As you’ve probably heard/seen/read in the news lately, a recent research result showed that taking vitamin D supplements “dramatically” reduced the risk of developing cancer. Puhleeze!
But, I am already digressing.
One would expect, or at least I would expect, influential national societies to exercise a modicum of caution before making recommendations that have the power to affect the health of large populations. Alas, the Canadian Cancer Society (and, I fear, other agencies around the globe) has, on the basis of that one study, already advised people to commence taking vitamin D supplements. Again, this was done on the basis of a SINGLE clinical trial, the quality of which was far from excellent. Indeed, it was actually a rather mediocre bit of research (which probably explains why such a “stunning” piece of news wound up being published in such a middle-of-the-road journal). Please allow me to explain.
The study in question involved exclusively white, predominantly obese (average body mass index was 29, with 30 being massive obesity), rural, older (over age 55), women. Hardly representative of the general population!
The study’s end-point of cancer incidence was actually a secondary one (fracture rate was the primary one). In other words, the study was not designed with particular attention to detecting, analyzing, or even studying cancer. Indeed, as a general rule, one doesn’t get too excited about secondary end-points of studies.
To underscore the importance of, and liability in, using the secondary end-point of cancer in this study, it behooves us to consider just how cases of cancer were sought for and identified. After all, wouldn’t such an important end-point merit strict attention to such details? So, how did this study look for and identify cancer? By waiting for the subjects’ family doctors to diagnose any new cancers and then to inform the study authors! Hmmm . . . might family docs have missed some cancers? Might they have neglected to inform the authors? Might they have miscalled some noncancerous conditions as cancers? Let’s hope not, because if they did any of those things, the entire study may be invalid (a distinct possibility when you’re talking about a mere dozen or so cases of cancer in the study groups!) (but more about that below).
The main result of the study, that vitamin D supplementation reduced the incidence of cancer, was of marginal statistical significance (p < 0.03). This is even more concerning when one realizes that, literally, dozens, if not hundreds, of statistical comparisons were made in the study, i.e. the more comparisons that are made in a study, the more likely that there will be a significantly positive result strictly as a result of random chance. To allow for this, corrections need to be made – all such corrections have in common the effect of ‘raising the bar’ as to what constitutes statistical significance. As an example, if one takes the standard p < 0.05 as being positive (i.e. p of 5%), then one in twenty results can be expected to pass that statistical hurdle simply by chance. So, if the study made, say, 100 comparisons, you’d likely have about five statistically significant results (at the p = 0.05 level) simply due to chance. Hence, there is a requirement to use a more conservative p value when multiple comparisons are made. Therefore, this study should have used a p value of 0.01, or less, as its threshold for significance. It made no such allowance.
Notwithstanding the statistical naiveté shown by the authors, it is informative to consider just how many cancers occurred during this study. Thousands? Hundreds? Twenty? No. In fact, there were fewer than 20 cancers per subject group! In some groups, there were eight cancers. Eight! And on the basis of such tiny numbers, some would advocate changes affecting millions of people. The mind boggles.
OK, maybe despite the flaws I’ve mentioned (and God knows how many more flaws an expert would have picked up), the study is still “valid”. Then, we’d still want to ask how does it compare to other studies? Well, we have one such study available – the monumental Women’s Health Initiative. It studied over 36,000 women, many times more than the current study’s 1000 women. And what did the WHI find? Nothing. Nada. Nil. Zilch. Which do you think is the more robust, sounder study? Regardless, when the various studies and trials of vitamin D supplementation are pooled (i.e. meta-analysis), and one day they will be, the negative WHI results will dwarf the results of the study at hand.
OK, OK, you may say. Even with all those issues, what harm is there in taking vitamin D? For one person, probably not much (so long as they follow directions and are vigilant for other sources of vitamin D). But, for the population as a whole, the law of large numbers guarantees that there will be trouble. And it can be serious trouble. Things like calcification of what should be, and remain, “soft” tissue, high blood pressure, and even heart rhythm disturbances.
The Canadian Cancer Society (and the media in general) should be embarrassed.