DOP20 Can lifestyle and psychosocial factors predict flares of IBD; an exploratory study using telemedicine

Abstract

Background Tight control of mucosal inflammation and prevention of disease flares are emerging treatment goals to prevent disease progression in inflammatory bowel disease (IBD). The state of the art clinical classification only marginally predict flare occurrence. Mounting evidence shows that psychosocial and lifestyle factors are associated with flares. Longitudinal monitoring of these factors has been made possible by implementing a web-based telemedicine tool, called myIBDcoach. This study is an exploration on the potential additive predictive value of data captured in a telemedicine system to predict flares. Methods Consecutive IBD patients (n = 393) were recruited from the prospective myIBDcoach telemedicine study cohort (ClinicalTrials.gov, NCT02173002). During a one-year follow-up, every 1–3 months, participants reported information on all variables displayed in Figure 1 via myIBDcoach. Variables were subdivided into two main categories; Baseline and myIBDcoach. The myIBDcoach variables were once again subdivided (Figure 1). The outcome of interest, a flare during follow-up, was defined as having clinical symptoms of disease activity (using the Monitor IBD At Home questionnaire) combined with either a faecal calprotectin > 250g/g, disease activity on endoscopy, MRI or CT. Stepwise group-Lasso logistic regression (G-LASSO) was used to estimate associations between flares and individual variables, as well as between flares and the different variable categories (Figure 1). G-LASSO performs variable selection, resulting in models that contain only the most relevant explanatory variables. Results Seven G-LASSO regressions were estimated and evaluated using the Akaike information criteria (AIC), the area under the curve (AUC) and stepwise importance using 10-fold cross-validated penalty parameter. Figure 2 shows that the overall model performance increased when consecutively adding extra variable categories, the largest increase occurred for psychosocial and lifestyle factors. All telemedicine variable categories were found to be more important compared with the baseline variable category, as shown in Figure 3. Results shown in Figure 4 indicate that the information from myIBDcoach was found to be more important than the baseline variables. Conclusion In this exploratory study, psychosocial and lifestyle factors monitored via telemedicine are of superior value to predict flares in IBD patients compared with clinical stratification. Holistic monitoring, including psychosocial and lifestyle factors, and targeted interventions are of interest for future trials and are a promising strategy to prevent flares and improve the outcome of IBD.