Hi There,
Would appreciate any suggestions with this problem.
I am designing a study to investigate the impact of a transient exposure (infection) on the incidence of a binary outcome. The exposure period will be 28days from the infection-related healthcare consultation. The plan is to compare the incidence of the outcome during unexposed periods versus exposed periods, within individuals, using a self-controlled case series design, adjusting for time-variant confounders.
The sample size is fixed based on the number of cases we can identify in the data-source. What is the most appropriate way to estimate the minimum difference we could detect with 80% power at the alpha=0.05 level, for this study design.
Thanks
Harry