Migraine trials: crossover or parallel group?

This paper evaluates the relative merits of the crossover design and the parallel group design for the evaluation of migraine therapy, focussing on prophylaxis. Three issues are highlighted which lead to a preference for the parallel group trial: changes in the treatment effect with time, the possibility of carry-over effects, and the handling of withdrawals and protocol deviations. These have to be weighed against the greater power of the crossover design. It is shown that the difference in power is relatively small when baseline or run-in assessments are available. It is therefore concluded that the parallel group trial is the design of choice for prophylaxis, and that it has a place in the evaluation of acute therapy.