Statistics from Altmetric.com
New therapeutic options for rheumatoid arthritis (RA) have become available during the past decade, rendering remission an achievable goal in clinical trials and also in daily practice.1 However, the percentage of patients achieving this goal is strongly influenced by the remission criteria used. A committee of representatives from the American College of Rheumatology (ACR) and the European League Against Rheumatism (EULAR) recently developed new provisional definitions of remission in RA.2 Two alternative definitions were proposed: either a Boolean definition (number of swollen and tender joints each ≤1, C-reactive protein (CRP) ≤1 mg/dl, and patient global assessment (PtGA) ≤1), or an index-based definition with a Simplified Disease Activity Index (SDAI) score ≤3.3. Since the new criteria were developed and evaluated using data from randomised clinical trials, the committee encouraged further evaluation with observational data. Meanwhile, it has been shown in observational studies that the major reason for not achieving remission according to the Boolean definition was PtGA>1.3–5 Moreover, these studies showed that PtGA was influenced by RA disease activity and by non-inflammatory factors—for example, fatigue, functional limitations or low back pain.3 ,6 Interestingly, in the BEST study7 no difference in Health Assessment Questionnaire (HAQ) scores between patients with early RA in SDAI or 28-joint count Disease Activity Score (DAS28) remission was found. At the same time, between 20% and 30% of patients classified as being in remission by any of the definitions showed significant radiographic damage over 2 years, while 50–60% of those not in remission by any criteria did not show radiographic progression.8
This study used data from a large epidemiological database to examine which unselected real-life patients at different stages of disease would meet the DAS28, the SDAI or the new Boolean remission criteria. The main study question was: Which of those patients who would have reached the study end point remission according to the old DAS28 criteria would not reach this end point when the new remission criteria were applied?
Which of the predictors of missing the new criteria can we influence with RA-specific treatments and which are independent background risks that we have to take into account when interpreting the results in order to avoid false conclusions? We investigated the contribution of patient-reported outcomes that are not part of the remission criteria such as functional status, pain, morning stiffness and fatigue, as well as physician-reported comorbidity to explain discrepancies between the criteria. Since prevention of functional disability is a major aim of remission-oriented treatment and the new criteria were explicitly selected for their ability to predict good functional outcome, we compared the functional ability of patients in remission according to the DAS28, SDAI and Boolean definition and also contrasted them with controls from a large survey which assessed the functional status in the general population.
Patients and methods
The national database of the German Collaborative Arthritis Centres is an ongoing prospective study which began in 1993 as a long-term monitoring system for German rheumatology.9–11 It contains annually updated clinical data and patient-reported outcomes for unselected outpatients with inflammatory rheumatic diseases. The database received study approval from the ethics committee of the Charité University Medicine Berlin (EA1/196/06).
In this study, data from the years 2007–9 were used. Patients who were seen more than once during this period were included with the first available visit. Only patients with valid data on all required remission items were analysed. For the Boolean remission criteria these items include the numbers of swollen and tender joints (28-joint count), the CRP (mg/dl) and PtGA (numerical rating scale (NRS) 0–10). The SDAI definition additionally includes physician global assessment of disease activity (‘What is your assessment of the patient's current disease activity?’ (NRS 0=no activity, 10=extremely high activity)). The DAS28 remission rate was calculated from the numbers of swollen and tender joints, PtGA and the erythrocyte sedimentation rate (ESR).
PtGA can be measured as global assessment of health or global assessment of disease activity. For the majority of patients in our study we had values only for global assessment of health. However, we were able to compare the two assessments for a subsample of 1660 patients. The wordings of both scales were ‘In general, how would you describe your health today?’ and ‘How would you assess the activity of your arthritis today?’. We found that 77% of the patients reported exactly the same rating for global health and global disease activity, and 14% patients exhibited a marginal difference of one unit. Only 9% of patients reported ratings that differed by two or more points on the two scales.
