The management of disease in patients with rheumatoid arthritis (RA) has dramatically altered in the last decade. The introduction of biologics has successfully modified the disease course of patients with RA with active disease that had not previously been controlled by conventional disease-modifying antirheumatic drugs (DMARDs).1 2 Furthermore, the entire strategy for the treatment of RA has also been altered. Treatment with DMARDs should be aggressively initiated at an early stage of disease to improve prognosis,3^–5 and some large trials showed extensive suppression of disease activity led patients to a better functional capacity.6 7 Although RA was regarded as a relentlessly progressive disease only 20 years ago, “remission” rather than “low disease activity” is achievable and a realistic therapeutic goal these days based on the development of RA treatment.

The ultimate goals in the treatment of RA generally are no disease activity, and no radiographic joint damage, and normalisation of functional disability. The Disease Activity Score (DAS),8 which consists of the Ritchie score, swollen joint count, erythrocyte sedimentation rate (ESR) and the patient global assessment score, and its modified version, the 28-joint DAS (DAS28),9 are commonly used for the assessment of RA disease activity in many clinical trials, while Health Assessment Questionnaire (HAQ)10 is currently the most used, reliable instrument for assessing functional disability in patients with RA. Several studies have reported on the relationship between functional disability, disease activity, and radiographic assessment in RA.11^–27 Functional disability in patients with RA is associated with multiple variables such as disease activity, joint destruction, disease duration, age, gender, psychological state, educational level and comorbidities.11^–27 Among these variables, disease activity is thought to be the most important factor in the loss of functional capacity, especially in the earlier phase of RA disease course.11^–13

Since maintenance of good functional capacity is the fundamental goal of RA treatment, “remission” is now an important concept and the accepted goal of management in RA. Despite this fact, RA suffers from the absence of a single “gold standard” quantitative measure, such as serum blood sugar and HbA1c levels in diabetes or blood pressure.28 Although the definition of “remission” has been controversial and various types of definition of “remission” have been proposed,29^–31 DAS28 remission at a cut-off level of 2.6 is often used in clinical practice and clinical trials.32^–34 However, association between a threshold DAS28 level and long-term functional disability has not been well studied. In addition, it is not well known whether tight control of RA disease activity leads to prevention of long-term disability in clinical practice.

The IORRA (Institute of Rheumatology, Rheumatoid Arthritis) study is a prospective observational cohort study established at the Institute of Rheumatology, Tokyo Women’s Medical University, Tokyo, Japan, since October, 2000.35 For this cohort, physician assessments, patient assessments and laboratory data were supplemented with various types of patient information including disability index, disease activity, comorbidity, and medications used, collected from patient questionnaire sheets from approximately 5000 patients with RA every 6 months. More than 99% of patients with RA in our institute were enrolled, and more than 98% of patients answered and mailed their questionnaire back to us every time. This cohort database is a powerful resource not only for clinical research, but also for extensively evaluating therapeutic strategies under real-life conditions.35^–42

In this study, we clarify the importance of suppression of disease activity in the prevention of long-term patient disfability using a single-institute based large observational cohort in Japan.

PATIENTS AND METHODS

Patients

Among 7511 patients with RA (female 81.7%, mean age 55.8 years, mean RA duration 8.5 years) enrolled in an observational cohort study at our institute (IORRA) from October 2000 to April 2005, 2775 patients with RA (36.9%), who had continuous enrolment for at least 3 years from October 2000 to April 2005, were selected as the cohort of patients with RA for this study. All patients were of Japanese origin and were examined by rheumatologists, and had been diagnosed with RA according to the 1987 classification criteria by the American College of Rheumatology (ACR, formerly, the American Rheumatism Association).43 DAS28 values (based on the tender joint counts, swollen joint counts, patient-oriented visual analogue scales (VAS) for general health, and erythrocyte sedimentation rate (ESR)) and the J-HAQ (Japanese Health Assessment Questionnaire, the validated Japanese version of HAQ36) were collected from all the patients at 6-month intervals (each phase). Functional disability status was assessed by J-HAQ.

