By Kosali Simon
Kosali: Jason, we hear a lot these days about genetics-based health research—can you define this field and tell us how social science/economics contributes to it?
Jason: While medical genetics involves research on what causes genetic disorders, how genetic conditions are passed on through generations, and how those conditions are diagnosed and managed, social science research in the genetics space examines how human behaviors and societal forces interact with biological factors to affect health and related outcomes.
Let’s take an example of genetics data called polygenic scores: our individual DNA carries information that can be summarized by a polygenic score that aims to predict our risk of a trait (e.g. a certain type of cancer) based just on our genetic information, not our environmental factors. One way that genetics data can be applied in economics research is to use polygenic scores as control variables, or as IVs (eg studying the causal impact of a cancer diagnosis on life expectancy)—of course, these measures run into challenges that all IVs face, such as knowing whether these genetic scores ONLY affect the trait in question, or could affect the outcomes through other pathways. These scores could also be an interaction when looking at the heterogeneous effects of a treatment on individuals according to dimensions related to genetics.
Kosali: You have published many papers related to genetics in health economics. Can you tell us how your interest in this area developed?
Jason: I was a graduate student at the University of Wisconsin when John Mullahy introduced us to scholars in the RWJF Health and Society Scholars program. We brainstormed how social scientists might use what was emerging as genetic data in studies such as Add Health. At the time, data was available on fewer than 10 genetic locations; today, we can use data on many millions of genetic locations. My first job was in a School of Public Health, and there, use of data on biologic measurements and explanations, and genetics, was “in the air” so it led naturally to my thinking of how to integrate health economics and genetics; I then became a RWJF Health and Society Scholar at Columbia University, working with faculty (Peter Bearman, Bruce Link, Julien Teitler) who encouraged me to continue thinking about this integration.
Kosali: How has the way you viewed the integration of health economics and genetics changed over time?
Jason: 15 years ago, there appeared to be a much larger role for specific genetic variants that we could identify—for example, we thought there was a “smoking gene” that played a large role in nicotine dependence. Now, we believe such links to be essentially completely untrue, except for a few medical diagnoses where single genes can be very strongly linked. For any outcome, there are thousands to hundreds of thousands of tiny genetic influences, so it has become a bit harder to think of how economics can contribute to genetics, but, we can still think of how genetics contributes to economics. For example, genetics will continue to contribute to economics to be able to measure aspects of humans in ways difficult to do so through surveys or administrative data. When we use genetic information (such as using the fact that identical twins are genetically more similar than others), we are trying to use genetics to help us control for levels of individual endowments we otherwise cannot (e.g., in a model of health capital formation), since our genetic features are the same across our lifetime. Societal questions about equity and efficiency tradeoffs from the use of genetics data is also an area where economists can contribute. Because large sample sizes are needed to establish relationships with statistical power, sadly, most research exists on majority populations in industrialized nations, who tend to be white. Expanding sample sizes of non-white populations is a high priority in this research. I also see continuing usefulness of contributing to cost effectiveness and calculating value of innovations and information.
Kosali: Can you point us to some trainings where economists can increase their knowledge of genetics research?
Jason: The Russell Sage Foundation has a good free-access online workshop with videos, readings lists and problem sets. University of Michigan also offers summer workshops for social scientists wanting to integrate more biology into their research. The mechanics of using the data are more straightforward than they used to be. You can just dive into the Health and Retirement Study and other studies, use the polygenic scores (which are deemed publicly available variables now), see how much they correlate with other measures, that will give you a good start, and look for more biology and genetics trained collaborators for further exploration.
Kosali: Further readings? What are outlets that have tended to publish econ/genetics research?
Jason: General science (Science, Nature, PNAS) tends to publish genetic discoveries. But downstream use of genetic data to answer social science questions are emerging, and are being published in health economics journals. Researchers look at differential treatment effects by genetics, or in the form of IVs. Using genetics measures also broadens your scope of journals more into public health and medicine, if you are so inclined. Economics has contributed to the development of methods (causal inference) using genetic data, even though genetics by nature is non-experimental data, and because data sets such as tissue banks are non-representative (often only include the diseased population). In a recent paper, a coauthor and I use polygenic scores as a predictor of educational attainment, even within siblings (who have arguably similar environmental factors). This also helps us understand if parents and others reinforce or compensate for baseline differences in endowments—this is an example of a question that uses genetics in a novel way that is more of interest to social science than to health sciences. Genetics is also involved in peer influence research. Economists may be able to squeeze more insight when using genetics data in their research than otherwise. A few other examples are linked.