I am intending to do some work on fasting serum insulin...my aim is to establish its reference range among people of my locality& I would be happy if i can get some useful advises about the sample size of my proposed work.
The general advice is to use 40 persons for an estimate and 120 persons for a more accurate establishment of a reference range (Westgard - Basic Method Validation) unless there is a gender effect.
The most difficult phase is to select exclusion criteria. Consider posing question like: How can I be shure that they are fasting. How can I be sure that there is no diabetes in the normal group. How to avoid physical or other stress that might influence the insulin level.
And last but not least, is your method well validated including sampling and sample storage/transport conditions.
I recommend to you to see the MSc thesis of Dr. Luay Al-Helaly in the Dept of Chemistry, College of Science, University of Mosul...as I remember he was estimated the reference values of certain biochemical parameters in Mosul city in starvation and normal...I think you will be interest with his work...
Sample size will be drawn from the population of your locality you intend to set a reference range for and decide a margin of error 5% or 10% easily through online sample size calculator . If you are talking about a Union Counsel, Tehsil or district or division or province.
Few suggestions I will add for you
1. Is there any need to set a reference range and why if international ranges are available ?
2. Condition that brings flaws in study as mentioned by Jos P M Wielders plus insulin resistance, and certain drugs do affect some conditions increase the production of insulin like insulinoma.
3. Insulin production is different from insulin levels in plasma because insulin undergoes hepatic and renal extraction so plasma levels are not accurate
4. Measuring the levels of plasma C-Peptide is a better index for insulin levels because it is produced along with insulin from beta cells of the pancreas attached to insulin
The problem of working with a cohort is that you can’t never be sure about how many people are going to be LFU. Usually the retention of individuals in a study depends on the site experience and a good case management. So, if you need an n=40, you may want to include extra cases for your ITT.