Jørn Olsen, Aarhus University / UCLA School of Public Health

Thank you for the invitation to come to this conference. I have long advocated for the IEA-EEF to be more visible in this part of Europe. I have only wished the meeting to have come earlier in the history of the IEA-EEF when I still had my work in Europe and where I could have taken a more active part.
 
This meeting is important because we need close collaboration among epidemiologists in Europe. In a time when epidemiologists are scattered into specialized areas of research; cancer, infectious diseases, toxicology, mental health, health care, we are often only a smaller part of international meetings. As such we have to accept the conditions given by others and there is limited time for advocating the discipline and for using time to learn what is new in the core part of epidemiology. What are the new methods and what are the research infrastructures we need to build to meet the needs for using those methods in public health epidemiology?
 
First of all, we have to recognize that epidemiologic research is often expensive, and without having access to research funds we can do very little. It is naïve to think this access is only a question of having good ideas and the necessary skills. Access to large funds is becoming more and more a function of being able to lobby for the cause at the right places; in Brussels and in other places where large research funds are being distributed. If we have no peers in the review process we will not get a fair review. We can use international meetings to discuss how we strengthen our position, and one part of this is of course to be a member of a society. Far too few epidemiologists in Europe are a member of the IEA, and there is no other international organization in epidemiology to speak our cause. By becoming a member of the IEA you also help supporting epidemiologists in the developing countries.
 
We should also recognize that access to personal data for research is still a major problem in many parts of Europe as well as in the US. Policy for these issues is now partly out of the hands of national government for member states of the EU. Some of us remember the threat to epidemiologic research (and clinical research) the first draft of the directive for data protection posed. Without a strong and coordinated effort we could have been set back decades in our ability to provide information on health hazards and performance in the health care sector. Similar things may well happen again since these directives are not written by people who know about research. We have strong national societies for epidemiology in many European countries but they need to pay attention to what happens outside their own geographical border. The IEA-EEF offers such a mechanism.
 
Concerning research options we have never lived in a more exciting time period. High through-put methods in laboratories, together with low costs, have provided access to biological information of a magnitude never seen before. Our possibilities to screen for thousands and even millions of gene variants in humans, animals, or microorganisms have given us unprecedented opportunities. Unfortunately these possibilities are still way beyond our biological understanding and have shifted part of epidemiology from a deductive science in to an inductive discipline. Combined firepower in the laboratories and the principle of “brutal force and ignorance” makes it possible to analyze correlations without limits or guidance by biological theory. It is not pretty to watch, but it has produced interesting results and generated huge funds to research. Epidemiologists should become part of these research teams, not only as data providers but as methodologists. We need to learn the skills of running these large operations and we should make sure our students learn these skills of large scale operations. Better biological understandings will generate new hypotheses that may be testable in epidemiologic designs.
 
Expectations were high when the human genome was mapped, also in public health and prevention. Results have so far been meager and disappointing, but many things indicate this is about to change. We may be able to detect powerful gene-environmental interactions in the next decade that can be translated to disease prevention. We may be able to identify modifiable environmental exposures when we study subpopulation of genetic susceptible. 
 
We may also be able to detect new epidemics at a stage where they can still be contained by using new monitoring tools. But we should not rely on these tools alone. It is worth noting the importance of basic epidemiologic monitoring as it is well illustrated by the recent H1N1 flu epidemic. Without proper denominators case-fatality cannot be estimated and these data were missing in the early phase. This shortcoming was perhaps leading to overreactions that may well turn into apathy if serious actions are needed at a later stage.
 
New theories also produce new research opportunities and most important of these is “the early origin of chronic diseases”. From the work of the pioneers of Forsdal and Barker who studied social determinants and fetal nutrition a whole set of new theories have emerged that include infections, fetotoxic exposures and lack of specific food items. Not only cardiovascular diseases may be “programmed” early in life, mental disorders, injuries, neurological diseases and autoimmune diseases may also be candidates for this research.
 
And, finally, within the core discipline itself, theoretical epidemiology, important advancements have taken place especially related to the concept of causality and how it corresponds with the way we plan our studies, analyze our data and interpret our results. Mackie’s causal field theories provided a concept of causation that did fit observations and was useful in making predictions. These theories are still being applied frequently to better understand many of the concepts we use although most of the epidemiologist are more familiar with Rothman’s contributions to this important part of epidemiology. Mackie’s bood from 1974 (the Cement of the Universe, OUP) is still worth reading as well as Judea Paerl’s book on causation from this year. His contribution has been to provide a set of graphical rules to depict webs of causation that has helped understanding concepts like confounding and bias and to select the proper statistical models to analyse data.
 
We are probably witnessing the death of “frequentism”, and the role of P-values is declining although this has been a much too slow death. We may also witness a brand new way of analyzing data using new tools brought to us by more powerful computers, but first of all much clearer ideas on what we really study.
 
These skills have to be spread to the users and conferences and meetings are important as well basic training at a pre- as well as a postgraduate level.
 
At the same time we should not forget that epidemiology is a population science like demography. We should use existing trends on incidence rates over time or between populations to generate hypotheses and we should not forget the big picture; what are the important determinants of diseases and how do we avoid them. We have the power to improve health for thousands but also to do the opposite. Ours studies on HRT and CVD illustrate how careful we have to be.
 
I wish you all will take an active part in a conference. I hope you will have a conference that will give you new ideas and new friends.
 
Jørn Olsen
 
Past president of the IEA