Comparison of Physician and Patient Global Assessments Over Time in Patients With Rheumatoid Arthritis

Abstract

In rheumatoid arthritis (RA), there is discordance between patient and physician assessments of disease severity and treatment response. This retrospective analysis of the RADIUS (RA Disease-Modifying Anti-Rheumatic Drug Intervention and Utilization Study) 1 cohort examined specific factors that influence differences in global assessments for therapeutic effectiveness of disease-modifying antirheumatic drugs made by physicians (physician global assessment [PhGA]) and patients (patient global assessment [PtGA]). The RADIUS 1 cohort consisted of primarily community-based private practice patients with RA requiring either the addition of or a switch to a new biologic or nonbiologic disease-modifying antirheumatic drug and who were followed for up to 5 years by their rheumatologists. Periodic assessments included PhGA, PtGA, Health Assessment Questionnaire-Disability Index (HAQ-DI), 28-item tender/painful joint count (TJC28), swollen joint count (SJC28), pain Visual Analog Scale (VAS), and acute-phase reactants. Among 4359 patients (mean disease duration, 7.3 years), PhGA most highly correlated with TJC28 (0.6956; 95% confidence interval [CI], 0.6881-0.7030) and SJC28 (0.6757; 95% CI, 0.6678-0.6834). Moderate overall correlations were observed for PtGA with TJC28 (0.5000; 95% CI, 0.4890-0.5108) and less 50 with SJC28 (0.3754; 95% CI, 0.3628-0.3878). Patient global assessment most strongly correlated with pain VAS (0.8349; 95% CI, 0.8305-0.8392) and moderately correlated with HAQ-DI (0.5979; 95% CI, 0.5886-0.6071). Acute-phase reactants poorly correlated with PhGA and PtGA. Low correlations between PhGA and acute-phase reactants suggest that these measurements have a limited contribution compared with the physical examination when physicians make global assessments. These results also suggest that physicians should consider patients' assessments of their disease activity (HAQ, pain VAS, and PtGA) and put joint counts into proper context.