Number Needed to Screen for Axial Spondyloarthritis: Preliminary Results from the Fastrax Cohort

Objectives Prevalence estimates for axial spondyloarthritis (axSpA) is 1%,[1] suggesting the number needed to screen (NNS) to identify 1 patient with axSpA is 100, assuming perfect sensitivity of the screening mechanism. FASTRAX is a model of care designed to screen patients with chronic low back pain for axSpA and utilizes Advanced Clinician Practitioner in Arthritis Care (ACPAC)-trained extended scope providers and rheumatology fellows as screeners. The aim of this study is to determine the NNS to diagnose 1 patient with axSpA through the FASTRAX model at its 3 Ontario sites (Toronto, Thunder Bay and Ottawa). Methods Adults (≥18 years) with low back pain >3 months duration, onset age <45 years were referred by their primary care provider, specialist (eg, ophthalmologist, gastroenterologist) or provincial low back pain program ( https://lowbackrac.ca ) for axSpA screening. Screeners performed standard of care assessment (ie, history, physical exam, imaging and laboratory investigations). Screened patients were reviewed by the attending rheumatologist with expertise in spondyloarthritis. Screeners indicated risk of axSpA on a 11-point scale (−5, high confidence not axSpA to +5, high confidence axSpA). Risk assignment was dichotomized to negative or positive risk of axSpA. The rheumatologist provided final diagnosis for all patients. Prevalence of axSpA was calculated as the number of axSpA diagnoses/total number of patients screened. Sensitivity and specificity were calculated for the screener’s assignment of risk against the rheumatologist’s diagnosis (gold standard). Number needed to screen was calculated as 1/ (prevalence of axSpA in the screening model * sensitivity of the screener). Results In total, 269 patients completed the screening process: mean (SD) age of 39.6 (10.5) years; 38% were male; mean (SD) duration of back pain was 11.5 (11.1) years. Referral sources varied by site. 36 patients were diagnosed with axSpA, for a prevalence of 13.4%. Overall sensitivity and specificity of the screener’s assignment of axSpA risk were 91.7% (95% CI 87.4% to 95.9%) and 75.0% (95% CI 68.3% to 81.7%), respectively, with 100% sensitivity at Thunder Bay and Toronto sites. NNS was 9 people with chronic back pain to identify 1 patient with axSpA and ranged from 5 to 19, depending on site (see Table 1). Table 1: Number needed to screen across all, and by, FASTRAX site. Conclusion The FASTRAX model for axSpA screening demonstrates the NNS for 1 patient diagnosed with axSpA is 9, ranging from 5 to 19 depending on screening site in Ontario. This is far fewer NNS based on current population prevalence estimates and demonstrates a valid and efficient pathway to identify patients with axSpA. References [1.] Reveille JD. Arthritis Care Res (Hoboken) 2012;64:905-10.