Multilevel Interventions in Health Care Conference: Discussant comments by Thomas Vogt, MD, MPH
Uploaded by NIHOD on 05.05.2011
Transcript:
>>>DR. STEVE CLAUSER: Our discussion for this session is
Tom Vogt. I can tell, he's got his hands full with this group.
But he's a former senior investigator at
Kaiser Permanente Center for Health Services Research.
His work is investigating improving prevention services
in medical care settings, the quality and cost of preventive
care, primary care organization,
and satisfaction with care across several states and
multiple managed care systems. And it looks like he's more
than prepared to deal with this group.
So, without further ado I'll let Tom go ahead.
>>>[APPLAUSE]
>>>TOM VOGT: Thank you. I'm not sure anyone is prepared
to deal with this group, but I'll start out.
I'm not going to show you slides.
I made some, and I looked at them and didn't like them
because as I thought about these papers,
I sort of - synthesis of them came together and so I am going
to do this verbally. I do have slides and they'll be in the
little thumb drive you've got, if you must see them.
But they're not my preferred approach at this point.
I want to start out by talking about what a practical
multi-level intervention should have.
What are the characteristics that we're shooting for here?
Well, I thought about categorizing in light of the
three presentations that you've just heard.
And when I started that I had notes all over the place.
And as I looked at those notes they sort of distilled down
into the following points. One, a practical multi-level
intervention should be broadly applicable.
It should apply to a lot of folks because otherwise you're
wasting a lot of time and money on the few that it
does apply to. It should be effective across multiple
levels, which is both a design issue and a synergy issue.
And we've been talking about both of those things.
It should be adopted across multiple levels,
which of course is synergistic approach.
It should be implemented faithfully, meaning that the
design of what you put in place should reflect the
knowledge base at the time you do the design of the study.
It should learn from the past. And then finally, it should
be maintained over a long period of time.
The time issue that Dr. Alexander was discussing.
And it should be maintained, not just in one level but
presumably in multiple levels, although that's a question
in itself. Now, if you think about the points that I just
made and you list them like I did and you redefine them,
they look a lot like reach, efficaciousness,
adoption, implementation and maintenance. Which using those
terms you may have heard because Dr. Glasgow is the
father and grandparent of the REAIM model,
and is sitting out there. And it seemed to me the REAIM
model applies not just to single interventions but it applies
to how you would approach the design and the evaluation
of multiple level interventions considering all those elements.
But it isn't sufficient; it's a place to start.
There are some badly needed things that we rarely discuss
and that I'm going to take the opportunity of standing up here
to mention, because I think that in some ways our approach
to behavioral interventions in general is sort of like the
story of the emperor's new clothes. And I should warn you,
I'm going to use two clichés in this talk and
this is the first one. The emperor's new clothes.
I think clichés are clichés because there's almost always
some truth to them. In the story of the emperor's new
clothes you will recall the emperor was convinced to
walk around naked and he told everybody how wonderful
and beautiful his new clothes were so everybody
believed him until some naive little child said,
he doesn't have any clothes on. And suddenly everybody
realized that the child was right. Well, I think in our
interventions there are some examples like that.
I think the most effective prevention intervention I have
encountered in my entire career was the implementation of HITA
standards for integrated health care systems in the
early 1990's. In those days integrated health care systems
were being dogged by various contractors to give data,
and they all wanted something different, so they finally got
together and agreed that NCQA would do these HITA
standards and everybody would be providing the same data.
In a single year immunization and screening rates increased
by 50 to 100% in nearly all the participating health
care systems. I challenge you to find any better intervention
in all of our studies, 50 to 100% in one year.
Why did that happen? It happened because the health care
systems involved made a budget for doing it.
And it happened because they made the people who were
supposed to do it accountable for doing it and doing
it right. And in one year everything changed.
Budget and accountability is something I never see in
behavioral studies. It's not measured, it's not a variable,
but it is crucial. If you don't have it in the budget,
even if it's implemented it will fade away. And I think
we've all experienced that. Besides budget and
accountability, the impact must be sustained over time.
I think that there are some opportunities right now that
are new for all of us as researchers, that have not
been available before that will help us do multi-level
interventions at least in health care systems,
which is the one we're talking the most about.
