American College of Rheumatology criteria for improvement in rheumatoid arthritis should only be calculated from scores that decrease on improvement

Objective Change in a patient's condition is expressed as a percentage of the baseline value for a core set of measures in the American College of Rheumatology (ACR) improvement criteria for rheumatoid arthritis (RA), and this is used as the basis to decide whether a patient has improved. The result is dependent on whether the underlying measure has a score that increases or decreases on improvement. We examined the importance of this effect in the application of the ACR improvement criteria. Methods Data were obtained from the COBRA trial, in which 155 patients with early active RA had been randomized to receive either combination treatment with step-down prednisolone, methotrexate, and sulfasalazine or sulfasalazine alone. Patient and physician global assessments were recoded to reflect decreasing scores on improvement. The effects of this difference in scoring systems were compared among 3 response criteria levels (20%, 50%, and 70%) that are currently being used to assess improvement in RA clinical trials. Results Analyses showed that the effects of a decreasing, versus increasing, score on the designation of improvement cannot be ignored, especially at higher percentages of improvement (e.g., 50%, 70%). Conclusion We recommend that percentage improvement in RA be calculated only on scores that decrease on improvement. When necessary, raw data should be recoded before the ACR improvement criteria are applied.