How to calculate sample size for a disease-specific survey?

If you want to find out the quality of life of diabetes patients, you would look for a large sample of diabetes patients, not for people who might or might not be prone to developing diabetes. You would also have to distinguish between diabetes I and II and between early and late onset of the affliction.

First you determine how you are going to measure quality of life. For example you are going to use a survey and generate a score from 1 to 100. Next, you need a reference set of people to compare to. Once you score them, you can see what a typical distribution looks like. If everyone scores a 50, you need a small sample size to see a difference. If the scores range from 20 to 80, you will need a big sample size. Once you see that, you need to determine what you want your margin of error to be. If you want a margin of error of one point, you will need more samples than if you want a margin of error of five points.