Study Showing Antioxidant Vitamins Increase Mortality Flawed

Reading time: 6 – 10 minutes

A study published in The Journal of the American Medical Association (JAMA) made headlines recently. The review, “Mortality in randomized trials of antioxidant supplements for primary and secondary prevention: Systematic review and meta-analysis”, assessed the effect of antioxidant supplementation on mortality in randomized primary and secondary prevention trials and concluded that beta carotene, vitamin A and vitamin E supplementation are positively correlated with death and may increase mortality.

antioxidants.jpgThe Copenhagen University Hospital research group searched electronic databases and bibliographies published prior to November 2005 and included 68 randomized trials (385 publications) with 232,606 participants. The authors have published two other articles recently using a similar methodology and demonstrated similar results [1-2].

The study, not a clinical trial itself, but a study of studies (i.e. a meta-analysis), has a number of problems, notably flawed methodology and over-analysis of the data.

Flawed methodology

According to the article, the goal of the analysis was:

… to analyze the effects of antioxidant supplements (beta carotene, vitamin A, vitamin E, vitamin C and selenium) on all-cause mortality of adults included in primary and secondary prevention trials.

However, the researchers specifically excluded 405 studies that reported no deaths during a trial period or follow-up.

In addition, 47 of the 68 trials included in the meta-analysis were secondary prevention trials conducted on subjects that were diagnosed with disease. The goal of these trials was to establish whether intervention would slow disease progression or reduce the risk of death from disease. Only 21 were primary prevention trials conducted on healthy subjects (the goal of these trials was to establish a benefit to health). Thus, not only did all studies included in the meta-analysis report deaths during a trial period or follow-up, but over two thirds of the studies were trials on people that were already diagnosed with a disease.

Over-analysis of the data

Trials were classified according to the risk of bias based on the quality of the methods used in the study, either low-bias risk (high methodological quality) or high-bias risk (low methodological quality). The abstract indicates that “randomization, blinding, and follow-up were considered markers of bias in the included trials”. Nevertheless, when all low- and high-bias risk trials of antioxidant supplements were pooled together, there was no significant association between supplement use and mortality.

Significant influences of dosage on mortality were found for beta-carotene, vitamin A, selenium and bias risk. Doses of vitamin A ranged from 1333 IU per day to 200,000 IU per day (mean value 20,219 IU), well above the upper tolerable limit. Notably, there was no increased risk for vitamin C or vitamin E in either single or combined regimen, duration of supplementation, or primary or secondary prevention. Selenium had a statistically significant protective effect by dose. However, when multiple variables were used in the meta-regression, dose of selenium for low-bias risk trials was associated with significantly higher mortality. Beta carotene and vitamin A, as well as the other antioxidants, failed to show increased risk.

On his Livejournal page, Phil explains the inverted-J-shaped response curve of vitamins. As with any medication, mortality will be observed if dosage exceeds toxic levels. In addition, he points out that all the antioxidants suggested to increase mortality (beta carotene, vitamin A and vitamin E) are fat soluble vitamins, which are stored in the body for long periods of time and can build up toxic levels when taken in high doses.

Regina over at Weight of Evidence did an extensive review of the study and summarizes the results for each antioxidant as follows:

  • Beta carotene by itself, bad; combined, nothing; exlude some study data and again it’s bad.
  • Vitamin A alone or in combination, nothing; exclude some study data, bad.
  • Vitamin E alone, in combination, in high dose – nothing; exclude some study data, it’s bad.
  • Vitamin C alone, in combination and when excluding some study data, nothing.
  • Selenium alone or in combination, nothing; data analyzed together singly or in combination, benefit; exclude some data, nothing.

The meta-analysis did not investigate causes of death, but it’s likely given that 69% of the trials analyzed were secondary prevention that the deaths occurring were principally due to previously diagnosed disease and not antioxidant supplementation. The authors also failed to analyze for potential outcome differences between primary and secondary prevention trials.

Thus, there are many problems with this meta-analysis. Unfortunately, the media has focused only on the results of the low risk bias group, which showed a 16% increase in mortality rate. Here’s the rest of the story:

  1. Intervention effect of antioxidant supplements vs placebo on mortality in trials with low risk of bias
      Antioxidants: 15,366 out of 99,095 participants (15.5%)
      Control: 9,131 out of 81,843 participants (11.1%)
      Relative to control: 15.5% – 11.1% = 4.4% (Not 16% as reported by the media)
      Relative risk: 1.05, mortality significantly increased
  2. Intervention effect of antioxidant supplements vs placebo or no intervention on mortality in trials with high risk of bias
      Antioxidants: 2,532 out of 36,940 participants (6.9%)
      Control: 1,027 out of 14,728 participants (7.0%)
      Relative to control: 6.9% – 7.0% = -0.1%
      Relative risk: 0.91, mortality significantly decreased
  3. Pooled low- and high-risk bias
      Antioxidants: 17,898 out of 136,035 (13.2%)
      Control: 10,158 out of 96,571 (10.5%)
      Relative to control: 13.2% – 10.5% = 2.7%
      Relative risk: 1.02, no significant effect on mortality

Note that there was a significant decrease in mortality in trials with high risk of bias.


The Council for Responsible Nutrition (CRN) has rebuked the meta-analysis review, noting several problems including:

  • The meta-analysis combined studies that differ vastly from each other in a number of important ways that compromise the results, including dosage, duration, study population and nutrients tested.
  • Many of the clinical trials included in the meta-analysis tested nutrients beyond those that were the focus of the article including lutein and zinc, making it difficult to appropriately evaluate the contribution of those trials to the overall meta-analysis.
  • The overwhelming majority of the clinical trials included in the meta-analysis tested for secondary prevention.
  • Many of the treatment trials had limitations, including the expectation that a simple antioxidant vitamin could be expected to overturn serious illness, such as cancer or heart disease.

Andrew Shao, vice president for science and regulatory affairs at CRN, said:

The study authors concluded that overall there was no effect of antioxidant supplements on all-cause mortality. It was only after the researchers divided the chosen clinical trials into ‘high risk bias’ and ‘low risk bias’ groups, using their own criteria, that they observed a statistically significant effect on mortality. This meta-analysis appears to be a predetermined conclusion in search of a method to support it.

On a final note, the study authors acknowledge that their results are in conflict with observational studies that have shown beneficial effects of supplemental antioxidants, even in secondary prevention trials [3-5].


  1. Bjelakovic et al. Antioxidant supplements for prevention of gastrointestinal cancers: a systematic review and meta-analysis. Lancet. 2004 Oct 2-8;364(9441):1219-28.
    View abstract
  2. Bjelakovic et al. Meta-analysis: antioxidant supplements for primary and secondary prevention of colorectal adenoma. Aliment Pharmacol Ther. 2006 Jul 15;24(2):281-91.
    View abstract
  3. Fleischauer et al. Antioxidant supplements and risk of breast cancer recurrence and breast cancer-related mortality among postmenopausal women. Nutr Cancer. 2003;46(1):15-22.
    View abstract
  4. Baron et al. Neoplastic and antineoplastic effects of beta-carotene on colorectal adenoma recurrence: results of a randomized trial. J Natl Cancer Inst. 2003 May 21;95(10):717-22.
    View abstract
  5. Whelan et al. Vitamin and calcium supplement use is associated with decreased adenoma recurrence in patients with a previous history of neoplasia. Dis Colon Rectum. 1999 Feb;42(2):212-7.
    View abstract
About the Author

Walter Jessen is a senior writer for Highlight HEALTH Media.