Periodic Reporting for period 2 - DONNI (Developmental Origins: exploring the Nature-Nurture Interplay)
Okres sprawozdawczy: 2022-03-01 do 2023-08-31
The project’s overall objective is to improve our understanding of how individuals’ early life environments interact with one’s genetic predisposition to causally shape individuals’ health and human capital outcomes in later life. That is, we are no longer viewing one’s life chances as a function of either nature or nurture. Instead, the project quantifies how the two causally interact and jointly shape individuals’ later life outcomes. This allows us to identify the long-term effects of short-term variations in early-life conditions and evaluate the extent to which such early-life environments exacerbate or reduce arguably unfair genetic inequalities in the population. As such, an important part of the project is the introduction of molecular genetic data within the economics and social sciences, and hence the project includes some more methodological papers exploring the best ways to do this. Another important part of the project involves the digitization of historical data sources and we combine these with existing individual-level datasets to characterise individuals’ early-life environment. An improved understanding of the role of gene-environment interplay in shaping individuals’ outcomes is crucial in today’s society, especially when designing current (government) policy and evaluating their impacts.
The project has made substantial progress since the start, with four papers published in high-ranking interdisciplinary journals, one book chapter, as well as ten research papers under review. In addition, we have contributed to a government report about the use of molecular genetic information in areas beyond health and health care and written five policy pieces about the role of the nature-nurture interplay in shaping individuals’ outcomes. The project has also made substantial progress on an exciting new data source, digitizing large amounts of historical local area-level data for England and Wales over a 40-year period.
More specifically, we have created software that has already allowed us to digitize over 40,000 pages of historical local area-level data from the 1940s to the 1970s. These data can in turn be merged with existing data at the individual and/or regional level. For example, we have shown the extent to which two UK measles vaccination campaigns introduced in the 1960s reduced measles rates across the country, how this reduction had some long-term impacts on the health outcomes of those exposed to the vaccination campaigns, and how these impacts partially depend on individuals’ genetic variation. Similarly, we have documented the long-term cardiovascular impacts of more general adverse early-life environments, captured by one’s infant mortality rate in the year and local area of birth. Our results are consistent with the original Barker hypothesis, but we show considerable genetic heterogeneity in this relationship: in areas with the lowest infant mortality rates, the effect of one’s genetic predisposition effectively vanishes.
Our research has also shown how the early life nutritional environment is crucial in shaping individuals’ economic and health outcomes in later life, and how it interacts with individuals’ genetic ‘predisposition’ for the economic and health outcomes. Our work highlights the long-term impacts of prenatal exposure to sugar confectionery on one’s later life health and educational outcomes. Our results are consistent with the developmental origins hypothesis, suggesting it is advantageous to be exposed to a prenatal environment that is more aligned with the postnatal environment.
Furthermore, we have quantified the adverse impacts of prenatal and early childhood exposure to severe pollution for individuals’ later life economic and health outcomes. We show that children who were exposed to excessive pollution in early life have substantially lower intelligence scores and worse respiratory health, with some evidence of a reduction in years of schooling. Furthermore, we show that these effects are stronger for those genetically ‘predisposed’ to these outcomes.
We have also shown the importance of parental investments in child development, and how this is moderated by individuals’ genetic ‘predisposition’. Following the literature, we use an individual’s birth order as a proxy for parental investments and interact that with one’s polygenic score for educational attainment. Using a sample of full siblings, we show that the additional parental investments associated with being firstborn are more effective for those who randomly inherited higher genetic endowments for education.
We have also explored the use of molecular genetic data from a more methodological perspective. We have highlighted two implications of the fact that polygenic scores are measured with error. First, it leads to bias in the effect estimates in regression models; second, it affects individuals’ ranking in the polygenic score distribution which can impact on clinical decision-making if treatment decisions are made (partially) based on individuals’ genetic data (i.e. personalised interventions). We show the importance of both empirically as well as via simulations and offer a solution to the issue of attenuation bias.
More specifically, we have created software that has already allowed us to digitize over 40,000 pages of historical local area-level data from the 1940s to the 1970s. These data can in turn be merged with existing data at the individual and/or regional level. For example, we have shown the extent to which two UK measles vaccination campaigns introduced in the 1960s reduced measles rates across the country, how this reduction had some long-term impacts on the health outcomes of those exposed to the vaccination campaigns, and how these impacts partially depend on individuals’ genetic variation. Similarly, we have documented the long-term cardiovascular impacts of more general adverse early-life environments, captured by one’s infant mortality rate in the year and local area of birth. Our results are consistent with the original Barker hypothesis, but we show considerable genetic heterogeneity in this relationship: in areas with the lowest infant mortality rates, the effect of one’s genetic predisposition effectively vanishes.
Our research has also shown how the early life nutritional environment is crucial in shaping individuals’ economic and health outcomes in later life, and how it interacts with individuals’ genetic ‘predisposition’ for the economic and health outcomes. Our work highlights the long-term impacts of prenatal exposure to sugar confectionery on one’s later life health and educational outcomes. Our results are consistent with the developmental origins hypothesis, suggesting it is advantageous to be exposed to a prenatal environment that is more aligned with the postnatal environment.
