reference interval determination

Hi all,

I was wondering if I could ask for some help from some more statistically inclined persons than myself.

We are trying to determine a reference interval for a new machine that measures a particular protein in the blood of horses.
This machine has never been used before in horses and no current literature exists on the subject in this species (and so no population data is available).

What statistical justification do I need to determine a reference range?
I have read in some papers that 120 subjects are required (, however other people have told me that I can use a small pilot study (e.g. 5 subjects) to determine how many total subjects I will need.

Would anyone please be able to gently walk me through how to determine how many subjects I will need?

Hope everyone is having a good start to the weekend.


Less is more. Stay pure. Stay poor.
Is there a gold standard you are comparing to or a disease state the values are predictive of? Or are you just try to establish the normal range, say a 2 standard deviation spread?