Title: Factors that affect tender and swollen joint counts in rheumatoid arthritis
Christine L. Amity1, Marisa Eckels1, Kenneth Gold1, Kelly A. Reckley1, Niveditha Mohan1, Stephen R. Wisniewski2, Elizabeth A. Schlenk3, Marc C. Levesque1 and Terence Starz1
1University of Pittsburgh School of Medicine; 2 University of Pittsburgh Graduate School of Public Health; 3 University of Pittsburgh School of Nursing, Pittsburgh, PA USA
Background: Quantitative joint counts are used to determine rheumatoid arthritis (RA) disease activity and are increasingly important for routine clinical practice to optimize patient care. Several factors make quantitative joint assessments challenging. The purpose of this study was to determine how osteoarthritis (OA), fibromyalgia (FM), body mass index (BMI), handedness and disease activity affect tender and swollen joint counts by examining their effects on the inter-rater reliability of tender and swollen joint counts in a usual care setting.
Methods: 72 RA patients (54 F, 18M) recruited from the University of Pittsburgh Rheumatoid Arthritis Comparative Effectiveness Research (RACER) registry underwent standardized 28-joint assessments performed by two registered nurses and two rheumatologists at a single clinic visit. A Manual Tender Point (TP) Survey to determine the presence of FM (≥ 11/18 TPs) took place during the same visit. Hand X-rays within 1 year of the clinic visit were graded for the presence of OA by a blinded, independent rheumatologist. BMI and disease duration were obtained from registry data. Subjects were classified according to Clinical Disease Activity Index (CDAI) as being in remission-low (CDAI ≤ 10) or moderate-high (CDAI > 10) disease activity. Intra-class correlations and 95% CIs were determined for tender and swollen joints, stratified by variables of interest.
Results: The overall agreement among raters was moderate for tender and swollen joints (ICC = 0.48 and ICC = 0.56, respectively). The agreement among raters for swollen joints was similar in subjects with and without OA, with disease duration < or ≥ 3 years, with BMI < or ≥ 30 and with remission-low or moderate-high disease activity (ICC = 0.17 to 0.56). For tender joints, agreement was also moderate between disease duration and BMI subgroups (ICC = 0.47 to 0.64). However there were significant differences in agreement for tender joints for subjects with OA versus without (ICC = 0.33 (0.12, 0.50) vs. ICC = 0.65 (0.55, 0.72), respectively), and for subjects in remission-low disease activity versus moderate-high (ICC = 0.13 (-0.03, 0.29) vs. ICC = 0.52 (0.39, 0.63), respectively).
Conclusions: Agreement between raters was only moderate for tender and swollen joint counts. The presence of OA and lower disease activity reduced inter-rater reliability significantly. These results indicate that new methods of assessing joint disease activity may be needed for many of the patients seen in usual care settings as opposed to clinical trials where subjects are generally younger, have less OA and higher levels of disease activity. Our future analyses will focus on the level of agreement between nurses and physicians, on the level of agreement at the individual joint level, and on multivariable analyses that incorporate all factors that may affect inter-rater agreement of tender and swollen joint counts.