Multilevel Interventions in Health Care Conference: Presentation by Jeff Alexander, PhD

Multilevel Interventions in Health Care Conference: Presentation by Jeff Alexander, PhD


>>>DR. STEVEN CLAUSER: Our next
speaker will be Jeff Alexander. He is Richard Carl Jelinek
Professor of Health Management and Policy at the School of
Public Health at the University of Michigan and holds positions
in the university’s Ross School of Business and
Institute for Social Research. His teaching and
research interests focus on organizational change in
the healthcare sector, multi-institutional systems,
governance and physician participation in institutional
management and policymaking. Dr. Alexander,
please take the podium.>>>[APPLAUSE]>>>DR. JEFF ALEXANDER: Thank
you, Steve. This was really an interesting paper to
write in several respects. So if you look at
the notion of time, on the surface it seems like a
fairly circumscribed and fairly narrowly, fairly
narrow in its coverage. But in fact, as we try to
illustrate in the paper, it has broad implications for how
multilevel interventions are conceptualized, how program
designs may be structured or designed, how research on
multilevel interventions is done and even how statistical
analyses and measurement is performed. So a fairly
circumscribed concept with broad implications. I guess the
sub-context for this slide would be things change and I
think that’s what the paper is essentially about. I don’t
care what level you are talking about, whether it is
the individual level, the organizational level
or the environmental level, things do not remain
in a steady state. The paper really addresses how
these changes over time might affect the impact of a
multilevel intervention at multiple levels and particularly
at the individual level. In some respects, time might be
viewed as the third dimension. It sounds like
the Twilight Zone, but the third dimension of
a multilevel intervention, the first two being what
happens at each level, the second being what happens
between levels or across levels and then the third, the third
dimension being time itself. So if we look at the vertical
relationships in MLIs and how they are affected
differently by change, we can come up with a couple
of illustrative examples. So for example, a particular
MLI might have differential effects on individuals,
depending on where that individual was in their disease
life course or where they were starting in the intervention.
So a take away here is that time can help us think about
when an intervention might be more or less effective,
how it should be designed to target particular populations
or sub-populations. And the same thing goes if you
are looking at higher levels, if you are looking at the
organizational level or environmental level. So for
example, very few interventions I think occur outside a
specific social context, the social context have a
way of changing over time. So for example, at the
organizational level, you might argue that depending
on changes in organizational priorities, resources,
availability of technical support, etc., etc., or simply
entropy, the interventions and their effects on individual
patient outcomes might differ through nothing to do
with the intervention itself, but the social context under
which the intervention operates. So in my simple
way of thinking about this, and this is for those of you
who are more kind of visually oriented, this is my schematic
of maybe a typical MLI. You have three
levels – environment, organization and individual
and then the second dimension illustrated here is obviously
the relationship between and across levels. And when
you add time, things get complicated in a hurry because
in the paper, as I argue, particularly the cross
level effects tend to take on a different shape and a different
meaning than if you were just to look at an MLI in a
cross sectional fashion. And I am kind of telegraphing
my big question at the end. But the take away here in this
slide is that particularly you are looking at the cross level
effects, I would submit that they will be very different
than if you were just to look at this in a static way.
So where does time fit in multilevel intervention,
both in terms of evaluation and program design? I think
it’s important for us to think about how time affects
patients and time is often viewed as a level, if you
will, or a dimension that is embedded within
individual patients. As the example I gave
previously suggests, it could operate at the
organizational level as priorities change
in the organization, the development of
that organization and the intervention itself might be
effective – affected, sorry. And at the environmental level,
similar changes could occur. Importantly again, what is
problematic to me is how these changes that occur over time
affect the cross level nature of the relationship between the
intervention and the target of those interventions. And the
way we approached this in the paper was dealing, it was by
dealing with four distinct topics, the first being the role
of disease life force and drug trajectories in
multilevel interventions. The second being a discussion
of approaches to incorporating time and research and
program design for multilevel interventions. The third
being analysis of time using multilevel data. Here we talk
about specific analytic techniques. And then finally,
and this is sort of a word of caution, the resource
considerations that have to be taken into consideration if you
are to expand from thinking about MLIs in a two dimensional
fashion to a three dimensional fashion. And my big question,
and I only have one, but it is a big one and what I would
like to hear from the various groups is something that I
have been struggling with. Is if you shift your
thinking to three dimensions, that is if you incorporate
time in your thinking about multilevel intervention, how
does that affect the way you think about how different
levels affect processes or outcomes at other levels? And
I think this is important not only for conceptual reasons,
but also for practical reasons. We were discussing at the
table this morning how MLI research is going to be
considerably more expensive in all probability than the
research we are used to. And we really do need to
think carefully and clearly about the theoretical
underpinnings of the relationship across dimensions.
And again, I think time only adds to that complexity.
So I will stop there and I will look forward to talking
about this in the small groups.>>>[APPLAUSE]

Leave a Reply

Your email address will not be published. Required fields are marked *