Abstract
Kim et al.'s cohort study on topical corticosteroids (TCS) and incident type 2 diabetes raises three methodological concerns. First, enrollment in 2012 with diabetes ascertainment from 2017 creates a five-year window when outcomes cannot occur, underestimating absolute risk. Second, exposure metrics are redundant (duration derived from prescriptions) and potency groups are pooled; a prescription-count gradient is needed. Third, propensity matching on age and sex ignores disease severity, explaining the spurious protective effect of low-potency TCS; an active comparator within a single indication is recommended.