Background/Purpose:

Two definitions of remission have been put forward by the ACR and EULAR: a Boolean based, requiring swollen and tender joint counts (SJC, TJC), C-reactive protein (CRP in mg/dl) and patient global assessment (PGA on a 0-10cm scale, VAS) to be ≤1; and an index based definition, requiring the simplified disease activity index (SDAI) to be ≤3.3. The patient global has been shown to be crucial in fulfilling the criteria (1).

In many settings a numerical rating scale (NRS) is preferred instead of a VAS. Here, we investigated whether the use of a NRS-like assessment would lead to different remission frequencies compared to a VAS.

Methods:

We obtained data of a random cross-sectional visit of RA outpatients from a longitudinal observational database. We used simulated (s)-NRS values, in which VAS values were rounded to the closest integer, for calculation of the SDAI. We compared proportions of patients in Boolean and/or SDAI remission between PGA by VAS and sNRS, evaluating their difference or agreement descriptively by Kappa and receiver operating curve analyses (ROC).

Results:

We identified 922 RA patients (80% female, 56% rheumatoid factor (RF) positive, mean disease duration 8 years). In the main analysis, 12.8% of patients were in Boolean remission using sNRS versus 11.3% using VAS (see table).

All patients in Boolean remission using VAS also were in remission by sNRS. Boolean remission using VAS and sNRS had a high agreement (κ of 0.93). SDAI remission frequencies were higher than Boolean remission frequencies with a good agreement (table), and showed a κ of 0.94 for the comparison of VAS and sNRS. In sensitivity analyses, in which we rounded all VAS values either up to the next higher integer, or down to the next lower integer to obtain the sNRS, confirmed the main analysis (table).

When using SDAI to test for Boolean remission in a ROC analysis, SDAI evaluated by VAS and by sNRS showed similar characteristics (AUC=0.99 for both), and sensitivity and specificity for the SDAI remission cutoff of 3.3 were 0.96 and 0.94 for VAS and 0.90 and 0.95 for sNRS.

Table. Frequencies of remission using different types of PGA

Conclusion:

As expected SDAI remission rates were somewhat higher than Boolean remission rates. This was consistent regardless of the method used to assess the patient global. While Boolean remission was about 15% more frequent using sNRS than VAS, SDAI remission was numerically lower with sNRS than VAS. Thus, using a sNRS instead of a VAS for PGA assessment seems to have little impact regarding the achievement of both SDAI and Boolean remission so that sNRS may be a valid tool for PGA assessment in the new remission criteria. However, this assumption will have to be tested using true rather than simulated NRS.

1.            Vermeer M, et al. The provisional ACR/EULAR definition of remission in RA: a comment on the patient global assessment criterion. Rheumatology 2012;51:1076-80.