From the Danish Epidemiology Science Centre, Aarhus University, Aarhus, Denmark.
Correspondence: Jorn Olsen, Danish Epidemiology Science Centre, Aarhus University, Vennelyst Boulevard 6, Building 260 DK-800, Aarhus, Denmark. E-mail: email@example.com.
Epidemiologic research has unavoidable errors. It is virtually impossible to get point estimates exactly right. At best, we can hope that we estimate effects in the right direction and within reasonable proximity to the true effect size for the population under study.
There are an endless number of choices epidemiologists must make when we select our research topic, design our study, analyze the results, and present the findings. We all trust that these decisions are made in good faith. We have to act like Kafka's truth-seeking dogs; the search for truth should be our only aspiration.
Yet, we know this is not always the case. We may put emphasis on elements in applications or in article that will increase our probability of being funded or being published. That kind of career-boosting is the dope of science and, like doping in sports, it destroys fair competition. Research results will have no value if they cannot be trusted within the limits with which we. The mechanisms that lead to minor unjustified manipulation of data may even lead to more severe and rare types of dishonesty, such as fabrication of data. Dishonesty may pay off in the short run but will hopefully be discovered with time.
Disclosure of conflicts of interest does not solve the problems with dishonesty in epidemiology or science in general. However, financial dependency may enhance the probability of deviations from good behavior and the truth-seeking path. It seems to be a reasonable to ask for information that may involve conflicts of interest.
A more radical solution would be to remove researchers from financial dependency on funding agents with special interests for some type of research. The world is probably not ready for such a step. Too many people benefit from the close contact between industry and science, and there is nothing wrong with such contact in general. The problems arise for research that addresses public health matters, such as estimating benefits and side effects of medicine or environmental exposures. If research results become biased because of this financial dependency, people will suffer. Patients will not receive the best treatment, and they may have side effects they could have avoided. One cannot rule out the possibility that this practice may kill people while we wait for science to be self-correcting. It is furthermore too optimistic to think that science always will be self-correcting, especially when we talk about treatments that may be on the market for a short time span.
Disclosure of conflicts is a small step in the right direction. Why not take this step? If we have nothing to hide, then why try to hide anything? In the sad history of long-term misinformation about the health effects of smoking, we would have avoided severe damage had we known that some of the colleagues who tried to dismantle sound scientific results were on the payroll of the tobacco industry.
Without any doubt, many epidemiologists do honest scientific work of good quality despite being on the payroll of a company that has a financial interest in specific results of the research. Many, and probably most, of our colleagues are not influenced by the pressure to stress or downplay certain results. In a situation in which public health aspects are at stake, we should remove as many obstacles or circumstances as possible that could lead to misleading results. To have a policy of openness concerning conflicts of interest is the first step.
In the future, we will probably look upon this time and wonder why we accepted obvious conflicts of interest in research of public health importance. We will find it hard to understand why we did not do our best to make research as reliable as possible. After all, without reliability we have nothing to offer to the public. Research results will have limited value if they are not done in good faith with truth-seeking as the main driving force.
ABOUT THE AUTHOR
JØRN OLSEN is now Professor of Epidemiology at the School of Public Health, UCLA. He was director of The Danish Epidemiology Science Centre. He is currently the president of the International Epidemiological Association.
© 2006 Lippincott Williams & Wilkins, Inc.