A social epidemiologic approach to health and aging provides a specific language for conceptualizing, investigating, and contributing novel understandings of the nature of aging. The integration of creative counterfactual thinking through a causal lens, life-course models, and cross-national comparisons allows social epidemiologists to critically examine the determinants of health and illness across diverse global contexts. These tools provide a foundation from which empirical evidence can be generated to reimagine what aging could look like in the future, under various revisions of global and local social structures around the world.
social epidemiology, life course, causal inference, counterfactuals, comparative research, health equity
Bodies tell stories that voices cannot. Aging bodies tell stories of accumulation; of years spent in struggle, with freedom, with plenty, with little, with kinship, in isolation, and with joy and sadness. It is in later-life—however we define this phase—that the totality of one’s experiences become physically inscribed in health status. Health, or lack thereof, is the ultimate expression of social inequality.
Epidemiology is the study of the distribution of health and disease in populations (Susser, 1973), and social epidemiology is a subfield of epidemiology. Social epidemiologists study the ways in which societies are organized to produce or impede population health (Berkman & Kawachi, 2014). Through its integration of sociological theory with mathematical approaches to causal effect estimation, social epidemiology provides a language for conceptualizing, discovering, and determining the universalities and local diversities in health and aging.
Social epidemiologists use data to tell stories of bodies that people cannot tell with their voices, whether because they lack the opportunity or platform to tell their stories, because they are forbidden from telling their stories, or because they lack the words with which to tell their stories. As with many other academic disciplines, social epidemiologists hold a precious form of power in their ability to use data to tell stories. The language with which we choose to tell these stories has real consequences for the perceptions, policies, and actions that will shape the futures of aging bodies to come.
Within and beyond social epidemiology, a predominant conceptualization of aging is that of decline, one that frames conversations around inevitable losses. Aging brings about increased health risks, including chronic diseases, in addition to declines in physical and cognitive function, all of which can lead to frailty, dependence, and loss of quality of life.
Theoretical models of “successful aging” and “healthy aging” implicitly retain this focus as they emphasize successful aging or healthy aging despite the inherent “risks” of getting older. But social epidemiology gives us tools with which to recognize nuanced heterogeneities in aging within and across societies, allowing a reframing of the narrative around aging and an uncovering of the realities of aging, from global to hyper-local scales.
A global perspective necessitates encapsulating global geographies and interconnections in society, and understanding universal commonalities that drive social processes and often lead to vast disparities (Finn, 2023). We also can use a global perspective when referring to human universalities, such as the need for food, water, shelter, companionship, love, dignity, and a sense of purpose.
A hyper-local perspective can add nuance, by telling us more about the day-to-day living conditions in which these needs are fulfilled or denied, providing contextually specific data to inform appropriate intervention and prevention strategies. Finally, a consideration of how experiences accumulate over time tells us the stories of aging bodies, and social epidemiologic data gives us the language and power to tell these stories.
This article discusses how social epidemiology provides a language for conceptualizing, discovering, and determining the universalities and local diversities in health and aging. Three main approaches are discussed: the use of a life-course approach, the practice of causal inference using quantitative data, and the creative leveraging around the world of cross-national comparisons to identify variations in risks and protective factors for health during aging. The article ends by summarizing how these approaches can contribute to reframing narratives about how people age around the world.
Social Epidemiologists, Aging, and the Life Course
Social epidemiologists use a life-course approach to conceptualize and investigate the stories of bodies, especially of aging bodies. While life-course theory originated in sociology, epidemiologists have refined it to include considering disease development and latency periods, approaches to causal effect estimation, and the integration of novel methodologies (Mayeda et al., 2021). Understanding the complexities of health and illness during aging requires a life-course approach, as risks for chronic health conditions such as cancers, cardiovascular disease, and Alzheimer’s disease and related dementias are understood to accumulate over time and express in clinically recognizable symptoms following long latency periods. The popular phrase “the long arm of childhood” was coined by demographers Mark Hayward and Brigid Gorman (2004) in a study of the relationship between early childhood social conditions and men’s mortality in the National Longitudinal Survey of Older Men in the United States.
