PROVIDING WIDELY-UNDERSTANDABLE PERSPECTIVE FOR RISKS REPORTED IN STUDIES

CV Phillips* and G Maldonado (Minnesota Center for Philosophy of Science, Minneapolis, MN
55455)

Results from health research are increasingly acted upon directly by individuals, based on popular
news releases or self-motivated Internet searches, without the involvement or interpretation of
reviewers, clinicians, or policy makers with relevant expertise. Among the troubling implications of
this is an over-reaction to minimal risks dues to people's tendency against discriminating between low
risks (e.g., 1%) and extremely low risks (e.g., one in ten million). This leads to behaviors like people
becoming extremely worried about some environmental chemical when even the largest estimates of
its effects are dwarfed by the risks from automobiles or saturated fat. Since there is likely no one else
to do so, it is important that researchers who report results that individuals may act upon directly
(e.g., information about diet, exercise, over-the-counter pharmaceuticals, recreational drugs) provide
a context for appreciating the absolute level of risk. This should take the form of both a direct
comparison to comparable risk (e.g., "as unlikely as getting hit by lightning") and comparison to a
major health risk (e.g., "about 1/100th the effect from not wearing your seatbelt"). We argue that such
comparisons should be added to research reports and press releases. We facilitate this by providing
absolute risk levels for various exposures and behaviors, including transport accidents, common
medical conditions, and recognizably rare events, for use as comparisons. We then provide examples
of this approach for recent high-profile research findings that affect public behavior, including the
risks from phenylpropanolamine and cellular phones.

American Journal of Epidemiology, 153(11):S272, 2001.