Supplements, Yea or Nay?


Today’s Preamble

First off, please visit the Canadian Grand Challenges website and vote for the project proposed by my friend and colleague Dr Ophira Ginsburg.  Go.  Now!

Second, there’s a new article over at, this time a review of the TV show Sanctuary.

Third, I would like to express my gratitude to my former student Jill, who nominated Deonandia for a “media excellence” award from the University of Ottawa President’s office.  We didn’t win.  And though I’m not receiving formal recognition by the University at a reception tomorrow night, I will be attending, and have at least been acknowledged by the jury.  Details are here.  Here’s the text of the email from President Allan Rock:

Dear Professor Deonandan,

Please accept my congratulations on being recognized for a special mention by the jury in the recent 2011 Excellence in Media Relations Award. We received many nominations this year and although your accomplishments were very impressive, committee members ultimately selected professors Pascale Fournier and Norman O’Reilly.

Your research and writing in the areas of epidemiology and global health are helping to facilitate a better understanding of how these issues impact on the world’s population. Your unique talent in integrating literature, art, ethics, law and social policy into the study of global disease burden has had a powerful impact on the field. Thank you for sharing your insight with the media, which helped to inform and educate the broader community, and will no doubt continue to raise awareness and eloquently advocate for change in the realm of global health.

Congratulations once again on this well-deserved recognition. It is because of exceptional scholars and researchers such as you that the University of Ottawa is highly regarded among Canadian universities.


Allan Rock

The concerning part is, of course, that now this blog is on the radar of people who employ me.  Given some questionable past content, this might not be such a great thing.  On the other hand, why stop now?  I do, however, feel the need to write more about epidemiology.  So now we move on to today’s real topic…


Are Supplements Good For You?

If you’re on Facebook (and, really, who isn’t these days?) someone has probably sent you a link to this article: “No evidence for benefits of most vitamin supplements.”  The article summarizes a study by the University of Minnesota, where researchers surveyed 39000 women about their use of multivitamins, vitamins A, C, D and E as well as beta-carotene, B vitamins and minerals such as calcium, copper, magnesium, selenium and zinc.  It’s unclear how health outcomes were assessed, since the original study is not well described by the journalist.  But I assume health status was assessed via the questionnaire.

The article suggests that the findings of the study fail to show a link between supplement use and the avoidance of death from ailments that are meant to be treated by those very supplements.  Many in the media (and by media, I include Facebook) are citing this study –or more precisely, this article about the study– as evidence that the supplement industry is not only without benefit, but possibly deliberately duplicitous, as well.

Now, I don’t know what the intentions of supplement manufacturers is.  And I don’t really know if supplementation is good or bad for you.  All I really know is epidemiology: in other words, how to assess whether a study is really able to conclude the things that it concludes.

So let’s break it down.  First, I have not read the original study.  I, like everyone who is citing this report, have only read the news article.  So any misrepresentations I may make of the study are due to the linked media article having reported such misrepresentations.

As already noted, it is unclear how health outcomes were assessed in this study.  If indeed it was via questionnaires, then how could death rate have been ascertained?  Dead people don’t do surveys.   Additionally, if other health states were also assessed via survey, then something called “recall bias” is at play.  For example, if some asked you, “how many head colds have you had in the past 12 months?”, could you recall enough of your suffering to answer accurately?  And yet that is typically how a lot of this kind of research is done.

What’s more likely is that the Iowa Women’s Health Study, which is what this study was a part of, ascertained real health outcomes from direct observation of patients, and measured associations with those real outcomes with self-reported supplement usage that was ascertained via questionnaire.  In that case, death rates could have truly been measured, and the recall bias would have been in the ways subjects remembered their supplement usage.

The article presents no information about the selection criteria for subjects entering the study.  For all we know, only women with serious underlying health issues were enrolled.  In such a case, higher rates of mortality and morbidity, relative to the general population, would be expected.  Additionally, women who are ill may be more likely to seek supplementation.  Thus, their underlying health states constitute what we call a “confounder”, which is a factor that affects both the exposure (tendency to seek supplements) and the coutome (mortality and morbidity rates).  The existence of a confounder –unless it is controlled either methodologically or statistically– renders all conclusions of a study moot.

Again, I am not saying that confounding was in play, only that the study, as described by the news piece, does not discount its possibility.

Lastly, the article makes it a point to insist that supplementation does not decrease death rates for deaths caused by ailments that are supposed to be addressed by supplements.  But death is a fairly extreme outcome.  Most people don’t take vitamins to epxlicitly avoid death, but rather to avoid illness.  A study that fails to detect an association with supplementation and reduced death rate, is also likely to be insufficiently sensitive to detect an association between supplementation and reduced illness rate.

So what to conclude from all this?  Well, first that the National Post sucks at describing scientific studies.  Second, that the results of the cited study, as described by the National Post, do not necessarily suggest that supplementation is useless, but merely that in the case of women who may or may not have serious predisposing conditions, supplementation with these selected nutrients does not prevent death.