Hi! I am considering a research project that would seek the answer the following question: Is someone who is over weight and looses weight then regains it more or less likely to develop diabetes in the future. This would be a retrospective study. I can pull reports on patient’s weight and hba1c (measure of blood sugar) from our EHR. It would be nice if I could boil an individuals BMI over time down to one or two variables that would tell me wether or not it went down a significant amount (>5% of initial) then back up (at least to initial) again.
If it helps to know more specifically what my reports look like column A - patient ID, B date of BMI, C - BMI
Column A will be the same for patient who have multiple instances of BMI recorded in the EHR.
That way I could easily group patients by those who have lost and regained and those who have stayed roughly the same.
Can anyone suggest analysis that could do this?