ABSTRACT
This reports a massive piece of work in the USA in which the incontinent use of the non-hormonal steroid diethylstilbestrol to prevent miscarriage in the 1950s was studied with respect to the longterm breast cancer risks on the mothers. (The daughters had been studied after a consequential and massive increase in the risk of adenocarcinoma of the vagina had become apparent quite quickly.) This study followed mothers, both given the drug and not given the drug during a pregnancy between 1940 and 1960, for 40 years. Such was the use of this drug in the practice styles of the time that it can be safely assumed that providers believed in its (ultimately shown to be ineffectual) effect on preventing miscarriage or they did not. Selection by intrinsic breast cancer risk seemed to be unimportant. But the risk of breast cancer was doubled among those exposed but not until 40 years after exposure. The period between exposure and 20 some years later is completely devoid of any excess risk among the exposed group. The possibility that such a profound, and unique, effect could be attributable to bias or confounding must be quite fanciful.
