Although confounding is an important problem of cohort studies, its effects can be minimised to enable valid comparison
Mamdani et al. (Thu,) studied this question.