You’ve heard the news reports: Vitamins are a waste of money. Heck, this story even says that vitamins have been dealt a “death blow”! Clearly, this means that I’m going to stop taking my vitamins, right? Of course, if there’s anything we’ve learned from news stories about studies, it’s that the stories are almost always full of crap. Okay, so most news stories about studies are bad, but that doesn’t mean we should just dismiss this out of hand. As with all things, we must look at the studies.
Now, I’m not going to go through every vitamin study ever published and point to issues here or there or where good science happened or whatever. Instead, I’ll be looking at the studies that supposedly deal a “death blow” to vitamins. Thankfully, there are still some news outlets that actually source their information. Unfortunately, the AoIM is hiding two of these studies behind a paywall, so I can’t talk about anything that isn’t discussed in the abstracts. One day, maybe I’ll be able to subscribe to journals for a single Monday Musings entry, but right now, that paywall is just too daunting. Okay, it would only cost $15 for this issue, but that’s still too rich for my blood.
Anyway, let’s get down to business. Let’s deal first and foremost with the study on which I’m sure everyone wants to focus. This particular one studied what happened when they gave multivitamins to patients who had suffered a miocardial infarction. Starting point: Miracles of miracles, but this is actually a clinical trial. I’m shocked, news outlets. Then again, I’m not that shocked, since it’s pretty close to impossible to show negative results for vitamin usage in observational studies (thanks to the effects of health consciousness, an effect shared by vegetarian diets). Now if we could only get the news outlets to report clinical trials in regards to vegetarian, low carb, and low fat diets… Anyways, that’s aside from the point. The point is that the study design is actually good. At least, as far as I can tell from the abstract. There is a problem with the study, however: The drop-out and non-compliance percentages are huge. This, too, was reported by the news outlets. They’ve reported all the facts, haven’t they?
Well… no. See, there’s one very important number that they didn’t report: 79%. What do I mean by that? Well, a (very) inaccurate way of explaining CI is this: When the relative risk’s 95% CI falls to the point where the larger number is less than 1, there’s a 95% chance that what they did actually did something positive. But what happens if, say, the larger number is 1.01? Do we just toss out something that shows a 90% chance of effectiveness? Oh, you say, but this didn’t end at 1.01! In fact, the 95% CI was .75 - 1.07. Clearly, this debunks any chance at all that the vitamins were effective, right?
Oh, you silly boy… or girl. Or intersexed/androgyne/genderqueer/etc.
See, the p value here represents the probability that the null hypothesis (the null hypothesis in this case being that vitamins don’t do anything) is true (see above about 95% CI). Now, I did my own math… with a calculator, and the p value they present seems a little… off. Not being a professional study writer, however, I can only assume that this is the standard way these are written or else the distribution is a little wonky. Either way, though, I’m going to go ahead and use their p value just because .21 works just as well for my point as .10 - .15 (depending on what standard deviation I used to describe that CI). What does that .21 mean? Well, it means that there was a 21% chance that the multivitamins either did nothing or were harmful… in comparison to a 79% chance that they were helpful. In the world of Texas Hold ‘Em, this is like holding a pair of kings against a pair of aces pre-flop.
And that 21% chance, to those editorialists and journalists, trumps the 79% chance that the opposite is true. I’m sorry, but I think I’m justified in calling bias here. Oh, don’t get me wrong; I’m not saying that this study proves beyond a shadow of a doubt that multivitamins help all-cause mortality or hospitalization due to the listed effects in the study in individuals that have previously experienced a myocardial infarction. Nonono. But to come to the opposite conclusion from these results is absurd at best. You want to know what these results should say? Take multivitamins for now, but we’ll keep doing research on them to find an answer for sure. Unfortunately, medicine doesn’t work like that because screw probability, we have bias.
This, by the way, is why I’m a firm proponent of the default null hypothesis. The default null hypothesis is this: Instead of the null hypothesis being that X doesn’t do anything, have it be that we don’t know if X does anything until it’s studied, at which point the null hypothesis favors probability, with a strength depending on how strongly the probably favors that side.
