The drumbeat of the anti-HRT war grows louder. Yesterday HRT was accused of causing breast cancer. Today it is accused of causing ovarian cancer -- on equally frivolous grounds. A PDF of the full journal article in "Lancet" is here. I reproduce a media report of it below. Rather than adding to my own comments of yesterday, I follow the media report with a reproduction of Prof. Brignell's comment on the nonsense. Prof. Brignell is a mathematician who campaigns against the ignorant and malicious misuse of statistics
HORMONE replacement therapy, a contested treatment for post-menopausal women that has already been linked to breast cancer, is also associated with ovarian cancer, a study in The Lancet said today.
Women who take HRT are on average 20 per cent likelier to develop and die from ovarian cancer compared to women who have never been on this treatment, according to the research. The evidence comes from a major British investigation into female health, the Million Women Study, covering 1.3 million British women from 1996-2001....
The British researchers assessed data from 948,000 post-menopausal women, who had been questioned and later given a follow-up exam some three years later. Around 30 per cent were current HRT users; 20 per cent had previously received HRT; and the remaining 50 per cent had never taken it. Across all three groups, a total of 2273 women developed cancer, and 1591 died from it....
The HRT link with breast cancer surfaced in 2002 [See my post of 14th. for a comment on THAT crap study], prompting many women in the US to drop the treatment - a trend that notably coincided with a sharp fall in new breast cancer cases in the US...
The empire strikes back
By John Brignell
Like the environmentalists, the epidemiologists do not like to have their hegemony over their corner of the media to be challenged. No sooner has their dangerous and destructive nonsense over breast cancer been thwarted than they come out with even more dangerous and destructive nonsense about ovarian cancer. Valerie Beral, a women noted for the size of her Trojan Numbers, has come out with a relative risk of 1.2.
There are at least two well known confounding factors to which such an observational study such as this are prey:
* If the therapy is successful then the patient will have a marked change of life style.
* The reasons for which the therapy was prescribed in the first place might well pose a risk factor.
The second of these can be eliminated in a properly conducted double-blind randomised trial, but the first cannot.
A personal anecdote will illustrate how this factor works. Last year your bending author was reduced to life as a housebound cripple by a marked increase in arthritic inflammation. Eventually, therapy with Diclofenac and Co-Codomol restored an element of normal living and the patient celebrated by going out and digging over a large allotment. In retrospect this was rather foolish, a such violent activity after a period of forced idleness would have exacerbated any incipient heart disease. Fortunately, survival indicates that there was none.
It seems, however, more likely that the second of these confounding factors would be more important in this case, but more haunting is the possibility of confounding factors we have not thought of.
Which words in the truism "Correlation is not Causation" do the epidemiologists not understand? There is no reason to suppose from these tacky observations that any women at all have been killed by HRT.
Is there anything more despicable than pinning your claim to fame on scaring millions of women out of using a hugely liberating therapy? As for Cancer UK, which we all know from the constant begging letters, it could put the product of its suppliance to better use by supporting science rather than nonsense.
Prof. Brignell is too contemptuous of the study to comment at length on it above but he is making essentially the same point that I do: The very low percentage of women apparently affected makes it highly likely that the result is random noise. That some of the results are statistically significant rules out only one source of random fluctuation -- small sample size. A large enough sample will make ANY observed effect statistically significant. Statistical significance does not and cannot rule out other random (or non-random) events, effects and influences.
Prof. Brignell has some links in his article that I have not reproduced above. See the original for those links
This whole anti-HRT campaign is quite despicable. It aims to get women to take large risks (of osteoporosis etc.) in order to avoid tiny risks (of cancer)
(For more postings from me, see TONGUE-TIED, EDUCATION WATCH, GREENIE WATCH, POLITICAL CORRECTNESS WATCH, FOOD & HEALTH SKEPTIC, GUN WATCH, SOCIALIZED MEDICINE, AUSTRALIAN POLITICS, DISSECTING LEFTISM, IMMIGRATION WATCH and EYE ON BRITAIN. My Home Pages are here or here or here. Email me (John Ray) here.)