OP0280-HPR VALIDATION OF A NEW GLUCOCORTICOID-SPECIFIC PATIENT REPORTED OUTCOME QUESTIONNAIRE (THE STEROID PRO)

Abstract

Background: The Rheumatoid Arthritis Impact of Disease (RAID) is a patient-reported outcome measure (PROM) originating from a EULAR initiative. It supports patient-centered care and shared decision-making between patient and rheumatologist regarding treatment and adjustment of medication. RAID could be a useful tool in remote monitoring, and its association with disease activity is thus of interest.[1] It is not known if RAID is responsive and discriminative to disease activity flares. Objectives: To explore the responsiveness of the RAID (sum score and single domain scores) to clinical disease flare and to assess its discriminative ability with regard to flare. Methods: We used data from the two Norwegian multicenter ARCTIC REWIND trials assessing tapering of TNFi and csDMARDs (ClinicalTrials.gov: NCT01881308).[2] Eligible participants had RA (ACR/EULAR 2010 criteria), were in sustained remission for ≥12 months on stable DMARDs with Disease Activity Score (DAS) remission combined with no swollen joints at inclusion. In the TNFi trial, patients were randomized to stable TNFi or tapering to discontinuation of TNFi (stable csDMARD comedication). Patients in the csDMARD trial were randomized to stable or half dose treatment for the first year. Study visits with assessment of disease activity were conducted every four months, if a flare was suspected between visits the patient was seen within a week. Flare was defined as a combination of DAS>1.6, an increase in DAS ≥0.6 units from the previous visit and minimum two swollen joints. If a patient did not fulfill these criteria, a disease flare could be recorded if both the patient and investigator agreed that a clinically significant flare had occurred. If a flare was confirmed, the full dose of the study medication was reinstated or (in the stable-dose group) treatment adjusted according to current recommendations. We evaluated the median RAID score and single domains at the last visit before flare, flare visit and first visit after flare in relation to the suggested ≤2 RAID threshold for patient acceptable symptom state (PASS), and assessed the changes between those visits using Wilcoxon rank-sum test.[3] The ability of RAID to detect disease flare was assessed using the area under the ROC curve (AUC) based on logistic regression models. For comparison, similar analyses were performed for patient global assessment (PGA), DAS and C-reactive protein (CRP). Results: of the 248 patients included, 159 (64%) were female. Mean (SD) age and disease duration were 56 (11.8) and 6.3 (5.7) years. For patients who experienced a flare the median (IQR) RAID score was 0.9 (0.3, 1.4) at last visit prior to flare with 86% of scores within the PASS range. At the flare visit median RAID was 2.6 (1.4, 5.6) with only 30% of scores within the PASS range (Figure 1). Similar changes were observed in DAS, PGA and CRP. All seven RAID domains increased significantly at the flare visit (p-values < 0.01), with the largest increase in pain (Figure 1). According to the AUC analyses the estimated accuracy in detecting flare was high for RAID, but higher for DAS and PGA. CRP was less accurate in detecting flare (Table 1). Conclusion: There was a clinically and statistically significant increase in median RAID score and all single domains of RAID at flare visit, supporting its responsiveness to clinical disease flares. Both RAID and PGA discriminated well between flare and non-flare, hence both PROMs might contribute in early identification of flare and patients who might need an adjustment of medication, in a regular clinical setting as well as in remote monitoring of disease activity.