Background Tenosynovitis occurs frequently in patients with rheumatoid arthritis (RA) and magnetic resonance imaging (MRI) assessment of tenosynovitis may have an important role as an outcome measure in RA clinical trials. A scoring system for tenosynovitis in RA-patients proposed by Haavardsholm et al. has shown high reliability. However, these results are based on scorings by experienced readers, and it is not known how well the scoring system can be transferred to inexperienced MRI-readers.
Objectives To investigate whether a 3-day tutorial is sufficient for an inexperienced MRI-reader to achieve a high reliability in scoring tenosynovitis.
Methods The 3-day tutorial was lead by a tutor with previous experience with the tenosynovitis scoring system proposed by Haavardsholm et al (1). The student had no previous experience of scoring MRIs for tenosynovitis.
Prior to the reading exercise, the scoring system was introduced to the student in a 2-hour session: First, the basic rules of the scoring system were explained. Then, the tendon sheaths on 2 image-sets were discussed and scored together, followed by 2 image-sets that were scored separately and discussed afterwards.
The exercise was divided into 3 parts: First, 20 paired image-sets were scored separately by both readers. Second, 5 of these image-sets were analysed and discussed in detail, and modifications were suggested by the tutor. Third, all 20 paired image-sets were re-anonymized and re-read by both readers.
The 20 image-sets were selected at 2 different timepoints to comprise a wide range of tendon sheaths involvement. On axial T1-weighted pre- and post-contrast images, 9 tendon compartments of the wrist and the flexor tendon sheaths adjacent to the 2nd-5th metacarpophalangeal joints were scored semiquantitatively for tenosynovitis; 0: No enhancement, 1: 0 to <1.5mm, 2: 1.5 to <3mm, 3: ≥3mm width. All images were blinded for patient data but not for chronology. Statistical calculations are explained and described in table 1.
Results The inter-reader intra-class correlation coefficients (ICC) were poor (0.30-0.60) at the first reading. At the second reading, the inter-reader ICC were very good for baseline status scores (0.87) and change scores (0.83), but poor for follow-up status scores (0.47). Excluding the 5 image-sets that were assessed 3 times, resulted in a lower but still good ICC (0.60-0.78). The SDC ranged from 8.2-9.2 at the first reading and from 5.5-5.7 at the second reading. The intra-reader ICC for status and change scores were very good (0.90-0.92) for the tutor reader and poor (0.42-0.58) for the student reader. Intra-reader SDC were 4.47 and 7.57 respectively. The poor intra-reader agreement between first and second reading shows that the student, who received training between the first and second reading, scored the images differently at the first and second reading.
Conclusions This exercise documents that an inexperienced reader, after a 3-day tutorial can perform MRI-scoring of tenosynovitis with high agreement with an experienced reader. MRI-assessed tenosynovitis is an easily learned potential outcome measure for RA clinical trials.
Haavardsholm EA et al. Introduction of a novel magnetic resonance imaging tenosynovitis score for rheumatoid arthritis: reliability in a multireader longitudinal study. Ann Rheum Dis. 2007;66:1216-20.
Disclosure of Interest D. Glinatsi: None declared, C. Martinez: None declared, M. Østergaard Grant/research support from: Abbvie, Centocor, Merck, Schering-Plough, Consultant for: Abbvie, BMS, Boehringer-Ingelheim, Lilly, Centocor, GSK, Janssen, Merck, Mundipharma, Novo Nordisk, Pfizer, Schering-Plough, Roche, UCB, Wyeth