When the panel convened in Chicago in May 2009, they were shown a scatterplot of their rankings of the cases to initiate discussions about differences of opinion. During this meeting, the panel identified four domains associated with an increasing probability of developing RA: joint involvement, serology, duration of synovitis, and acute phase reactants. These later became the key domains in a 10-point score. Consensus panel discussions also generated what they considered “mandatory requirements” to fulfill the classification of RA: evidence of synovitis in at least one joint, as validated by an expert clinician or imaging, and signs or symptoms that could not be explained by other diseases or conditions.
To evaluate the weights, and thus the scores, attached to each of the categories within the four domains, the panel used a series of paired discrete choices and a novel decision analytic software, 1000 Minds. Those weights then determined the scores assigned to each category; these category scores were then summed to produce an overall patient score, which represented the patient’s probability of developing persistent or erosive RA (from 0, very low probability, to 100, very high probability). The program deduced that involvement of more than 10 joints, including small joints, had a very strong impact on a patient’s score of probable RA. The next highest influence was high positive serology, following by longer duration of synovitis and positive acute phase response. The results of the consensus meeting were remarkably congruent with findings from the Phase One analysis, said Dr. Hawker.
The task force presented its preliminary results at the 2009 EULAR meeting in Copenhagen. Following revisions in response to feedback, the panel refined their definitions of synovitis, joint involvement, high- and low-titer serology, and acute phase reactants, and determined the strategy for classification of patients with erosive changes. They then rescored the original case scenarios using a 100-point scale to verify rank order. In their final phase, they derived cut points for definite RA (60 out of 100 or 6 out of 10) and simplified the scoring system, from a 100-point to a 10-point scale.
At the end of her October 2009 presentation at the ACR annual meeting, Dr. Hawker walked her audience through both a tree algorithm and the 10-point classification scale, both of which are now available as part of the published criteria and on the ACR website at www.rheumatology.org/practice. Later, during the question-and-answer period, presenters urged audience members to test the criteria in a prospective manner. “Once the criteria are published, our call to everyone is to go back to their own cohorts and test these new criteria in all possible ways,” said Dr. Aletaha.
