The author responds:
Professors Kogevinas and Stayner1 do not address the central point of my commentary2 but they do pursue an issue on which I welcome the opportunity to elucidate. As a student of epidemiology myself, I appreciate professors who are more critical of their discipline. Self-satisfaction and complacency cannot fuel the necessary innovation and evolution needed for the new challenges in the field of epidemiology. By all means, celebrate success but do learn from the mistakes. If the professoriate does not raise questions, then others, including thoughtful science writers, will do it for them. Fortunately, awareness of the many problems facing observational epidemiology is increasing.3-8
Ten “successes” are mentioned (there are many more), and here are 10 examples (there are many more) in which observational epidemiology provided, at least for a period of time, false-positive estimates of health effects, with some unintended consequences:
- Bendectin/Debendox and congenital malformations: removal from market of effective safe morning sickness pill with no equivalent replacement.
- Breast implants and immune diseases: threat to postsurgical reconstruction options.
- Electromagnetic fields and cancer: school closures and loss of home equity.
- Environmental dioxin and many diseases: dislocation of entire communities.
- HIV: incorrect identification of risk factors before experimental virologists found the virus.
- Induced abortion and subsequent breast cancer: used as evidence to reduce reproductive choice.
- Nuclear power plant radiation and cancer in surrounding populations: unnecessary alarm.
- Psychological stress and gastric ulcers: decades of therapeutic mismanagement.
- Radar and cancer: threatened important safety technology.
- Spermicides, intrauterine devices, and other contraceptives and congenital malformations: discontinued contraceptive use and unwanted pregnancies.
Observational epidemiology is an imprecise science where false-positive signals concerning the environment and health have particular potential to cause social disruption. The low-hanging fruit that observational epidemiology has harvested with some success in the last century has been replaced by the challenge of correctly associating rare exposures with infrequent outcomes. Whether our science is up to the task, without leaving a long trail of error, remains a question that our students will have to address. Give them the critical faculties that they will need to do their job.
Michael B. Bracken
Center for Perinatal, Pediatric and Environmental Epidemiology
Yale University Schools of Medicine and Public Health
New Haven, CT
1. Kogevinas M, Stayner L. Response to “Preregistration of Epidemiology Protocols: A Commentary in Support” [letter to the editor]. Epidemiology
2. Bracken MB. Preregistration of epidemiology protocols: A commentary in support. Epidemiology.
3. Bracken MB. Why are so many epidemiology associations inflated or wrong? Does poorly conducted animal research suggest implausible hypotheses? Ann Epidemiol.
4. Ebrahim S, Davey Smith G. Mendelian randomization: can genetic epidemiology help redress the failures of observational epidemiology? Hum Genet.
5. Ioannidis JP. Why most published research findings are false. PLoS Med.
6. Ioannidis JP. Why most discovered associations are inflated. Epidemiology.
7. Kraft P. Curses-winner's and otherwise-in genetic epidemiology. Epidemiology
. 2008;19:649–651; discussion 657–658.
8. Yusuf S, Collins R, Peto R. Why do we need some large, simple randomized trials? Stat Med.