Editor—GlaxoSmithKline's recent legal troubles resulting from not publishing negative results of clinical trials on the antidepressant paroxetine are just part of a larger problem of publication bias in modern research.1
There has been evidence that the literature that is published is more likely to be positive than chance alone might predict.2 There have even been links shown between positive trials and industry sponsorship.3 But these cases, although they are alarming, should not blind us to the general problem of negative results not being reported. Proper analysis of new medical treatments requires properly weighing the evidence for the new treatment.4
The proper solution to the under-reporting of negative results is to track all clinical trials so that we can ensure that the results of such trials are properly reported. It is important for both journals and investigators to work together to ensure that this occurs. Had such a system been in place, it would have been much more difficult for GlaxoSmithKline to conceal these results.
Competing interests: None declared.
References
- 1.Dyer O. GlaxoSmithKline faces US lawsuit over concealment of trial results. BMJ 2004;328: 1395. (12 June.) [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Felson DT, Glantz L. A surplus of positive trials: weighing biases and reconsidering equipoise. Arthritis Res Ther 2004;6: 117-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Bhandari M, Busse JW, Jackowski D, Montori VM, Schunemann H, Sprague S, et al. Association between industry funding and statistically significant pro-industry findings in medical and surgical randomized trials. CMAJ 2004;170: 477-80. [PMC free article] [PubMed] [Google Scholar]
- 4.Kleijnen J, Knipschild P. Review articles and publication bias. Arzneimittelforschung 1992;42: 587-91. [PubMed] [Google Scholar]