Background/Purpose: The identification of risk factors associated with flares in RA is challenging because of the use of different composite disease activity scores and a lack of consistency in the definition of a flare. The Clinical Disease Activity Index (CDAI) is a composite disease activity score that incorporates both a patient and evaluator global assessment (PGA and EGA, respectively). We hypothesized that predictors of flares would vary based on the contribution of clinically important worsening of PGA or EGA to the CDAI and identified predictors of RA flare.

Methods:

We utilized data from the Veterans Affairs RA (VARA) registry- a longitudinal multicenter cohort of US veterans who fulfill the 1987 American College of Rheumatology (ACR) criteria for the classification of RA. Only patients with low disease activity (LDA) defined by CDAI score (</=10) were included in analyses. A CDAI flare was defined as an increase in CDAI to >/=10 and change in CDAI of >/=2. PGA and EGA LDA flares were defined as an increase to >10 and a change >/=10 mm in the PGA or EGA. We used multivariable logistic regression incorporating generalized estimating equations to evaluate for independent predictors of flare by the next visit. Covariates tested included age, sex, race, disease duration, smoking status, common comorbidities, sustained low activity, serostatus, current medications, recent medication changes, inflammatory markers, function status (MD-HAQ), and length of interval between visits.

Results:

Data from 1,582 observations with LDA were included [mean age 63 (±11), 90% male, 82% Caucasian, 81% rheumatoid factor (RF) positive]. Predictors of subsequent CDAI flare, EGA flare and PGA flare are outlined in Table 1. Factors associated with subsequent CDAI flare (N=437) included higher tender joint count, swollen joint, and MD-HAQ score. CDAI flare was less likely among those with sustained CDAI LDA at the prior visit. EGA flare (N=505) was associated with the presence of RF, black ethnicity, elevated CRP, and greater TJC. EGA flare was less likely among patients using methotrexate, a sustained LDA by CDAI and higher patient global score. The strongest predictors of PGA flare (N=542) were a greater MD-HAQ score, black ethnicity, and RF positivity.

Conclusion:

Sustained LDA is a strongly associated with a reduced risk of subsequent CDAI flare and EGA flare but not PGA flare. Predictors of increase in both PGA and EGA included black ethnicity and RF positivity. However, while the TJC and CRP predicted EGA flare, the MD-HAQ was the strongest predictor of PGA flare. Understanding the factors that influence evaluator and patient perception of disease flare is important, as this may have implications for treatment decisions.

Table 1: Factors associated with subsequent CDAI, EGA or PGA flare from low disease activity. *p<0.05, **p<0.01, ***p<0.001, ns – non-significant