EULAR DAS28 Response Criteria Interpreter

The response criteria that need two facts, not one: how much the DAS28 fell AND where it has landed. The same fall of 1.5 is a good response from 4.5 and only a moderate one from 6.5.

Last updated 8 Oct 2026

EULAR response

2 DAS28 values → response category

DAS28 at baselineThe earlier value. Both values must come from the SAME version of the DAS28: the ESR and CRP versions are different formulas and published mean differences between them run from 0.30 to 0.72, which is most of the 0.6 and 1.2 thresholds these criteria turn on. DAS28-ESR and DAS28-CRP

DAS28 nowThe later value. This number does double duty: the criteria test both the fall from baseline and the value itself, which is what makes them different in kind from the ACR response criteria.

Moderate responseExample

DAS28 6.0 at baseline, 4.0 now

A 3×3 table over two facts

Good response = improvement > 1.2 AND present DAS28 ≤ 3.2
No response = improvement ≤ 0.6, OR improvement ≤ 1.2 with present DAS28 > 5.1
Moderate response = everything else

two facts, not one: every other response criterion on this site is a function of the change alone — the ASDAS thresholds, the ACR percentages. These criteria read the change AND the destination, which is why the same fall can be a good response or a moderate one
> 1.2 and > 0.6: the improvement thresholds, and both are STRICT. An improvement of exactly 0.6 is not an improvement greater than 0.6, and an improvement of exactly 1.2 cannot produce a good response however low the current value
≤ 3.2 and ≤ 5.1: the DAS28’s own activity band edges. Sources differ on whether the bands own their upper bounds; this page follows the inclusive reading used in the response criteria themselves, which is also what DAS28-ESR does
a floating-point caution: neither 0.6 nor 1.2 is exactly representable in binary. A DAS28 reported to two decimals can give a difference one unit in the last place either side of a threshold, so a patient whose fall is exactly 0.60 or 1.20 may be classified differently by two calculators that both implement the published rule correctly. This page subtracts first and compares afterwards
one version at both visits: published mean DAS28-ESR minus DAS28-CRP differences are 0.30 in a UK cohort, about 0.55 in pooled early-RA trial data and 0.72 in a Japanese cohort of 3,073 patients. Any of those is a large fraction of the 0.6 threshold, so mixing versions manufactures a response or hides one
not a remission criterion: a good response is not remission, and a patient can be a good responder while still having active disease by the ACR/EULAR definitions. Boolean and index-based remission criteria computes those

Worked example

DAS28 6.0 at baseline, 4.0 now
Improvement: 6.0 − 4.0 = 2.0, which is greater than 1.2
Current value: 4.0, which is above 3.2
A good response needs BOTH a fall greater than 1.2 and a current value of 3.2 or less. The fall qualifies; the destination does not, so this is a moderate response
THE SAME FALL, TWO ANSWERS: 4.5 → 3.0 is a fall of 1.5 to a current value of 3.0, which is a good response. 6.5 → 5.0 is the same fall of 1.5 to a current value of 5.0, which is only a moderate response. Identical improvement, different category
A BIG FALL THAT IS STILL ONLY MODERATE: 8.0 → 5.5 is a fall of 2.5, and 5.5 is above 3.2, so it is a moderate response
A REAL IMPROVEMENT THAT IS NO RESPONSE: 6.0 → 5.2 is a fall of 0.8 — more than 0.6 — but 5.2 is above 5.1, so it falls in the no-response cell
THE STRICT THRESHOLDS: a fall of exactly 1.2 to a current value of 0.1 is a moderate response and not a good one, because the criterion is ‘greater than 1.2’. A fall of exactly 0.6 is no response, for the same reason

The published table, in full

Nine cells, three outcomes, and only one cell gives a good response. Reading down the first column shows what makes these criteria unusual: the same large improvement is a good response, a moderate response or a moderate response depending only on where the patient has landed. Reading across the bottom row shows the other half: a patient still above 5.1 cannot be called a responder on anything less than a 1.2-point fall.
Present DAS28	Improvement > 1.2	Improvement > 0.6 and ≤ 1.2	Improvement ≤ 0.6
≤ 3.2	Good response	Moderate response	No response
> 3.2 to 5.1	Moderate response	Moderate response	No response
> 5.1	Moderate response	No response	No response

The same patient on two DAS28 versions

These figures are carried from this site’s own DAS28-ESR page, where they are cited in full. They matter here because every one of them is a large fraction of the 0.6 improvement threshold: a baseline measured on ESR and a follow-up measured on CRP can manufacture a moderate response out of nothing, or hide one. Use the same version at both visits.
Study	Population	Mean DAS28-ESR − DAS28-CRP
Hensor et al, 2019	UK cohort	0.30 (0.35 in women and over-50s)
Fleischmann et al, 2017	COMET and PRIZE early-RA trial data, n = 834	about 0.55
Matsui et al, 2007	Japanese observational cohort, n = 3,073	0.72 (mean 4.31 against 3.59)

Why these criteria need two numbers, and what the moderate cell hides

Van Gestel and colleagues published the EULAR response criteria in 1996, a year after the ACR published its own, and they took a different shape on purpose. The ACR definition asks only how much a patient has improved. The EULAR criteria ask that and also where the patient has got to, and they will not call a response good unless both answers qualify: the DAS28 must have fallen by more than 1.2 and must now be 3.2 or less. A patient who falls from 8.0 to 5.5 has improved by 2.5 points and is a moderate responder, because 5.5 is still high disease activity. A patient who falls from 4.5 to 3.0 has improved by 1.5 and is a good responder. The second patient improved less and responded better, which is exactly what the criteria are built to say.

