Women’s Health
BMJ: Doubt cast on Million Women Study, implicating HRT in breast cancer risk
[Does hormone replacement therapy cause breast cancer? An application of causal principles to three studies Part 4. The Million Women Study Online First 2012 doi: 10.1136/jfprhc-2011-100229]
Findings from the Million Women Study, which were used to establish that hormone replacement therapy (HRT) causes breast cancer, do not in fact, prove a causal link, concludes a review of the evidence published online in the Journal of Family Planning and Reproductive Health Care.
The Million Women Study (MWS), which has produced four reports (2003, 2004, 2006 and 2011), and is the largest study of its kind on the topic, was one of three major pieces of research that prompted a rethink of the long term safety of HRT.
The MWS drew on new cases of breast cancer arising in women invited for breast screening in the UK between 1966 and 2001. The estimated risk levels it found for the disease were higher than those found in either of the other two studies---the collaborative reanalysis (CR) and the Women’s Health Initiative (WHI).
Given its key role in influencing the regulatory authorities and public perception, the current authors assessed generally accepted causal criteria applied to scientific research, such as biases and biological implausibility, to review the findings of the MWS.
Their analysis highlighted several design flaws that would have skewed the findings.
For example, cancers detected within a few months of the study’s start would have already been present when the women were enrolled in the research, but these were not excluded.
Using study participants taking part in a breast screening programme, and the invitation to the study itself, would have increased the participation of HRT users already aware of breast lumps/pre-cancerous changes, so leading to higher numbers of cancers being detected (detection bias), say the authors.
This is borne out by a higher rate of cancers among the study participants than in the general population, irrespective of whether or not they used HRT.
Breast cancer can sometimes grow slowly, and remain undetected for years. Even a small bias resulting in the detection of 1 to 1.2 extra cases each year among 1000 women who had used HRT for five years, or an extra 1.5 to 1.9 cases in those using HRT for 10 years, would have invalidated the overall findings, say the authors.
The average time from recruitment to the MWS to detection of breast cancer of 1.2 to 1.7 years, and the likelihood of a fatal breast cancer being 22% higher among HRT users within this short time frame, which the study found, was biologically implausible, they claim.
For the most aggressive forms of cancer, it is generally accepted that it takes a decade for a tumour to reach the size of 1 cm, which is the smallest size that can be diagnosed clinically.
Crucial data were often also missing, say the authors. In the third MWS report, follow up data on HRT use was not available for 57%-62% of the study participants.
“The name ‘Million Women Study’ implies an authority beyond criticism or refutation. Many commentators, and the investigators, have repeatedly stressed that it was the largest study of HRT and breast cancer ever conducted,” write the authors.
“Yet the validity of any study is dependent on the quality of its design, execution, analysis and interpretation. Size alone does not guarantee that the findings are reliable,” they continue.
“The MWS was an observational study, and it has the attendant problems and uncertainties intrinsic to such studies. If the evidence was unreliable, the only effect of its massive size would have been to confer spurious statistical authority to doubtful findings,” they conclude, adding that the MWS evidence “was indeed unreliable.”
Source: British Medical Journal
