Article Text
Abstract
Introduction The new American College for Rheumatology (ACR)/European League Against Rheumatism (EULAR) remission criteria are based on the assessment of 28 joints. A study was undertaken to study the consequences of remission misclassification due to residual disease activity in the feet on physical function and joint damage in the subsequent year in an observational early disease cohort.
Methods All patients with rheumatoid arthritis at inclusion or at 1-year follow-up in the early arthritis cohort of the Jan van Breemen Institute, The Netherlands were included. ACR/EULAR remission definitions for trials and clinical practice were calculated twice, once using a 28-joint count and once using a 38-joint count that included the 10 metatarsophalangeal joints. Disease stability was defined as stable x-ray scores over 1 year (change ≤0 in Sharp/van der Heijde scores) and stable and low scores on the Health Assessment Questionnaire (HAQ change ≤0 and HAQ score consistently ≤0.5), all during the second year after inclusion. Analyses comprised residual disease activity (swollen or tender joints >0) in the feet of patients who fulfilled the candidate remission criteria using a 28-joint count and likelihood ratios of remission definitions to predict disease stability.
Results Of 421 patients, 9–15% reached remission at 1 year using a 28-joint count. Of these, 26–40% showed activity in the feet. Misclassification due to reduced joint counts was observed in 2–3%. A state of remission increased the likelihood of stability of both x-ray and HAQ, with similar likelihood ratios for definitions using 38-joint counts and those using 28-joint counts.
Conclusion The ability of remission definitions with 28-joint counts versus 38-joint counts to predict long-term good radiological and functional outcome is similar. This confirms that inclusion of ankles and forefeet in the assessment of remission is not required, although inclusion of these joints in the examination is recommended.
Statistics from Altmetric.com
Rheumatoid arthritis (RA) is a chronic inflammatory condition. Since no curative therapy is known, the current treatment of RA aims for a state of lowest possible disease activity, ideally remission.
An important aspect of disease activity is the amount of swelling and pain in joints that are typically involved in RA. Current methods evaluate up to 68 joints for tenderness and up to 66 joints for swelling, but reduced joint counts such as the 28 tender and swollen joint count are widely used in clinical practice and in composite measures of disease activity. The most critical joints not included in the 28-joint count are the ankle and foot joints. The 28-joint count correlates well with the full joint count, addresses most of the frequently involved joints, eliminates the frequently comorbidity-related tenderness and swelling of ankles and forefeet and is more feasible in both clinical trials and clinical practice, explaining its popularity in both settings.1,–,4
The assessment of remission in RA has always involved evaluation of joints. Until recently, remission could be defined in many different ways, ranging from strict definitions allowing no disease activity whatsoever to relaxed definitions allowing up to 21 swollen joints.5
In order to establish uniformity, the American College for Rheumatology (ACR) and the European League Against Rheumatism (EULAR) together with the Outcome Measures in Rheumatology initiative have redefined remission in RA.6 7 The new definition for RA allows, among other measures, a maximum of one swollen and tender joint based on a 28-joint count or remission based on the simplified disease activity index (SDAI) in which the sum of swollen and tender joints cannot exceed a total of two (box 1).
Box 1 American College of Rheumatology/European League Against Rheumatism definitions of remission in rheumatoid arthritis clinical trials
Boolean-based definition
At any time point, patient must satisfy all of the following:
Index-based definition
At any time point, patient must have SDAI ≤3.3§
Boolean-based suggestion for clinical practice
At any time point, patient must satisfy all of the following:
Index-based suggestion for clinical practice
At any time point, patient must have CDAI ≤2.8§
↵* For tender and swollen joint counts a 28-joint count may miss active joints especially in the feet and ankles and it is preferable to include feet and ankles also when evaluating remission.
↵† The following wording and response categories should be used for global assessment: “Considering all of the ways your arthritis has affected you, how do you feel your arthritis is today?” Verbal anchors for the response can range from ‘asymptomatic’ to ‘severe symptoms’.
↵§ SDAI, simplified disease activity index is defined as the simple sum of the tender joint count (28), swollen joint count (28), patient global assessment (on a 0–0 scale), physician global assessment (on a 0–10 scale) and C-reactive protein (mg/dl). CDAI, clinical disease activity index is the same as the SDAI minus C-reactive protein.
Although this new definition can be regarded as strict, theoretically a patient with considerable activity in the feet and ankles but none in other areas could be defined as in remission if a 28-joint count is used.