The national database comprises further physician-derived information, such as onset of symptoms, comorbidities and rheumatoid factor. The patients assessed pain and fatigue on NRS from 0 (best) to 10 (worst status), morning stiffness (duration in minutes) and functional status. Functional status was assessed with the Hannover Functional Status Questionnaire (FFbH), which is widely used in Germany and is highly correlated with the HAQ. FFbH scores range from 0 to 100 to indicate the percentage of full functional ability, and these scores can be transformed into HAQ values.12
We first compared clinical and patient-derived data from patients fulfilling the three remission criteria with patients not in remission by any criteria. We then investigated which parameters were responsible for patients in DAS28 remission missing one of the new criteria.
By univariate logistic regression analysis we selected variables that predicted that patients in DAS28 remission would violate either the Boolean or the SDAI remission criteria (see online supplement 1). Based upon the results of this analysis, in a multivariate logistic regression we determined the adjusted effects of these preselected variables. The constituent parts of the three remission definitions were excluded in order to identify which other variables were predictive of failing the criteria, being aware that some of them were highly correlated with the criteria parameters.
Finally, we compared the functional status of patients in remission with an age- (1-year steps) and sex-matched population sample. The population sample included 9263 randomly selected people who were investigated for their physical disability in the framework of a population study on back pain.13 The 12-item version of the FFbH14 that was applied in the population was also calculated from our data. The correlation between the 18-item and the shortened 12-item FFbH versions was high, with a correlation coefficient of r=0.99.
SPSS V.19.0 was used for data analysis.
Comparison of patients in remission according to the three criteria
A total of 6864 patients with RA according to the 1987 ACR criteria15 were analysed. Seventy-five per cent of these were female; they had a mean age of 61.5, a median disease duration of 9.1 years and a mean DAS28 of 3.4.
Table 1 shows characteristics of patients in remission by any of the criteria compared with those not in remission. Twenty-eight per cent of the patients were in DAS28 remission while only 7% fulfilled the Boolean criteria and 11% the SDAI remission criteria (For agreement between the remission criteria see supplement 2.) Patients in remission had an approximately 3 years shorter disease duration than patients who were not.
There were only small differences between the three remission groups in swollen or tender joint counts and moderate differences concerning acute-phase reactants or physician global assessment. However, large differences were seen in patient-reported outcomes: while in the Boolean group, for PtGA, according to definition only values of 0 or 1 are allowed, 37.7% of patients in SDAI and 75.8% of those in DAS28 remission had scores >1. Further, approximately 95% of the patients who were classified as being in remission according to the new criteria had little or no pain (0–3 on a rating scale from 0 to 10) compared with 70% of patients in DAS28 remission. This was also reflected in fatigue, morning stiffness and functional status. The percentage of patients with potentially disabling comorbidities (osteoarthritis, osteoporosis and degenerative spine disease) was remarkably higher in patients in DAS28 remission than in the two groups fulfilling either of the new criteria.
Comparison of patients in DAS28 remission by fulfilment of the new criteria
Of the 1931 patients in DAS28 remission, 21% were also in Boolean remission, 34% also in SDAI remission (table 2). Fifty-eight per cent of the patients in index-based remission also fulfilled the Boolean criteria.
The most striking differences were seen for PtGA, pain, fatigue and function. Ninety-six per cent of the patients who missed the Boolean criteria and 95% of those who missed the SDAI criteria had a PtGA>1. Similar differences were seen for patient-reported outcomes that are not included in any of the definitions.
Patients in DAS28 remission not meeting the new criteria had significantly lower physical function and more morning stiffness beyond 30 min than those who met any of the new criteria. They also had more chronic musculoskeletal disorders, but there was no difference in other comorbidities, including coronary heart disease.
Using logistic regression analysis we examined which variables predicted that patients in DAS28 remission would miss the new criteria. For the Boolean definition, longer disease duration, higher age, male gender, presence of spine disease and higher scores for pain and fatigue increased the risk of not achieving remission (table 3). The likelihood of not achieving SDAI remission was increased by the same variables, with the exception of age and male gender. In addition, positive rheumatoid factor, morning stiffness and poor function were associated with missing SDAI remission.
