Article Text


THU0185 Reliability of ra mri scoring
  1. MN Lassere1,
  2. F McQueen2,
  3. M Ostergaard3,
  4. P Conaghan4,
  5. R Shnier1,
  6. C Peterfy5,
  7. M Klarlund3,
  8. N Stewart2,
  9. P O’Connor4,
  10. P Bird1,
  11. J Edmonds1,
  12. P Emery4,
  13. H Genant6
  1. 1Department of Rheumatology, St. George Hospital, Sydney, Australia
  2. 2Department of Molecular Medicine, Auckland, New Zealand
  3. 3Department of Rheumatology, University of Copenhagen, Copenhagen, Denmark
  4. 4Rheumatology Research Unit, University of Leeds, Leeds, UK
  5. 5Synarc Inc., San Francisco, USA
  6. 6Department of Radiology, University of California, San Francisco, USA


Background At OMERACT5 recommendations were developed for acquisitions/scoring of MRI synovitis/bone destruction of hand/wrist in RA: OM5 RA MRI Score (Om5 RAMRIS).1 These were based on results of a multicentre inter-reader agreement studies of two earlier MRI scoring methods, developed at EULAR 1999 and ACR 1999.2

Objectives Determine whether reader calibration, standardisation of imaging protocols and more precise definitions of lesions improved inter-reader agreement of MRI scoring.

Methods MRIs (1.5T, cor and axial T1& T2, FS, iv Gd) of wrist (8)& 2‑5 MCPJ(10) scored by experienced readers in 5 centres. Synovitis determined by Gd enhancement in synovial compartment by 2 methods: global score (0‑3)(wrist 3 sites, MCPJ 4 sites) and direct measure (mm) of maximal thickness of enhancing tissue (wrist 4 sites, MCPJ 2 sites). Bone erosion (wrist 15 sites, MCPJ 8 sites) was a bone defect with sharp margins, with a cortical break. Bone defect and bone oedema also defined. All measured 0‑10 by volume of defect as proportion of assessed bone volume by 10% increments judged on all images.

Results Reliability for most sites for all lesions was good. Poor intraclass correlation coefficients for bone defect (but excellent agreement) were due to predominance of zero scores. Table 1 shows results of summated scores). Om5 RAMRIS performed better than our previous scoring methods.

Abstract THU0185 Table 1

Conclusion We found Om5 RAMRIS has acceptable inter-reader reliability. Whether it can be further improved and degree to which it’s sensitive to change will need to be determined by its performance in longitudinal/intervention studies.


  1. Conaghan, et al. J Rheum, in press

  2. Ostergaard, et al. J Rheum, in press

Statistics from

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.