Play the Health Policy (Simulation) Game

By Alina Hsu

If payers cut reimbursement rates, what is the effect on healthcare cost over time? On morbidity and mortality?

How about if we improve quality but make no other changes to the system?

Or if we simply move to universal coverage?

What if we try combinations of interventions?

HealthBound, a simulation game available on the CDC website, is based on a system dynamics model of causal relationships between components of the US healthcare system, including outcome measures. Models are simplified and abstracted representations of complex realities. For complex systems, especially considering unintended consequences and time lags between interventions and results, it can be difficult or impossible to intuit what the results will be. The only ways we can learn about the results are by implementing changes in the actual systems, or by running simulations. When the risks or costs of making changes in the actual systems are very high, we can experiment with various options and use what we learn to guide policy. Interventions in the real world also don’t permit us to go back and try something different. HealthBound also provides a neutral, consistent framework for considering alternative approaches to healthcare reform.

The model builders, Bobby Milstein, Jack Homer, and Gary Hirsch, consider this a work in progress, and are using it to engage a widening circle of stakeholders in learning and discussion.

The interface is easy to use, so it’s quick to get started. The learning is about healthcare system dynamics, not about how to use the interface.

Here are their answers to the first two questions above, from pages 4-5 of their model summary:

Cut reimbursement rates: Reducing reimbursements to office-based providers offers the promise of lowering health care costs. But it also has the tendency to diminish quality of care. The reduction in quality quickly leads to greater morbidity and mortality, particularly among those with chronic diseases. The game suggests that the increase in morbidity minimizes the net cost reduction from the outset, making it much smaller than one might anticipate. Also, the loss of income to providers causes their numbers to dwindle over time, leading to further worsening of morbidity and mortality. As a result, any initial reduction in costs is ultimately negated. (This will not come as a surprise to the Lean folks out there!)

Improve quality: The game can simulate the effects of greater adherence to guidelines for effective chronic and preventive care by office-based primary care physicians and specialists. That improvement in quality results in a quick reduction in morbidity and mortality. In fact, the reduction in morbidity is greater than in the universal coverage scenario, because the quality improvement applies to a much larger segment of the population. Despite the reduction in morbidity, improved quality also leads to a significant increase in health care costs, due to increased physician visits and use of medications (again, disease management is not cost saving). Also, quality improvement worsens health inequity. Quality improvement requires that providers give more routine care per patient, and this leads to some shortage of primary care providers for both the advantaged and the disadvantaged. In response, many of the advantaged are able to shift to specialists and maintain the same improved quality of care, whereas the disadvantaged cannot afford to do so. As a result, a greater proportion of the advantaged than then disadvantaged end up benefiting from the improvement in quality.

The model interface lets you explore the impact on any component of the interventions you have chosen. A recent survey of ED administrators by the Schumacher group shows that 66% expect that reforms will cause ED volume to increase, and 64% expect that they will see more patients who cannot access primary care physicians. Both ED volume and primary care physician availability are included in the model, so those of you who are concerned about the impact of reform on the ED can experiment with the model.

Let me (and the model builders) know what you think of it!