Methodologic considerations in the design and analysis of nested case-control studies: association between cytokines and postoperative delirium