Breast cancer: reporting results with inflationary arithmetic.

Numeric data are almost always generated as final measures in scientific experiments, but the publications of these «hard» data is rare. Reductions of the mass are always required and many techniques have been developed. Data reduction may take many forms and the selection of which form to present in a publication reflects an author's bias-it is inescapable since almost all forms are scientifically valid even though each carries its own subliminal message. This editorial reviews several of the means whereby data derived from breast cancer research or observation of outcomes may be presented. Unstated observations, the use of untested premises, selection of the denominator in establishing ratios, graphic display, the concept of «cure» and calculations of harm/benefit are discussed