Portal de Periódicos da CAPES

Work that Matters

2015; Lippincott Williams & Wilkins; Volume: 26; Issue: 2 Linguagem: Inglês

10.1097/ede.0000000000000240

1531-5487

Howard Frumkin,

Health disparities and outcomes

In 1983, during the Wade Hampton Frost Lecture at the American Public Health Association meeting, Bill Foege introduced the term “consequential epidemiology.” The context was an ongoing debate over whether epidemiologists should engage in the policy process as advocates, or, as many believed, just do the science and let advocates and policymakers worry about implementation. Foege’s view was clear: epidemiology was “a tool to change the world, not merely to study the world.”1 A decade later, Willard Cates reintroduced the term “consequential epidemiology” in his 1994 Society for Epidemiologic Research presidential address.2 The context was different: epidemiology was under challenge. The iconoclast and gadfly Petr Skrabanek, at Trinity College Dublin, had inveighed against “the poverty of epidemiology,” charging not that epidemiology was too apolitical, but that it was “a misnomer for scaremongering made respectable by the use of sophisticated statistical methods,” a problem resulting from “a high prevalence of epidemiologists when the incidence of problems soluble by epidemiological methods is low.”3 The limits of epidemiology were spotlighted—perhaps unfairly, at times bombastically, but very publicly.4 Critics, as described by Cates, charged that epidemiologists tortured their data until some—frequently obscure—associations or risk factors were found, and that these findings rarely led to public health action.2 Cates anchored his response in the health care system, arguing that the future of epidemiology in the United States depended on adapting to ongoing health care reform. He held that consequential epidemiology should be aligned with “outcomes research” that consumers and providers would demand.2 This entailed two key questions: “so what?” (i.e., do our inferences work to change people’s lives?) and “how much?” (i.e., what do interventions cost for the benefits they provide?). Other leading epidemiologists have cautioned that infatuation with methods can distract from focus on impact. Lewis Kuller observed that “…over time, a lot of epidemiology has become data analysis and data dredging and highly sophisticated statistical modeling but without any emphasis on the application of epidemiology on public health and preventive medicine.” (quoted in 5). In 2013, Sandro Galea, as president of the Society for Epidemiologic Research, revived the term, as “consequentialist epidemiology”.6 He reviewed a year’s output of four leading epidemiology journals, as well as the content of 14 leading epidemiology textbooks, finding an overwhelming preoccupation with etiology and causal inference, and little emphasis on intervention. “We do not,” he concluded, provide either a framing for or an orientation around what approaches to the improvement of population health are most relevant in particular contexts and why, how we can assess what approaches matter and when, how and where we may best intervene, and the role of epidemiologists in both framing these questions and in helping lead public health science to their answers.6, p1186 It is not difficult to find explicit examples of Galea’s conclusion. In fact, ePIDEMIOLOGY informs prospective authors that “We give highest priority to etiologic health research and the methods that undergird it…”,7 a sharp contrast with giving highest priority to improving public health, and warns that “Policy implications of research results … may not be included in research reports” (http://edmgr.ovid.com/epid/accounts/ifauth.htm)—a formidable barrier to addressing public health improvement. Galea called for a consequentialist epidemiology, which …would be centrally concerned with improving health outcomes. We would be much more concerned with maximizing the good that can be achieved by our studies and by our approaches than we are by our approaches themselves. A consequentialist epidemiology inducts new trainees not around canonical learning but rather around our goals. Our purpose would be defined around health optimization and disease reduction, with our methods as tools, convenient only insofar as they help us get there.6, p1187 Consequential environmental epidemiology, I suggest, is a special case of consequential epidemiology. Nine features can be identified. Of these, the first six are common to consequential work in other fields of epidemiology, whereas the last three are unique to a subspecialty defined by its focus on environmental exposures. First, consequential environmental epidemiology addresses widespread causes of suffering or premature death, for which there are plausible environmental contributors. In the 2010 Global Burden of Disease study (http://www.healthdata.org/gbd), ischemic heart disease, pneumonia, stroke, and diarrhea rank among the major causes of disability-adjusted life year loss; other high-ranking causes with relevance to the environment include malaria, road injuries, and