Differentiating the efficacy of 5-HT(1B/1D) agonists.
OBJECTIVE: To examine, for a set of published clinical trials of serotonin (5-HT(1B/1D)) agonists as acute treatments for migraine, whether transformation of efficacy data into therapeutic gain (TG) or number needed to treat (NNT) is useful. BACKGROUND: Pivotal clinical trials of 5-HT(1B/1D) agonists in migraine use a primary end point of change in pain score from 3 or 2 to 1 or 0. Placebo response rates among such studies are variable. Meta-analytic comparisons of 5-HT(1B/1D) agonists often employ TG and NNT as efficacy measures. METHODS: Data from US product labeling or published sources were converted into TG (TG = active response rate [%] - placebo response rate [%]) and NNT (NNT = 1/TG). Pivotal clinical trial data were compared before and after transformation. RESULTS: Therapeutic gain ranged from 17.5% to 51%. The transformation of TG into NNT yielded no clinically significant difference in efficacy estimate for the range of 17.5% to 47% (N = 29 clinical trials). However, NNT and TG had a nonlinear relationship for some secondary end points. When the relationship between the standard primary and secondary end points was compared, the correlation of TG with clinical disability (Pearson coefficient R = 0.93) was stronger than for NNT. Placebo response rates correlated more strongly with NNT (R = 0.66) than active response rates (R = 0.42; N = 29 clinical trials), although both TG and NNT were sensitive to placebo response rate. CONCLUSIONS: Transforming efficacy rates into TG or NNT adds no new information to placebo-controlled trials. The variables, TG and NNT, should not be used to compare members of this class of drugs. Migraine therapies can only be compared using well-designed head-to-head studies and not by meta-analysis. Broader measures of efficacy should be used to describe and compare 5-HT(1B/1D) efficacy.