A study in the Journal of the American Medical Association (Mar. 10) debunks the myth that reducing dietary fat intake decreases breast cancer risk.
The study was done with data from the Nurses Health Study -- 88,795 women followed since 1980.
Although the Nurses Health Study is not such a good data set for identifying small increases or decreases in risk, it is certainly good enough for showing the absence of an increase or decrease in risk.
What is most interesting about this study is that a 1990 meta-analysis of 12 case-control studies reported a statistically significant 46 percent increase in breast cancer risk for women with high dietary fat intake.
Those who follow this page know that meta-analysis -- a statistical technique for combining small studies into a single large study to produce a statistically more robust result -- is inappropriate for case-control epidemiology. Case-control studies invariably are too different in data collection and analytic techniques to combine into a single study; it's like mixing apples and oranges. Meta-analysis is well-suited for tightly controlled, randomized clinical trials.
The most famous example of a meta-analysis gone awry is the EPA's 1992 risk assessment for secondhand smoke. In that case, the EPA combined 9 case-control studies with 2 cohort studies to arrive at a statistically insignificant 19 percent increase in lung cancer risk.
This page has continually pointed out flaky nature of the secondhand smoke risk assessment. This new study helps make the case.
This one cohort study has blown away a 12-study meta-analysis that is statistically stronger than the EPA secondhand smoke risk assessment.
One can only imagine what a large, cohort study like the Nurses Health Study would do to the secondhand smoke-lung cancer myth.
Comments on this posting?
Click here to post a public comment on the Trash Talk Bulletin Board.
Click here to send a private comment to the Junkman.