Skip to main content
. 2022 Aug 3;6(1):443–459. doi: 10.3233/ADR-220002

Fig. 1.

Fig. 1

Immortal time bias. a) The entire follow-up duration, including immortal time, is classified into the exposure group, leading to immortal time bias, which can incorrectly show metformin having a protective effect on dementia risk. b) Immortal person time in the exposure group was classified into the unexposed group, showing a proper method to classify exposed and unexposed groups at time zero.