Statistical methods
Generalized Estimating Equation (GEE) models were used to analyze the data (IBM SPSS statistics version 22). The GEE model was generated using a Poisson distribution for Emax and linear for logED50. The exchangeable correlation matrix was then selected and Wald’s estimate for mean and 95% Confidence intervals generated.
The power calculation was based on previous cough challenge data, where one doubling dose of the challenge agent is generally accepted to be a clinically relevant difference in cough reflex sensitivity. Thus, with 15 participants treated, this study had approximately 90% power to detect a one doubling dose change in cough reflex sensitivity of 0.3 for lesogaberan compared with placebo, assuming a standard deviation of ± 0.29 (9).