I am studying the role of a specific type of smokeless tobacco in the development of oral cancer, for my doctoral project, so originally we calculated a sample based on the exposure to this specific tobacco product among the cases and controls, however we are also planning to write a paper and see the effect of other risk factors such as diet, smoking, sunlight exposure and SES. Is it acceptable that we go ahead with the same sample size or do we have to revise the sample size.
If your primary hypothesis relates to exposure to smokeless tobacco, and the other exposures are secondary, then you can get away with the original power calculation, on the understanding that you may be underpowered for some of the secondary exposures. However, if you want to assess the odds ratio for smokeless tobacco after adjusting for sunlight etc, you should really take this into account and redo your power calculation. Many of the sample size software programs, eg nQuery and PASS have procedures for multiple logistic regression. These are expensive, so your best bet is to find a friendly biostatistician who can do the calculation for you.
Adrian
Dear Khan
In order to calculate sample size of the case-control study, you should know the prevalence of the disease (oral cancer) in the study population, then choosing the Confidence level, yo will be able to calculate it by online calculators like suggested by Ishag Adam.
Hope it helps.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling