Question 5: How do the results of the study compare with the sample size calculations?
The percentage developing disease in the control group is considerably less than the postulated 30%. This is why they increased the planned sample size to maintain the power. The proportion in the treated group is similar to what the study was designed to detect. The confidence interval for the difference just includes the 15 percentage point difference which the study was designed to have 80% power to detect and just excludes zero. Either the population difference is smaller than they planned for or they were unlucky and their sample was one with a difference considerably smaller than they planned for. They were lucky, of course, in that the difference was just statistically significant and they got into the Lancet.
Back to Exercise: estimation of sample size.
This page maintained by Martin Bland.
Last updated: 3 February, 2020.