Multilevel Interventions in Health Care Conference: Discussant comments by Maria Fernandez, PhD

>>>DR. STEVEN CLAUSER: Okay, I would like to move now to our
discussant for our other program and introduce Maria Fernandez.
Dr. Fernandez is Associate Professor in Health Promotion
and Behavioral Science at the University of Texas and has
been very involved in this field for some time.
>>>[APPLAUSE]
>>>DR. MARIA FERNANDEZ: Thank you. First I'd like to thank
the two Steves for the invitation to participate and
also as such, to have a preview of three outstanding papers.
I am going to just highlight quickly some of the
key points of each paper, since you just saw those,
discuss some cross-cutting themes,
issues and challenges and, recommend some questions for
discussion, including the questions that you just saw at
the end of each paper, presentation, but also a few
additional ones. So in the first paper we saw that the authors
provided a framework for understanding multilevel
influences and identifying intervention targets and
importantly, provided some definition of key terms.
The intent is both to better understand and to influence the
interplay of forces at these multiple levels.
Paper 1 also proposed desired measures of success based on
the AHRQ's and IOM's definition of quality of care across the
cancer continuum and articulated a key research
question as we move forward -- "how do the levels of influence
affect each other and the processes of healthcare
delivery and health outcomes?" In Paper 2, Jane Zapka and
her colleagues presented case studies and hypothetical
interventions at different levels.
I think this paper does an excellent job in helping us see
what an individual experience is,
as well as the different levels that influence the quality of
their care. They used these cases to illustrate both the
complexity and the types of multilevel interventions and
identified intervention targets at various levels.
And then finally they discussed challenges in designing studies
to determine the impact of different components at
different levels. In Paper 3 the authors did an outstanding
review of the literature looking at articles from
cancer control, chronic disease and the prevention literature
and identified opportunities related to design,
analysis and translation and especially for interventions
that create synergy across the levels and adapt over time.
In the paper, they described four examples and show the
potential for how this might be addressed.
They conclude that disciplinary,
participatory research can generate knowledge to reduce
the cancer burden and I will talk a little bit more about
that in a minute. So now to the cross-cutting issues,
themes and challenges that really were presented in all
the papers. First, I think, important to point out the
opportunities for research on multilevel interventions
across the cancer care continuum and why cancer is a good
model for evaluating multilevel influences.
Part of the reason, and we've heard a lot about this,
is that it includes several types of care,
as well as transitions between the types of care,
from prevention through end of life. And as such, it lets
us look at contextual influences at these different stages.
So the types of care, screening, diagnosis and treatment include
process steps, as Steve talked about, within each of these
stages of care and individuals across the levels that
influence and interact in different ways depending on
where the person is in that stage of care.
So the challenge, and I believe this is brought up in paper
three, is to design interventions that improve care
for patients at different points across the continuum,
so they may be very different for each individual patient,
yet are sustainable and flexible and generalizable
enough to impact population groups.
So some of the gaps, and a lot of this came from Paper 3 in
the literature review, was across the cancer control
continuum, most multilevel interventions have focused on
prevention screening and end of life but seldom on diagnosis,
treatment, and surveillance. There are few detailed reports
of how multilevel interventions have been implemented and how
they have become successful or unsuccessful.
And part of the reason for this that is highlighted in the
paper may be issues in funding, and this is an important
challenge that I hope you'll discuss at your tables,
how do we fund these types of interventions and intervention
research, and also publication bias is another issue that was
brought up in the paper - most of empirical reports are RCTs
and we may need to move beyond this. Another important gap
is that theories, models and interventions are not well
integrated in reports and studies.
Other gaps include that while many studies use ecological
systems and complexity models, few of them describe how they
have applied this theory to inform interventions and very
few measured impacts at multiple levels.
Again, you heard that while multilevel interventions are
contextual, context is inadequately reported.
There is more need for measuring these effects.
There is no clear guidance on the types of intervention
strategies at various levels. And probably, most importantly,
as while we all think that this is a good idea,
and there is some evidence that it might work, effectiveness
of multilevel interventions is not yet evidence based.
I wanted to talk a little bit about theory because theory is
sort of mentioned across the papers.
And in particular, although I think Steve and others point
out that there really isn't a unifying theory that can help
answer all the questions we have as we try to understand
multilevel influences and intervene,
there are some theories that can be helpful.
