Maxwell C
Br J Clin Pharmacol. 1981 Jan;11(1):15-8. doi: 10.1111/j.1365-2125.1981.tb01095.x.
1 Eighteen our of twenty non-significant placebo controlled trials are shown to have trends favouring imipramine and are consistent with a further ten significant at 0.05. 2 The problems of falsely positive and falsely negative results are discussed. A plea is made to medical journal editors to publish 'negative' or null difference trials by title only; rather than not at all.