Robinson et al. report the effect of adherence and medication switches on relapse. Non-adherence to inflammatory bowel disease (IBD) medication is multifactorial, and associated with increased flares and costs.[2-4] We recently demonstrated that medication beliefs, but not dosing regimens, predict adherence to 5-aminosalicylates (5-ASAs). There has been anecdotal evidence that nonclinical changes in 5-ASA preparations lead to poor adherence and/or flares.
The authors examined switches within a prescription database, without access to clinical data. By using a proxy measure for relapse (doubling of 5-ASA dose), the authors introduce a nonvalidated tool. While doubling of oral 5-ASA dose is one way of treating flares, some flares are only treated with additional topical 5-ASA. In addition, moderate-to-severe flares requiring glucocorticosteroids will have been missed altogether. As such, the proxy measure cannot be relied upon. Linkage with clinical data is the only way to record flares accurately.
It seems arbitrary to simply monitor the effect of switches away from Asacol, but not other 5-ASA preparations. While statistical power is important, the data should have been analysed and reported in full for all 5-ASA preparations. This restriction brings in considerable selection bias.
Switches in adherent patients led to an increase in flares, but adherence apparently remained unchanged. Patient confidence in medication is important for adherence, and may be adversely affected by switches, but are the authors suggesting that flares were rather due to pharmacological effects?
Perhaps the most important message is that medication switches in stable patients may potentially exert an adverse effect. The evidence is however too weak to draw firm conclusions. It is high time for funding bodies to support prospective, high-quality, large-scale studies addressing adherence and switches. Switches intending to save costs could inadvertently lead to poor outcomes, and thus higher costs if adherence were to be adversely affected.