Is enough enough?

3/31/2014 01:22:00 PM
Strictly speaking, a cost analysis of vitamins should take into account that some of them taste like candy, and people pay good money for candy. But, alas, let's just restrict our attention to the health effects.
You may have heard of a recent review in the Annals of Internal Medicine titled "Enough Is Enough: Stop Wasting Money on Vitamin and Mineral Supplements" which argues that all the evidence says vitamin supplements are worthless, at least for people not diagnosed with a particular deficiency. The most important article reviewed is this one, which is a meta-analysis of trials on vitamin supplements. But do the results actually show that a daily multivitamin regimen is a total waste of money? No!

Figure 1 in the paper summarizes the studies of the impact of multivitamins on all-cause mortality. First, I want to say that the cited studies are weaker than I imagined--I was looking for RCTs of a plain over-the-counter multivitamin like Centrum, but that's not what the studies cited under "multivitamin" actually are. So, while these studies are relevant to certain clinical considerations, I'm not sure they've actually even examined a question that's relevant for the general public, but whatever, let's role with it. In the combined meta-results, they estimate a 95 percent confidence interval for the relative risk ratio of [0.89, 1.01], which is then presented not as inconclusive, but as conclusive evidence that vitamins are a waste of money. For clarity, a relative risk less than 1 means that they reduced the mortality rate, while greater than 1 means it increased mortality.

But what basis do they have in arguing that this means vitamins are a waste of money? Let's do the cost-benefit calculus. The internet tells me that 200 pills of Centrum costs $13.99. Taken once a day for 40 years (roughly the number of years remaining in a so-called "statistical life"), that's a lifetime present discounted cost of $1,021.27. A reasonably low estimate of the value of a statistical life taken from the VSL literature is $3,000,000. That implies that multivitamins are cost-effective if they decrease the mortality rate by at least 0.34 per 1,000. The mortality rate in the meta-analysis was 113.7 per 1,000, so that implies that vitamins are cost-effective at a relative risk of less than or equal to 0.997, which falls well within the estimated confidence interval.

What all this says is that the vast majority of vitamin users aren't going to see any benefits from their vitamin use, but that the insurance value of a daily multivitamin may well exceed it's tiny cost.

I'd also add that contrary to the impression you'd get reading the multivitamin literature, a daily multivitamin regimen is actually extremely cheap. If we can't at least acknowledge that, then we simply aren't having a reasonable discussion.