For those of you who haven't, some of you in this room
have been participating in these. But those who haven't,
I think you may not realize the degree to which progress
has been made in standardizing and collapsing across health
care systems longitudinal electronic medical records data
to ask questions at reasonable cost that we've never been
able to ask before. Some of my colleagues and I have been
doing this for more than ten years in the Kaiser Permanente
system, which is just one member of the National HMO
Research Network, which now has a virtual data warehouse
that covers more than 15 million people across 18 health
care systems. And it isn't easy, but it is now possible to
collect data on total populations of health care
systems over long periods of time, a decade or more.
And that offers extraordinary opportunities to create
historical prospectus studies in ways that we've never been
able to do before. Those don't prove necessarily that
they're correct, but they give us a new avenue for
developing hypotheses and looking at questions that we
simply can't look at in randomized trials because
they're too expensive and too long term.
So having said that, I would advise those of you who are
interested to get in touch with some people who have worked
with those kinds of data. Jane Zapka sitting down here
has done that, and so have several others in this room.
Diana Buest, I saw her somewhere. And I would like
to finish by going back to my second cliché which
is the old story about the drunk who is crawling
around on the ground and the policeman walks up and says
what are you looking for? And he said, my keys, I lost my
keys. And the policeman says, well where did you lose them?
And he said, well over in the alley over there.
And the policeman says, why are you looking for your
keys here under the streetlight, and he said because it's dark
in the alley. And the point I am going to bring up with that,
and I think this is crucial crucial if we want to do
effective multi-level interventions, if we want to
address the problem that Dr. Quisney raised this morning
that we ranked 36th in the world in life expectancy,
but that was in the year 2000, and in 2009 we ranked 52nd.
So that's the direction that we're going. And that is,
we have to acknowledge as researchers even if our
benefactors at the National Cancer Institute or
other NIH institutes are not allowed to acknowledge,
that the problem when we want to look at cost effectiveness
and cost containment that is responsible for poor and
overly expensive U.S. health care, is largely not scientific
but social and political. And it was the original proposals
from the Obama Administration to reform health care
included clear attempts to reduce costs.
For example, the outrageous notion that Medicare might be
able to negotiate drug prices. Congress refused to allow
them to do that. All of the cost cutting elements of the
original proposals were stripped.
Which is pretty much what killed the 1994,
1993 attempt at reform in the Clinton Administration.
Don't researchers have an obligation to start talking
about the social and political contributions to
ineffective and overly expensive care? Thank you.
>>>[APPLAUSE]
Tom Vogt. I can tell, he's got his hands full with this group.
But he's a former senior investigator at
Kaiser Permanente Center for Health Services Research.
His work is investigating improving prevention services
in medical care settings, the quality and cost of preventive
care, primary care organization,
and satisfaction with care across several states and
multiple managed care systems. And it looks like he's more
than prepared to deal with this group.
So, without further ado I'll let Tom go ahead.
>>>[APPLAUSE]
>>>TOM VOGT: Thank you. I'm not sure anyone is prepared
to deal with this group, but I'll start out.
I'm not going to show you slides.
I made some, and I looked at them and didn't like them
because as I thought about these papers,
I sort of - synthesis of them came together and so I am going
to do this verbally. I do have slides and they'll be in the
little thumb drive you've got, if you must see them.
But they're not my preferred approach at this point.
I want to start out by talking about what a practical
multi-level intervention should have.
What are the characteristics that we're shooting for here?
Well, I thought about categorizing in light of the
three presentations that you've just heard.
And when I started that I had notes all over the place.
And as I looked at those notes they sort of distilled down
into the following points. One, a practical multi-level
intervention should be broadly applicable.
It should apply to a lot of folks because otherwise you're
wasting a lot of time and money on the few that it
does apply to. It should be effective across multiple
levels, which is both a design issue and a synergy issue.
And we've been talking about both of those things.
It should be adopted across multiple levels,
which of course is synergistic approach.
It should be implemented faithfully, meaning that the
design of what you put in place should reflect the
knowledge base at the time you do the design of the study.
It should learn from the past. And then finally, it should
be maintained over a long period of time.
The time issue that Dr. Alexander was discussing.
And it should be maintained, not just in one level but
presumably in multiple levels, although that's a question
in itself. Now, if you think about the points that I just
made and you list them like I did and you redefine them,
they look a lot like reach, efficaciousness,
adoption, implementation and maintenance. Which using those
terms you may have heard because Dr. Glasgow is the
father and grandparent of the REAIM model,
and is sitting out there. And it seemed to me the REAIM
model applies not just to single interventions but it applies
to how you would approach the design and the evaluation
of multiple level interventions considering all those elements.
