I will have longitudinal data in which I hope to compare samples at "surveillance" and "disease" time points. However, "disease" can occur more than once, and "surveillance" occurs before "disease", such that each "disease" sample can be paired with a unique "surveillance" sample that occurs before disease.
If I were to conduct a paired-t-test between paired surveillance and disease samples, my pairs/differences would not be independent. Do I need to worry about this? If so, I've thought about using linear mixed-effects/GEE models to control for individual. The issue with this is that there might be confounding that would effect my results if I didn't keep the exact pairing, such as the season of disease. If I were to adjust for season, would that address the potential "unmatched" issue?
Please let me know if there's anything I can do to clarify my question.