Several studies investigated the amount of residual disease activity in the ankles and feet of patients that were classified as in remission using limited joint counts, with conflicting results. Landewé et al showed that remission defined by the disease activity score (DAS) using 28 joint counts (DAS28) is inferior to the original DAS definition owing to residual swelling and tenderness in the ankles and feet.8 Mäkinen and colleagues agree that DAS28 should not be used to define remission in clinical practice or clinical trials,9 and recent work by van der Leeden et al shows that 40% of patients in DAS28 remission had activity in the feet.10 Kapral et al studied the whole range of disease activity and showed that, while there were some differences between the extended versus limited joint count, this had only little relevance for the definition of remission by composite disease activity assessments such as the DAS28 and SDAI (both using 28-joint counts)2 11 because other components of these measures such as patient global assessment or erythrocyte sedimentation rate (ESR) would mostly be higher in patients with foot joint involvement, thus exceeding thresholds of remission. The authors therefore concluded that, although assessment of ankles and feet is important, reduced joint counts are appropriate and valid for disease activity assessment.12
Although missing potential activity in feet and ankles is inevitable when a 28-joint count is applied, this has only proved a problem for the DAS28 remission definition. In trial datasets studied during the development of the new remission definition, the effects of missing residual disease activity in the ankles and feet appeared limited. In patients with activity in the feet, other measures in the definition also frequently showed increased levels that precluded remission.
However, it is unclear whether the experience in highly selected trial patients can be translated to routine clinical practice. The aim of the current study is to study the consequences of remission misclassification due to residual disease activity in the feet on physical function and joint damage in the subsequent year by comparison of likelihood ratios in an observational early disease cohort.
Methods
Patients
All patients fulfilling the 1987 ACR criteria for RA13 at inclusion or at 1-year follow-up in the early arthritis cohort (EAC) of Reade (formerly the Jan van Breemen Institute), Amsterdam, The Netherlands were selected. Only patients who had never used biological agents before or during the observation period with complete data records were included in the analyses. Patients with early RA have been included in the EAC since 1995 and return to the clinic for follow-up at least yearly. Data collection is done by trained research personnel following a strict protocol, ensuring consistency of the accrued data.
Definitions of remission and outcome
Joint counts were performed every year and included 28 and 38 joint counts. The 28-joint count comprises the metacarpophalangeal (n=10), proximal interphalangeal (n=8) and interphalangeal thumb joints (n=2), shoulder (n=2), elbow (n=2), wrist (n=2) and knee (n=2). The 38-joint count additionally evaluates the 10 metatarsophalangeal joints in the forefoot (but not the ankles). Joint assessments were performed by trained research nurses.
Laboratory data comprised baseline rheumatoid factor and anticitrullinated protein antibodies, annual ESR and C-reactive protein (CRP). Patient pain, patient global disease activity and physician global disease activity were all measured at yearly intervals on a visual analogue scale ranging from 0 to 100 mm.
Remission status was evaluated at 1-year follow-up. We calculated the new ACR/EULAR remission definitions for trials and clinical practice, presented in box 1. Usually, SDAI should be calculated using a CRP value of 0.5 as the upper limit of normal, even if high sensitivity CRP measurement allows for obtaining lower levels. In our study we did not transform CRP levels that were <0.5 to a minimum of 0.5. Nevertheless, no patient in SDAI remission had more than two swollen or tender joints.
For calculation purposes, visual analogue scale scores were recoded where necessary from a 0–100 scale to a 0–10 scale and CRP from mg/l to mg/dl.
Residual disease activity was defined as one or more swollen or tender metatarsophalangeal joints.
Long-term outcome included the Health Assessment Questionnaire (HAQ) and x-rays of hands and feet scored by one of two independent trained researchers (intraclass correlation coefficient 0.90) according to the Sharp/van der Heijde method.14 15
Consistent with the methodology of the remission development paper, disease stability (the good outcome) was defined in three ways: (1) as stable x-ray scores (change ≤0 in Sharp/van der Heijde); (2) as stable and low scores on the HAQ (HAQ change ≤0 and HAQ score consistently ≤0.5); and (3) as both stable x-rays scores and stable and low HAQ scores as defined above, all during the second year after inclusion.
