Although officially it shouldn't, I do think convenience is a factor. After all, you will need to get ethical approval from the relevant board and also be able to access participants.
It is hard to comment on other factors without knowing more about your subject. Obviously generalisability would be something to think about but that is not a straight forward issue in qualitative research (some people say this is not achievable, others claim that it is). It might be worth thinking about whether you want to sample similar practices (e.g. all rural small practices, district hospitals) or a wide range of practices (inner city huge practices, large teaching hospitals, tertiary care centres in addition to the above rural examples). I would imagine that partly depends on the qualitative methodology you wish to use.
I think you need to consider exactly what your research question is, what analytical methodology you want to use and maybe this will help guide your sampling strategy.