Statistical Issues in designing a large-scale reliability exercise in ultrasonography of the joint synovium

Meeting report
Meeting slides

This meeting brought together statisticians, clinicians and other methodologists for discussion of common statistical issues in imaging studies. Challenges were illustrated by a clinical example, designing a large-scale reliability exercise in ultrasonography of the joint synovium, but they are also widely applicable to other imaging modalities.

• Dr Richard Wakefield (Senior lecturer and honorary consultant in rheumatology, NIHR Leeds Musculoskeletal Biomedical Research Unit and Leeds Institute of Rheumatic and Musculoskeletal Medicine)
• Dr Liz Hensor (Statistician, NIHR Leeds Musculoskeletal Biomedical Research Unit and Leeds Institute of Rheumatic and Musculoskeletal Medicine)
• Dr. Wakefield overviewed the clinical challenges in this area, and Dr Hensor introduced the statistical challenges.

The meeting featured small group discussion of statistical issues including:

• Design of reliability studies
• Image selection for reliability assessments
• Pros and cons of stored image vs in vivo imaging reliability assessments
• Handling of multiple images/scores per patient
• Considerations when constructing summary pathology scores