higher PHQ-9 depression scores (median, 4.0 vs. 1.0, P = .04) compared with patients in the concordant group. In the MHDA category, patients in the discordant group had higher median scores for fibromyalgia WPI (8.0 vs. 4.0; P = .005) and polysymptomatic distress (13.5 vs. 7.5; P = .01). In order to explore the different effects of depression and fibromyalgia on discordance depending on disease activity states, the analysis was also performed on the overall study population (Table 4). In the LDA category, a higher prevalence of depression in this analysis by provider diagnosis was evident in the discordant group (47% vs. 27%, P = .006), but in the MHDA category there was no difference in the prevalence of depression between the discordant and concordant groups (30% vs. 24%, P = .46). As compared to the abovementioned observations for the fibromyalgia WPI, a higher prevalence of fibromyalgia by provider diagnosis was observed in the discordant groups in both the LDA (15% vs. 3%, P = .002) and the MHDA (20% vs. 6%, P = .015) categories. Comparison of patient characteristics between the discordant and concordant groups according to clinical disease activity level in the overall retrospective study populationa Multivariable logistic regression analysis in the total retrospective population showed that diagnoses of fibromyalgia (adjusted odds ratio (OR) 3.06, 95% confidence interval (CI) 1.87–8.00), depression (adjusted OR 1.79, 95% CI 1.02–3.15), and lack of erosions (adjusted OR 0.56, 95% CI 0.32–0.97) were independently associated with patient-provider discordance (Table 5). The associations of body mass index and osteoarthritis with discordance did not reach statistical significance in the overall population. However, osteoarthritis was significantly associated with discordance in the LDA category (adjusted OR 3.36, 95% CI 1.35–8.34) but not in the MDHA category (adjusted OR 0.86, 95% CI 0.33–2.27). The addition of use of glucocorticoids, conventional or biologic DMARDs, opioids, fibromyalgia medications, and antidepressants or anxiolytics to this model did not reveal any significant associations. Pain VAS was not added to the final model for the purposes of this study due to colinearity with the patient global assessment as well as other variables in the model (e.g., fibromyalgia). Overall, the model showed strong performance, with a C statistic of 0.694. Multivariable logistic regression model of patient-physician discordance compared with concordance in assessments of global disease activity Comparison of Ultrasound 7 scores between concordant and discordant groups was stratified by disease activity categories (Fig. 2). The ultrasound studies were performed for research purposes as part of the prospective substudy and were not available to the primary rheumatologist who completed the provider global assessment, so the Ultrasound 7 scores were independent of the determination of patient-provider discordance. No statistically significant differences were found in continuous scores for GS synovitis, PD synovitis, GS tenosynovitis, PD tenosynovitis, or GS erosions between the groups in the LDA or MHDA categories (P > .10 for all comparisons). Among patients with LDA, active GS synovitis (≥ 2) was detected in 60% of the concordant and discordant groups (P = .97), and active PD synovitis (≥ 2) was detected in 30% of the concordant group and 14% of the discordant group (P = .27). Among patients with MHDA in the concordant