and as such, this proxy for flare is likely to represent the majority of flares, but certainly not all. For example, steroids are often initiated in hospital for a number of patients, but these formulations would not appear in this pharmacy database. Despite being unvalidated, as Dr Selinger has highlighted, we are encouraged that the use of the proxy supports the findings from other studies, namely that non-adherence is linked to increased relapse. 6 The switch analysis was conducted in a cohort of patients receiving Asacol only. Asacol prescribing represents the majority of prescriptions for oral 5-ASA products, and in this analysis, there were insufficient patients on alternative formulations to enable an analysis of sufficient statistical power.This study shows that switches in a subgroup of adherent patients lead to an increase in flares, but the study was not designed to measure how adherence changed after a flare. Furthermore, the authors are not suggesting that flares are solely due to pharmacological effects, rather that pharmacological effects are one of a number of potential contributing factors.The limitations of the study were acknowledged in the paper. However, the study sheds useful light on the relationships among adherence, switches and flares, and we certainly hope, in the future, that further prospectively designed studies can increase our understanding of these relationships.
ACKNOWLEDGEMENTThe authors' declarations of personal and financial interests are unchanged from those in the original article.