Algorithm development for corticosteroid management in systemic juvenile idiopathic arthritis trial using consensus methodology

Background: The management of background corticosteroid therapy in rheumatology clinical trials poses a major challenge. We describe the consensus methodology used to design an algorithm to standardize changes in corticosteroid dosing during the Randomized Placebo Phase Study of Rilonacept in Systemic Juvenile Idiopathic Arthritis Trial (RAPPORT). Methods: The 20 RAPPORT site principal investigators (PIs) and 4 topic specialists constituted an expert panel that participated in the consensus process. The panel used a modified Delphi Method consisting of an on-line questionnaire, followed by a one day face-to-face consensus conference. Consensus was defined as >= 75% agreement. For items deemed essential but when consensus on critical values was not achieved, simple majority vote drove the final decision. Results: The panel identified criteria for initiating or increasing corticosteroids. These included the presence or development of anemia, myocarditis, pericarditis, pleuritis, peritonitis, and either complete or incomplete macrophage activation syndrome (MAS). The panel also identified criteria for tapering corticosteroids which included absence of fever for >= 3 days in the previous week, absence of poor physical functioning, and seven laboratory criteria. A tapering schedule was also defined. Conclusion: The expert panel established consensus regarding corticosteroid management and an algorithm for steroid dosing that was well accepted and used by RAPPORT investigators. Developed specifically for the RAPPORT trial, further study of the algorithm is needed before recommendation for more general clinical use.