‘What would the world have looked like if we rolled back the clock and altered some aspect of reality?’
In social epidemiology, and in epidemiology more broadly, applying a life-course approach largely involves crafting one’s research question as informed by a life-course theoretical model (Mayeda et al., 2021). A thoughtful application of the life-course approach requires knowledge of the pathophysiologic process underlying the chronic disease of interest so that appropriate life-course periods of exposure can be identified for investigation, as well as the appropriate induction period, which is the time elapsed between exposure and disease onset.
For example, a social epidemiologist using a life-course approach might ask, “Do the effects of exposure to air pollution on the risk of dementia depend on the timing and duration across the life course during which air pollution exposure is experienced?” They may then examine different types, sources, and concentrations of air pollution exposure at different times and durations across the life course, as informed by empirical evidence or theory on the biological effects of air pollution on the brain. Life-course approaches to health and aging in social epidemiology invite us to ask empirical questions with improved theoretical and empirical rigor and appropriate temporality between exposures and outcomes, all of which supports the validity of effect estimation and ultimately the planning of effective interventions to support health during aging.
The Role of Data in the Storytelling of Bodies: Association Is Not Causation
A key goal of social epidemiology is to understand the causes of health and disease to inform effective interventional strategies that promote population health and health equity. To achieve this goal, social epidemiologists must perform causal inference, the process of determining whether a causal relationship exists between an exposure (say, air pollution) and a subsequent health outcome (say, later-life dementia). Causal inference is not unique to epidemiology; many disciplines share a rich philosophy and history of theoretical frameworks and applications of causal inference (Imbens & Rubin, 2015; Rothman & Greenland, 2005; Sobel, 1996). Many academics across quantitative science disciplines are familiar with the phrases “association is not causation” and “absence of evidence is not evidence of absence,” which allude to some challenges inherent in the practice of causal inference.
Despite these challenges, causal thinking in social epidemiology invites us to imagine counterfactuals—what would the world have looked like if we rolled back the clock and altered some aspect of reality to see what would have happened in this alternative, counterfactual timeline?
When thinking about health and aging, causal thinking invites us to ask what would have happened to the distribution of health and illness in this population, in this place, at this time, had historical exposures occurred in a different way, counter to reality?
In the air pollution example, the use of counterfactual thinking combined with a life-course approach would encourage us to frame our research question in terms of the potential population distribution of dementia outcomes that would have been observed, had the subset of individuals classified as exposed to air pollution been, counter to fact, not exposed. The definition of air pollution exposure and the study design used to answer this question would be selected to test hypotheses about different types and sources of air pollution experienced at different life-course stages in relation to later-life dementia risk, as informed by theory or empirical evidence of the pathophysiological effects of air pollution exposure on the brain.
The clear conceptual and mathematical definitions of counterfactuals are a cornerstone of causal effect estimation in epidemiology. Forcing ourselves to imagine and articulate a counterfactual world when estimating the effect of an exposure on a health outcome is a creative exercise in which we consider how things might have been under a different exposure scenario, whether that exposure scenario is on a micro or a macro scale.
We find one of the strongest clues that counterfactual worlds in which humans achieve longer and healthier lives are possible in global life expectancy data. Life expectancy is a useful indicator of the overall well-being of a population. Around the globe, as of 2020, life expectancy at birth ranged from 53 years (Chad) to 85 years (Hong Kong, Macao, and Japan; World Bank, 2022).
Life expectancy at birth is highly sensitive to infant mortality rates, such that other informative metrics, especially when thinking about older populations, are remaining life expectancies at age 50 or at age 65. The existence of wide variation in life expectancies around the world tells us that they have the potential to be lengthened, pushing us to ponder the counterfactual social, economic, and political conditions under which longer life expectancies could be achieved.
