Harmonization of real-world studies in multiple sclerosis: Retrospective analysis from the rirems group

Background: Worldwide multiple sclerosis (MS) centers have coordinated their efforts to use data acquired in clinical practice for real-world observational studies. In this retrospective study, we aim to harmonize outcome measures, and to evaluate their heterogeneity within the Rising Italian Researchers in MS (RIReMS) study group. Methods: RIReMS members filled in a structured questionnaire evaluating the use of different outcome measures in clinical practice. Thereafter, thirty-four already-published papers from RIReMS centers were used for heterogeneity analyses, using the DerSimonian and Laird random-effects method to compute the between-study variance (τ2). Results: Based on questionnaire results, we defined basic modules for diagnosis and follow-up, consisting of outcome measures recorded by all participating centers at the time of diagnosis, and, then, at least annually; we also defined more detailed/optional modules, with outcome measures recorded less frequently and/or in the presence of specific clinical indications. Looking at heterogeneity, we found 5-year variance in age at onset (ES=27.34; 95%CI=26.18, 28.49; p<0.01; τ2=4.76), and 7% in female percent (ES=66.42; 95%CI=63.08, 69.76; p<0.01; τ2=7.15). EDSS variance was 0.2 in studies including patients with average age <36.1 years (ES=1.96; 95%CI=1.69, 2.24; p<0.01; τ2=0.19), or from 36.8 to 41.1 years (ES=2.70; 95%CI=2.39, 3.01; p<0.01; τ2=0.18), but increased to 3 in studies including patients aged >41.4 years (ES=4.37; 95%CI=3.40, 5.35; p<0.01; τ2=2.96). The lowest variance of relapse rate was found in studies with follow-up duration ≤2 years (ES=9.07; 95%CI=5.21, 12.93; p = 0.02; τ2=5.53), whilst the lowest variance in EDSS progression was found in studies with follow-up duration >2 years (ES=5.41; 95%CI=3.22, 7.60; p = 0.02; τ2=1.00). Discussion: We suggest common sets of biomarkers to be acquired in clinical practice, that can be used for research purposes. Also, we provide researchers with specific indications for improving inclusion criteria and data analysis, ultimately allowing data harmonization and high-quality collaborative studies.