Stability of clinical outcome measures in rheumatoid arthritis patients with stable disease defined on the basis of the EULAR response criteria.

Abstract

Natural variation also known as measurement error is assessed in individuals in "steady state." The study aimed to examine inter-visit variations in clinical outcome measures in rheumatoid arthritis (RA) patients with stable disease defined on the basis of the EULAR response criteria. Two hundred thirty-three RA patients with stable disease defined as a change in Disease Activity Score (DAS28-CRP) ≤0.6 between two consecutive visits were identified in the Danish rheumatology registry for biological treatment (DANBIO). Clinical data from a single set of such two visits were extracted for each patient. Using the Bland-Altman method, lower and upper 95% limits of agreement (LLoA; ULoA) between the consecutive assessments and the bias were calculated for each measure. Associations were characterized by Pearson's r-values and standard errors of estimation (SEE). The mean change in DAS28-CRP was 0.0 ± 0.3. Agreements between the assessments were close on the group level but poor on the individual level. For example, LLoA; ULoA [bias] for patient global assessment (0-100) was -28.3; 29.7 [0.7], for fatigue (0-100) -38.1; 36.3 [-0.9] and for 28 swollen joint count -3.3; 3.3 [0.0]. Inter-visit differences were poorly explained by the baseline values (r [SEE] ranging from 0.15 [12.6] for fatigue to 0.58 [1.4] for 28 tender joint count) and by changes in other outcome variables. In conclusion, outcome measures fluctuated substantially and unpredictably in individual RA patients considered in steady state. The observed between-visit differences may be interpreted to reflect natural variation or measurement error that should be taken into account when monitoring patients in the daily clinic.