Of course, life is worth little without quality of life, and it will take a substantial restructuring of the conditions under which people age for populations to live not only longer, but also healthier lives. The COVID-19 pandemic and subsequent declines in life expectancy in several countries prompt us to examine how national responses to the pandemic, not only in terms of public health control measures but also changes to political institutions, social protection policies, and the social fabrics of societies have impacted the health and well-being of populations (Finn & Kobayashi, 2020).
Cross-national Comparisons Can Uncover Global Universalities and Local Diversities in Aging
The social epidemiologist Geoffrey Rose’s (2001) seminal work “Sick Individuals and Sick Populations” articulates how the drivers of differences in health outcomes between individuals may not be the same as the drivers of differences in health outcomes between populations. The causes of ill health may be masked on the individual level if they are homogeneously distributed within a given population. Rose’s example is that of the distribution of systolic blood pressure among middle-age men in two populations: Kenyan nomads, and London civil servants. Rose explained that while an analysis of the determinants of high blood pressure within either population alone may help to identify individual-level drivers of high blood pressure—answering the question, “Why do some individuals have higher blood pressure than others?”—such an analysis would miss the more important public health question, “Why is hypertension absent in the Kenyans, and common in London?”
‘The causes of ill health may be masked on the individual level if they are homogeneously distributed within a given population.’
To answer the latter question, Rose (2001) proposed we need to study the characteristics of populations, rather than the characteristics of individuals. Zooming out to the scale of countries, to think not just about individual bodies but about the body politic at large, allows us to identify population-level levers that may be pulled to alter the macro-level contexts shaping health and disease risk.
Cross-national or cross-population comparisons are a valuable yet under-utilized tool for social epidemiologists to identify population-level drivers of the prevalence and incidence of health conditions. When integrated with a life-course approach to health and aging and a careful consideration of the conditions necessary for causal effect estimation, cross-national comparisons can be powerful for identifying sources of variation in the health of older populations.
The 2020 Lancet Commission on Dementia concluded that 2.3% of all dementia cases were attributable to air pollution exposure, yet the evidence upon which this conclusion was based came from a single cohort study in Canada, a country with some of the lowest air pollution levels in the world (Livingston et al., 2020). A social epidemiologic approach that leverages cross-national data, drawing from Rose, could provide a more comprehensive picture of this potential dementia risk factor that better represents the real living conditions of people who experience air pollution exposure. Exposure to air pollution is dependent upon many factors, namely physical proximity to a range of sources such as indoor open flame cooking, outdoor fires, road traffic, and agricultural or industrial production.
While the study of health effects of air pollution is often overlooked by social epidemiologists, exposure and susceptibility to air pollution are linked to poverty and social marginalization and are thus highly inequitable within and across societies (Hajat et al., 2015). Further, the long-term health effects of air pollution exposure may only express themselves in later-life, in health conditions such as chronic respiratory problems, cardiovascular disease, lung cancer, and dementia risk, necessitating a life-course approach.
A study of air pollution and dementia risk in the United States may conclude that only certain sources of emission prevalent in the United States are linked with later-life dementia risk, and that physical residential proximity to these emission sources is inequitably distributed across racial and ethnic groups and by socioeconomic position (Gonzalez et al., 2023). A restricted geographic scope of study limited to the United States or other high-income nations may also conclude that air pollution contributes to a relatively small share of dementia cases, given the prevalence of exposure in these settings. In contrast, studies of air pollution and dementia risk in many low- and middle-income countries would likely expand their focus to include the indoor combustion of household fuels for cooking, lighting, and heating, as well as open waste burning.
The nature of air pollution exposure around the world is variable and likely different from that of the experience of high-income older populations that are represented in current dementia evidence. This is a major inequity in the dementia evidence base that limits policy to support the healthy aging of populations worldwide, which social epidemiologists are well-poised to address.
