Sample size determination in case-control studies: The influence of the distribution of exposure