next session has really kind of transitioned us into really
talking about getting into some of the real applications, on the
challenges and opportunities, as it is called, for multi-level
intervention research. And in particular, we are
going to be looking at practical ways of trying to
figure out do we select interventions, what are
some of the conceptual and theoretical issues about
timing and other things and thinking about appropriate
interventions. And then talk about some of the
research methods that are very appropriate for thinking about
getting into this field in a way to really address some of
the extraordinarily challenging conceptual questions that we
talked about this morning. So my first speaker is
going to be Brian Weiner, he is professor of the
Department of Health Policy and Management at the School
of Public Health at the University of North Carolina
Chapel Hill. Dr. Weiner is a good friend and he’s also
Director of the Program of Healthcare Organizations at
the Cecil Sheps Center for Health Services Research. His
research focuses on such things as adoption, implementation and
sustainability of innovations, very appropriate for this
meeting, in healthcare organizations and he earned
his PhD in organizational psychology from the
University of Michigan.>>>DR. BRIAN WEINER: So I
would like to begin with a Sufi teaching story that
goes something like this. You think that
because you understand one, that you also understand two
because one and one make two. But you forget that you
must also understand “and.” And I think that actually
does a nice job setting up this particular talk and the paper
that we have wrote about the search for synergy. And in
the interest of disclosure, we are going to shift in this
particular ten minutes to a different model of thinking
about multi-level interventions from the biopsychosocial model
that Steve Taplin represented to the social ecological model,
which was one of the three that he mentioned. So if you
feel confused, that’s probably because this is a good example
of where different people using the word multi-level have
different ideas about what that means. So I just want
to alert you to that. The social ecological
perspective provides a very compelling justification
for multi-level intervention. While there are numerous
variants of the social ecological perspective,
they share three – excuse me, two basic principles. The
first is that human health results from the complex
interaction of personal factors and multiple aspects of social
and physical environments. The second principle is that
these multiple factors that influence health
are interdependent, that is they mutually
influence each other. On the basis of
these two principles, proponents of the social
ecological perspective contend that multi-level interventions
should be more effective than single level interventions.
The key to designing an effective multi-level
intervention, however, they note, is to select and combine
interventions that work in complimentary or synergistic
ways. All good advice. Conspicuously absent, however,
are discussions of how, when or why interventions at
different levels of influence work together in
mutually reinforcing ways. The problem is that
without such guidance, multi-level intervention
designers run the risk of combining interventions
that produce scattered, redundant or even potentially
contradictory effects. So using a causal
modeling framework, we described in the paper,
and I will briefly touch on it during this ten
minute presentation, five strategies for increasing
the potential complimentarity or synergy among interventions
that operate at different levels of influence. Given the
importance of interdependence in the social ecological
framework, we focus on two types of causal relationships,
mediation and moderation to illustrate the potential
strategies for increasing synergy or complimentarity. We,
in the paper, and as you’ll see in the slides that follow,
focused on multi-level interventions designed to
improve the quality of treatment for locally
advanced rectal cancer. In the accumulation strategy,
interventions at different levels produce a cumulative
impact on a common mediating process or pathway. Note here
that the effect of each intervention in this particular
strategy is not conditional on the other interventions.
