We read the paper by A. Fournier et al. that was published online on 18 November 20041 with interest. This large study (54,548 postmenopausal women and a total of 948 breast cancer cases) showed an increased RR for breast cancer in HRT users compared to nonusers. Estrogen-mono therapy and estrogens plus micronised progesterone were not associated with an increased RR for breast cancer.
However, the article raises some questions that need to be addressed. Using simple arithmetic, we found 372 breast cancer cases in women using HRT (Table IV). However, in total 948 breast cancer cases were reported, where are the other 576 missing cases? Were these cases in the nonusers cohort? If yes, how can the RR for breast cancer of 1.2 be explained (576 breast cancer cases in 25,128 nonusers vs. 372 breast cancer cases in 29,420 HRT users)?
Additionally, we would question whether it is appropriate to compare 2 study groups that have significant differences in all baseline characteristics (Table I). Could these differences (confounding), although adjusted for some of these factors (but interestingly not adjusted for the year of birth), have impacted on the results?
There are some further points that need to be considered. For example, 71% of the nonusers and 84% of the HRT users reported a BMI ≤ 25. Even for French women, this seems rather unlikely for postmenopausal women at that age.
And then, for what indication were progestins used for ≥ 5 years?
Finally, the study period was 1990–2000 (and included women who had used HRT prior to January 1992). However, we would question if it is proven that the HRT compounds containing special progestins, e.g., nomegestrol, were marketed in sufficient quantities in the early 1990s for the indication HRT to draw conclusions of the far reaching nature as discussed in the current paper.