Background Diagnosis and/or classification of a patient with axial Spondyloarthritis (SpA) generally relies on the presence of structural damage of the sacroiliac joints (SIJ) assessed by plain X-rays. However, the psychometric properties of the evaluation of such abnormalities by plain X-rays are not well known.
Objectives Evaluate the intra and inter-reader agreement for the scoring of the sacroiliac SIJ by pelvic postero-anterior plain X-rays. Evaluate the discriminant capacity of pelvic X-rays for the detection of structural damage of the SIJ suggestive of SpA.
Methods Observational transversal mono-centric study. 77 patients were included: 29 patients Ankylosing Spondylitis (AS) according to the modified New York (mNY) criteria (from medical file) and 48 control patients with mechanical back pain (MBP). A pelvic X-ray was performed on all patients.:all 77 X-rays were anonymised and read by two readers (blinded to diagnosis) in a random order to assess inter-reader agreement. Image reading consisted of scoring from 0 to 4 for each SIJ according to the modified new York criteria for the 77 pelvic X-rays. Thereafter, psychometric properties for the detection of structural damage of the SIJ suggestive of SpA were assessed by one reader (blinded to diagnosis) considering the medical record as the gold standard (mNY sacroiliitis yes/no).
Statistical analysis: a) Inter- and intra-reader agreement for sacroiliitis detection: unweighted kappa (mNY sacroiliitis yes/no, where mNY sacroiliitis was defined as a grade ≥2 at both SIJ) b) Inter-reader agreement at the SIJ level: Weighted kappa for each grade of the mNY criteria at each SIJ. c) Sensitivity, specificity and positive likelihood ratio for the presence of structural damage of the SIJ suggestive of SpA.
Results 29 AS patients (mean age 50.2±14.3 years, symptom duration 25.0±12.7 years, 24 (80%) men, and 48 MBP controls (mean age 49.1±10.7 years), 9 (22.5%) men) were included.
Intra-reader and inter-reader agreement for the mNY sacroiliitis was 0.40 [0.19-0.99], and 0.36 [0.09-0.63] by unweighted kappa, respectively.
Inter-reader agreement for the mNY scoring at the sacroiliac level was 0.62 [0.48-0.75]
Sensitivity of pelvic X-rays for the detection of structural damage of the SIJ suggestive of SpA was 0.76 [IC95%, 0.58-0.88], with a specificity of 0.84 [0.71-0.91] (positive likelihood ratio was 4.65).
Conclusions This study confirms the challenge of sacroiliac joint evaluation by pain X-rays, which reveals modest agreement; nevertheless, the specificity of such imaging for structural damage evaluation of the SIJ suggestive of SpA was good (84%). Development of imaging techniques that allow more reproducible detection of structural damage is necessary.
Disclosure of Interest None declared