Risk of bias in observational studies of interventions: the case of antipsychotic-induced diabetes | Publicación