A study of referral bias in NMOSD and MOGAD cohorts

Background: Neuromyelitis optica spectrum disorder (NMOSD) and myelin oligodendrocyte glycoprotein antibody–associated disease (MOGAD) are rare disorders often seen in highly specialized services or tertiary centres. We aimed to assess if cohort characteristics depend on the origin of the referral catchment areas serviced by our centre (i.e. local, regional or national). Methods: Retrospective cohort study using a national referral service database including local (Oxfordshire), regional (Oxfordshire and neighbouring counties), and national patients. We included patients with the diagnosis of NMOSD, seronegative NMOSD or MOGAD, followed at the Oxford Neuromyelitis Optica Service. Results: We included 720 patients (331 with MOGAD, 333 with aquaporin-4 antibody (AQP4)-NMOSD, and 56 with seronegative NMOSD. The distribution of diagnoses was similar across referral cohorts. There were no significant differences in the proportion of pediatric onset patients, sex, or onset phenotype; more White AQP4-NMOSD patients were present in the local than in the national cohort (81 % vs 52 %). Despite no differences in follow-up time, more relapsing MOGAD disease was present in the national than in the local cohort (42.9 % vs. 24 %, p = 0.029). Conclusion: This is the first study assessing the impact of potential referral bias in cohorts of NMOSD or MOGAD. The racial difference in the AQP4-NMOSD cohorts likely reflects the variation in the population demographics rather than a referral bias. The over representation of relapsing MOGAD patients in the national cohort probably is a true referral bias and highlights the need to analyze incident cohorts when describing disease course and prognosis. It seems reasonable therefore to compare MOGAD and NMOSD patients seen withing specialised centres to general neurology services, provided both use similar antibody assays.