How good is the agreement between clinical diagnoses and classification criteria fulfilment in axial spondyloarthritis? Results from the SPARTAKUS cohort

Objectives To study the agreement between clinical axial spondyloarthritis (axSpA) diagnoses and fulfilment of the Assessment of SpondyloArthritis international Society (ASAS) axSpA and modified New York (mNY) classification criteria, and to compare disease/health status between axSpA subtypes. Method Patients with prevalent, clinical axSpA attending a rheumatology clinic were enrolled in a cross-sectional study. Assessments included physical evaluation, laboratory testing, questionnaires, and appropriate imaging, allowing classification. Standard axSpA outcome measures were compared between patients fulfilling mNY/radiographic versus non-radiographic axSpA (r-axSpA/nr-axSpA) criteria. Results Of 239 consecutively included patients, 141 fulfilled ASAS r-axSpA and/or mNY criteria, while 57 fulfilled nr-axSpA criteria. The agreement between r-axSpA and mNY criteria fulfilment was 94%. The positive predictive value (PPV) of a clinical ankylosing spondylitis (AS) diagnosis for mNY criteria fulfilment was 71%; the PPV of an undifferentiated axSpA (u-axSpA) diagnosis for fulfilment of nr-axSpA criteria was 30% and 40% for mNY criteria. Patients with r-axSpA/AS were older, more often men, and had longer disease duration, more uveitis, and worse spinal mobility than nr-axSpA patients, who had more enthesitis and dactylitis. Conclusion We found an overall good concordance between clinical axSpA diagnoses and classification criteria fulfilment, with 83% fulfilling ASAS axSpA and/or mNY criteria. Regarding axSpA subtypes, the concordance was weaker, and although the ICD-10 code for AS correctly identified patients meeting mNY criteria in 71% of cases, one-third of mNY-positive patients lacked an AS diagnosis. Moreover, clinical u-axSpA diagnoses could not serve as a proxy to identify nr-axSpA, highlighting the importance of thorough classification in research on axSpA subtypes.