Problem of immortal time bias in cohort studies: example using statins for preventing progression of diabetes

Abstract

Immortal time in observational studies can bias the results in favour of the treatment group, but it is not difficult to identify and avoid