Lupus patients who never had zoster were eligible as controls and were matched 2:1 to cases for age, race, sex, and survival status.