Alex Broadbent, a young British-trained philosopher of science with a faculty appointment at the University of Johannesburg, clearly believes in the potential value of philosophy for epidemiology, and has set out to prove it. His forays into the field are now evident from the work he has published as guest editor of a themed-section of peer-reviewed essays in Preventive Medicine on Philosophy and Epidemiology, to which eminent figures from epidemiology and statistics as well as philosophy contributed, and as principal author of a new report from the PHG foundation entitled “Epidemiology, risk, and causation—Conceptual and methodological issues in public health science”.

These publications are not solo contributions but rather the products of a series of workshops which Broadbent helped to organize in 2009-10 on Epidemiology, Risk, and Genomics hosted at the University of Cambridge, Department of History and Philosophy of Science and sponsored by the PHG Foundation. He has also organized an international conference on the topic in December 2011, with funds from the South African government.

Doubts About Philosophy

In the introduction to his PHG report, Broadbent is candid about the skepticism which he believes may confront his endeavors in the epidemiology community. He states, “Doubt is often expressed when philosophers purport to contribute to the scientific enterprise…Epidemiology is a useful activity, and there is ample evidence in epidemiological journals and text books that philosophical problems arise in the course of doing it. The motivation of this project was to identify some of these problems, and to begin the process of solving them.”

Philosophical Problems

Before enumerating the problems of interest in epidemiology, Broadbent was asked to clarify what constitutes a philosophical problem. He told the Epidemiology Monitor that such problems are those which people disagree about, where the disagreements are persistent yet reasonable, and for which further empirical evidence will not get you an answer. He stated that ethical problems are a type of philosophical problem that are easier to recognize than other types because we have pre-existing ideas about right and wrong.

According to Broadbent, where concepts such as causation, risk, uncertainty, and causal inference have seemed too vague to use, epidemiologists have developed well-defined substitutes. However, it is not always clear how to relate these well-defined concepts back to the vague ones in which we think when deciding on courses of action or making evaluative judgments. Moreover, inconsistencies in our innate, “Stone Age” thinking are often brought out when epidemiologists seek to apply and extend the Stone Age concepts. Evidence cannot be brought to bear to resolve the difficulties, which are conceptual. It is in these areas where a philosophical approach could be helpful.

Philosophers Training

If philosophical problems have no definite solutions, why should philosophers be expected to be more helpful than epidemiologists? According to Broadbent, philosophers are trained in tackling such questions and have skills which better enable them to identify the problems to begin with, make distinctions, and clarify them. Given these skills, philosophers may not settle issues definitively, however, their approach can help “to close down some options and avoid inconsistencies,” says Broadbent.

Epidemiologists Not Trained

Broadbent told the Epidemiology Monitor that he has observed epidemiologists struggling with thorny philosophical issues in epidemiology. In one case, he observed an epidemiologist author taking a very authoritarian approach to explaining causation, asserting “this is how it is”. Philosophers would take a completely different approach, said Broadbent, a non-authoritarian, more exploratory one in which coherent options or alternatives are identified and discussed. For these types of questions which may not have a final answer, this is “a better approach” according to Broadbent.

Hitting Too Hard

Broadbent contrasts the philosophical approach with the one taken by scientists in yet another way. “Scientists tend to hit these problems too hard,” according to Broadbent, “closing off options and moving on when it is too soon to close options down without really understanding the position you are taking.” He added that, in contrast to scientists, philosophers are not typically trying “to get on with something”. [AB1]  Philosophy can pay off for epidemiology in   a number of ways: by reducing methodological inconsistencies, raising awareness of methodological error, making epidemiological concepts more useful, and making epidemiological evidence and arguments more compelling to a non-epidemiological audience, such as health policy makers or courts of law.

An Example

As an example of how philosophical thinking can be useful to epidemiology, Broadbent cited the topic of prediction. He said that philosophers have not given a lot of thought to what makes a good prediction as they have for what makes for a good causal inference. How to assess a predictive claim is a philosophical problem, but one which would have practical value for epidemiologists, according to Broadbent.

His paper published in Preventive Medicine creates a heuristic—having to explain “what could possibly go wrong”—to achieve a robust prediction about the outcomes of a health intervention in the real world (as opposed to in a study population). Just as Bradford Hill’s heuristic for causal inference is helpful for that function, so Broadbent is hoping his heuristic for prediction will have practical value.

As another example, he has also recently published an article making detailed recommendations on the use of epidemiological evidence to prove causation in law.

Philosophical Topics

Based on the discussions at the workshops he has helped organize, Broadbent has developed a preliminary list of topics which have either practical or philosophical significance and which could benefit from further philosophical work. They are:

1. The need for conceptual clarity in the use of health statistics

2. The need for clarity in the use of statistical significance testing in particular

3. The difficulty of causal inference and its continued resistance to formal methods.

4. The importance of distinguishing between internal and external validity and the difficulties of applying study results to a wider group

5. Continued methodological development in epidemiology as a young science

6. Understanding the live methodological and conceptual debates within the discipline

7. The complexity of the sorts of claims epidemiology makes about general causation

8. The paradox of prevention

9. Reconsideration of philosophic views of science in light of unique features of epidemiology as a science.

10. The need for a philosophy of epidemiology within the philosophy of science.

Final Word

Broadbent told the Epidemiology Monitor that the need for a philosophy of epidemiology is the sign of a growing and maturing discipline with an increasing impact on society. When asked what response he would like to obtain to his work from the epidemiology community, Broadbent said it would be beneficial if epidemiologists could be more accepting of the types of problems he is seeking to identify. He understands that epidemiologists might not always be inclined to spend precious time discussing problems that “you cannot shut down with a sharp answer”. According to Broadbent,  the existence of philosophical problems in epidemiology is not a sign of weakness, but a corollary of the maturing and conceptual development of the discipline. They are a sign of the difficulty the human mind has of getting a hold of the  concepts epidemiologists use and the phenomena they study. It would be beneficial for epidemiologists to be open to the possibility that a philosophy of epidemiology could be useful, says Broadbent.

Another view, not intended to be critical of epidemiologists, is that epidemiologists should not rely solely on other epidemiologists to resolve these philosophical questions. There is a tendency in every field for senior scientists to wade into the philosophy pool, according to Broadbent, but some papers on philosophical topics by senior epidemiologists are “just not good”, he said.

Metaphors

He told the Epi Monitor he does not intend his remarks to be belligerent, but simply a call for epidemiologists to recognize the expertise of others in this domain. “You don’t expect birds to be good ornithologists,” said Broadbent, and scientists should not expect to be good philosophers of science, any more than philosophers of science can expect to be good scientists. It is difficult for persons to avoid producing just hot air on these types of problems, according to Broadbent. You have to stop yourself. It is hard. Recognizing that philosophers have relevant skills and training, would be a useful, according to Broadbent.

Links to a free download of the PHG report and to the relevant issue of Preventive Medicine are provided below.

Also, read two related articles on the Philosophy of Epidemiology in this issue.

http://www.phgfoundation.org/reports/10491/

http://www.sciencedirect.com/science/journal/00917435