Functional status of patients in remission according to the three sets of criteria compared with a population sample
Since chronic musculoskeletal comorbidity played an important role in preventing patients from meeting the new remission criteria, we compared the functional status of the patients with RA in our sample and that of the general population. Data on functional status were available from a large German population study.13 We randomly matched the patients in each of the three remission groups with one person from the population sample of similar age and sex. We achieved matches for 1618 patients in DAS28 remission, 412 patients in Boolean and 628 in SDAI remission.
Figure 1 shows the cumulative frequencies of FFbH values for patients fulfilling or not fulfilling the respective criteria compared with the matched population controls. Patients in DAS28 remission had a functional ability that was identical to that of the age- and sex-matched general population. Forty-five per cent of the patients in DAS28 remission had FFbH12 values >90, which indicates minimal or no limitations in the activities of daily living, similar to 49% in the matched population sample. Patients who fulfilled one of the new sets of criteria had a considerably higher functional ability than their population counterparts: 72% of those in Boolean remission had more than 90% of full function compared with 48% in the population. The same was observed for the SDAI group: 67% of the patients had more than 90% of full function compared with 47% from the matched population.
Comparison of the preliminary ACR/EULAR criteria with DAS28 remission showed that both of the new sets of criteria were more stringent, classifying significantly fewer patients as in remission. Among the new criteria the SDAI was more likely to be met. Patients achieving Boolean remission had a better PtGA as well as lower pain and fatigue scores than those in SDAI remission.
For patients in DAS28 remission, disease duration, pain, fatigue and comorbid spine disease were significantly associated with not achieving both new sets of remission criteria. FFbH and age were significantly associated with remission failure in univariate analysis, but since age and function are highly correlated, only one of them was included in each of the multivariate models.
Women achieved DAS28 remission less often than men, which has also been found by others and can be attributed to their generally higher ESR and more painful joints.16 ,17 The gender bias in the DAS28 definition seems to be counterbalanced in the new approaches.
The finding that patients rated global disease activity very similar to global health (see ‘Methods’ section) is in agreement with a recent study by Dougados et al18, who found that the technique of collecting PtGA has only minimal influence on the DAS28. Patients may have difficulty in distinguishing between the impact of RA and other comorbidities on their well-being, especially when other diseases also affect the musculoskeletal system. In addition, PtGA on one single scale may not be sufficient to reflect a patient's experience of remission and treatment satisfaction. As stated by the ACR/EULAR committee, development of a concept of patient-assessed ‘absence of disease’ could improve the capture of relevant patient dimensions.2
Using data from the BEST trial, Klarenbeek et al7 reported that during the first year of observation, 8% of patients reached SDAI and 23% DAS28 remission. This is similar to the 11% and 28% in our data. Differing from our results, the mean HAQ was nearly equal in the DAS28 and the SDAI remission groups. This can be explained by the fact that the BEST study had enrolled patients with early RA while our data included patients with RA at all stages of disease. At disease onset, the correlation between activity and function is stronger than in later phases.19 ,20
Also with data from very early RA (median disease duration 14 weeks), Vermeer et al6 reported recently that 20.1% reached ACR/EULAR Boolean remission at 6 months when a 28-joint count was applied. In a further 21%, all criteria were fulfilled except for the PtGA≤1. Even of those patients with no disease activity according to joint counts and CRP, 41% had a PtGA>1. These patients also had considerable pain and fatigue, which is in agreement with our data. Nevertheless, we are not convinced that the PtGA is falsely high in patients with RA. Table 2 shows that patients not fulfilling the new criteria had more tender and swollen joints, higher CRP and were also assessed by the physician to have a higher disease activity. From a clinical point of view, these patients might have good reason to assess their disease activity a worse than 1.
