BIOANALYSISPAPER
Rheumatoid arthritis has had one goal for decades: remission, judged by a score that mixes lab markers, joint counts and how the patient feels.
Rheumatoid arthritis is an autoimmune disease in which the immune system attacks the joint lining, and it erodes bone over time. There is no cure, so care is a long game of measuring inflammation and holding it down, and the measure has been one yes-or-no question for decades: remission, or not remission. The paper in the November 2026 issue of the Annals of the Rheumatic Diseases replaces that question with three numbers, and the gap they open is wide: in the Veterans Affairs Rheumatoid Arthritis registry, 44.0% of patients met the new low-inflammation threshold, against 32.1% who met remission, with the same ordering in all three trial datasets.
The problem underneath those numbers is the measuring instrument, and the American College of Rheumatology's plenary abstract for the work names it: scores "can be influenced by comorbidity and chronic pain, do not always reflect active inflammation," and providers "often do not escalate therapies in the context of elevated scores." The standard disease activity scores mix laboratory markers, joint counts and the patient's own rating of how they feel, and chronic pain, osteoarthritis, spine degeneration and mood disorders all inflate that last input. One number can flag a patient whose pain is genuine but whose inflammation is quiet. It can also pass a patient whose formal remission sits alongside erosions and nodules.
The scores are modified versions of the standard ones, built by the same group in 2014 after the tender-joint count and the patient's own global rating were dropped, because those components tracked MRI-detected joint inflammation and damage on imaging less well than the rest of the score, according to the consortium's earlier Arthritis & Rheumatology paper. In the new study the team set the cut-points by asking which modified score best separates traditional remission at month six, using receiver-operating-characteristic analysis in one trial of 530 patients, then tested the thresholds against what happened next: joint damage on imaging, cardiovascular events, death. The cut-points landed below 1.4 on the modified 28-joint disease activity score, 3 or below on the modified simplified disease activity index, and 2 or below on the modified clinical disease activity index.
The lower bar does not cost joint protection. In the Abatacept in Inadequate Responders to Methotrexate trial, the Genant-Sharp measure of joint-space narrowing rose by 0.88 units among patients who met the new threshold, against 1.86 among those who did not, a difference the conference abstract reports as -0.97 with a 95% confidence interval of -1.80 to -0.14. Traditional remission did not separate the same group: 1.39 units against 1.68, a difference of -0.28 whose interval runs from -1.29 to 0.73. In the registry, the new category also predicted cardiovascular events and early death better than remission did, with a hazard ratio for death of 0.65 (95% CI 0.57 to 0.73) against 0.75 (0.65 to 0.85). Those figures sit in the meeting abstract; the journal abstract gives the same direction without the numbers.
And the patients who clear the new bar but still fail the old one are not hiding more arthritis. In the registry they carried higher rates of comorbidity, including anxiety and depression, and lower rates of seropositivity, the antibodies that mark this form of arthritis, and of erosive disease, damage already eaten into bone. Baker described the same split in the meeting coverage: "But when you look at those that are in RALI but that are not in traditional remission, you see high rates of spine degeneration, osteoarthritis, mood disorder, and seronegativity." A score built partly on how a patient feels will read those people as active disease. That is a design problem in the measurement, not a personal failing.
Three limits deserve the same space as the numbers. The thresholds were derived and validated by the same consortium, in its own trial and registry data, with no external or prospective validation cohort reported; what has been shown is internal consistency and prognostic performance, not that the categories are natural truths. The registry population is predominantly male US veterans, so the size of the 44.0% versus 32.1% gap may not travel to a typical outpatient population, even though the ordering held in all three trial datasets and the meeting coverage describes the comparison as spanning four independent data sources. Nothing here tests a treatment: the study shows that patients who met the new threshold had less joint damage and fewer cardiovascular events and deaths, not that aiming at the threshold changes any of that. The publisher's metadata lists the Department of Veterans Affairs and the National Institute of General Medical Sciences as funders, and the meeting abstract's disclosure list shows Baker reporting a financial tie to Formation Bio.
So the numbers are computable from data a rheumatologist already collects, and the paper concludes that they work as "a point-of-care clinical target to inform management." The rulebook has not moved: as of the November 2026 issue, the 2022 ACR/EULAR revised remission criteria and the 2021 ACR and 2026 EULAR rheumatoid arthritis recommendations do not mention these thresholds. The open decision is whether a care plan that meets the new bar counts as control, or whether it belongs to a patient with low inflammation, depression, back pain and osteoarthritis who needs a different plan than a medication change.