Epi Wit & Wisdom Articles

On the Future of Epidemiology

(3 of 3)

Following are brief summaries of the major points made by each of the speakers at the First International Panum Seminar held in Copenhagen on January 25, 1999

Stig Wall

Umea University, Sweden

“Four-Wheel Drive Epidemiology”

The first speaker at the Seminar was Stig Wall, Professor and Head of the Department of Epidemiology and Public Health at Umea University in Sweden. Dr Wall focused his presentation on two main topics—the role of epidemiology in assessing public health interventions and the need for epidemiology to tackle global health issues. “A nice paper and discussion of international work” according to Zena Stein, Columbia University professor and seminar participant. Wall stated that “there is potential for epidemiology in prevention activities and a challenge lies in bringing the development of epidemiological theory and methods nearer to public health efforts.” More specifically, Wall called for epidemiologists to 1) give greater importance to structural and macro-epidemiology that analyzes the effect of societal, economic, political and environmental changes on morbidity and mortality; 2) develop theoretical and methodological measures of outcome in preventive programs; 3) account for the exposure intensity of interventions they study;      4) analyze the social consequences of community-oriented health work, its benefits, and possible harm for public health; and 5) assess the potential for prevention with varying levels of public participation in intervention strategies.

Reflecting his global concerns, Wall closed by saying that “for one billion people living in the world’s poorest countries, where the burden of disease is highest, those who are born or who die are still not counted. I propose this to be the challenge for future epidemiologists.” In discussing his talk with the Epi Monitor, Wall expanded on his remarks by stating that many of the official statistics in developing countries are based on haphazard surveys or outdated information and we present data as if a country was homogeneous when in fact there are inequities, differences and variations in rates. “Instead we perceive the situation as ‘poverty all over’. We need data to detect the differences and I would like to see more epidemiologists involved in developing countries. Sweden with its good data can follow people from cradle to grave and we are well-known for our ‘armchair’ or ‘desktop’ epidemiology. I would like to see more ‘four-wheel drive epidemiology’,” he said.

Hans-Olov Adami

Karolinska Institute, Sweden

“Rescue from Methodologic Misery”

Participants of the Seminar who spoke with the Epi Monitor agreed that the remarks made by Hans-Olov Adami put him in the minority and at odds with the general thrust of remarks made by Wall and most of the other speakers who called for more emphasis on public health epidemiology. Adami acknowledged this and said that his background is that of a surgeon rather than a public health professional. He focused his talk on his vision for the future which includes a merger between clinical research and epidemiology. He believes this would be enormously fruitful because clinical research is currently living in “methodologic misery.” Adami believes that merger would be of mutual benefit because epidemiology would be enriched biologically and clinical and laboratory research methodologically. Interestingly, Adami stated “the epidemiology community has been unnecessarily defensive in the realization that epidemiology has now become a basic discipline for clinical medicine as it has long been for public health.” To accomplish the merger, the barriers between the disciplines should be disregarded and the spectrum of research from clinical to descriptive, to analytic, to clincal trials should be looked upon as one continuum on validity. Investigators should master the broad repertoire of techniques needed to investigate all along the spectrum.

Adami believes epidemiologists will have a need to understand molecular epidemiology in order to do etiologic research well. This is an enormous challenge. It is not realistic to also expect epidemiologists to have competence in areas related to modifying human behavior, political lobbying or interacting with the media, all of which skills are desirable for implementation of public health programs. “It is a disservice to public health if we think we can,” he said. It is this view limiting the role and concerns of epidemiologists that most set Adami apart from the other speakers.

Adami said his views also differ from those who believe risk factor epidemiologists are working only with minutiae or small relative risks. He gave the example of cervical cancer where recent RR estimates increased from 2.0 when sex partners were studied to 20.0 or even 50.0 when human papilloma virus was added, and he cited RR’s of 40 - 50 for reflux symptoms as risk factors for esophageal cancer. “Other dramatic risk gradients remain to be discovered,” he assured. On the other hand, he did point out that discoveries are unpredictable and “we need to accept that science is an end in itself... Investigator initiated research should be let go with no rigid priorities,” he said. He cited as an example of this serendipity the hypothesis which emerged from their work on hip implants. He and his colleagues found no evidence that these implants caused sarcomas in the bone or soft tissue but they did find a 50% lower risk of stomach cancer in these patients and are now pursuing the hypothesis that this reduction may be caused by the prophylactic antibiotics the patients were given which may have decreased infections with h. pylori.