depression. These outcomes should be high priorities for environmental epidemiologists. In some instances, of course, rare or minor conditions merit study. But there is an opportunity cost to such research—the foregone consequential questions that skilled researchers could tackle instead—and epidemiologists should carefully assess its value in public health terms. Second, consequential environmental epidemiology extends beyond etiologic studies to test and document solutions. Galea’s analysis of leading epidemiologic journals, mentioned earlier, found a strong preponderance of etiologic studies over intervention studies; although similar data are not available for the environmental epidemiology literature, the results would likely be similar. We need to know not only that traffic-derived air pollution harms health but also what transportation policies most effectively promote health. We need to know not only that pesticides affect reproductive function but also what agricultural methods yield both robust crop yields and healthy workers and consumers. Environmental epidemiologists should be lighting the way to solutions. Third, and related, consequential environmental epidemiology answers questions that decision-makers need to have answered. This recalls the concept of “user-driven research,” defined by the National Research Council as “applied research that supports decision-making.”8, p5 User-driven research has obvious relevance to product development, in fields such as computer science, digital media, engineering, and biotechnology, but is equally relevant in fields that serve the public good, such as sustainability science and health services research.9 In environmental health, decision-makers include people in many roles: regulators; leaders in conservation, environmental and community organizations; leaders in industry; elected local and state officials; and officials in sectors such as energy, transportation, and housing. Environmental epidemiologists will be best positioned to do consequential work if they craft their research agendas in consultation with, and in response to the needs of, such decision-makers. Fourth, the methods of consequential environmental epidemiology are varied and flexible, not purist or doctrinaire. Ross Brownson et al, in a recent commentary entitled “From Epidemiology to Policy,” wrote eloquently of the unintended harm to policy making that results from relying on a fixed hierarchy of evidence, one well-suited to establishing biologic cause but ill-suited to many other issues in the formation of public policy. For biologic cause, the hierarchy typically ranks experimental above observational data and prospective above retrospective data, when nearly always a mix is required to make good policy. Experiment serves to eliminate confounding but seldom mimics real world application. Prospective data often do not exist or take too long to gather. Similarly, observation of effects on individuals (e.g., body mass index) outrank more upstream end points (e.g., presence of a strong policy) for understanding biology, yet for many policy concerns, the natural unit of observation is made not at the individual level but instead at multiple levels of an ecologic framework.10, p409 Epidemiologists need to be more elastic than scholastic; methodologic heterodoxy needs to trump orthodoxy. Ultimately, consequential environmental epidemiology will be marked less by innovative exposure assessment or elegant analytical methods than by conclusions that translate into advancing the public’s health. Fifth, consequential environmental epidemiology does not beat dead horses. When enough research has been done to provide a clear basis for action, the customary closing benediction of research reports—“More research is needed”—may become untrue.11 Do we really need more evidence that lead damages children’s neurologic development? That asbestos causes mesothelioma and lung cancer? That particulate air pollution increases cardiovascular mortality? Again, there is an opportunity cost to such research. Sixth, consequential environmental epidemiology results are well communicated. Public health science, including environmental epidemiology, is a public good—not only because such a large part of our training and research is publicly funded but also because public health is fundamentally about serving the greater good. Accordingly, researchers carry an obligation to get consequential results to those who need to know and act on them—decision-makers, affected communities, the general public. A well-written peer-reviewed paper in a scientific journal is a pleasure to read, but few would confuse it with effective public communication. Consequential environmental epidemiology also needs to be disseminated through public presentations, op-ed pieces, blogs, TED talks, and social media.12 Seventh, consequential environmental epidemiology addresses consequential environmental conditions or trends. How do we recognize such “exposures”? They are widespread, and there is plausible biological reason to think they threaten health on a large