• Multivitamins? Really?

    A longtime reader of the blog pointed me to this. It’s blowing my mind. From JAMA, “Multivitamins in the Prevention of Cancer in Men The Physicians’ Health Study II Randomized Controlled Trial“:

    Context  Multivitamin preparations are the most common dietary supplement, taken by at least one-third of all US adults. Observational studies have not provided evidence regarding associations of multivitamin use with total and site-specific cancer incidence or mortality.

    Objective  To determine whether long-term multivitamin supplementation decreases the risk of total and site-specific cancer events among men.

    Design, Setting, and Participants  A large-scale, randomized, double-blind, placebo-controlled trial (Physicians’ Health Study II) of 14 641 male US physicians initially aged 50 years or older (mean [SD] age, 64.3 [9.2] years), including 1312 men with a history of cancer at randomization, enrolled in a common multivitamin study that began in 1997 with treatment and follow-up through June 1, 2011.

    Intervention  Daily multivitamin or placebo.

    Main Outcome Measures  Total cancer (excluding nonmelanoma skin cancer), with prostate, colorectal, and other site-specific cancers among the secondary end points.

    Here’s the gist. They rounded up more than 14,000 doctors 50 years or older in 1997 and randomized them to get a daily multivitamin or placebo, and then they followed them through June of 2011. Otherwise, they did nothing to these participants, so there’s every reason to believe they were otherwise treated similarly. They wanted to see if the two groups developed cancer at different rates. They did.

    Men who took a daily multivitamin had a statistically significant lower rate of cancer than those who took the placebo (17.0 versus 18.3 events per 1000 person-years). Although mortality was lower as well, it wasn’t statistically significant (4.9 versus 5.6 events per 1000 person-year).

    This was an extremely large study, well done, with amazing follow-up. You can’t dismiss it easily.

    Multivitamins are cheap. You can buy them by the barrel at Costco. There are few harms or side effects. And, unlike many of the things I often roll my eyes at, they appear to really reduce the risk of cancer. Plus, you get the nutritional benefits. I’ve been lazy about taking one for a while, but I think I may reconsider.


    • Cheap and easy! Now if we could just get something to say exercise is cheap and easy and has a myriad of benefits. Oh wait…

    • A few comments:

      *Note the barely significant. hazard ratio [HR], 0.92; 95% CI, 0.86-0.998; P = .04. If there were a few (or even just one) more cancer cases in the multivitamin group would have made the result unsignificant . However 0.05 isn’t some magic number regardless, there would be a very small potential benefit either way.

      *They did a nice sensitivity analysis where they excluded prostate cancers which have 99%+ 5 year survival rate in the PSA era. They saw a similar effect among these cancers: HR, 0.88; 95% CI, 0.79-0.98; P = .02.

      *Considering cancer is a very heterogeneous disease and the HR was very modest (0.92), we should consider studies that further investigate the mechanism through other intermediate endpoints (ex: adenomas).

      *In addition, it’s worth looking at the results by the amount of fruit and vegetable intake (I didn’t see this in the paper). The largest benefit could be among those with low fruit/veggie intake (maybe there would be a stronger result (more benefit) in a lower SES population).

      *We should be careful how strong to advocate use of multivitamins based on this trial. Observational studies show a stronger reduction in many cancers with increased fruit and veggie intake (I remember 30+% reductions vs. a 8% reduction here). If individuals lowered their fruit and veggie after taking a multivitamin there could actually be increase of cancers on the population level. It’d be interesting to see a study that considers if the initiation of multivitamin use affects the rate of fruit/veggie intake especially in certain populations (non-doctors).

      *It’d also be interesting to see some cost effectiveness analyses using this data. Should insurance policies cover multivitamins based on this data? (not that insurance companies consider cost anyway)

      • I’m fine with most of your points (although the next to last one would need to be proven to me for me to consider it).

        But let’s be clear. I didn’t say everyone should go out and get MVIs immediately. Nor did I advocate that insurance should cover them.

        I did say they they have known nutritional benefits. I did say they are cheap. I did say they have relatively few harms. Add in a potentially real chance to reduce a risk for cancer, and I think I might try them again. What’s the downside?

        • Thanks for the reply. I didn’t mean to infer that you were aggressively advocating for MVIs. I thought this post was fine…I was just adding my two cents.

