“I know”, said A. “Let’s do a parallel group design”.  “No, no, no. Not a subject consuming parallel group design!”  B cried out in a high voice “This is obviously a text book example of when a cross-over design should be used”. C looked at them both in disbelief. Did they not follow the latest developments? Did they not receive the weekly e-issue of “Stats for You and Me”? “Well”, he said feeling quite pleased with himself, “I suggest an adaptive design”. He did not really know what it meant, but he knew it was the new in-word.  “You what?” said A and B in chorus. “What do you mean?” C did not feel quite as comfortable any longer and wished his friend and statistician, Ritch Monde was there. “Well”, he said again and took a deep breath.

“We can start with a parallel group design, and then we may cross over to a factorial design if the first 30 subjects included show a lower between subject variability than the next 20 subjects we may or may not include. That of course depends on the 92.5% confidence interval which we will adjust according to the Brown-Darling method, and how many p-values we want to present”. “To be absolutely certain we should carry out a couple of regression models analyses, and perhaps top up with an analysis of variance. Just in case.” “What do you reckon?” “Hello, where is everybody gone?”

Do you feel confused and not certain how to design and analyse your clinical trial? Our senior statisticians have many years of experience from a wide range of therapeutic areas, statistical designs and models. Nothing is too simple or too complex.

Challenge us!