact,
that is one of the basic assumptions is that we
need to do better. We take that for granted.
We also need to think carefully, and Arnie has already alluded
to this, about whether the approach we've been taking and
the reductionist approach to finding simple little steps
that are exquisitely designed and exquisitely and validly
proven, whether that particular approach can continue to work
in our system. It takes a long time to get those little pieces
into place even when we are convinced they work, and
then they are not always adopted and they are not always used.
So we need to begin to get out of that approach.
That's an assumption. We take that as an assumption, as a
starting point. We also think we can do better in pursuit
of health, in rethinking our questions about delivering
care and in considering multilevel interventions.
Now the consideration of multilevel interventions is not
necessarily completely new at all. People have been
talking about multilevel interventions for quite awhile.
My colleague Erica Breslau and Helen Meissner,
in 2004 in a Cancer supplement, looked back at twenty years of
screening history in which they were identifying effective
interventions for twenty years in screening,
and at the end of that twenty year period concluding that
they are not all being adopted. There are efficacious
interventions and screening that are not being used
and we continue to generate more. And that maybe we
need to step back from the reductionist approach
and ask the question what is the context in which
these interventions are being implemented,
how do we begin to understand and address that context and
begin to think about that before we are faced with the
task of implementing. And so we are taking that as
a basic assumption that in fact, the reductionist approach
has not worked completely well. It certainly has contributed
many things to what we know, but we want to step
back, as Arnie says, rotate 90 degrees perhaps, look at the
runway in a different way and begin to think about
what we are doing in a different way.
So let's define a couple of terms here because we have
been throwing them around already.
One is "multilevel interventions."
So what exactly is an intervention?
And when we are talking about an intervention,
we are saying "specific strategy or set of strategies
designed to change the knowledge,
perceptions, skills and behavior of the individual
groups or organizations with the goal of improving
patient health outcomes." So we are trying to think about
the context directly linked to patient health outcomes
and we are looking at whole sets of strategies to do that.
We're also defining "levels" in a particular way and we are
saying that they first are patient-centered so we want to
be, as we said, linked to health outcomes.
We want to look at a measure at the patient level,
but then target at least two other levels of a multilevel
model of improving health status,
which I will explain in just a second.
And then we also want to measure effects at each of
these levels. Now we set out with this definition at the
beginning as a sort of stake in the road, a stake
in the ground, a place to begin.
And we then started to look for literature, could we
find examples in the literature beyond cancer, anywhere
that we could find that would meet those definitions.
And we found a very sparse literature that would meet
those definitions. We actually, looking back I think fifteen
years, we actually could find only three which were clear
multilevel interventions in which there was a clear
intervention at at least three levels and these are the
three tests. And I think you all, if you have been around,
and I can tell that some of you are gray,
that you've been around a little bit,
that COMMIT, ASSIST and CATCH and all pieces that we have
heard about. They have addressed three levels,
individuals, organizations, communities.
They have measured some outcomes - smoking cessation in
the community or cholesterol levels and BMI at an individual
level, but none of them actually addressed all three
interventions with measures at all three interventions.
And so we decided that we needed to do more and that we
would do that not based on the fact that there was a clear,
compelling evidence that multilevel interventions
worked. In fact, it is not there, but because the
reductionist approach has not worked completely well for us.
It has developed little bitty steps, but we need to be
stepping back, we need to rotate 90 degrees and
look at the runway in a different way.
So we decided to convene a group of people to start,
many of them are in the room, to begin to generate the issues
and generate a set of recommendations,
and we convened in June of 2009.
And as a result of that meeting, we gathered together
a set of topics to talk about and that is the basis
of the supplement that Steve has already alluded to
that will be coming out, we hope, in the fall of 2011.
The supplement topics are the focus here and as Arnie has
already alluded to, we're going to first,
section one, talk about definitions.
We will also talk about, Jane is going to elaborate
on a couple of cases that use multilevel thinking
and then Kurt Stange is going to tell us about
the state of the art in that area.
Section two, we are going to be looking at little bit harder
at conceptual issues. Brian Weiner is going to talk about
how you conceptualize a synergistic intervention.
Looking at time, time is a critical issue.
Paul Cleary is going to talk about design and you'll see
down there Marty and Joe are going to add their thoughts.
Section three, we'll get back to reality a little bit and focus
on what's actually happening in care here in the United States.
So we look forward to an exciting set of papers.
And we look forward to your comments.
As you have also heard, we want your input, we want your
critical thought, we want your skepticism. This is the time
to step up and we look forward to hearing from you.
So session one, multilevel influences in interventions
across the cancer care continuum.
Now I have defined levels and I have defined what we mean by
interventions, but let me get a little deeper into levels and
acknowledge that the way that I have already alluded to in
terms of levels is only one way to do it. There are other
ways. There is ecological and psychological models
that talk about inter-personal and policy level.
There is a system model, individuals and groups,
organizations, economic and social systems.
These are all ways that are out there in the literature that
talk about multilevel interventions.
We have chosen a biopsychosocial model developed
by a person from Rochester, New York,
which is where I trained as a family doc,
and that is George Engel. He looks at individuals, groups,
organizations, communities in the nation,
really looking at the aggregates of individuals for
me, for us, as potential targets of interventions.
And so, really, we take that model as our model.
A layered context of care is what we call it.
