A conditional analysis for two-treatment multiple-period crossover designs with binomial or poisson outcomes and subjects who drop out.

We propose a conditional analysis for outcome data on numbers of recurrent symptoms arising in a two-treatment, multiple-period crossover trial. Conditioning on subject-specific totals removes any dependence on mean subject-specific symptom rates and permits the use of standard software to perform regression analysis to examine treatment, period, carryover and interaction effects. The addition of offsets to the regression equations allows the incorporation of data from subjects who do not complete all periods in the trial. We apply the proposed method to data from a crossover trial and discuss its advantages and disadvantages.