Modeling Long Run Population Health Costs and Outcomes

By David A. Kindig, MD, PhD

The broad population health model that underpins this blog and the County Health Rankings must have long-term relevance because quick fixes are few and far between. There is an imperative to invest now in factors we know to be strong drivers of long-term health, such as early childhood interventions (see Can We Afford to Wait for Better Evidence on Improving Child Health? and Business Investment in Early Childhood: Making Future Workers Happier). There is also a need for more research which can better estimate such long-term impacts; the IOM’s recent report on Public Health Measurement and Accountability calls for advancing the use of predictive and system-based simulation models to understand the health consequences of the underlying determinants of health.

One such model does exist, and was reported on in the May 2011 Health Affairs issue on Environmental Health discussed last week. The article, authored by Bobby Milstein, Jack Homer and colleagues, focuses on the HealthBound policy simulation model which has been developed by the CDC over the past decade. This work seems and is ambitious, as any such analytic and projection tool has to be, but it also simplifies the sheer complexity of the U.S. health system into a tractable form that can be understood and studied by diverse stakeholders. It is built on the methodology of systems dynamics modeling, using several hundred interacting elements and differential equations tied to ten national data bases and many key reference studies.

There have been various iterations of the model over time and several published reports, including an earlier overview in the American Journal of Public Health.  The most recent Health Affairs article reported the results from simulating three strategies over 10 and 25 years to reduce deaths and improve the cost effectiveness of interventions:

1. expanding health insurance coverage
2. delivering better preventive and chronic care, and
3. protecting health by enabling healthier behavior and improving environmental conditions.

The main finding was that each would alone would save lives and provide economic value, but the combination of all three was likely to be more effective. For example, adding protection to the coverage and care scenarios would save 90% more lives and reduce costs by 30% in year 10, but by year 25 the protection investment could save 140% more lives and reduce costs by 62%.

Of course, what goes into the model determines what comes out, and it is difficult for someone not intimately familiar with the inputs and equations to evaluate strengths and weaknesses. Because of the complexity of all the multiple inputs and outcomes, precise estimates for every independent interaction over time do not exist, and ranges and sensitivity analyses are usually required.  The authors have extensive experience with these cutting-edge methods; however they caution that “better data…would help narrow uncertainties and yield even stronger policy insights.” While the Health Affairs article placed more emphasis on health care elements with some behaviors and physical environmental elements in the protection scenarios, planners with more interest in these could simulate the “Pathways to Advantage” intervention, which summarizes intervention research around things like education, living wage, and job training.

While causal understanding of how such factors operate is perhaps more incomplete than those in the health care realm, the authors remain open to incorporating new findings as they emerge. They are also very interested in creating practical and relevant applications for policy makers, which is in part why a HealthBound game also exists to gain hands on familiarity with the methods and results.

The HealthBound tool is are not sufficiently precise to identify local opportunities to invest in specific programs and policies; however, other groups are moving the field in that direction. For example, the ReThink Health Initiative has begun to create simulation models and games that focus specifically on regional investments to transform health system performance. Policy makers need and want such guidance and these tools offer tremendous potential for population health policy – especially as they become increasingly robust and reliable over the coming decade.

For further reading, check out the Robert Wood Johnson Foundation’s interview with Bobby Milstein about his article on NewPublicHealth.org.

David A. Kindig, MD, PhD is Emeritus Professor of Population Health Sciences and Emeritus Vice-Chancellor for Health Sciences at the University of Wisconsin School of Medicine and Public Health.