Figure 1.

Nested case–control design. At the time, a case (diamond end caps) occurred, matched controls (circle end caps) were selected from all persons under follow-up (i.e., from insurance policy start [PS] date to policy end [PE] date). Only exposures prior to the case occurrence were counted for matched controls (i.e., purple for risk set 1 [dashed lines] and both purple and blue areas for risk set 2 [dotted lines]).