In Reply Donovan and colleagues raised concerns regarding the definition of the nonexposed group in our study.1 As they noted, the variable “PAP [positive airway pressure] prescription” was derived from a set of branched questions, so that the control group may indeed result in a mixed group of individuals. The control group comprised 311 participants, among whom 305 answered “no” to the “sleepapnea” variable and 6 participants were coded “no” to the “sacpap” variable. Hence, the 305 participants either had no access to a physician or were not given information about their obstructive sleep apnea (OSA) by their physician, and the 6 participants probably either declined PAP therapy and/or used other modalities to treat their OSA.