In my opinion, the Age-Related Eye Disease Study (AREDS) investigators
promoted a nonsignificant result into a conclusive recommendation.1 Here is how they did it.
The primary study outcomes for AREDS are explicitly stated in the "Participants
and Methods" section of the article: (1) progression to advanced age-related
macular degeneration (AMD) and (2) a 15-letter decrease in visual acuity.
These outcomes were to be evaluated in all patients by independent tests of
significance of the 2 primary treatments. This carefully specified primary
analysis led to 4 tests, none of which was statistically significant. One,
testing the effect of zinc on progression to advanced AMD, achieved a level
of significance defined by the investigators as suggestive.
Seigel D. AREDS Investigators Distort Findings. Arch Ophthalmol. 2002;120(1):100. doi: