The need to strive for impartiality in the evaluation of evidence must be stressed, partly because there are strong forces encouraging subjectivity. Among the most vital, exciting aspects of epidemiology are its value in understanding how the world we live in operates to affect health and the applicability of epidemiologic evidence to policy. Epidemiologic research bears on the foods we eat, the medications we take, our physical activity levels, and the most intimate aspects of our sexual behavior, emotional ties, and whether there are health benefits to having pets. Putting aside the scholarly arguments made in this book, I am sure every reader "knows" something about what is beneficial and harmful, and it is difficult to overcome such insights with scientific evidence. (I don't need epidemio-logic research to convince me that there are profound health benefits from owning pet dogs, and I am equally certain that pet cats are lacking in such value.) Epidemiologic evidence bearing on such issues is not interpreted in a vacuum, but rather intrudes on deeply held preconceptions based on our cultures, religions, and lifestyles. Judgments about epidemiologic evidence pertinent to policy inevitably collide with our political philosophy and social values. In fact, suspicion regarding the objectivity of the interpretation of evidence should arise when researchers generate findings that are consistently seen as supporting strongly held ideology.
Beyond the more global context for epidemiologic evidence, other challenges to impartiality arise in the professional workplace. On a personal level, we may not always welcome criticism of the quality of our own work or that of valued colleagues and friends, or be quite as willing as we should be to accept the excellent work done by those we dislike. The ultimate revelation of an ad hominem assessment of evidence lies in the statement that "I didn't believe it until we saw it in our own data." Such self-esteem may have great psychological value but is worrisome with regard to objectivity. Epidemiologists may also be motivated to protect the prestige of the discipline, which can encourage us to overstate or understate the conclusiveness of a given research product. We may be tempted to close ranks and defend our department or research team in the face of criticism, especially from outsiders. Such behavior is admirable in many ways, but counter to scientific neutrality.
Perhaps most significant for epidemiologists, who are often drawn to the field by their strong conviction to promote public health agendas, is the temptation to promote those public health agendas in part through their interpretations of scientific evidence (Savitz et al., 1999). The often influential, practical implications of epidemiology, the greatest strength of the discipline, can also be its greatest pitfall to the extent that it detracts from dispassionate evaluation. The implications of the findings (quite separate from the scientific merits of the research itself) create incentives to reach a particular conclusion or at least to lean one way or another in the face of true ambiguity. The greatest service epidemiologists can provide those who must make policy decisions or just decide how to live their lives is to offer an objective evaluation of the state of knowledge and let the many other pertinent factors that bear on such decisions be distilled by the policy maker or individual in the community, without being predigested by the epidemiologist.
For example, advocates of restrictions on exposure to environmental tobacco smoke may be inclined to interpret the evidence linking such exposures to lung cancer as strong whereas the same evidence, viewed by those who oppose such restrictions, is viewed as weak. A recent review of funding sources and conclusions in overviews of the epidemiologic evidence on this topic finds, not surprisingly, that tobacco industry sponsorship is associated with a more skeptical point of view (Barnes & Bero, 1998). Whereas judgment of the epidemiologic evidence is (or should be) a matter for science, a position on the policy of restricting public smoking is, by definition, in the realm of advocacy—public policy decisions require taking sides. However, the goal of establishing sound public policy that advances public health is not well served by distorting the epidemiologic evidence.
Fallible epidemiologic evidence on the health effects of environmental tobacco smoke may well be combined with other lines of evidence and principles to justify restricted public smoking. Believing that public smoking should be curtailed is a perfectly reasonable policy position but should not be used to retrofit the epidemiologic evidence linking environmental tobacco smoke to adverse health effects and exaggerate its strength. Similarly, strongly held views about individual liberties may legitimately outweigh epidemiologic evidence supporting adverse health effects of environmental tobacco smoke in some settings, and there is no need to distort the epidemiologic evidence to justify such a policy position. As discussed in Chapter 2, epidemiologic evidence is only one among many sources of information to consider, so that limited epidemiologic evidence or even an absence of epidemiologic evidence does not preclude support for a policy of such restriction nor does strong epidemiologic evidence dictate that such a policy must be adopted. Cleanly separating evaluation of epidemiology from applications of that evidence to policy encourages a more dispassionate assessment of the epidemiology and ultimately more rational, informed policy.
A primary goal of this book is to help make the evaluation of epidemiologic evidence more objective, in large part by making the criticisms and credibility of those criticisms more explicit, quantitative, comprehensive, and testable through empirical evaluation. Even when scientists disagree about the proper course of action, which they inevitably will do, just as nonscientists disagree, they may still agree about the key sources of uncertainty in the epidemiologic literature and the direction and magnitude of the potential biases.
At first glance, revealing epidemiology's "dirty laundry" by exposing and dwelling on the sources and magnitude of error may be seen as threatening to its credibility among other scientists and the public at large. Elevating the debate to focus on concrete, testable hypotheses of bias is more likely to have the beneficial by-product of enhancing the image of epidemiology in the broader scientific community. There seems to be the impression that epidemiologists have limitless criticisms of every study and thus they are unable to present a clear consensus to other scientists, policy makers, and the public. Such criticism and debate should not be restrained for public relations purposes, but to be useful the debate should focus on important issues in the interpretation of the evidence, explain why those issues are important, and point toward research to resolve those concerns. If those who held opposing viewpoints were better able to reach agreement on the specific points of contention that underlie their differences, work to encourage the research that would resolve their disagreements, and accept the results of that improved research, other scientists and the public could better understand that epidemiologists engage in the same process of successive approximations of the truth as other scientific disciplines. The disagreements would be clearly seen as constructive debate that helps to refine the study methods and reach greater clarity in the results and encourages the investment of resources in the research to resolve important controversies, not as personal bickering or petty disagreements over arcane, inconsequential issues. The ultimate test of the value of the disagreement is in whether it leads to improvements in the research and advancements in knowledge.
Was this article helpful?