The Evidence Base

Informing Policy in Health, Economics & Well-Being
A collaboration with
USC Dornsife Center for economic and social research

Predicting the Economic Impact of Changes to Population Health

So much of health care policy is decided by the science of the educated guess, with “what-if” forecasts at the heart of any analysis.

How much will the nation be spending on health care in two decades if the Affordable Care Act goes untouched? What if the ACA is scrapped entirely?

If, by a miracle, diabetes were cured tomorrow, what would happen to life expectancy in the United States?

The Roybal Center for Health Policy Simulation, housed within the Schaeffer Center, delves into these major health care questions and the consequent implications upon the nation’s health and economics.

To that end, Roybal Center researchers have developed two simulations: the Future Elderly Model and Future Americans Model (now Future Adult Model). The FEM represents Americans ages 51 years and older, the FAM pulls from a younger demographic of 25 years of age and older. Using the FEM and FAM, researchers can simulate a nation improving its collective health by ensuring everyone has insurance or eliminating use of tobacco. Or perhaps the hypothetical country is pursuing incremental change, such as developing treatments that are slight improvements on those now available.

The Future Americans

Predicting characteristics of a population in the future is based in part on characteristics present today.

“In one sense, we are just continuing the trend lines,” said the Roybal Center’s Director of Simulation and Data, Bryan Tysinger. “Our simulations are based on data sets that are longitudinal panel surveys.”

“But you couldn’t just naively continue those survey responses into the future,” he said.  At the core of the models, Tysinger continued, is a “complicated machine” revising the projected population for every year in the future to be simulated. If a FEM forecast is set in 2036, that’s 20 years of people turning 51 or dying – and sometimes both. And all the while, the models adjusts accordingly the populace’s characteristics in matters of health (e.g., body-mass index, smoking rates) and otherwise, such as education level.

After the models are calibrated to the experts’ satisfaction, the simulations they run produces invaluable research when planning for an uncertain future in fields ranging from nursing-home care to FDA policies to medical breakthroughs in Alzheimer’s disease.

The first paper to emerge from the models was published in 2003. In the 13 years since then, over 30 peer-reviewed publications followed, including two special issues in Health Affairs, the scientific journal dubbed “the Bible of health policy” by the Washington Post.

Predictions of the Future Allow for Policy Changes Now

Recently, the Roybal Center supported a study by the National Academies of Sciences, Engineering, and Medicine on the gap in life expectancy between the highest and lowest earners in the U.S. Not only is the gap growing over time, exacerbating the inequity in life expectancy between the highest and lowest earners is the subsequent trend of high earners disproportionately receiving larger lifetime benefits from government programs such as Social Security and Medicare.

Etienne Gaudette, the Roybal Center’s Policy Director, said it’s almost the norm for the models to deliver shocking results.

Among the biggest surprises he recalled was a forecast stating that medical innovations that lower disease incidence and improve quality of life actually would, over the long term, increase health-care costs nationally.

“It really makes sense, because if you save people from heart disease they get to live longer and incur medical costs for more years, not to mention that they may end up with another costly disease like Alzheimer’s or cancer three years later.” Gaudette said.

It was through these types of projections that Dana Goldman, Principal Investigator at the Roybal Center, came to the conclusion America would be better served not by science trying to eliminate individual medical conditions but by attempting to delay the body’s aging process as long as possible.

Modeling Local-Level Populations

The models also can go local, using data provided by the Los Angeles County Department of Public Health. Among the topics studied have been the health benefits of encouraging sodium reduction among LA County consumers.

With the models’ ability to look backward as well as forward, Tysinger thought it would be fascinating to ask: What would’ve been the health consequences if, a quarter century ago, the decision

was never made to clean up the air? It was in 1991 when the state forbade the sale of leaded gasoline.

“That’s a natural,” Tysinger said. “You could look at asthma rates or chronic lung disease rates or whatever, and say, ‘this is what we think it might have been; what would that have cost?’”

Several groups of researchers are currently working to expand the FEM to include other countries so predictive simulations of population change may be shared and compared across borders.