Point: Immortal Time Bias—What Are the Determinants of Its Magnitude?. Issue 6 (27th March 2019)

• Record Type: Journal Article
• Authors: Harding, Barbara N; Weiss, Noel S
• Is Part Of: American journal of epidemiology. Volume 188:Issue 6(2019)
• Page Start: 1013
• Page End: 1015
• Publication Date: 2019-03-27
• Subjects: bias -- cohort studies -- epidemiologic methods -- immortal time bias -- pharmacoepidemiology; Epidemiology -- Periodicals; Public health -- Periodicals; 614.4
• DOI: 10.1093/aje/kwz067
• Languages: English
• ISSNs: 0002-9262

Abstract

Abstract: In cohort studies, "immortal time" bias refers to a portion of time during which events cannot occur for a particular group of participants. Typically, immortal time bias occurs when: 1) Exposure can be initiated after follow-up of cohort members has begun; and 2) analytically, the preexposure experience is combined with that which takes place following exposure, rather than (correctly) as part of the experience of nonexposed individuals. Using the example of a cohort study of mortality in relation to receipt of cataract surgery, we sought to describe those study design and population characteristics that influence the magnitude of immortal time bias, so as to aid readers in gauging its impact on published research findings. These characteristics include the mean interval between cohort entry and when exposure criteria are met, the proportion of exposed study participants, and the length of study follow-up.