P104 The relationship between patient and clinician assessments of disease activity in rheumatoid arthritis: observational study using electronic health record data

Abstract Background/Aims Assessing disease activity underpins rheumatoid arthritis (RA) care. Several electronic patient-reported outcome measures (ePROMs) exist for patients to self-assess disease activity. Largely evaluated in research settings, the extent they relate to clinician disease activity assessments in routine care is uncertain. We evaluated this using routinely collected electronic health record data at the Midlands Partnership University NHS Foundation Trust. Methods Patients with RA are invited to complete online ePROMs, including RAPID3, and patient 28 swollen (SJC) and tender (TJC) joint counts and global assessments in the week before their rheumatology appointment, with clinician SJCs/TJCs performed in clinic. People not completing ePROMs at home are invited to complete them in clinic on a digital tablet (excluding joint counts). ePROMs are captured using the “Haywood Arthritis Portal” online NHS platform, which provides written information/videos about self-completing joint counts. We used data from patients with RA with self-assessed and clinician-assessed disease activity captured at the same appointment (ethical approval obtained to use anonymised data for research). Correlation coefficients (r) compared: (a) patient and clinician TJCs/SJCs; (b) patient and clinician DAS28-CRP; and (c) RAPID3 and clinician DAS28-CRP. Patient versus clinician DAS28-CRP were also compared using limits of agreement. Disease activity categories (remission REM/low LDA/moderate MDA/high HDA) from clinician DAS28-CRP were compared to patient DAS28-CRP categories and RAPID3 categories using weighted kappa (κ) statistics. Analyses were conducted overall and stratified by age/gender/deprivation. For individuals with multiple clinic visits with available data, the first was used (where the patient was most naïve to the process). Results Of the 751 patients included, 69% were female; their mean age was 66.8 years (95% CI: 65.9, 67.7); and mean DAS28-CRP 3.30 (3.21, 3.40). Patient and clinician TJCs had moderate/strong correlation (r = 0.59) but SJCs had weak correlation (r = 0.37). Despite this, patient and clinician DAS28-CRP had strong correlation (r = 0.67), although patients self-assessed their DAS28-CRP 0.77 points higher than clinicians (limits of agreement: -1.33, 2.87). Patient and clinician disease activity categories had moderate agreement (κ = 0.52), although this was greater when patients self-assessed themselves as having REM/LDA (with 84% of the 81 patients in self-assessed REM/LDA also in clinician-assessed REM/LDA) than MDA/HDA (with 32% of the 180 patients in self-assessed MDA/HDA, in clinician-assessed REM/LDA). Correlation (r = 0.47) was low and agreement (κ = 0.16) very low between RAPID3 and clinician DAS28-CRP categories. For all comparisons, correlation/agreement was better in people aged 74 than those ≥75, but similar across genders/by deprivation. Conclusion Electronic patient-completed DAS28-CRP has moderate/strong correlations and agreement with clinician-completed DAS28-CRP in routine care, with the caveat that patients tend to score their disease activity levels higher than their clinician. As correlations/agreement are lower in older people, this needs consideration when implementing the use of electronic patient-completed DAS28-CRP in routine settings. Disclosure I.C. Scott: Grants/research support; ICS is funded by an NIHR Advanced Research Fellowship NIHR300826.. N. Daud: None. N. Cox: None. L. Gray: None. S. Muller: Grants/research support; SM is partly funded by the NIHR Applied Research Collaboration West Midlands..