I am doing an analysis on the prevalence of a drug resistance mutation in a particular country and how prevalence changed over time. Samples were collected from four sites at five different time points (a few location/time combinations are missing) and each collection is about 35-50 samples. Each sample was analyzed to see whether it had the drug resistance mutation; the outcome is binary (yes or no).

In all four of the sites, the prevalence of the mutation dropped over time, but the rate of decline seems to vary between the sites. I am interested in whether the rate of decline is correlated with another variable, frequency of exposure in each of the sites. Unfortunately, we don't have precise estimates of the exposure in each of these sites; all we have is a ranking (highest>high>moderate>low or something like that). So, the question is whether decline was more rapid with increasing exposure.

I have background in probability theory, but not in regression analysis, so I am at a bit of a loss for how to do this, at least without reducing each site to a single measurement of prevalence, which discards a lot of information. I do all my analysis in R.

Thanks in advance for any help you can provide.