Background New axial SpA classification criteria have recently been established in which MRI of the sacroiliac joint (SIJ) is key in the imaging arm. The ASAS/OMERACT MRI group have described and defined MRI findings that are essential for the diagnosis of sacroiliitis in patients with axial SpA.

Objectives We proposed to train rheumatologists online on SIJ MRI reading and to assess the impact of this training on their SpA sacroiliitis (SI) diagnosis accuracy.

Methods 168 French rheumatologists, not members of ASAS or OMERACT, were invited to participate to this nationwide prospective randomized controlled trial. Study design: After a baseline reading test (test 1), rheumatologists were randomized in 2 groups: Group 1 had access to 3 e-learning sessions: on elementary MRI lesions, active SpA SI diagnosis (including images from the ASAS/OMERACT consensus), and differential diagnoses. Two weeks after the e-learning session, rheumatologists were asked to redo an online test (test 2). Group 2 was the control group, without e-learning before test 2. The reading test consisted on 15 MRI files including 7 SpA SI, 6 differential diagnosis and 2 normal SIJ. During test, rheumatologists were asked to click on areas they considered as being an MRI lesion according to the ASAS/OMERACT consensus: bone marrow edema/osteitis, synovitis, enthesitis, capsulitis, sclerosis, erosions, fat deposition and ankylosis. Then, they were asked to give their diagnosis (SpA active SI yes/no) and to grade their diagnosis confidence on a 0-10 scale. Test 1 and 2 file presentation orders were different. The main outcome criterion was the rate of adequate diagnosis for SI at test 2 compared to the diagnosis made by an experts panel. Analysis: The rate of adequate diagnoses at test 2 was compared between groups using a multiple linear regression adjusted on baseline value. Elementary lesions associated to SI diagnosis were evaluated using a multivariate logistic regression.

Results Answers of 161 rheumatologists (male: 52%, mean age: 43±11 years) were analysable (Group 1: 80 and Group 2: 81). Characteristics of both groups were similar. At test 1, 69±12% diagnosed SpA active SI were adequate and mean rate of adequate identification of the elementary lesions was 32±23%. Elementary lesions that were significantly associated to SpA sacroiliitis diagnosis were bone marrow edema (OR=6.91), ankylosis (OR=6.13), enthesitis (OR=4.24) and synovitis (OR=3.45). Elementary lesions significantly associated to a reject of diagnosis of SpA active SI were fracture (OR=0.00), soft tissues inflammation (OR=0.06) and fat deposition (OR=0.50). At test 2, the proportion of adequate diagnoses was equal in both groups: 72±11%. Considering diagnosis with a level of confidence ≥9/10, the proportion of adequate diagnoses was statistically higher in the e-learning group (Group 1: 24±2% and Group 2: 19±2%; p=0.044).

Conclusions French rheumatologists already correctly diagnosed SI on MRI. E-learning sessions reinforced the level of confidence in SI diagnosis in rheumatologists that adequately made that diagnosis.

Disclosure of Interest T. Pham Consultant for: Abbott, C. Hacquard-Bouder Employee of: Abbott, F. Roux Consultant for: Abbott, A. Cotten Paid Instructor for: Abbott, D. Loeuille Paid Instructor for: Abbott, J. Malghem Paid Instructor for: Abbott, L. Dunogeant: None Declared, P. Claudepierre Consultant for: Abbott, P. Goupille Consultant for: Abbott, D. Wendling Consultant for: Abbott, F. Tubach Consultant for: Abbott