Statistical analysis
We first analysed the prevalence of remission in the study population according to the two trial and two practice definitions. Next we calculated the prevalence of residual disease activity in the feet of patients who were classified as being in remission using a 28-joint count. We then evaluated the proportion of patients misclassified—that is, those patients declared in remission according to the 28-joint count but not in remission according to the 38-joint count. Subsequently we calculated the likelihood of patients in remission at 1 year to experience disease stability in the subsequent year compared with patients not in remission. Specifically, the likelihood ratio expressed the proportion of patients in remission having the good outcome to the proportion of patients not in remission having the good outcome. The likelihood ratios of the 28-joint and 38-joint count definitions were then compared using a χ2 statistic in the setting of a replacement test. Finally, we evaluated whether radiological progression in patients in remission with activity in the feet was different from that in patients without activity in the feet by looking at progression in the hands and feet separately.
Results
Patients
A total of 421 patients with RA could be analysed. Most were women (69%) and rheumatoid factor positive (56%), and the mean disease duration at inclusion was <1 year (table 1). Most patients had active disease with a mean (SD) DAS28 at inclusion of 5.1 (1.2); 94% of patients started disease-modifying antirheumatic therapy, in 43% of patients methotrexate was started, in 32% sulfasalazine was started and in 12% hydroxychloroquine was started.
Total damage scores increased from 3.4 at baseline to 6.1 at year 1 and 9.1 at year 2. One hundred and eighty-two patients (43%) showed radiological progression (mean 7.2 points) between years 1 and 2 of follow-up. HAQ scores decreased from 1.2 at baseline to 0.6 at both 1 and 2 years of follow-up. A total of 269 (64%) patients showed progression of HAQ scores and/or had a HAQ score of >0.5 points. Accordingly, 239 patients (57%) had a good outcome in terms of radiographic stability and 152 (36%) in terms of HAQ stability; 85 patients (20%) had a good outcome in both measures.
Remission and residual disease activity
One year after inclusion in the cohort, 9% (Boolean) to 14% (SDAI) of patients reached remission according to the trial definitions of remission by 28-joint counts. For the practice definitions, the prevalence of remission was 10% (Boolean) to 15% (clinical disease activity index (CDAI)) by 28-joint counts.
When we calculated the prevalence of remission according to one of the four definitions using 38-joint counts instead of 28-joint counts, the percentages decreased from 9% to 7% for the Boolean trial definition, from 14% to 10% for the SDAI, from 10% to 7% for the Boolean practice definition and from 15% to 10% for the CDAI (table 2).
Table 3 shows the disease characteristics of patients in 28-joint count remission and of those not in remission. The characteristics of patients in 28-joint count remission but not in 38-joint count remission were similar to the characteristics of patients in 28-joint count remission (except for disease activity in the feet) and are therefore not shown. In sum, for the Boolean trial definition this concerned a group of nine patients with a mean (SD) ESR of 9 (7), swollen 38-joint count of 2.9 (2.3) and tender 38-joint count of 2.6 (2.4).
Residual activity in the feet was seen in 26% of patients classified in remission by the Boolean definition and 36% of patients classified by the SDAI definition (Figure 1). In these patients about one-third had only painful joints, one-third had only swollen joints and one-third had both, with a mean (SD) tender joint count of 2.4 (2.3) and swollen joint count of 2.6 (2.3). As a consequence, application of the 38-joint count reduced the prevalence of remission by 2–3% in the Boolean definitions. The higher prevalence of remission in the SDAI (and CDAI) definitions compared with the Boolean definitions appeared to be mostly explained by residual disease activity in the feet. While the data—when interpreted as here—slightly reduce the prevalence of remission, it is worth noting that the mean residual swollen and tender joints when applying the 38-joint count are well in line with the threshold level of core set measures for remission according to a survey done among committee members of the ACR/EULAR remission task force.6 7
Remission and long-term disease stability
A state of remission at 1-year follow-up defined by any of the tested remission definitions was significantly associated with stable long-term outcome, with 35–44% of patients in remission showing stable long-term outcomes compared with 14–19% of patients not in remission.
The likelihood of stability of both x-ray and HAQ scores for patients in remission compared with patients not in remission at 1-year follow-up was increased for all definitions and varied between 2.4 and 3.2 (table 4). There was no statistically significant difference between the likelihood ratios for 38-joint and 28-joint definitions, showing that there is little consequence for the long-term outcome in groups of patients with RA when remission definitions with reduced (28) joint counts are applied.
Looking in more detail at the progression of damage in hands and feet separately, we found that there were also no differences in damage progression in hands or feet between patients in remission with activity in the feet versus patients in remission without activity in the feet. Radiological progression in patients in remission without activity in the feet (n=28) was 0.79 (1.7) compared with 0.80 (1.3) for patients in remission with activity in the feet (n=10).
