Preventing Patients with Crohnʼs Disease from Discontinuing Infliximab: A Dynamic Survival Model

Abstract

Purpose: Infliximab (IFX) treatment has been shown in the literature to have a low rate of persistence, although there has been limited information on causes for discontinuation of IFX therapy. We reviewed charts and utilized dynamic modeling to predict the effect of a variety of interventions on the ability for patients to continue IFX in the community setting. Methods: Records of all 120 Crohn's disease patients treated with IFX at the NorthShore University HealthSystem from March 17th 2003 to September 1st 2010 were reviewed. Nine categories of reasons for discontinued treatment were found: 1) ongoing disease activity (13.3%), 2) acute infusion reactions (9.2%), 3) complications of therapy (9.17%), 4) chronic infusion reactions (6.67%), 5) health insurance reasons (5.83%), 6) patient intolerance (5.00%), 7) surgery (4.17%), 8) patient lost to follow up (4.17%), and 9) patient improved (0.83%). Reasons 1, 2, 4, and 5 were deemed points of intervention. Dose increases or changing medication could improve reason 1, premedication or changing medication could improve reasons 2 and 4, while attention to insurance could improve reason 5. We created a model to simulate IFX therapy persistence using STELLA (isee, Hanover, NH). Rates of discontinuation and timing of these occurrences used in model were determined from the collected data and simulated with Weibull distributions. Data was expressed as a Kaplan-Meyer survival analysis to show the instantaneous percent of people on the drug where infliximab persistence, represented as “survival”. We then simulated various degrees of improvement of the four previously described points of intervention and their effect on survival curves. Results: Intervention produced little effect of persistence of therapy. For example, with a 25% reduction in reasons 1, 2, 4, and 5, no improvement in overall persistence was seen (Figure 1). With a 90% reduction in all infusion reactions, only a 3-10% increase in IFX persistence is seen over 5 year period.Figure: No Caption available.Conclusion: Dynamic modeling is a constructive method for characterizing the effects of interventions to improve IFX therapy for Crohn's disease. Although improvement in persistence of IFX therapy can be achieved with aggressive interventions, marked enhancement of multiple factors are required to produce relatively small changes in the high discontinuation rate seen within the community setting.Table: Effect of changes in rate of infusion reactions on IFX persistence