Walter Holland

London School of Economics and Political Science, London

“Glorifying Science, Neglecting Application”

Walter Holland, formerly professor at St. Thomas’ Medical School and currently visiting professor at the London School of Economics, said epidemiologists have three real tasks—1) to be concerned with knowledge of the causes and associations of disease and abnormal function and with this knowledge to contribute to the formulation of preventive strategies; 2) to be involved with the testing of etiological hypotheses and the efficiency/effectiveness/efficacy of interventions; and 3) to be involved in decisions on planning, evaluation and organization of services to communities and individuals. In considering the future, Holland referred to the work of Beaglehole and Bonita who considered possible directions for public health—a broad focus on the underlying social causes of health and disease or towards a narrow medical focus. They urged public health to adopt a broad focus and Holland endorsed this view for epidemiologists as well. According to Holland, “it is crucial for our subject that we tackle broad issues and do not restrict ourselves to narrow expressions or concerns. The great dangers that we face are that we become concerned only with methodology, the secondary analysis of data, or the assessment of diagnostic tests or interventions.” In seeking to forecast a future for epidemiologists, Holland identified concerns and major areas the field should tackle, including: 1) problems of our own making, many of which are traceable to the shortcomings of the case-control design;   2) use of large databanks collected for other purposes without doing proper quality control of the data, without using biologically plausible hypotheses, or without being concerned about how one can use the findings; and 3) the intense study of some topics at the expense of others that are very neglected, e.g., mental health.

In conclusion, Holland stated, “I believe that if we are to be concerned with the future then it is not a choice between population or molecular methods or increasing emphasis on one or the other. It is that the appropriate methods be applied to identify the causes and mechanisms of conditions which influence or affect the health of populations and then to determine how these can be used in order to control or mitigate the effects of the condition. As a discipline, we have tended to emphasize and glorify the scientific aspects of our subject, but have neglected its application. Holland called for epidemiologists 1) to develop additional methods to test the effectiveness of control measures; 2) to expand the number of conditions studied; and 3) to become more concerned with the ability to apply the findings rather than only the scientific merit or nicety of the study.

David Hunter

Harvard School of Public Health, Boston

“Complexity Before Simplicity”

Another speaker at the Seminar was David Hunter, Director of the Harvard Center for Cancer Prevention. He labelled the question about the relationship between epidemiology and molecular biology “a very timely question because there is an opportunity for a marriage of disciplines.” Dr. Hunter focused on the role of molecular biology in helping epidemiologists to sort out the relative importance of genes and environmental factors in disease causation. While he believes that greater understanding of genetics will ultimately be “simplifying,” he warned that for the next 5 - 20 years we are in for a period of great complexity and confusion in sorting out what percentage of variation in disease is attributable to environmental causes and how much is due to inherited susceptibility. We might discover that specific mutations are like fingerprints which leave a recognizable telltale sign, but aflatoxin and UV light mutations are the only two examples where this has been clear-cut so far and we might have expected more instances by now. There may not be a specific mutation for each exposure, Hunter said. We have discovered the high penetrance genes and now it is hard to study low penetrance genes outside of a familial context... “This is very exciting and will challenge our science logistically, analytically and intellectually. There will be low RR’s and many false positives which will create a major problem in interpreting studies. It is important for public health people to get cross trained and to be at the table. If we anticipate, maybe we can deal with the coming complexity in a logical way,” he said.

Another concern is with the ethical and informed consent challenges this work will pose. The models we have for addressing these concerns in the classic studies simply do not apply in large population studies. For example, there is not the option for 1-on-1 counseling in large studies.

Neil Pearce

Wellington School of Medicine, University of Otago, New Zealand

“Population Epidemiology”

Neil Pearce, Director of the Wellington Asthma Research Group, entitled his talk “Epidemiology as a Population Science.” He labelled it a complement to his 1996 paper on the limitations of epidemiology published in the AJPH. Pearce, more than any other speaker at the Seminar, used his presentation to discuss the current debate in epidemiology between those who focus on risk factor epidemiology and those who wish to focus more on public health epidemiology. He took exception with the way the debate has been characterized most recently in the Lancet article (September 5, 1998) by Ken Rothman, Hans-Olav Adami, and Dimitrios Trichopoulos as a debate between those who wish to be activists and those who wish to be scientists. He said the real debate is between population research and research at the individual and micro-level. According to Pearce, the issue is the level at which the research hypotheses are conceptualized. He stated that epidemiologists have been able to “add value” principally through their unique population focus which is hypothesis-generating. He credits “population epidemiology” with success in discovering risk factors while micro-level studies have contributed by establishing etiologic mechanisms. He appeared to support Susser’s concept of “ecoepidemiology” which outlines a multi-level epidemiology that recognizes the value of epidemiology undertaken at all levels.

