P122 Conducting federated treatment effectiveness analyses for systemic therapies in atopic dermatitis across country borders: methodological considerations from the DREAM TO TREAT AD project

Abstract There is a lack of evidence on treatment outcomes in patients with moderate-to-severe atopic dermatitis (AD) in the context of regular healthcare provision. While registers are a valuable tool for data collection in the context of regular care, treatment outcomes may not always be collected on the visit dates predetermined by the study protocol. Therefore, time windows around those dates have to be defined for statistical analyses. In addition, uptake on some systemic therapies is gradual, resulting in relatively small numbers of patients in some country registers. The DREAM TO TREAT AD (D2T AD) project is a European register-based study of abrocitinib and conventional systemic therapies in AD. It aims to develop and validate a federated statistical analysis methodology to evaluate treatment effectiveness across national AD registers. The D2T AD consortium consists of A-STAR Ireland, A-STAR UK, SCRATCH Denmark, TREATgermany and TREAT NL/Belgium. The federated analysis of treatment outcomes is conducted from patient enrolment through the follow-up period using the secure DataSHIELD environment. Outcomes recorded include Eczema Area and Severity Index (EASI), (Children’s) Dermatology Life Quality Index [(C)DLQI], Patient-Oriented Eczema Measure (POEM),and Peak Pruritus Numerical Rate Scale (PP-NRS). This consists of individual register data nodes and an analytical hub. To evaluate the treatment effectiveness of abrocitinib and conventional systemic therapies in DataSHIELD, changes in EASI, POEM, (C)DLQI and PP-NRS scores between treatment initiation (baseline) and follow-up timepoints (from week 16 until 3 years) are assessed using a linear regression model with adjustment for the baseline outcome score. The data structure includes two outcome values paired with distance in days collected to each visit date – one before and one after – at treatment initiation and every planned follow-up visit. It is possible to select different time windows at the analysis stage in the DataSHIELD environment and produce statistical results depending on the data included in the selected time windows. This approach allows for the analysis to be done without creating several datasets for each specific time window to assign a treatment outcome. The methodology has been validated using synthetic data in the DataSHIELD environment to evaluate achievement of ≥ 50%, ≥ 75% and ≥ 90% improvement in EASI score at week 16 from baseline. A novel federated statistical analysis approach for evaluating treatment effectiveness in patients with moderate-to-severe AD treated with abrocitinib and conventional systemic treatments has been developed and validated using synthetic data. As a next step, the analysis will be applied across the D2T AD consortium registers to generate valuable effectiveness results including sensitivity analyses regarding choice of visit windows.