A response to the commentaries: look away now or face up to the profound problem of publication bias in drug education research

摘要

Publication bias is known to be a profound problem in the wider health sciences. Another recent example occurred in the relatively highly regulated sphere of drug company trials in the United States, where deposition of the raw data is a necessary part of getting a drug to market. According to the published literature, 94% of the trials of antidepressant drugs in the treatment of depression were positive, yet when the raw data were examined only 51 % of trials were found to be so . The effect on subsequent meta analyses was to increase the apparent effect size by 32%. This study follows on from previous controversy about the effect of publication bias on estimates of the efficacy of these drugs in children .