Fillers POEM

Antioxidants don't prevent GI cancers, but increase overall mortality

BMJ 2004; 329 doi: (Published 09 December 2004) Cite this as: BMJ 2004;329:0-f

Question: Do antioxidants prevent gastrointestinal cancers?

Synopsis: This Cochrane review follows their usual rigorous methods of searching, identification of unpublished data, and data extraction. The authors included all trials that randomised participants to supplementation with antioxidants (β carotene; vitamins A, C, and E; and selenium, as different combinations or separately) versus placebo, and that reported the incidence of gastrointestinal cancers. They assessed the methodological quality of trials and calculated whether the findings were consistent across trials. A total of 14 randomised controlled trials with 170 525 patients were evaluated. The number of patients in each trial ranged from 226 to nearly 40 000. Half the studies of cancer incidence were of good quality; seven of the nine that also reported mortality were of good quality. None of the supplements protected against oesophageal cancer, gastric cancer, colorectal cancer, or pancreatic cancer. In the high quality studies, antioxidants increased overall mortality (8.0% v 6.6%). This translates to a number needed to treat to harm of 69 for one additional death (95% confidence interval 58 to 85). Four trials of selenium (three with unclear or poor methodology) reduced the incidence of gastrointestinal cancer (odds ratio 0.49; 0.36 to 0.67). Selenium should be evaluated in randomised trials with sound methods.

Bottom line: Antioxidants do not prevent gastrointestinal cancers. In fact, in pooled results of high quality studies, antioxidants increased overall mortality.

Level of evidence: 1a (see systematic reviews (with homogeneity) of randomised controlled trials


  • Bjelakovic G, Nikolova D, Simonetti RG, Gluud C. Antioxidant supplements for prevention of gastrointestinal cancers: a systematic review and meta-analysis. Lancet 2004;364: 1219-28. ©infoPOEMs 1992-2003

  • * Patient-Oriented Evidence that Matters. See editorial (BMJ 2002;325: 983)

View Abstract

Log in

Log in through your institution


* For online subscription