Evaluation and documentation of rheumatoid arthritis disease status in the clinic: which variables best predict change in therapy.

To determine in clinical practice which rheumatoid arthritis (RA) clinical status variables are most associated with a change in disease modifying antirheumatic drug (DMARD) therapy, we studied 26,240 observations from 1905 RA patients occurring over 25 years. Variables included tender joint count, erythrocyte sedimentation rate (ESR), grip strength, visual analog scale for pain, global severity, fatigue and sleep, Health Assessment Question functional disability scale (HAQ), anxiety, depression and morning stiffness. Only the tender joint count required a physician. Observations at which a change in DMARD therapy occurred were compared to those where a change did not occur using generalized estimating equations (GEE) and classification and regression tree analysis (CART). Tender joint count, pain, global severity, and ESR were the 4 variables most strongly predictive of DMARD change. CART modeling indicated a special role for fatigue and sleep disturbance in some patients. These data add support in clinical practice for the ACR core set and the DAS set of variables. In addition, they validate the use of these variables in a practice setting. We suggest a minimum set of evaluations comprising: joint count, ESR or CRP, measures of pain and/or severity, a fatigue scale (fatigue being a surrogate for sleep disturbance), and a measure of function such as the HAQ or modified HAQ. Because only joint count requires physician participation, these evaluations are practical for the clinic, and allow quantitative measurement of RA status. With the use of quantile charts, the comparative status of RA and the change in RA status can be determined easily.