While many may agree with the importance of epidemiologic analysis carried out at multiple levels, that does not make every level the equal of other levels. Pearce seems to recognize some hierarchy of epidemiology types because he says we must still possess a means of allocating resources between types and decide how our hypotheses will be generated and conceptualized. This funding concern which raises the priority issue is reminiscent of the point raised by Jorn Olsen, the main organizer of the Seminar in his remarks to the Epi Monitor—we can all agree to do studies at different levels, but which level do we select when it comes to priorities for research?

For Pearce, the population-level analysis is the most important because it is the most fundamental in achieving the goal of epidemiology—the control of disease in populations. He said, “if our ultimate aim is to find out which factors are most important at the population-level, then it is essential that we should start at and continually refer back to the population-level for a ‘reality check’. Pearce recognizes that for some epidemiology there is a generic method for measuring the occurrence of disease and the word population is not part of the definition. “If epidemiology is merely a system of measurement, then it can never claim to be a science... The key feature of science is not measurement...but understanding... Epidemiologists need to learn to think in a multi-level way rather than just adding multi-level modelling to their analytical toolkit,” he told the participants.

Should epidemiologists be social scientists or molecular biologists? They should be both and they should be neither, said Pearce. In brief, they should be epidemiologists. “In his view, they should address the most important public health questions at the population-level, and use appropriate methods to address them at whatever level of analysis is most appropriate. In doing so, they may utilize methods from statistics, social science, molecular biology, and a host of other research disciplines in developing and testing population-level hypotheses... The problem with modern epidemiology is that it lacks a coherent substantive theory and is based on methods that are inadequate for studying the distribution and determinants of health-related states or events in specified populations. The current danger for epidemiology stems not from the use of new techniques, or from the use of individual and micro-level analyses, but from the fact that these techniques may narrowly define the questions that epidemiologists find acceptable for study. If the population perspective is lost, then epidemiology will simply become a measurement tool for testing hypotheses developed by other researchers. We need to rediscover the population perspective, and continue the development of epidemiology as a population science.”

Mervyn Susser

Columbia University, New York

More on “Ecoepidemiology”

Mervyn Susser gave the last presentation of the day and it was generally considered the highlight. He began with the same question given to others about being social scientists or molecular biologists and answered that to survive as a discipline, epidemiology must encompass both social science and molecular epidemiology, that is, epidemiologists must “comprehend and deploy the basic premises and the nature of the information these other disciplines yield.”

Susser said risk factor epidemiology is threatened by the fact that 1) it relies steadfastly on a single level of organization, the study of individuals almost always disconnected from each other, and 2) it tends to ignore mediators. Put another way, “linkages between exposure and outcome tend to be dispensed with. In so far as they are sought at all, the motive tends to be not the explanation of sequences in the causal process, but the testing for and protection against bias and confounding,” he said.

But in addition to the threat to risk factor epidemiology, Susser also has a keen historical perspective and is able to see a threat to the integrity of the discipline coming from the allure of molecular biology. “The micro-level exerts the same forceful attractions as did the germ theory at the end of the 19th century. The germ theory eclipsed the considerable power and prestige of the miasmatic epidemiology that preceded it... Now molecular epidemiology offers the same kind of illusion as did the germ theory. It is the illusion that an unarguable definitiveness and specificity of this extreme biological-level can explain everything—which is to say, everything at all the successive and increasingly complex levels of organization above. That was never true then and is not true now, so long as our concern is with the dynamic disease process as it occurs within and across populations,” he said.

In a nutshell, Susser’s main points were that risk factor epidemiology and its underlying multiple risk factor one-level paradigm are becoming exhausted and must be replaced with a multi-dimensional model of causality. This model will encompass successive levels of organization rising from micro-levels within individuals to societies beyond them. Does this “multi-level epidemiology” really exist or can it be expected to develop? Susser admitted that “As yet, I can rise to no more than a sketch or meager outline of what multi-level epidemiology is and can do...in order to preserve our public health tradition, we shall have to make a conscious choice to broaden and deepen our approach in the future. And if we try, my own experience of the risk factor era confirms the historical evidence—the demands of the scientific endeavor will in themselves produce the techniques and methods that we now lack.” 

Published February 1999  v

Privacy  |  Terms of Use  |  Sitemap

Digital Smart Tools, LLC