Power Study: Planning a Power Analysis for a Genetic Case-Control Study, Assignments of Community Health

Download Power Study: Planning a Power Analysis for a Genetic Case-Control Study and more Assignments Community Health in PDF only on Docsity! Power Study: Some more discussions on power Question: How do I plan a power study for my case-control genetic study design? What do I usually vary? What can I fix? Answer: For each type of factor/parameter, the following comments deals with how to vary them and under what conditions to vary them: Parameter/factor to vary Comment Type I error Usually you have a liberal type I error of 5% and then adjusted type I error of 5%/n, where n is the number of analyses for a conservative Bonferroni adjustment. In many cases, you will have to specify how you will adjust for multiple testing in your grant, especially if you are implementing a genome-wide analysis. N (number of cases) Unless you already have your sample size fixed, it is good to vary this number from the worst case sampling scenario to the optimal one. Control : Case ratio Usually given your recruitment and matching case-control design you can fix this ratio. High risk allele frequency (A) You will have to vary this unknown value from common to rare (i.e., from 0.50 to 0.01) Prevalence Usually you can fix this value if you can estimate it from a pilot study or previous study. Genotype relative risk of Aa You will have to vary this information depending on the size of the effect (usually a moderate size that might have been estimated from previous studies) and the inheritance model you are imposing (additive, dominant, recessive, and co-dominant). Genotype relative risk of AA You will have to vary this information depending on the size of the effect (usually a moderate size that might have been estimated from previous studies) and the inheritance model you are imposing (additive, dominant, recessive, and co-dominant). D-prime Usually this is set to 1 (to be liberal) or 0.80 (to be conservative). Marker allele frequency (B) You will want to vary this value dependent on the marker panel you use. 1