The social cognitive theory, for example, informs how
individuals and their social and built environment
interact through a concept called reciprocal determinism.
Organizational development theories and organization
theories illustrate how organizations innovate by
establishing goals, programs and ideas.
Network theories describe webs of linkages between
organizations and the theory of complex adaptive systems
suggests that effects travel in multiple directions and it is
not necessarily linear or hierarchical.
The practical integrated systems model or PRISM combines
elements of both organizational theory and partnership models
and then RE-AIM helps us understand targets and
considerations for increasing your reach and implementation
and maintenance. In Steve's paper, he also mentions how
Precede-Proceed may be useful in planning interventions,
multilevel interventions and I would also add that
intervention mapping may also be useful in planning as well
because it helps us consider individuals and their behaviors
at multiple stages, and I will talk just a little bit about
that in a minute. So essentially though, no single theory
has been developed to explain complex relationships between
contextual factors and behaviors of those providing or
seeking care - an important gap.
Across the papers they identify design and analysis challenges,
including the fact that while the current paradigm in which
level specific theories drive level-specific interventions
and measures is inadequate and what is needed is measurement
across levels and statistical methods that consider
inter-dependence of the outcomes.
I think it was in Paper 3 that Zapka and colleagues pointed
out that like multi-faceted interventions where we know
they work, but we're not exactly sure what part of that
intervention worked, it may not be possible to determine which
elements or levels of multilevel interventions are
the most important or effective and that is something that we
should consider as we begin to design multilevel intervention
research studies. Ultimately again, studies are needed to
test whether multilevel interventions have a greater
effect than intensive interventions at a single level
and cost effectiveness also should be a consideration.
Some of the opportunities, I'll go through this quickly because
I believe that Kurt covered a lot of this.
Dynamic adaptive designs that evolve over time are promising.
Multilevel statistical methods and models that allow
measurement and comparison at different levels.
And the use of multi-method approaches, both qualitative
and quantitative methods is promising, as well
as complex systems and dynamic simulation modeling.
I think importantly, and this was highlighted in a couple of
the papers, training, research partnerships need to be more
trans-disciplinary, inclusive, democratic and participatory.
And I think that much of what we know about participatory
planning in the community setting can be applied to
planning interventions at multiple levels.
And then lastly, as sort of a cross-cutting issue and
observation, I think that there is a lot of overlap between
multilevel intervention research and dissemination and
implementation research. And some of these have been pointed
out and includes the observation that discovery of improvements
and isolated steps of care, that the reductionist approach that
Steve talked about, doesn't necessarily lead to their
incorporation of care. And part of that reason is because when
you try to incorporate interventions that have
been developed in tightly controlled settings, they may
lead to unexpected results because context wasn't
taken into consideration. So conceptual models and
frameworks that have been used and that have been suggested
for informing research on multilevel interventions are
also used for dissemination and implementation research,
and those include like RE-AIM, like I mentioned.
However, I would like to point out that I think that there's
some fundamental differences between multilevel intervention
research and dissemination and implementation research.
While some of the issues, like adaptation and fidelity are
important in both, I think that when we are considering making
changes at particular levels in multilevel interventions,
and if you think about organizational levels,
providers, that those behaviors and activities may be quite
different than those behaviors or activities or processes for
adopting and implementing interventions.
So this is a much longer discussion,
but I do think that there are some fundamental differences,
although there is overlap. Finally, as a behavioral
scientist that is sort of exploring these different
models and areas as we consider multilevel interventions,
I think that - I may disagree a little bit with Arnie about
forget everything you learned before.
Because when I think that when you look at these levels,
you realize that yes, there's policy levels and there is
organizational and community, but fundamentally there's
people in these levels and there's people that need to
make specific changes in order to change their environments.
Even policymakers, even directors of clinics,
they have to make specific decisions,
specific changes in their own behaviors in order to make
those changes in the environment. So I think to
address the complexity of multilevel influences, we have
to develop intervention approaches that ultimately
are changing the behavior of people in the environment.
I am not going to read these questions.
We'll put them up, again, afterwards. Is that okay?
Some of them you have already heard and now it's
time to talk amongst yourselves.
>>>[APPLAUSE]