A key goal in treating rheumatoid arthritis (RA) with conventional disease-modifying anti-rheumatic drugs (DMARDs) and biologic agents is reducing inflammatory synovitis, which reduces disability and maximises quality of life. Reductions in synovitis with treatment are assessed using the disease activity score for 28 joints (DAS28) [1]. The European League Against Rheumatism (EULAR) classifies good responders as having DAS28 scores of 3.2 or less with reductions in DAS28 of more than 1.2 [2]. Patients with DAS28 scores over 3.2 who have reductions in DAS28 scores of more than 1.2 have moderate responses. Patients with reductions in DAS28 of less than 0.6 are non-responders. The treat-to-target initiative recommends increasing treatment until patients achieve remission or low disease activity [3].

High-cost biologic treatments such as tumour necrosis factor inhibitors are used within agreed guidelines. When patients achieve good EULAR responses there is strong rationale for maintaining treatment. Similarly, when patients do not respond there is no reason to continue treatment. However, there is uncertainty in patients with incomplete responses. The British Society for Rheumatology (BSR), recommends patients with active RA, who have not responded to methotrexate and another DMARD, should be considered for biologic therapy [4]. Provided they achieve reductions in DAS28 of over 1.2 by 6 months they should continue receiving biologic treatments. This approach was accepted by the National Institute for Health And Care Excellence (NICE) in its health technology appraisals of biologic treatments for RA [5, 6]. Initially NICE recommended only continuing treatment if DAS28 is reduced by 1.2 or more [5], though recent guidelines recommend patients achieve at least a moderate EULAR response for treatment to continue [6].

The impact on disability and quality of life of reducing disease activity are assessed using the Health Assessment Questionnaire (HAQ) [7] and EuroQol-5D (EQ5D) [8]. There are doubts about the value both to patients and the funders of healthcare from maintaining high-cost treatments if patients only achieve moderate EULAR responses. We hypothesised that patients with good EULAR responses have substantially larger improvements in HAQ and EQ5D scores than patients with moderate EULAR responses, who do not achieve low disease activity or remission, and in whom the final DAS28 scores remain over 3.2. We tested our hypothesis in secondary analyses of two published randomised controlled trials (RCTs). One trial, cost-effectiveness of treatment strategies using combination of disease modifying anti-rheumatic drugs and glucocorticoids in early rheumatoid arthritis (CARDERA), studied intensive DMARD treatment in early RA [9]. The other trial, tumour necrosis factor inhibitors against combination intensive therapy (TACIT), studied intensive treatment with DMARDs and biologic agents in established RA [10]. In both trials we compared changes in HAQ and EQ5D scores in patients achieving good EULAR responses with patients achieving reductions in DAS28 scores over 1.2, and who had only moderate EULAR responses.

Our secondary analyses of two large English RCTs in early and established active RA shows that patients with good EULAR responses have larger clinically meaningful improvements in disability and quality of life than patients with moderate EULAR responses. These findings support the concept of treat-to-target. They indicate that achieving low disease activity or remission states should be the treatment goal. The benefits of good EULAR responses are seen in both early and established RA. The findings were similar if we used the earlier NICE criterion of an improvement in DAS28 of more than 1.2.

The reliance of current NICE guidance on changes in DAS28 of more than 1.2 has been criticised by Jerram et al [14], who point out that this threshold was not used in clinical trials. This concern is ameliorated by the new NICE guidance. Judging patients’ responses to DMARDs and biologic therapy is challenging. EULAR response criteria were designed for use in individual patients and trials. The other major composite response criteria, the American College of Rheumatology (ACR) criteria, are only intended for trials. These two validated response criteria have been compared in several studies [15, 16]; they measure similar changes. However, their use in routine clinical practice and relating their findings to patients’ perceived responses requires more research. Gulfe et al. showed that when response criteria are used at the individual patient level the results are difficult to interpret [17]. Ward et al. [18] found patients’ perceptions of improvement may differ from conventional assessments like ACR20 responses, and that thresholds of minimal clinically important improvement may be larger than previously thought, including the need for a DAS28 improvement of at least 1.2 [19]. Among patients receiving treatment for active RA, three months may be the optimal time point to increase the intensity of treatment [20]. However, when assessing response to biologic therapy, real-world evidence suggests a six-month time point is needed to optimise patients’ responses [21].

Low disease activity optimises key patient-related outcomes like HAQ and EQ5D scores [22–24]. Our own findings mirror these previous reports. Remission is one potentially important outcome, although only a minority of our patients achieved this. Interestingly some other trials in early RA, such as the FIN-RACo trial [25], reported higher remission rates; the reasons for this heterogeneity are uncertain. When patients achieve sustained remission, they have less long-term work disability [26]. Some patients with moderate EULAR responses in our trials had good reductions in disability and improvements in quality of life. However, many did not and the best management of such patients is uncertain. Continuing to prescribe high-cost biologic therapy for patients with modest improvements in disease activity and little or no reduction in disability seems questionable. It may be preferable to continue biologic treatments in patients with moderate EULAR responses who have also had clinically meaningful reductions in disability.

Our report has several strengths. The two trials were large, there was reasonable patient retention on treatment, and they were based across many English specialist centres with a broad geographic spread. We found similar effects in trials of early and established RA. Our findings are therefore likely to be generalisable in guiding clinical practice decisions.

Our analyses also have limitations. First, we only considered differences at the trial endpoints. Different effects would have been seen by evaluating all time points. Second, choosing specific cutoff points, such as DAS28 of 3.2 or less may have over done similarly well. Third, some of the treatments used, particularly ciclosporin, which was used in the CARDERA trial, are not commonly prescribed in routine practice. Fourth, we combined the effects of a range of different treatments, and the relationship between disease activity and HAQ and EQ5D scores may be different across drug classes. Finally, our inference that patients who do not achieve low DAS28 scores with one treatment strategy may do better with another could be incorrect; some patients may achieve poor outcomes with all treatment strategies.

Achieving good EULAR responses to treatment with DMARDs and biologic agents leads to substantial improvements in HAQ and EQ5D scores. Patients who achieve such responses should continue on treatment. However, continuing patients on intensive treatment regimens, particularly using high-cost biologic agents, is more challenging if they only achieve moderate EULAR responses. Evidence of additional clinically important benefits in measures such as the HAQ should also be sought to justify continuing treatment in these patients.

ACR, American College of Rheumatology; CARDERA, Cost-effectiveness of treatment strategies using combination of disease-modifying anti-rheumatic drugs and glucocorticoids in early rheumatoid arthritis; CI, confidence interval; DAS28, twenty-eight joint disease activity score; DMARD, disease-modifying anti-rheumatic drug; EQ5D, Euroqol-5D (a standardized instrument for measuring generic health status); EULAR, European League Against Rheumatism; HAQ, Health Assessment Questionnaire; MCID, minimal clinically important difference; NICE, National Institute of Health and Care Excellence; RA, rheumatoid arthritis; RCT, randomized controlled trial; TACIT, tumour necrosis factor inhibitors against combination intensive therapy