If I have rightly construed, you wish to conduct study from PHC & Hospitals. For proper representation, multistage sampling would be appropriate in which PHC, CHC, secondary care & tertiary care hospitals as health institutions. So, state, district, block & sector wise systematic random sampling could be considered which would represent PHC CHC & hospitals.

How many primary health facilities and hospitals do you have in your region?

As mentioned above, first consider the PHC centers, then you can take 30 centers. Actually I don't know why the WHO usually recommends the number 30 as a representative sample. So I think it is reasonable and convenient to take 30 centers from approximately 300 centers. Then you can use the systematic sampling method to choose the centers. First make a list of the centers ranked according to their geographic location, and then take the systematic sample, so all the areas would be represented in your sample. Regarding the hospitals, which are less in number, you can stratify according the the level whether secondary or tertiary care, and then take a sample, or maybe you are going to take all of them if their number is small.

I do not get a sense of what you want to know about the surveillance system. Are you looking for a qualitative assessment of a representative sample of facilities or for quantitative results about certain characteristics of the facilities in your area?

I found this formula I don't know is it possible to be used here or not?

Using the following formula: n = [[ ( z2 * p * q ) + ME2 ] / [ ME2 + z2 * p * q / N ]]*d where: n = sample size z = confidence level at 95% (1.96) ME = margin of error (15%) p = the anticipated proportion of facilities with the attribute of interest (.5) q = 1-P N = population size d = design effect

Thanks@ Ebrahim

Thanks@ Patrick

@Stephen C

It's aquantitative study