Assessing the ACR/EULAR Criteria’s Performance
In addition to not having been derived from patients in routine practice, the ACR/EULAR definitions of remission may also show variability from physician to physician, and unstable reproducibility within individual patients.
Last November, researchers from the Arthritis and Rheumatology Clinics of Kansas in Wichita and their collaborators published an analysis that applied the ACR/EULAR provisional remission definition to patients in two cohorts: 1,341 patients from the Department of Veteran’s Affairs, and 1,153 patients in a cohort assembled at the Arthritis and Rheumatology Clinics of Kansas. The two ACR/EULAR criteria identified about 5%-10% of patients as being in remission at any one time, depending on the cohort and specific definition applied, they reported (Arthritis Rheum. 2011;63:3204-15).
However, the probability of a specific patient being in remission at two or more visits ranged from 2%-5%. The analysis also showed "substantial evidence to indicate that inter-rater reliability is poor with respect to the examination of tender and swollen joints." The authors estimated that the probability of remission using the ACR/EULAR definitions in these two cohorts could vary by twofold depending solely on the physician examiner and independent of disease activity.
The researchers concluded that "problems with reliability and agreement limit the usefulness of these criteria in the individual patient." But several rheumatologists interviewed noted that variability among physicians in scoring swollen and tender joint counts is expected. The finding simply underscores that the best way to serially monitor joint status in a patient is for the same physician to do it every time. They also commented that the low rates of remission seen in this study highlight just how challenging it is to maintain RA patients at a very low level of disease activity.
"These criteria were developed for clinical trials, and I agree that they’re reasonable for trials," said Dr. Frederick Wolfe, the senior author on this study and a member of the ACR/EULAR definition panel. "Until now, there were lots of different ideas of what was remission, which limited its use. We wanted to make a standard for remission [in trials], and did a pretty good job. But in the clinic, physicians know more than they can get from questionnaires," he said in an interview.
For example, Dr. Wolfe cited a study he and his associates ran in which patients were asked what influenced their global self rating. "We found that some patients who all physicians would agree were in remission gave themselves poor scores because they had pain in other regions, mostly back pain, or because of fatigue, or other things" not related to RA. "In clinical practice, we’d ask patients to ignore that, or to explain their high global self-assessment score, so in the clinic you can decide whether patients are in remission," said Dr. Wolfe, director of the National Databank for Rheumatic Diseases in Wichita, Kan.
For measuring RA activity in routine practice, Dr. Wolfe said that he favors the CDAI "as an overall measure of disease activity." He also endorsed patient-centered measures of pain, such as the Health Assessment Questionnaire (HAQ), global self-assessment, RAPID3, and the PAS.
A more positive assessment of the ACR/EULAR definitions in clinical practice came from a recent study that applied them to a cohort of 535 RA patients at Brigham and Women’s Hospital in Boston followed for 2 years by serial joint radiographs. The authors examined the ability of the definitions to predict good radiographic and functional outcomes over time in these clinical-practice patients. They found that 30 patients (6%) of the 535 met the four-part ACR/EULAR definition of remission at baseline, while 26 (5%) met the CDAI definition, 37 (7%) met a SDAI definition of remission, and 106 (20%) met a DAS28-CRP definition. The best of these measures for predicting freedom from radiographic joint progression during 2 years of follow-up was the CDAI, with a positive likelihood ratio of 2.8, followed closely by the four-part ACR/EULAR definition, with a positive likelihood ratio of 2.7. The ratio for SDAI was 2.1, and for DAS28-CRP 1.5 (Ann. Rheum. Dis. 2011 Oct. 12 [doi:10.1136/ard.2011.154625]).
"Our findings strengthen the applicability of the ACR/EULAR criteria in clinical practice," the researchers concluded, and also noted that "the findings might have been different in a patient population of early RA patients." The cohort they examined had been diagnosed with RA for a median of 11 years, with 75% diagnosed for 4 years or longer.
The authors also said "a significant proportion of patients who do not fulfill the remission criteria still experience a good outcome of their disease, especially for radiographic joint damage. ... A byproduct of more stringent remission criteria as a treatment target will be more patients in low disease activity with good disease outcome, potentially risking overtreatment."