Here is a good illustration why newspapers desperately need proper science reporters:
Women who regularly take multi-vitamin pills face a much higher risk of breast cancer, a study has found.
The Swedish study, which looked at more than 35,000 women aged between 49 and 83 during a 10-year period, found that women who take daily multi-vitamin pills are nearly 20 per cent more likely to develop breast cancer.
Let’s try a slightly more truthful headline “Somewhat flakey cohort study suggests there may be a vague link between breast cancer and multi-vitamins supplementation but authors would like to do a lot more research before saying anything as inflammatory as the above headline”.
It is probably just as well that the MacDoctor runs a blog rather than a newspaper.
This article is not completely contentless, so it can’t join my spam collection. On the other hand, the redoubtable Mr Lincoln Tan manages to misunderstand almost every aspect of the study – partially assisted by the study’s authors who say:
“These results suggest multi-vitamin use is associated with an increased risk of breast cancer. This is of concern and merits further investigation,” researchers told the American Journal of Clinical Nutrition.
Medical authors should know better than to couch their abstract in such alarmist terms. Sometimes it seems to me that the term “peer-reviewed” cannot possibly mean that someone with a modicum of sense read the paper. While I know that getting funding for further research is difficult, I do think that scaring women witless on fairly slim evidence is not particularly ethical. It also allows journalists to be “authoritatively” alarming, even though the facts of the study do not support the conclusions.
Firstly, the overall risk of women developing breast cancer is fairly low (even though it is a relatively common cancer) so a 20% increase in overall risk is not “much higher” , just slightly elevated. In addition, during their attempts to identify and account for confounding factors (other things that might cause breast cancer) they identified so many possible ones that their eventual 95% confidence limits are so wide they virtually encompass the null hypothesis. Translated into non-statistics jargon: There were so many other possible contributing factors to developing breast cancer that they were not able to tell for certain if multivitamins made a significant contribution.
But the largest failure of the study was that vitamin usage was determined by questionnaire virtually on a yes/no basis. Any multivitamin use above 1 tablet a week for a year was counted. In Sweden the choice of multivitamins is limited and most carry 300-400mcg of folate (a reasonably solid dose). Unfortunately, the study also established that women who took multivitamins were 4 times more likely to take folate supplements as well. Readers will recall the controversy over the fortification of bread with folate. There is some evidence that excessive folate supplementation may increase the incidence of several cancers. Failure to eliminate excessive folate supplementation is therefore a fatal error in this study, as it may well be a major confounding factor (especially given the wide margin of error).
While I would certainly agree that the findings in this study would suggest further research is needed, I tend to view cohort studies like this with some suspicion. There are dozens of similar studies in the US that say the exact opposite of this, including the famous Nurses’ Health Study. One has to take this sort of information in context and wait for more detailed prospective studies that can address confounding factors without guesswork and creative statistics. I would certainly not recommend that all women stop taking vitamin supplements simply on the grounds of a single study of this nature.