Viewing Health as a System
Posted March 07, 2014 by Charlie Homer, MD, MPH
More than a decade ago, the Institute of Medicine declared that the purpose of the US healthcare system was to continuously improve the health of the American people. Yet, for a long time the focus of many remained narrowly within the constraints of healthcare—addressing themes such as patient safety, clinical effectiveness and patient centeredness that are critically important but of themselves not likely to change the overall health and well being of the population.
Gradually, and now with increasing force and pace, a movement is building that seeks to recognize the broader influences on health, often summarized as “social determinants of health.” For example, The Robert Wood Johnson Foundation is reframing its strategic focus to emphasize a “culture of health” and the National Quality Forum has several working groups on both measuring population health and moving the nation to health. Hooray!
NICHQ of course has long had a broad focus on health, strongly driven by our work on addressing childhood obesity. In this work we recognized early on that clinical care is an extremely important element AND that addressing childhood obesity required coordination and integration between the clinic and community to change the context and provide effective services. I suspect that addressing the obesity epidemic is part of what has driven not just NICHQ but the broader health and healthcare community to see the need to bridge healthcare and health in a new way.
What Is the System that Produces Health?
Improvement science teaches us to view outcomes—such as health—as the inevitable product of a system, with the implication that achieving improved outcomes requires changing the system itself. A deep understanding of the system and how it functions can enable smarter decisions about selecting high leverage changes in order to improve system performance.
In this case, what is the system that produces the health of a population? How might we describe it and so choose promising points of intervention?
The most common framing that I see is that originally developed by McGinnis and modified by Kindig in creating county health rankings. In this model, the influences on health are broken down into a few simple categories—you might even call them “drivers.” The relative impact of these drivers are then estimated, with healthcare merely 10-20 percent, health behaviors a much higher 30-40 percent, and other factors in between.
All models are simplifications of reality; good models are useful in enabling understanding and driving action. The Kindig model is useful in broadening focus and, presumably, investment from health care to community or, more thoughtfully, assuring that community actions—from highway construction to food pricing—include health considerations if we want to alter societal health.
Yet as a system thinker and a pediatrician inherently oriented to think longitudinally and developmentally, I find the model inadequate. At least in its typical graphic representation, the model fails to emphasize the interaction and interdependence of the factors that when attached to percentages seem independent. Social factors clearly influence the physical environment in which one lives (how many bus depots are adjacent to luxury housing?). Similarly health behaviors—such as healthy eating—are strongly influenced by economic factors and, under ideal conditions, at least marginally influenced by high quality health care.
Envisioning a More Complete Model
So how can we improve the model? I’m early in the process of thinking about this, and thought I’d share my thinking and get some crowd-source reactions and feedback at this early stage.
I initially started to add elements. For example, if we put “personal resilience” in as a driver, we can start to see how supportive relationships can drive better health. If we add a driver for “earlier health status” we can begin to recognize the longitudinal nature of health. Here’s a snapshot of my white board brainstorm:
But, ultimately the linear driver framework seems to be an insufficient illustration of the system to truly help set priorities for action. This type of model doesn’t emphasize the interactions among the drivers, nor does it truly address the importance of timing and trajectory (there’s that developmental thinking again!).
Seeking a Systems Model
System diagrams and system modeling may be a more effective approach to framing the complexity of the influences on health, and especially to incorporate the critical role of development and what system modelers might call lagged effects, i.e., the effect of an intervention at one point in time on outcomes at a much later point in time. System modeling has been applied to community health, but the models I have seen don’t adequately account for the later or long-term effects of interventions at earlier points in time, particularly the protective effects of interventions at critical points in development as well as the cumulative effects over time.
Here’s my first rough attempt at a system diagram for health outcomes, presented for others to comment on and improve:
Cleaned up, it looks like this, more legible but still a draft:
What I’ve represented as the outcome is “health now.” A key influence of health now is health at an earlier stage, with this earlier stage health influenced by numerous drivers—many of which are not dramatically different than those in the McGinnis/Kindig model. The distinction is the emphasis on the interaction of these drivers and of the critical impact of health at an earlier stage on later health—in either a virtuous or vicious cycle.
Even this graphic model doesn’t adequately emphasize the particular importance of influences on health at particular times—such as infancy and late adolescence/early adulthood. Yet it does start to elevate the importance of interventions to improve health at an early stage in life—interventions both through health care (e.g., for those at great biologic vulnerability such as extreme prematurity) AND through enhancing the economic and social conditions and capabilities of parents and community.
Getting this right isn’t just an academic exercise. Without attention to time, policy makers may focus all of their efforts on behaviors and conditions at a late (adult) stage and fail to achieve the desired health impact that earlier childhood interventions might have. Similarly, without attention to the interaction among these factors—such as the impact of environmental exposures on epigenetics or the potential for healthcare to influence health behaviors—the potential benefits of some interventions or potential harms of exposures will be vastly underestimated.
I look forward to your help in improving the model and truly focusing all of our efforts on improving health.