Of 1118 patients with RA enrolled in the Brigham RA Sequential Study, 24.9% met the DAS28 remission and 8.8% the Boolean criteria at baseline, which again is similar to our data.21 A total of 18.2% remained in sustained DAS28 remission and 4.3% in sustained Boolean remission after 1 year. Of those in sustained DAS28 remission, 11.9% had clinically significant pain (>3) at baseline, while this applied to none of the patients fulfilling sustained Boolean remission.21 This is in agreement with our finding that patients in DAS28 remission who miss the ACR/EULAR criteria have considerably higher pain scores than those meeting any of these criteria.
In a review of various observational studies and trials, Mäkinen et al22 showed in 2006 that comorbidity, fatigue and morning stiffness—which may be present also in people without RA of the same age group—prevent many patients from meeting the traditional ACR remission criteria. Further, painful or swollen joints may reflect the sequelae of diseases other than RA or permanent joint damage caused by RA. By comparing the functional status of patients in remission with the general population, we found that the new criteria selected patients with a better functional status than their age- and sex-matched population counterparts. This may indicate that the new criteria preferentially select patients who are not only free from active RA, but also free from other disabling musculoskeletal diseases. This assumption is in line with the findings from Masri et al.3 Schmidt et al state that 20% of the general population testified severe or disabling back pain.13 Given the high correlation between PtGA and pain, most of them would not have fulfilled the remission criteria. This is also supported by the fact that patients in Boolean or SDAI remission had far less degenerative joint or spine disease than those in DAS28 remission and patients not in remission had the highest comorbid burden. Since clinical trials usually do not include patients with severe comorbidities, this confounding by comorbidity may not be of much relevance for randomised controlled trials. However, if these criteria are applied in daily practice to guide treatment decisions, clinical judgement will be essential to avoid overtreatment. Other reasons for a PtGA>1, such as pain from comorbid conditions, have to be considered and adequately treated—that is, by treatments other than immunosuppressive agents.21
On the other hand, patients with RA whose RA is inactive may overestimate their overall functional ability. This is supported by the finding from Heiberg et al23 that all patients in remission or with low disease activity, and most patients with moderate disease activity, found their state acceptable.
Our study has strengths and limitations. It is the largest study so far investigating the concordance of remission according to the DAS28 and the two new sets of criteria in unselected real-life patients. In addition, population data were available to put the results for functional status into perspective.
A weakness is that we had remission status at only one point in time. Since the aim is sustained remission with little or no radiographic progression, we need studies that analyse the ability of each of the remission criteria to predict favourable radiographic and functional outcome in the long term. Another potential shortcoming is the 60% response rate in the population survey, which might have led to an overestimation of the prevalence of back pain. The prevalence estimate is, however, within the range of results from other European studies.24
To conclude, we have to keep in mind that the new criteria, owing to their construction, may classify a considerable proportion of patients with clinically inactive RA as not in remission. This may translate into better radiographic outcome in those achieving the new criteria, even though patients in remission according to any criteria can experience radiographic progression.8 On the other hand, in order to avoid overtreatment of RA in daily practice, comorbidity has to be considered and adequately treated in patients with unfavourable PtGAs and report pain and fatigue despite clinically inactive RA.
The authors acknowledge the invaluable contributions and the enthusiasm of all German consultant rheumatologists who have contributed data of their patients with inflammatory rheumatic diseases to the national database since 1993. In particular, the authors would like to acknowledge the significant contributions of H Burkhardt (Frankfurt/Main), R de la Camp (Erlangen), T Eidner (Jena), K Fischer (Greifswald), U von Hinüber (Hildesheim), G Hoese (Stadthagen), K Karberg (Berlin), I Kötter (Tübingen), A Krause (Berlin), U Müller-Ladner (Bad Nauheim), W Ochs (Bayreuth), M Schneider (Düsseldorf) and S Wassenberg (Ratingen). In addition, we would like to thank Joachim Listing for helpful discussions and methodological advice in the preparation of this paper.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.