Background/Purpose:   

Clinical trials and epidemiological rheumatoid arthritis (RA) studies often recruit patients from multiple centers. We studied the proportion of variance in the American College of Rheumatology (ACR) core set measure, Disease Activity Score 28 (DAS28, representative physician and laboratory measure derived composite index), and Routine Assessment of Patient Index Data 3 (RAPID3, representative PRO derived composite index) explained by between-center differences in a multinational study.

Methods:  

7568 patients receiving usual care from rheumatologists in 83 centers located in 30 countries were recruited using standard protocol in the Quantitative Standard Monitoring of Patients with RA (QUEST-RA) study. Mixed-effects analyses of covariance (ANCOVA) models were used to model each ACR core set measure as functions of demographic, medical characteristics and remaining ACR core set measures.  Demographic variables included age, race (white, other races), gender, and education > 12 years (yes, no); while medical characteristics included RA duration, rheumatoid factor status, patient’s  fatigue score (0-10), Psychological Health Assessment Questionaire score (Psych HAQ, 0-3), morning stiffness (0 , 1-60, and >60 minutes), comorbidity burden, body mass index, fibromyalgia (yes, no), osteoarthritis (yes, no), and chronic back pain (yes, no). DAS28 model  included patient’s pain score (0-10 cm) and HAQ, while the RAPID3 model included tender joint count (TJC), swollen joint count (SJC) and erythrocyte sedimentation rate (ESR) in addition to demographic and medical characteristics. The patient recruiting center was included as a random effect to estimate the amount of the residual variance explained by it. MIXED procedure in SAS was used.

Results: Patient reported outcomes had lower proportion (3.25-6.87%) of residual variance that was explained by between recruiting center differences compared to clinician and laboratory derived measures (10.36-14.86%) of RA activity (Table). Similarly, DAS28 had a much larger proportion of variance explained by between recruiting center difference than RAPID3.

Conclusion: Patient reported outcomes variance accounted for by between different centers is considerably lower than clinician and laboratory derived measures after adjusting for potential demographic, medical and RA related characteristics. This is despite patient recruitment in QUEST-RA study from several countries with widely different cultural and socio-economic differences. These results highlight the need for implementation of procedures to standardize RA activity assessment by clinicians involved in multi-center studies.

Table. Variance of RA disease activity measures accounted for by between recruiting center differences.

	Residual Variance			% Residual Variance
Variable	Clinic	Unexplained	Total	Explained by Clinic
PTGL	0.0942	2.8039	2.8981	3.25
Pain	0.0629	2.527	2.5899	2.43
HAQ	0.0169	0.2291	0.246	6.87
MDGL	0.3463	1.9836	2.3299	14.86
SJC	1.7836	14.4741	16.2577	10.97
TJC	4.5332	27.762	32.2952	14.04
ESR	48.329	418.33	466.659	10.36
DAS28	0.3979	1.3654	1.7633	22.57
RAPID3	0.9806	17.4026	18.3832	5.33

PTGL, patient’s global assessment of RA activity; HAQ, health assessment questionnaire, MDGL, clinician’s global assessment of RA activity.

---

« Back to ACR/ARHP Annual Meeting 2012

ACR Meeting Abstracts - https://acrabstracts.org/abstract/patient-report-outcomes-variance-between-centers-is-much-lower-than-physician-and-laboratory-assessed-measures-of-rheumatoid-arthritis-activity-results-from-a-multinational-study/