Before I start today’s note, a couple of comments on communication. We have established SPH Today/This Week/This Month as our communication channels in order to communicate to our entire community of staff, faculty, students, and alumni consistently and clearly. If it’s not in SPH Today/This Week /This Month, it’s not happening. We have a lot of work to do together to achieve our aspirations and, as we move forward, I am committed to sharing information with the whole community in a timely and transparent manner. If you ever in doubt about something you hear, please ask me, anytime.
Relatedly, I very much enjoyed our first school-wide strategic thinking day on Tuesday this past week. I thought the day was a terrific opportunity to build links among faculty, staff, students, and alumni, and it generated ideas that can well inform our strategic thinking moving forward. Thank you to all who participated, to all members of the Strategic Thinking Steering Group who led us through the day, and to the staff who made the day happen.
Now, on to today’s topic. I have previously argued that an approach to population health that focuses on the central drivers of what matters most to the health of the public must lean on organizing frameworks to help us wrap our arms around the plethora of factors—potential causes—of health, on which we might choose to intervene. There are several good examples of efforts to bring theory to our understanding of population health science, and public health frequently borrows theoretical perspectives from other disciplines to guide specific aspects of our work. Here, I want to comment on two simpler frameworks that I find immensely useful in my own work: life course and multilevel frameworks, each of which aims to help organize our thinking in population health science. In the interest of keeping the Dean’s Notes more readable, I shall, as I have done before, divide this into two parts, starting today with a life course perspective.
A life course approach to health is based on the understanding that multiple factors, including biological, social, psychological, geographic, and economic, shape health over the life course through risk mechanisms that are independent and cumulative and interact over time. As John Lynch and George Davey Smith succinctly put it:
A life course approach to chronic disease epidemiology explicitly recognizes the importance of time and timing in understanding causal links between exposures and outcomes within an individual life course, across generations, and on population level disease trends. (p. 1)
The scope of specific factors covered within this approach includes, for example, physical growth, social mobility, behavior changes, physical environment, and life role transitions. Centrally, life course approaches attempt to assess how exposures arise and produce health throughout life, and how we make sense of these interconnected temporal processes. Life course approaches also extend beyond the life of any one individual to suggest connections in health across generations. Therefore, a life course approach guides us, for example, to this question: How does childhood exposure to a traumatic event change the risk of poor mental health in adulthood? Importantly, a life course perspective suggests that we cannot ignore this question—that is, unless we understand the traumatic event experienced during one’s childhood, our understanding of poor mental health in adulthood is going to remain limited and incomplete.
This raises the conceptual and analytic bar, suggesting that we must take into account factors throughout life, and across generations, to better understand the health of populations at any given moment. This may indeed make our job harder, but it also points to approaches that can yield compelling answers and help us move beyond the welter of contradictory findings that unfortunately characterize much of population health science literature.
The advent of formal thinking about life course approaches in population health science is relatively recent and emerged principally in the realm of chronic disease, although further work has well shown how this approach extends to psychiatric and substance use disorders, infectious disease, and oral health. The most recent precursor to the formalization of a life course approach in population health science came via professor David Barker and colleagues, who found a link between birth weight and lifetime risk for coronary heart disease. Known as the “fetal origins hypothesis,” this work focused on how prenatal programming may influence later health. Prior to this work, it was not entirely clear whether prenatal exposure mechanisms were linked to adult disease only through their correlation with later life exposures, or whether these early exposures mattered entirely on their own. The work of Barker and colleagues showed that these early-life exposures did matter on their own, above and beyond any measured confounding variables. Barker’s example opened the way to the formal introduction of life course thinking in the field in the coming decades.
As our thinking about life course exposures has sharpened, several authors have articulated key mechanistic models that may explain how exposures over the life course shape subsequent health. Key models in this regard are the critical period, sensitive period, accumulation of risk, and chains-of-risk models. The critical period model emphasizes the timing of an exposure during specific periods of unalterable biological development, with the understanding that the exposure can affect that development. One example of this is fetal exposure to teratogens [see Figure 1], which links directly to our understanding of human embryonic development to illustrate how fetal exposure to a particular event or agent can result in subsequent alterations to normal human development.
The sensitive periods hypothesis posits that there are sensitive periods throughout the life course, that are not temporally fixed, during which exposure can have a greater impact than at another time. An example is the effect of poverty on mental health during a period of social transition such as divorce.
In an accumulation of risk model, the total amount of exposure is what matters, rather than specific exposure time points. Nutrition and cancer risk provides an illustrative example [see Figure 2].
Finally, the chains-of-risk model emphasizes the sequence of exposures and assumes that one exposure increases the risk of, or triggers, another exposure. An example of this model is nicotine exposure potentiating cocaine addiction.
Perhaps the greatest challenge in adopting a life course approach rests on how one may operationalize such an approach to our analytic ends. One simplification rests on thinking about discrete life course stages and then considering how each stage can represent causes of later disease, and manifest consequences of prior exposure.
By way of example, we can turn to an area in which I have done a reasonable amount of work: substance use. We can consider a life course epidemiology of substance use by thinking of five life course stages: in utero, infancy, childhood, adolescence, and adulthood.
In utero exposure to smoking is a cause of increased risk of lifetime tobacco dependence and also carries the immediate consequence of low birth weight. Low family socioeconomic status and marital status changes during infancy predict early onset of smoking, while exposure to parental smoking during infancy is associated with sudden infant death syndrome. Childhood neglect and abuse are associated with binge drinking in adolescence, while maternal drug use during childhood predicts early onset of the same drug use in children. In adolescence, drinking is associated with alcohol dependence later in life, while multiple substance use is a consequence of prior physical and sexual abuse. Finally, in adulthood, low income is positively associated with increased risk of substance use disorders, while injuries are a consequence of alcohol intoxication.
Each of these illustrations well make the point that a life course perspective suggests links across phases of life, and that absent an understanding of these links it will be difficult to understand any particular “one point-in-time snapshot” of population health. These illustrations, however, will also suggest to the reader that this approach, while helping us better understand the determination of population health, raises substantial methodological and conceptual questions that might open up new scientific vistas and challenge dominant paradigms. At the simplest level, why should an exposure in childhood influence health in adulthood? Clearly some process, perhaps biological, perhaps social, must link these life stages. Even more provocatively, why should exposures for one generation influence the health of a subsequent generation? The recent emergence of epigenetics as a potential explanatory has provided some promise in these efforts, although that too perhaps opens up as many questions as it answers. Methodologically, a life course perspective calls for approaches that rise above our typical deterministic approach—that can take into account both long-term temporal influences and the dynamic, discontinuous, and non-linear influences that these approaches likely suggest. In a future Dean’s Note, I will comment on how a complex systems methodological approach may help in this regard.
I shall discuss multilevel perspectives in an upcoming Dean’s Note.
I hope everyone has a terrific week. Until next week.
Sandro Galea, MD, DrPH
Dean and Professor, Boston University School of Public Health
Acknowledgement: I would like to acknowledge the work of Gregory Cohen, MSW, on this Dean’s Note and the second Dean’s Note in this series.
Previous Dean’s Notes are archived at: https://www.bu.edu/sph/category/news/deans-notes/