Should you take your vitamins or not?

Earlier this month, an article appeared in the New York Times entitled Don't Take Your Vitamins. The article described a number of recent studies which have found that vitamin supplementation may be harmful, rather than helpful. For example, as the article reports, a 2005 Cochrane review of randomised-controlled trials documented that regular supplementation with vitamins A, C and E may increase the risk of mortality. This finding should obviously be qualified by saying that the direction of the effect of supplementation on mortality will depend on dosage, frequency of use, and overall nutritional status. For example, provision of vitamin A supplements can be invaluable for preventing child deaths in developing countries. And, as I've argued before, there's really no such thing as an unhealthy food (or vitamin).

Furthermore, my impression is that most people who supplement with vitamins in developed countries do not to take vitamin A or vitamin C individually, but instead take a multivitamin. Compared to individual-vitamin supplements, these contain much smaller quantities of a much larger number of micro-nutrients. So what does the evidence say as far as multivitamins are concerned? Might taking them regularly also be dangerous for people in developed countries?

In a paper published this year in The American Journal of Clinical Nutrition, Macpherson et al. set-out to answer the above question. They carried out a meta-analysis of 21 randomised-controlled trials, each of which had looked at the impact of multivitamin supplementation on one or measures of mortality. Their total sample comprised >91,000 people. Overall, they found no effect; the risk of mortality among those taking multivitamins was 98% of the risk among those taking placebos, and the 95% confidence interval around the estimate included 1. They also considered mortality from cancer and mortality from heart diesase seperately, and--again--found no effects. Their conclusion is that "the level of alarm" generated by "highly publicised reports from several recent epidemiological studies" may be unwarranted.