Relative contributions of the components of the American College of Rheumatology 20% criteria for improvement to responder status in patients with early seropositive rheumatoid arthritis

Objective. To evaluate factors that influence the responses defined by the American College of Rheumatology (ACR) 20% criteria for improvement in rheu matoid arthritis (RA). Methods. ACR 20% and 50% response rates were calculated from data collected for the intervals 0-6, 0-12, and 0-24 months for 180 RA patients participa ting in the Western Consortium of Practicing Rheumatologists long-term obse rvational study of early seropositive Ri (mean +/- SD duration of RA at stu dy entry 6.0 +/- 3.4 months). Analyzable cases were patients with paired da ta for tender and swollen joint counts plus at least 3 of the following cri teria: physician's and patient's global assessments of disease activity and patient's score for pain (by visual analog scale), physical function score on the Health Assessment Questionnaire (HAQ), and levels of an acute-phase reactant. Response rates were then recalculated by 3 different methods: 1) using only cases with complete paired data for all criteria, 2) sequential ly assuming no improvement in each of the 5 secondary criteria, and 3) subs tituting grip strength for HAQ scores. Results. Using 464 paired observations for all analyzable cases, ACR 20% (5 0%) improvement rates were 52.6% (33.0%), compared with 55.6% (34.8%) for 3 65 paired observations from the cases with complete data. Decreases in ACR response rates when secondary criteria were sequentially set at "no improve ment" ranged from 11.7% (pain at 0-6 months) to 1.2% (C-reactive protein at 0-12 months), but these were not statistically different by the kappa stat istic. Overall numerical rankings of the relative contributions of the seco ndary criteria to the ACR 20% or 50% response rates were physician's global assessment, pain, HAQ, patient's global assessment, and acute-phase reacta nt. Only 7.8% of paired grip strength observations showed greater than or e qual to 20% improvement, compared with 71% of paired HAQ observations. Conclusion. The use of all "analyzable" cases (paired data for tender and s wollen joint counts plus greater than or equal to3 of the 5 secondary crite ria) increases the number of subjects and only slightly decreases the ACR r esponse rate compared,vith analyses limited to cases with complete data. Th e contributions of the secondary criteria are not statistically different, supporting their equal weighting in the ACR definition of improvement. The ACR 20% response rates are higher when the HAQ, rather than grip strength, is used to measure physical function.