Discussion
In this observational early RA cohort, we looked at the consequences of remission misclassification due to the use of reduced joint counts. We show that a considerable proportion of patients (approximately 25%) classified as in remission using a reduced (28) joint count can have residual pain or joint swelling in the feet. Although substantial when related to the number of patients in remission, misclassification is limited when related to the total population. In addition, it appears likely that its impact on function or structure is limited, since our patients with residual activity in the feet had x-ray progression and functional status similar to those without such residual activity.
Assessment of ankles and feet is more difficult than that of other joints, leading to more variability in the results. In addition, other causes of pain and swelling in ankles and feet frequently confound the assessment.1 4 These considerations may help explain the lack of progression in the feet in the presence of pain and swelling, where joint swelling is usually associated with progression of damage in the same joint.16
Although several groups have studied the residual joint count in feet and ankles of patients in remission using a reduced joint count,1 4 8 12 this is the first study to investigate the long-term consequences on damage and physical function of the use of reduced joint counts, on the one hand, and the relationship with the new ACR/EULAR definition of remission on the other. A reduced (28) joint count is widely used, especially in observational cohort studies and clinical practice where feasibility of a measurement tool is of major importance. Trials select patients for extreme disease activity, while a general RA population as studied here has a different disease activity profile since it does not need to fulfil particular minimal disease activity criteria.
In contrast to the trial datasets used to develop the ACR/EULAR definition of remission, the current study was performed in an early disease cohort in which prediction of future outcome is very important. Despite the differences in patient selection, this study confirms the conclusion of the ACR/EULAR remission committee—namely, that the consequences of misclassification are limited.
A major limitation of this study is that swelling and tenderness of ankles were not assessed in this cohort. Data on ankles might have influenced the results in such a way that the misclassification and subsequently the impact of the use of reduced joint counts might have been larger than reported here. However, as mentioned above, this same analysis was done in two of the trial datasets used to develop the ACR/EULAR remission definition. These datasets did have data available on both metatarsophalangeal joints and ankles and showed very similar results. Since the results obtained for the feet in this study are fully in line with those derived in that study, one can infer that this would be similar for the ankles. Furthermore, the effect of missing the ankles is unlikely to affect the current assessment of damage since ankles are not included in the Sharp/van der Heijde score. Finally, the evaluation of radiological progression in the hands and feet separately was hampered by the small sample of patients reaching remission (n=38) and still showing activity in the feet (n=10), so progression in patients with truly inflamed feet joints is still to be expected.
Another limitation of the current study is that the patients came from a single Dutch centre; however, the cohort studied reflects patients seen in many other similar clinics and practices and the results are therefore generalisable to practices with similar assessments and treatment strategies for early RA worldwide. Only patients who had not used biological agents were included in the current study in order to study the effect of remission on long-term progression of damage, a relationship that might go unrecognised when patients use biological drugs. However, due to the disconnect between disease activity and damage progression in patients using biological agents, residual disease activity in the feet has limited long-term consequences, making our results generalisable to patients using biological agents.
The committee charged with redefining remission in RA stated that inclusion of ankles and forefeet in the assessment of remission is not required, although inclusion of these joints in the examination is recommended.
Although the current analysis in an observational early disease cohort confirms that the number of patients that are misclassified when using a 28-joint count is considerable (9 of 38 (24%) patients in remission and 421 (2%) of the total population), it also confirms the limited consequences of misclassification on outcome in a routine clinical practice population, thus validating and expanding the committee's conclusion. We stress that quality of care mandates the assessment of all relevant joints in the individual patient, and that these joints are often found in the feet.
Conclusion
Patients in remission according to the new definition as assessed by a 28-joint count can still show pain and swelling in the feet, resulting in misclassification of 2–3% of the total population studied. However, the ability of remission definitions with 28-joint counts versus 38-joint counts to predict subsequent good radiological and functional outcome appears very similar, suggesting that, at a population/group level, the involvement of the feet in patients defined as in remission by the new ACR/EULAR criteria in part reflects causes not related to RA and has limited impact on RA outcome.
References
Supplementary materials
Web Only Data
Files in this Data Supplement:
Footnotes
-
Competing interests None.
-
Patient consent Obtained.
-
Ethics approval This study was approved by the Institutional Review Board.
-
Provenance and peer review Not commissioned; externally peer reviewed.