But it isn't sufficient; it's a place to start.
There are some badly needed things that we rarely discuss
and that I'm going to take the opportunity of standing up here
to mention, because I think that in some ways our approach
to behavioral interventions in general is sort of like the
story of the emperor's new clothes. And I should warn you,
I'm going to use two clichés in this talk and
this is the first one. The emperor's new clothes.
I think clichés are clichés because there's almost always
some truth to them. In the story of the emperor's new
clothes you will recall the emperor was convinced to
walk around naked and he told everybody how wonderful
and beautiful his new clothes were so everybody
believed him until some naive little child said,
he doesn't have any clothes on. And suddenly everybody
realized that the child was right. Well, I think in our
interventions there are some examples like that.
I think the most effective prevention intervention I have
encountered in my entire career was the implementation of HITA
standards for integrated health care systems in the
early 1990's. In those days integrated health care systems
were being dogged by various contractors to give data,
and they all wanted something different, so they finally got
together and agreed that NCQA would do these HITA
standards and everybody would be providing the same data.
In a single year immunization and screening rates increased
by 50 to 100% in nearly all the participating health
care systems. I challenge you to find any better intervention
in all of our studies, 50 to 100% in one year.
Why did that happen? It happened because the health care
systems involved made a budget for doing it.
And it happened because they made the people who were
supposed to do it accountable for doing it and doing
it right. And in one year everything changed.
Budget and accountability is something I never see in
behavioral studies. It's not measured, it's not a variable,
but it is crucial. If you don't have it in the budget,
even if it's implemented it will fade away. And I think
we've all experienced that. Besides budget and
accountability, the impact must be sustained over time.
I think that there are some opportunities right now that
are new for all of us as researchers, that have not
been available before that will help us do multi-level
interventions at least in health care systems,
which is the one we're talking the most about.
For those of you who haven't, some of you in this room
have been participating in these. But those who haven't,
I think you may not realize the degree to which progress
has been made in standardizing and collapsing across health
care systems longitudinal electronic medical records data
to ask questions at reasonable cost that we've never been
able to ask before. Some of my colleagues and I have been
doing this for more than ten years in the Kaiser Permanente
system, which is just one member of the National HMO
Research Network, which now has a virtual data warehouse
that covers more than 15 million people across 18 health
care systems. And it isn't easy, but it is now possible to
collect data on total populations of health care
systems over long periods of time, a decade or more.
And that offers extraordinary opportunities to create
historical prospectus studies in ways that we've never been
able to do before. Those don't prove necessarily that
they're correct, but they give us a new avenue for
developing hypotheses and looking at questions that we
simply can't look at in randomized trials because
they're too expensive and too long term.
So having said that, I would advise those of you who are
interested to get in touch with some people who have worked
with those kinds of data. Jane Zapka sitting down here
has done that, and so have several others in this room.
Diana Buest, I saw her somewhere. And I would like
to finish by going back to my second cliché which
is the old story about the drunk who is crawling
around on the ground and the policeman walks up and says
what are you looking for? And he said, my keys, I lost my
keys. And the policeman says, well where did you lose them?
And he said, well over in the alley over there.
And the policeman says, why are you looking for your
keys here under the streetlight, and he said because it's dark
in the alley. And the point I am going to bring up with that,
and I think this is crucial crucial if we want to do
effective multi-level interventions, if we want to
address the problem that Dr. Quisney raised this morning
that we ranked 36th in the world in life expectancy,
but that was in the year 2000, and in 2009 we ranked 52nd.
So that's the direction that we're going. And that is,
we have to acknowledge as researchers even if our
benefactors at the National Cancer Institute or
other NIH institutes are not allowed to acknowledge,
that the problem when we want to look at cost effectiveness
and cost containment that is responsible for poor and
overly expensive U.S. health care, is largely not scientific
but social and political. And it was the original proposals
from the Obama Administration to reform health care
included clear attempts to reduce costs.
For example, the outrageous notion that Medicare might be
able to negotiate drug prices. Congress refused to allow
them to do that. All of the cost cutting elements of the
original proposals were stripped.
Which is pretty much what killed the 1994,
1993 attempt at reform in the Clinton Administration.
Don't researchers have an obligation to start talking
about the social and political contributions to
ineffective and overly expensive care? Thank you.
>>>[APPLAUSE]