The two value thresholds are not new numbers: 3.2 is the top of the DAS28’s low-activity band and 5.1 the top of its moderate band, so the criteria are simply the activity bands read alongside the change. That also imports the DAS28’s own boundary ambiguity. Sources differ on whether a DAS28 of exactly 3.2 is low activity or moderate, and this page follows the inclusive reading used in the response criteria themselves, as DAS28-ESR does. The improvement thresholds are strict rather than inclusive in the other direction: a fall of exactly 1.2 cannot produce a good response, because the criterion is written as greater than 1.2, and a fall of exactly 0.6 is no response at all. There is a genuine floating-point caveat there — neither 0.6 nor 1.2 is exactly representable in binary, so a patient whose fall is exactly 0.60 or 1.20 can be classified differently by two calculators that both implement the rule correctly.

The moderate cell is where the criteria are coarsest, and it is worth knowing what it contains. Five of the nine cells are moderate, reached by two quite different routes: a large improvement that has not arrived, and a middling improvement in a patient who was never that far from the target. A fall from 8.0 to 5.5 and a fall from 4.0 to 3.3 share a label. The no-response cell has its own trap in the opposite direction: an improvement of a whole point counts as no response if the patient remains above 5.1. And the criteria do not distinguish improvement from deterioration at the bottom end — there is no worsening category, so a rise of 3 points carries the same label as a fall of 0.5.

Both values must come from the same version of the DAS28, and the published differences between the ESR and CRP versions are large relative to the thresholds here: 0.30 in a UK cohort, about 0.55 in pooled early-RA trial data and 0.72 in a Japanese cohort of 3,073 patients. A baseline on one version and a follow-up on the other can manufacture a 0.6-point improvement out of nothing. Finally, a good response is not remission: a patient can be a good EULAR responder and still fail every ACR/EULAR remission definition, which Boolean and index-based remission criteria computes. The American equivalent of this page is ACR20, ACR50 and ACR70, a conjunction over seven measures rather than a table over two; the axial equivalent is ASDAS improvement and worsening, which is a pure change with no destination term at all. A composite index is not a diagnosis, and a cut-off chosen in one cohort is not a probability for one patient. This states what the instrument measured in a named cohort; the clinician in front of the patient decides what follows.

Frequently asked questions

What are the EULAR response criteria?

A good response requires an improvement in DAS28 of more than 1.2 AND a current DAS28 of 3.2 or less. No response means an improvement of 0.6 or less, or an improvement of 1.2 or less in a patient whose current DAS28 is still above 5.1. Everything else is a moderate response. They were published by van Gestel and colleagues in 1996.

Why do the criteria need the current DAS28 as well as the change?

Because improving a great deal and arriving somewhere acceptable are different achievements. A fall from 8.0 to 5.5 is an improvement of 2.5 and only a moderate response, because 5.5 is still high disease activity. A fall from 4.5 to 3.0 is an improvement of 1.5 and a good response. The criteria judge the destination as well as the distance.

Is a good EULAR response the same as remission?

No. A good response needs a current DAS28 of 3.2 or less, which includes patients in low disease activity rather than remission, and DAS28 remission itself (below 2.6) is looser than the ACR/EULAR remission definitions. A patient can be a good responder and meet no remission definition at all.

Can I mix DAS28-ESR and DAS28-CRP between the two visits?

No. Published mean differences between the two versions are 0.30 in a UK cohort, about 0.55 in pooled early-RA trial data and 0.72 in a Japanese cohort of 3,073 patients — each a large fraction of the 0.6 improvement threshold. Mixing versions can manufacture a moderate response or hide one.

Is a fall of exactly 1.2 a good response?

No. The criterion is an improvement GREATER than 1.2, so a fall of exactly 1.2 cannot give a good response however low the current value. The same applies at 0.6: a fall of exactly 0.6 is no response. Note that neither threshold is exactly representable in binary, so a patient on the boundary may be classified differently by different calculators.

Do the criteria have a category for worsening?

No. Any change that is not an improvement of more than 0.6 falls into no response, so a rise of 3 points carries the same label as a fall of 0.5. The ASDAS criteria for axial spondyloarthritis do have a separate worsening threshold, at a rise of 0.9 or more.

Related calculators

References

van Gestel AM, Prevoo MLL, van ‘t Hof MA, et al. Development and validation of the European League Against Rheumatism response criteria for rheumatoid arthritis. Arthritis Rheum. 1996;39(1):34–40.
OMERACT. EULAR response criteria. OMERACT criteria pages. (Reproduces the 3×3 table of present DAS28 against improvement from baseline and cites it to van Gestel 1996.)
Felson DT, Smolen JS, Wells G, et al. American College of Rheumatology/European League Against Rheumatism provisional definition of remission in rheumatoid arthritis for clinical trials. Ann Rheum Dis. 2011;70(3):404–13.
Studenic P, Aletaha D, de Wit M, et al. American College of Rheumatology/EULAR remission criteria for rheumatoid arthritis: 2022 revision. Ann Rheum Dis. 2023;82(1):74–80.
Aletaha D, Smolen J. The Simplified Disease Activity Index (SDAI) and the Clinical Disease Activity Index (CDAI): a review of their usefulness and validity in rheumatoid arthritis. Clin Exp Rheumatol. 2005;23(5 Suppl 39):S100–8.

Not medical advice. For healthcare professionals and education. Reference intervals vary by laboratory and assay — always use your own laboratory's. Never base a dose or a treatment decision on this page alone. Full disclaimer at calcengines.com/disclaimer/