Text this: Observational studies on glucocorticoid harm should always consider confounding by indication: comment on the article Katz et al