Rather, the interventions exhibit what scholars and
organization science called pooled interdependence, meaning
that the intervention, each intervention makes a
discrete contribution to the outcome through the mediating
variable or pathway without being dependent
upon each other. In the amplification
strategy, by contrast, the effect of one or more
interventions is conditional on another intervention. One
intervention increases the target audiences sensitivity
to or receptivity to the other interventions. That is,
one intervention amplifies the effect of other interventions
on the mediating process or pathway. And so you can
see in this example the use of reimbursement is used
to increase the sensitivity to or receptivity to changes
in public reporting or interventions that are designed
for public reporting or an opinion leader intervention is
trying to change social norms around the provision of chemo
radiation therapy for patients with locally
advanced rectal cancer. For those of you who are
curious, this particular strategy exemplifies a form
of mediated moderation. In the facilitation strategy,
the effect of one or more interventions is, again,
conditional on another intervention. However, instead
of boosting the signal, the conditional intervention,
in this case the clinical reminder, clears the
mediating pathway for the other interventions to
produce the desired outcome. In other words, one intervention
removes the barriers or facilitates the effect of
the other interventions. This strategy is another
form of mediated moderation. In the cascade strategy, an
intervention at one level affects the desired outcome
in and through one or more interventions at other
levels of influence. The interventions demonstrate
what scholars refer to as sequential interdependence
meaning that the outputs of an intervention at one level
become the inputs of an intervention at another level.
By linking multiple mediating processes into an
integrated causal pathway, cascading interventions
create a circuit through which the effects of
interventions combine and flow. And I must say, I have a little
bit of concern about calling this the cascade strategy
because it implies a trickle down effect, but in fact it is
quite possible for the effect to trickle up. Many of who
have worked, for example in the community care network
study or in community empowerment studies or in
community based research know that sometimes in order for
interventions at lower levels of influence or smaller units
of human organization to be effective, change has to
occur at a higher level of influence or a higher level of,
a higher unit of analysis. And then finally, in
the convergence strategy, interventions at different
levels mutually reinforce one another by altering the
patterns of interaction among two or more target audiences.
The interventions in this particular strategy exhibit
what scholars call reciprocal interdependence meaning that the
outputs of some interventions become the inputs of other
interventions and vice versa, so that’s the difference
between the sequential and the reciprocal forms
of interdependence. They are not necessarily
linked in a chain, but rather there is this constant back
and forth. I Know that is just a gloss of what we
talk about in the paper and I certainly hope that you will
be intrigued enough to read it. I think what we have tried to
do in this paper is to develop a general framework to guide
people’s thinking about the causal logic for multi-level
interventions. And this can be useful not only for intervention
designers, but also for reviewers of multi-level
intervention proposals. I think what we wish to do
is to avoid a kitchen sink approach, which can be very
wasteful and expensive and perhaps burn through a lot of
goodwill among stakeholders and communities, healthcare
delivery systems and the patients and families. Theory
and research clearly play a critical role in clarifying
the causal logic for combining interventions at multiple
levels and this raises several questions for discussion. The
first is, do we have theories that explain how determinants
at multiple levels of influence produce health
or other outcomes? I might be over-exaggerating,
but it seems to me that we have psychological theories that
look at intra-personal factors, we have organizational theories
that look at organizational factors and we have political
theories that look at political factors, but they don’t
necessarily look at how factors at multiple levels of influence
interact in order to produce health and other outcomes.
The second question is do we have enough cross level
research that examines the interdependence of variables
or determinants at multiple levels of influence?
What I think we need more of is sort of
cross level research. So if we are going to stick
with the innova framework and think about multi-level
modeling, we need to not only focus on which
level explains the most variants or which factors at
any given level explain the variants at that level, but we
really do need to be looking at cross level effects in order to
understand the interaction of causal factors at
different levels of influence. And then finally, do we have a
sufficient grasp of the causal mechanisms through which
many commonly employed interventions produce
their effects? Now I have been struck by
several papers that have been published in Implementation
Science and other journals that have looked at the intervention
of audit and feedback which, as many of you know, has been
studied in hundreds, or at least tens, if not hundreds
of times in various places. But it’s only really in the
last two or three years that we have begun to ask the
question how or why does audit and feedback work? In
other words, what is the theory about the causal mechanism
through which this works? And some very interesting
work is now occurring; it is just surprising to me that
we had to do forty or fifty studies before somebody finally
wrote a paper about that. So those are the three
discussion questions that I would like to encourage the
individuals at the tables to take up when we have our
discussion time. Thank you.>>>[APPLAUSE]