I don’t know about you, but I’ll play my aces against their kings any day of the week.
Okay, but that’s not the only study. In fact, the next one isn’t actually a study, but a systematic review. That’s one where they take a bunch of studies and re-figure the statistics, which means that, say, 5 studies that found no statistically significant results could end up in a systematic review where, all together, they do produce results. These are a heck of a lot better than individual studies. We may actually have something that provides conclusive results. Right?
…In some ways yes, but in some ways… no. First of all, the majority of studies reviewed dealt with specific vitamins, not multivitamins, and at least one in particular (vitamin E) is already known to be harmful when supplemented at high doses (note that that article was published two and a half years ago).
There were some studies they reviewed that studied multivitamin usage, however, and, in spite of most of the multivitamins actually only having a few vitamins or minerals involved… we find the same thing there as with the previous study: The probability is just shy of 95% that multivitamins had a positive effect on all-cause mortality. On cancer, it was actually within that 95% CI for male patients and the reviewers noted this. And yet, it spite of this, the advice the editorialists and journalists give is still to not take vitamins.
As for the last study that’s mentioned in these articles, I’m not going to try to debunk it or show probability issues. Truth is… I don’t think I can. Even if I knew what SU is (full disclosure: I don’t), the results look pretty reasonable to me in the claim that mulitvitamins don’t do anything to aid in cognitive function… in those studied. See, here’s a major issue with vitamin studies: Most vitamins are taken in the morning, which is also when most people have breakfast… breakfasts that often consist of fortified cereal (note that, as someone eating a whole food diet, that doesn’t apply to me). What’s the significance of this? Well, B12 is the vitamin in most multivitamins that arguably has the largest effect on cognitive function, and thanks to intrinsic factor (seriously, WTF is with that name?), your body can only effectively absorb 17-33% of your RDA of vitamin B12 in one sitting (of course, I imagine that the RDA is about twice what your body needs in a day to reflect the poor absorption even with intrinsic factor). This means that, if you’re eating fortified cereal with 100% of your RDA of vitamin B12 per serving and have perfect absorption, measuring abilities, and a completely homogenous cereal (yay physics!), you’re actually getting around 1.18 mcg of B12. Add another 100% on top? You’re now getting 1.26 mcg. Whoopdie friggin’ doo. If, on the other hand, you’re only getting a small amount of B12 in the morning, that extra B12 might help. We can’t say because studies of this nature are severely limited in this way. This, by the way, doesn’t take into account the absorbability of the supplements or the fortified cereal.
Now, that’s not to say that any old multivitamin will do. I take a multivitamin that I chose because of its price and composition. As many of you may know, my mother and I share our meals, and thus also food. On her latest trip to the store, I asked her to pick up another bottle of my multivitamins, which her store apparently does not sell. She instead got a 60 capsule bottle of two-a-day vitamins (one month’s supply) for $8. I immediately looked at the label and saw the bad news. These vitamins had 50 mcg of B12 (833% RDA). Remember that talk we had above? Perfect absorption would mean 1.6 mcg over a multivitamin with 100%, which would be 1.18. Vitamin C, which is water soluble and thus large amounts become expensive urine, is in these at a rate of 300 mg (500% RDA). But it gets worse. Remember that talk we had about excess vitamin E? Yeah. These have rather high amounts, as well as other vitamins that shouldn’t be taken in high amounts. And all this so they can charge people that think that more is better $8/month. Yeah, I went ahead and got my own brand when I went to the store.
My own brand, by the way, costs $6.49, is one/day, contains a reasonable amount of vitamins and minerals (you won’t find, nor would I want, 100% of every vitamin and mineral in a supplement), and comes with 250 capsules. That’s 2.6¢/day. Look at all that money that I’m wasting.
You might not want to take multivitamins for one reason or another. That’s fine. But for 2.6¢/day, I’m going to play my aces against their kings. Every single day.