Significance levels for use in the analysis of combination drug trials

Proof of efficacy of fixed combination drugs under current guidelines requires that the combination be statistically significantly superior to each component and that, to insure model validity, each component be statistically significantly superior to a placebo. Simulations indicate that if each of these four tests is performed at a significance level of 0,05 then the actual test of effectiveness of the fixed combination drug is done at a significance level that can be as small as 0.0001. This would seem to be excessively conservative. Further simulations indicate that significance levels of approximately 0,10 to 0.20 should be employed for all tests except in unusual circumstances.