Study of UK Cancer-Moderate-Consumption Paper Shows Fundamental Flaws In Data, Sample Selection, Methodology
Even a cursory examination of “Moderate Alcohol Intake and Cancer Incidence in Women” (published in the Feb. 24 online issue of the Journal of the National Cancer Institute) reveals flawed methodology and other contradictions that call into question the validity of its conclusions.
In the study, Naomi Allen, D.Phil., of the University of Oxford, U.K., and her co-authors, based their conclusions on their analysis of data from the Million Women Study, which included 1,280,296 middle-aged women in the United Kingdom.
SELF-SELECTED SAMPLES UNREPRESENTATIVE
The Allen paper said that their data came from questionnaires completed by 1.3 million, middle-aged women who attended breast cancer screening clinics in the U.K.
The method for selecting the study sample could be its largest flaw.
The Million Women Study web site shows that it is a voluntary, self-selecting study. Self-selecting samples represent a subset of any given population: those sufficiently motivated (for unknown reasons) to take the time and effort to participate. Self-selecting sample populations can often have little connection to reality because they are unrepresentative of a population as a whole.
Indeed, there is no way to know whether or not the women most motivated to participate in this study were those with existing or potential hereditary health problems.
QUESTIONAIRES: EXTENSIVE AND COMPLICATED
According to another page of the web site, “Women were sent the initial Million Women Study questionnaire with their invitation to attending the National Health Service Breast Screening Programme at participating centres across the country.
“Approximately 3 years after completing the first questionnaire, participants were sent a second “update” questionnaire to keep track of their health and to ask new questions about factors such as diet, family history, vitamins and supplements and early life experiences.”
In addition to being voluntary, self-reporting and spaced over three years, the questionnaires themselves are long, involved, require a substantial span of concentration and only peripherally address the issue of alcohol consumption.
The time- and thought-consuming nature of the questionnaires could make the ultimate sample even less representative.
SELF-ADMINISTERED QUESTIONNAIRES UNRELIABLE
One top data research expert who has served as head of research for one of the “Big Three” American television networks, looked at the Million Women Study questionnaires and told Wine Industry Insight that: “The other problem, obviously, is that it is a self-administered Q (questionnaire), which has virtually zero reliability. Not that people don’t use them, they just aren’t as reliable as one filled out by a live interview.”
ABSTAINERS EXCLUDED FROM SAMPLE
According to the paper, “Women who reported drinking no alcohol at the time they joined the study may be a biased group as some may have stopped drinking due to ill heath. Therefore, women who reported drinking some alcohol, but less than or equal to 2 drinks per week, were taken as the reference group.”
The issue of the “sick abstainer” has been well known in research circles for more than two decades and has effectively been corrected for in landmark studies in The Lancet, British Medical Journal, New England Journal of Medicine and others. The inability of the Million Women Study to comply with the established methodology leaves it without a valid “zero consumption” baseline.
NO DISCUSSION OF KEY SOCIOECONOMIC FACTORS
While the paper makes two cursory mentions of adjustments for socioeconomic factors, it devotes no discussion at all to significant factors related to breast cancer: namely income, education, body mass and onset of first pregnancy.
While the paper notes that:
“Women who drank alcohol were likely to be younger, leaner, more affluent, and to do strenuous exercise more frequently than nondrinkers.”
And “Women who drank wine exclusively also tended to be more affluent, to be leaner, or to take strenuous exercise more frequently and were less likely to be current smokers compared with other drinkers.”
Current research has consistently found that the more educated and affluent women are, the more likely they are to have no children or to have their first pregnancy at an older age. And a delayed onset of the first pregnancy is a documented risk for women’s breast cancer.
The authors’ failure to discuss how they handled this issue casts yet another shadow on the study’s conclusions.
Wine Industry Insight will continue to examine the original article.
2 Responses for “Study of UK Cancer-Moderate-Consumption Paper Shows Fundamental Flaws In Data, Sample Selection, Methodology”
Leave a Reply
RSS FEED
Subscribe to our feed
NEWS FETCH
(Subscribe)
Sponsored By:
About The Editor
WII Editor Lewis Perdue
is a former Washington, D.C. investigative reporter and has written 20 published books. He founded Wine Business Insider in 1991 followed three years later by Wine Business Monthly. Both were sold in 1997, but remain the dominant trade publications for the North American wine business. In 1997 he became a correspondent for CBS Marketwatch and a columnist for The Street.Com. Later, he worked as a columnist for the Wall Street Journal Online before founding an Internet payments company which went public. He has also been a negociant and a wine importer/distributor. Click here for more on Lew.
Thanks for continuing to bird dog this. I suspected the science was sketchy when I read the article Wednesday.
A friend of mine teaches post-grad mathematics and told me that something like 80% of studies reach erroneous conclusions due to the authors’ inability to properly apply statistical methods!
Sandy Szwarc does a pretty thorough analysis of the evidence at her site Junk Food Science. She actually breaks down the numbers.
http://junkfoodscience.blogspot.com/2009/03/in-vino-veritas-part-two.html
It was a study simply showing associations, not causations. And the actual study showed a negative result for alcohol and all cancers. Non-drinkers had 0.4% higher cancer rates than drinkers of up to 15 drinks a week. The reporting is just selective for cancers that had higher associated rates, while ignoring all the cancers that had lower associated rates.
The study actually looked at 21 cancers, but to hear PBS reporting you wouldn’t know.
“They found that drinking just one alcoholic drink per day increased the risk of breast, liver and rectal cancer. For women who also smoked, drinking increased the risk of mouth and throat cancer as well. PBS ”
Duh.