I work in a hospital setting and I had a researcher ask me a question about sampling for a control group. We are looking at infection rates throughout the hospital and want to test predictions based on co-morbidities. I understand we want to compare those with infections against those without but how to sample those without infections? The people with infections come from all over the hospital and they are an extremely small number compared to the entire patient population. Should I compare the infection group with hospital averages or should I do a simple random sample of a hundred or so patients? I also thought about stratifying the sample so the control group would have the same age distribution and match other demographics. Your input is much appreciated.