scale. The leading environmental trends now shaping our planet, and therefore shaping the human condition, certainly qualify: climate change; urbanization; biodiversity loss; environmental loading with persistent, bioaccumulative, and toxic chemicals; more frequent disasters. Surely environmental epidemiologists should be flocking to study this phenomena.13 Eighth, consequential environmental epidemiology utilizes the comparative advantage of environmental epidemiologists. Exploiting comparative advantage is a general strategy for maximizing impact. Two examples are illustrative. The first example is exposure assessment—a core part of the environmental epidemiology skill set. Consider the hypothesis that nature contact promotes health. Reported benefits include reduced stress, improved attention in children, smoother surgical recovery, reduced obesity, and even better birth outcomes.14 These are consequential observations, as evidence-based urban planning and landscape architecture could yield widespread health benefits, at relatively low cost, with few adverse side-effects, and with many co-benefits. But in studying the benefits of nature contact, what are the right exposure metrics? Do you need trees, or do shrubs suffice? Do the trees have to be in leaf, or do bare deciduous trees in the winter confer a benefit? Is it enough to view trees through a window, or do you have to walk among them? For how long is exposure necessary? Environmental epidemiologists bring salient research design skills to such questions, a comparative advantage in doing consequential work. As another example, environmental epidemiologists are more accustomed than most other epidemiologists to thinking outside the biomedical arena, to such diverse upstream factors as industrial production methods, traffic patterns and air quality, ecosystem function, and groundwater movement. Many consequential epidemiologic research questions require this skill set, and so, again, environmental epidemiologists have a comparative advantage in doing consequential work. Finally, consequential environmental epidemiology yields collateral benefits such as promoting environmental sustainability. Humanity faces epic challenges in the breaching of planetary limits15 and the degradation of earth systems: climate change, biodiversity loss, ocean acidification, and more.16 Environmental epidemiology, while focused on human health outcomes, is well positioned to explore and document impacts outside the health sector. Consider, for instance, findings that a shift from single-occupancy vehicle travel to walking, cycling, and transit use can reduce cardiovascular and respiratory disease.17 That is consequential information—made even more consequential by the companion finding that the change would substantially reduce regional air pollution and carbon dioxide emissions. Environmental epidemiology can advance such full-benefit accounting, amplifying the consequentiality of its results. Just as these nine features characterize consequential environmental epidemiology, some statements regarding research may signal the opposite, epidemiology at risk of being inconsequential: “My advisor did it this way.” “I’ve always done it this way.” “This work is fundable.” “This work is really interesting to…me.” Consequential environmental epidemiology has implications for training. An inclination toward such thinking, and not only quantitative skills, should be among the qualifications for graduate study in environmental epidemiology. Training should emphasize not only methods, but larger questions of benefit and impact. Faculty should teach and model identifying consequential questions and investigating them. The incorporation of ethics into epidemiologic training provides a useful model for such pedagogy.18–20 The argument for consequential environmental epidemiology has limits. It is not a claim that environmental epidemiologists should be advocates—a topic for a separate discussion.21–23 Nor does it deny the value of pure research, niche research, and replication research, all of which might seem vulnerable to a charge of inconsequentiality, but each of which has a proper and necessary role. (For instance, a well-established finding may need replication in a jurisdiction where decision-makers demand local data before taking action.) Indeed, research consequentiality is a continuous, not a categorical variable, and one that is not always easy to measure. I propose it not as a doctrine or a litmus test but as a heuristic, a set of questions, a framework that I hope helps in setting priorities for environmental epidemiology investments. Woodrow Wilson wrote that “To work for the common good is the greatest creed”—a widely applicable maxim. Ultimately, the success of environmental epidemiology lies in its impact in improving the human condition. ACKNOWLEDGMENTS This article is adapted from a plenary talk at the 2014 Annual Conference of the International Society for Environmental Epidemiology in Seattle, WA. I thank David Savitz, Manolis Kogevinus, Maria Feychting, and Miguel Hernán for useful comments.