Another bit of language social epidemiologists contribute to how we can think and talk about aging is consistency of cause. Differences in air pollution as a health risk exposure across global contexts can be seen as a violation of the consistency assumption of causal inference in epidemiology. The consistency assumption is the idea that there should be only one version of the exposure, and if any aspects of the exposure vary between individuals, they must be ignorable with respect to differences in their effects on a given outcome (Cole & Frangakis, 2009). In other words, different exposures require the estimation of different causal contrasts, and these exposure-specific contrasts should map onto hypothetical interventions that could be designed in the future.
While “air pollution” is a broad enough descriptor that any reasonable study of its health effects would need to be as specific as possible about the type and source of the pollutant(s) of interest, it can be difficult to ascertain the origins of airborne particulate matter and measures of ambient particulate matter in different contexts. Measures of fine particulate matter taken in the streets of urban Delhi, the surrounds of the Agbogbloshie electronic devices toxic waste site in Accra, in suburban areas near petrochemical sites in Texas, or following wildfires after periods of extreme heat may capture the same molecules but in different combinations and concentrations, from different sources, and with different health implications.
‘Social epidemiologists are aware of consistency assumption violations with respect to social exposures such as education or income.’
Social epidemiologists are aware of consistency assumption violations with respect to social exposures such as education or income (Rehkopf et al., 2016). Rather than abandoning a research topic due to consistency violations, social epidemiologists can creatively leverage consistency violations to use empirical data to identify and learn from heterogeneities in risk and protective factors for health during aging across diverse global contexts.
A related concept that allows social epidemiologists to identify variations in risk and resilience to health outcomes during aging across contexts from national to local scales is that of effect modification. Effect modification occurs when the effect of a risk factor on a given health outcome differs across population subgroups, usually due to a third factor or attribute held or experienced by that subgroup that confers additional susceptibility or resilience to that risk factor. While air pollution is a risk factor for dementia, identifying certain population subgroups or regions in which older adults appear to be resilient to the impacts of air pollution on dementia risk would be informative for designing policies to protect other populations.
Such investigation of effect modification would additionally contribute to the reframing of predominant narratives about inevitabilities of risk associated with aging. Effect modification in the case of the air pollution–dementia example could be any form of risk mitigation system that dulls the impact of exposure, such as air filtration or masking, or even a certain genetic polymorphism that blunts the exposure’s biological effect. Careful investigation of effect modification allows social epidemiologists to uncover diversities in aging at local scales.
A key goal of social epidemiology is to understand the causes of health and disease to inform effective interventions that promote population health and health equity. The integration of creative counterfactual thinking with life-course models and in a cross-national comparison framework allows social epidemiologists to critically examine the determinants of health and illness across diverse global contexts. These tools provide a language for describing global universalities and local diversities in health and aging, as they have the potential to illuminate where, when, and under what circumstances aging-related health outcomes may be malleable through intervention.
Hence, social epidemiologists can help us to re-imagine what aging could look like in the future under alternative population distributions of risk and protective factors for health, and alternative organizations of societies that structure such exposure distributions for their populations. It is up to social epidemiologists—and other quantitative population health science researchers—to use their linguistic tools responsibly to help improve the health of older adults worldwide and address global issues of health equity among the most marginalized older adults.
Lindsay C. Kobayashi, PhD, is the John G. Searle Assistant Professor of Epidemiology and Assistant Professor of Global Public Health at the University of Michigan School of Public Health in Ann Arbor, Michigan. She may be contacted at firstname.lastname@example.org.