We also call it the onion. And when we look at that
layered context of care, we expect that the effects of
those layers will improve; that if we address factors
in those layers, we can affect quality and we can
affect outcomes. But it's also clear this is a cross-sectional
view and the outcomes and care is not a single event.
That in cancer, care is a continuum.
That there are a number of types of care,
from looking at people at risk in the population to thinking
about primary preventions, so changing diet, changing
physical activity, thinking about detection or diagnosis.
And each of these types of care need to be addressed in order
to move somebody all the way across the cancer continuum.
So there is a time element, there is a longitudinal
element, but there is also an element and a fact that I think
any practioner in the audience knows that it's the connection
between the steps as much as the steps that matters.
And if you are in the middle of the care process,
it's how you link people up and link organizations and link the
steps that actually makes it possible to traverse that
entire continuum. So we think also about the transitions
in care and it's the transitions and the types of care,
it's those two pieces together, that address the quality
measures that we all know so well from the IOM.
And that those quality measures then affect patient outcomes
and we can look at patient outcomes at the individual
level, their biologic outcomes, their quality of life or we can
look at the population level, the summary across all
individuals, to look at mortality or look at morbidity
or look at cost effectiveness. And when you break down
that continuum and think about that cross sectional
association or think about the links between, the more
you look, the more complicated it becomes. There are, in fact,
multiple little steps between types of care. So you can
think about how results reporting occurs. How is that
the result of a test actually gets to the individual?
Who is doing that? How is the test getting to the person
that is supposed to get it to -- the individual?
And as you begin to take apart the care process,
there are, in fact, lots of little steps,
lots of little places where the organization and the context in
which care happens can affect that step.
So it is a way of thinking about care as a longitudinal
rather than a single event; a longitudinal rather than a
didactic discussion between a doc and a patient, but as
a process overall. And that is a critical step in understanding,
we think and I think, in understanding the process of
care, and understanding the potential influence of these
multiple layers on that process of care.
And then when you string out that process of care,
you begin to think that again, in fact,
it is a long series of events and as a person enters that,
they are then affected by family and social support all
the way across that continuum. They are affected by providers,
a set of providers, all the way across that continuum.
They are affected by the organizations all the way
across that continuum. And they are affected by the
communities in which they live and the communities in
which that care occurs, all the way across the continuum.
But of course, it's not that simple. In fact, if you look
at it, there is reason to believe that the effects
are different at different points in the continuum.
So family and social support may be very important at the
beginning and providers may play less of a role.
But as you move into the detection and diagnosis and
treatment part, providers become a bigger factor.
They are more important to actually what happens,
and family are still there, but may play a lesser role.
And if you get to the end, it may jump back.
Families and social support may vary again and become higher,
so that the factors are not monotonic; they are varying
across the continuum and we haven't even thought about
organizations and communities which also are going to
affect and going to change their influences across care.
So we are beginning to create a very complex process,
it is clear, multiple layers affecting particular processes
and care and a longitudinal effect over time with
varying levels of effect. And how do these
levels affect each other? It's another critical question and
I think it's one of the questions, at least where I
started to read the literature, where I got a
little bit confused between what is a level
and what is a mechanism of influence.
And so our structure is policy level and there are papers out
there that use policy as a level,
or is policy just a way of one level affecting another?
That, in fact, policy can be at a national level,
can be at a state level, could be at the community level,
could be in a team that can set policy. So I tried to
separate out the factors that are levels, human sets
of aggregation, and factors that are ways that those levels
influence each other and think about structure, policy,
incentives, interpersonal interactions to affect people.
Those are all mechanisms and potential ways that you would
organize an intervention. So care occurs in context.
And it's a complicated context and we need to think about.
Any provider who has worked within this context knows the
richness and knows, I think has experienced,
coming back from a continuing medical education program,
excited about something new you have learned and then try to
put it into practice and figure out that there is resistance.
It doesn't quite go the way you think it is going to go.
And so I think it's important to get a good feel for that and
to think about how the multilevel context
affects us every day in delivering care.
So care is a process. We can begin to think about national
level policy and medical reimbursement all
affecting care. We were quite concerned that no matter
what we did in this meeting today, if they shut
down government, we would be going home tonight.
That didn't happen, thank heaven.
There is state health policy, medical reimbursement,
there's local community environment, community level
resources, local hospital and cancer services,
local professional norms all affecting what happens within
that community and how those processes get delivered.
There are practice setting differences,
how does the physical structure within the practice allow
communication or inhibit communication?
And there are team factors, do people actually practice in
teams or are they just parallel play in groups? There is
also the family, how you connect to the family, whether
you incorporate the family in the discussion or not.
And then there is ultimately the individual patient and how
we think about talking with them and interacting with them.
So it is a complicated world. We can build from the
individual and go to the national.
We want to think about all of those levels,
but focus, as Arnie said, primarily in the individual,
family, social support, provider and community
organizational settings. So for discussion at the tables,
we hope that you will ask the question what is a "level."
Not to reflect necessarily -- or you can reflect what we said
-- but to look critically at the question that I have raised
and ask is the level really, is a unit of organization an easy
way, a good way, a useful way or are the alternative ways
better, worse, why? So think carefully about level -- how is
the level distinguished from the mechanism of influence,
how do levels affect each other, what difference
does it make in terms of how we design levels
and how we design interventions?
And I look forward to your discussion.
>>>[APPLAUSE]