Furthermore, we have quantified the adverse impacts of prenatal and early childhood exposure to severe pollution for individuals’ later life economic and health outcomes. We show that children who were exposed to excessive pollution in early life have substantially lower intelligence scores and worse respiratory health, with some evidence of a reduction in years of schooling. Furthermore, we show that these effects are stronger for those genetically ‘predisposed’ to these outcomes.
We have also shown the importance of parental investments in child development, and how this is moderated by individuals’ genetic ‘predisposition’. Following the literature, we use an individual’s birth order as a proxy for parental investments and interact that with one’s polygenic score for educational attainment. Using a sample of full siblings, we show that the additional parental investments associated with being firstborn are more effective for those who randomly inherited higher genetic endowments for education.
We have also explored the use of molecular genetic data from a more methodological perspective. We have highlighted two implications of the fact that polygenic scores are measured with error. First, it leads to bias in the effect estimates in regression models; second, it affects individuals’ ranking in the polygenic score distribution which can impact on clinical decision-making if treatment decisions are made (partially) based on individuals’ genetic data (i.e. personalised interventions). We show the importance of both empirically as well as via simulations and offer a solution to the issue of attenuation bias.
Our research is highlighting many ways in which molecular genetic data can be used in non-health settings, with a focus on applications within the economics and social sciences. We have advanced the state of the art of the economics and econometrics of gene-environment interplay by developing an economic model of decision making with genetic heterogeneity and using this to demonstrate how economic theory can guide empirical gene-environment interplay analyses and help in the interpretation of gene-environment findings. We have argued how incorporating genetic data into economic and social science analyses can shed new light on old questions, and be of interest even if one does not have a direct interest in genetics.
For example, we are among the first to empirically link a model of gene-environment interplay to economic theory of human capital production, allowing us to test the complementarity between endowments and private education investments and between endowments and public health investments. We have emphasized the importance of family data when estimating direct genetic effects, we highlight the intricacies of interpreting an empirical model that seeks to estimate gene-environment interplay, and provide a systematic categorization of different types of gene-environment interplay analyses, discussing the direction and nature of any bias. Methodologically, we have highlighted various implications of the fact that polygenic scores (or polygenic indices) are measured with error, affecting not only the regression estimates, but potentially also impacting on clinical decision making. We have suggested a solution to the former.
Empirically, we have shown how ‘nature’ (i.e. genetic variation) and ‘nurture’ (environments) causally and jointly shape many of the individual-level outcomes that are of interest to economists and social scientists. We have demonstrated the interaction between genetic variation and multiple environments, including nutritional, toxicological, health, and home environments. This in turn provides a strong antidote against arguments of genetic or environmental determinism; the evidence of which we are disseminating widely to academic as well as non-academic audiences.
We expect a range of additional results by the end of the project, including new local-area level monthly data on unemployment rates immediately after the second world war in England and Wales, and more evidence on the importance of the early-life economic environment for later life health and education outcomes, as well as how this is moderated by one’s genetic ‘predisposition’. We are also doing research on the accuracy of location-reporting and highlight the implications of this for empirical research. We are describing historical trends and social gradients in maternal smoking over time and illustrate how these vary by individuals’ genetic predisposition, linking it to the many government and policy interventions, as well as changes in societal smoking norms. Regarding pollution, we are modelling local dispersion using specific polluting sources, and we are digitizing novel data on historical regional pollution levels. We link these to government policy that aimed to reduce pollution and estimate the long-term impacts on individuals in older age.
For example, we are among the first to empirically link a model of gene-environment interplay to economic theory of human capital production, allowing us to test the complementarity between endowments and private education investments and between endowments and public health investments. We have emphasized the importance of family data when estimating direct genetic effects, we highlight the intricacies of interpreting an empirical model that seeks to estimate gene-environment interplay, and provide a systematic categorization of different types of gene-environment interplay analyses, discussing the direction and nature of any bias. Methodologically, we have highlighted various implications of the fact that polygenic scores (or polygenic indices) are measured with error, affecting not only the regression estimates, but potentially also impacting on clinical decision making. We have suggested a solution to the former.
Empirically, we have shown how ‘nature’ (i.e. genetic variation) and ‘nurture’ (environments) causally and jointly shape many of the individual-level outcomes that are of interest to economists and social scientists. We have demonstrated the interaction between genetic variation and multiple environments, including nutritional, toxicological, health, and home environments. This in turn provides a strong antidote against arguments of genetic or environmental determinism; the evidence of which we are disseminating widely to academic as well as non-academic audiences.
We expect a range of additional results by the end of the project, including new local-area level monthly data on unemployment rates immediately after the second world war in England and Wales, and more evidence on the importance of the early-life economic environment for later life health and education outcomes, as well as how this is moderated by one’s genetic ‘predisposition’. We are also doing research on the accuracy of location-reporting and highlight the implications of this for empirical research. We are describing historical trends and social gradients in maternal smoking over time and illustrate how these vary by individuals’ genetic predisposition, linking it to the many government and policy interventions, as well as changes in societal smoking norms. Regarding pollution, we are modelling local dispersion using specific polluting sources, and we are digitizing novel data on historical regional pollution levels. We link these to government policy that aimed to reduce pollution and estimate the long-term impacts on individuals in older age.