Photo credit: Shutterstock/Jacob Lund
Berkman, L. F., & Kawachi, I. (2014). A historical framework for social epidemiology: Social determinants of population health. In L. F. Berkman, I. Kawachi, & M. M. Glymour (Eds.), Social Epidemiology (p. 0). Oxford University Press. https://doi.org/10.1093/med/9780195377903.003.0001
Cole, S. R., & Frangakis, C. E. (2009). The consistency statement in causal inference: A definition or an assumption? Epidemiology, 20(1), 3. https://doi.org/10.1097/EDE.0b013e31818ef366
Finn, B. M. (2023). Pandemic urbanization: How South Africa’s history of labor and disease control creates its current disparities. Journal of Urban Affairs, 45(3), 616–29. https://doi.org/10.1080/07352166.2022.2101468
Finn, B. M., & Kobayashi, L. C. (2020). Structural inequality in the time of COVID-19: Urbanization, segregation, and pandemic control in sub-Saharan Africa. Dialogues in Human Geography, 10(2), 217–20. https://doi.org/10.1177/2043820620934310
Gonzalez, D. J. X., Nardone, A., Nguyen, A. V., Morello-Frosch, R., & Casey, J. A. (2023). Historic redlining and the siting of oil and gas wells in the United States. Journal of Exposure Science & Environmental Epidemiology, 33(1), 76–83. https://doi.org/10.1038/s41370-022-00434-9
Hajat, A., Hsia, C., & O’Neill, M. S. (2015). Socioeconomic disparities and air pollution exposure: A global review. Current Environmental Health Reports, 2(4), 440–50. https://doi.org/10.1007/s40572-015-0069-5
Hayward, M. D., & Gorman, B. K. (2004). The long arm of childhood: The influence of early-life social conditions on men’s mortality. Demography, 41(1), 87–107.
Imbens, G. W., & Rubin, D. B. (2015). Causal Inference for statistics, social, and biomedical sciences: An introduction. Cambridge University Press. https://doi.org/10.1017/CBO9781139025751
Livingston, G., Huntley, J., Sommerlad, A., Ames, D., Ballard, C., Banerjee, S., Brayne, C., Burns, A., Cohen-Mansfield, J., Cooper, C., Costafreda, S. G., Dias, A., Fox, N., Gitlin, L. N., Howard, R., Kales, H. C., Kivimäki, M., Larson, E. B., Ogunniyi, A., … Mukadam, N. (2020). Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. The Lancet, 396(10248), 413–46. https://doi.org/10.1016/S0140-6736(20)30367-6
Mayeda, E. R., Binder, A. M., & Kobayashi, L. C. (2021). Life course epidemiology and analysis. In R. Detels, Q. A. Karim, F. Baum, L. Li, A. H. Leyland, R. Detels, Q. Abdool Karim, F. Baum, L. Li, & A. H. Leyland (Eds.), Oxford textbook of global public health (p. 0). Oxford University Press. https://doi.org/10.1093/med/9780198816805.003.0043
Noppert, G. A., Hegde, S. T., & Kubale, J. T. (2023). Exposure, susceptibility, and recovery: A framework for examining the intersection of the social and physical environments and infectious disease risk. American Journal of Epidemiology, 192(3), 475–482. https://doi.org/10.1093/aje/kwac186
Rehkopf, D. H., Glymour, M. M., & Osypuk, T. L. (2016). The consistency assumption for causal inference in social epidemiology: When a rose is not a rose. Current Epidemiology Reports, 3(1), 63–71. https://doi.org/10.1007/s40471-016-0069-5
Rose, G. (2001). Sick individuals and sick populations. International Journal of Epidemiology, 30(3), 427–32. https://doi.org/10.1093/ije/30.3.427
Rothman, K. J., & Greenland, S. (2005). Causation and causal inference in epidemiology. American Journal of Public Health, 95 Suppl 1, S144–50. https://doi.org/10.2105/AJPH.2004.059204
Sobel, M. E. (1996). An introduction to causal inference. Sociological Methods & Research, 24(3), 353–79. https://doi.org/10.1177/0049124196024003004
Susser, M. (1973). Causal thinking in the health sciences: Concepts and strategies in epidemiology (Issue 3). Oxford University Press.
World Bank. (2022). World Bank open data. https://data.worldbank.org