News and Press Releases

Title:
Oestrogen plus progestin and lung cancer in postmenopausal women [WHI trial]: a post-hoc analysis of a randomised controlled trial

Date Written:
03 Oct 2009

This paper was published in The Lancet recently. When learning basic statistical analysis and I confess my knowledge is somewhat rudimentary, I was always taught to treat the results of post hoc analyses with caution. Many considered this type of analysis to be “data dredging” and since it was not one of the primary aims of the trial the results must not be over-interpreted. Admittedly, in the middle of the text this is acknowledged in regard to the paper reporting and increased likelihood of death from lung cancer in those on HRT, but, the press statement that was issued at the time of publication would suggest that the authors were very keen to over-interpret their findings.

The new analysis by Chlebowski et al concerns the incidence and death from lung cancer in women who took part in the Women’s Health Initiative. The incidence was not statistically different between the two groups although there was a statistically greater chance of death in those who had been on HRT. One should emphasise the point that these were older women taking HRT, heavy smokers and the numbers are extremely small. Something else that I could not find was the actual numbers of women who had continued to be followed up since this was a self-reported problem although verification of diagnosis was made thoroughly. There were three additional cases per 10,000 women of lung cancer in the hormone users and it would appear all were heavy smokers. In the Press release a comment was made that “these results...seriously question whether hormone replacement therapy has any role in Medicine today” which seems a rather extreme conclusion from a paper such as this. Surely it would be of infinitely more value to the health of the women if strategies were put in place to stop them smoking. It is always the case with risks associated with hormones when there is a greater risk posed by a lifestyle issue such as smoking or obesity that the recommendation is that the HRT is stopped because this is so much easier. These data do not appear to be terribly relevant to women in their 50s who are the age group most likely to take HRT and they should not be particularly concerned by the findings.

From a clinical perspective, the authors worked very hard to justify their findings from a scientific perspective and I look forward to see if any of these ideas are developed further.

--
Professor M.A. Lumsden
University of Glasgow
Division Developmental Medicine
Reproductive & Maternal Medicine