          I’m also considering adding a multivitamin now , but the few downsides I can think of are:
          1. Chance of wasting a modest amount of money. Even if it’s cheap taking a pill everyday for the rest of your life has to add up a bit.
          2. The chance it could decrease fruit and vegetable intake – I agree this hasn’t been tested and hopefully wouldn’t be true. However, I think it is a hypothesis worth testing.
          3. Small chance there are harms in select populations with other conditions. This study was among a homogenous population (old and probably mostly white doctors). The benefit is so small here that even a small harm could easily outweigh the chemopreventive benefit .

          Think of it this way – aspirin has some fairly strong evidence it can prevent cancer (observational studies suggestion a fairly large reduction – probably 5 or 10 times larger than what is seen here). However, we’re still not running out and recommending aspirin for everyone given it’s side effects in a select population. MVIs aren’t likely to have major side effects like aspirin given their wide use, but even a modest side effect would easily change the scales.

          • Agree, but since I was drawing conclusions for myself, a population of mostly white doctors is appropriate. 🙂

            I think there are likely far fewer downsides to a multivitamin than there are for aspirin. There have been many large studies of MVIs with few adverse effects. And there’s the nutritional benefit. They aren’t good just for the cancer prevention. That’s icing.

            But I suspect we likely largely agree, and this is quibbling over minutiae.

    • Very interesting. Omega-3, vitamin d, b-complex, and milk thistle make up my current daily regimen. Guess I’ll work a multi-vitamin back in! Like you said, few side effects, so I why not?!

    • It seems this research somewhat contradicts older research showing that vitamins can INCREASE cancer risks. In 2007, oncology researcher Dr. Larry Norton, gave a talk to the Aspen Ideas Festival discouraging vitamin supplementation: http://www.aspenideas.org/session/cure-breast-cancer-road-ahead And not to plug myself, but I cover that and some other research showing the risk of vitamins on my blog: http://danbraganca.com/2011/10/14/vitamyth/ This new study seems to contradict much of what I have previously wrote. Is this a case of new science trumping old or is it just something that needs further confirmation? I emailed Dr. Norton to get his opinion and will share if I receive a reply. Curious what your opinion is too.

      • Those are reports on studies of megadoses of individual vitamins. I am opposed to those (see my books). This was a study of a simple multivitamin, which does not carry the same risks.

        It’s like eating just enough food (good!) versus binging (bad!).

        • Hey Aaron, I do own your book and I’ll check out that section when I get a chance, but I’m still seems like there might be an apparent contradiction. The study I referenced did look at multivitamins not just mega-doses of individuals: http://archinte.jamanetwork.com/article.aspx?articleid=1105975 “In multivariable adjusted proportional hazards regression models, the use of multivitamins (hazard ratio, 1.06; 95% CI, 1.02-1.10; absolute risk increase, 2.4%)” That particular study is linked in the references of this new study, but I’m not sure how it deals with it. Maybe because one is about women and one is about men that accounts for the different results?

          Also, I’m paraphrasing, but Dr. Norton argues that most people get sufficient nutrients from a normal diet so even a multivitamin leads to dosing above the recommended amount. Of course, maybe this recent research overturns that thinking.

          • That was also a cohort study. All kinds of biases, and there’s no way that it can test for causation.

            Now I’m not dismissing cohort studies out of hand. But the results of the study you link to should make one pause and say, “There appears to be an association between MVI use and mortality. Maybe we should test that with an RCT.”

            They did one. It’s the study I talked about. It showed that MVIs reduce mortality. This was huge RCT with years-long follow-up.

    • What multivitamin you would recommend 🙂

    • Bunches of stuff on the benefits of vitamin D. And of course with bones in post-menopausal females we Rx a lot of mega D, 50K U/Wk and monitor their levels.

      Mega D is the only supplement I take.


    • And then there is this:

      Conclusions Treatment with beta carotene, vitamin A, and vitamin E may increase mortality. The potential roles of vitamin C and selenium on mortality need further study.

      Though I think a multivitamin is harmless, I think I will wait for further confirmation before shout this study from the house tops.

      • Those were studies of mega-doses of individual vitamins, which I agree can be harmful.

        Dismissing this study because too much was bad for you would be like not eating because over-eating is unhealthy. Or like refusing the correct dose of a drug because an overdose is bad for you.

        I know those are extreme examples, but the studies you link to are not the same as this one.