“Avg-Dscore” and “Avg-DAS28”

DAS28 values were calculated at each phase (once every 6 months) for all patients and a value less than 2.6 was considered as a tight control. We set up two new variables for each patient, “Avg-Dscore” and “Avg-DAS28” (average of DAS28 values over all phases), based on the transition of each patient’s DAS28 value. The Avg-Dscore was computed taking the threshold DAS28 level of 2.6 into consideration. The Avg-Dscore had a minimum score of 0 and a maximum score of 1; the DAS28 value of each patient at each phase (once every 6 months), was scored either as 0 if the DAS28 value was less than or equal to 2.6, or as 1 if it was greater than 2.6. The average of these scores over all the phases for each patient was calculated, and this gave the Avg-Dscore. The lower the Avg-Dscore, the higher the number of phases, in which the DAS28 value was under 2.6.

The Avg-DAS28 was the average of original DAS28 values, over all phases, computed for each patient. A larger Avg-DAS28 indicates a higher disease activity.

Statistical analysis

SAS software (V9.1; SAS Institute Inc., Cary, North Carolina, USA) was used for database management and statistical analysis. The relationship of “Avg-Dscore” and “Avg-DAS28” with the functional disability level (“Final J-HAQ score” and “Change in J-HAQ score” (final J-HAQ–initial J-HAQ)) was examined using Spearman correlation coefficients. Multiple linear regression analysis was carried out to investigate the relationship of DAS28 and functional disability. In models 1–6, the “Final J-HAQ score” was used as dependent variable, and in models 7–12 the “Change in J-HAQ score” was the dependent variable. The independent variable used in models 1, 2, 3, 7, 8, 9 was “Avg-Dscore” and in models 4, 5, 6, 10, 11, 12 it was “Avg-DAS28”. Models 2, 5, 8 and 11 were adjusted for baseline J-HAQ, and models 3, 6, 9 and 12 were adjusted for baseline DAS. All the models were also adjusted for age, gender, RA disease duration, time of enrolment and seasonal effect. Since we found a definite seasonal fluctuation in disease activities of patients with RA, which tend to show improvement from Spring to Autumn while worsening from Autumn to Spring in our observational cohort study,44 we adjusted for seasonal effect when performing multiple linear regression analysis. Considerable changes in RA therapy occurred during the 2000–2003 time period, so all the models were adjusted for time of enrolment. Differences in DAS28 and J-HAQ data between first phase and last phase observations were tested by Student paired t test or Wilcoxon signed-rank test where appropriate.

Based on the “Avg-DAS28”, 2775 patients with RA in this cohort were divided into following three groups: “Under 2.6”, “Between 2.6 and 3.2” and “Above 3.2”. The significance of differences in DAS28 and J-HAQ between first and last phase observations were assessed individually in each of the above groups using Student paired t test and Wilcoxon signed-rank test as appropriate.

RESULTS

Baseline clinical features of 2775 patients with RA

The baseline characteristics of the 2775 patients with RA in this study sample are shown in table 1. Overall disease activities in these 2775 patients with RA were relatively low, as reflected in the mean tender joint count, swollen joint count, C-reactive protein (CRP) titre and ESR. The percentages of DMARD and steroid users during the follow-up period (average follow-up period 3.6 years, minimum 3 years, and maximum 5 years) in 2775 patients with RA were 97.3% and 65.5%, respectively. The mean DAS28 decreased from 4.0 at study entry to 3.7 at study exit (p<0.001 using paired t test), while median J-HAQ scores were 0.63 and 0.63 respectively. The means of “Avg-Dscore” and “Avg-DAS28”, over all the patients, were 0.81 and 3.75, respectively.

We have also examined the baseline features of 4736 patients with RA who were not included in this study due to their lack of continuous follow-up for 3 years. The baseline features of these excluded patients with RA, such as age, RA duration, RA disease activity (mean tender joint count, swollen joint count, CRP titre and ESR, initial DAS28) and J-HAQ scores were quite similar to those of 2775 patients included in this study cohort (data not shown).

Relationship of RA disease activity with the functional disability by Spearman correlation coefficients

The relationship of disease activity (“Avg-Dscore” and “Avg-DAS28”) with the functional disability level (“Final J-HAQ score” or “Change in J-HAQ score”) using Spearman correlation coefficients is shown in table 2. There was a statistically significant correlation between “Avg-DAS28” and “Final J-HAQ score” (r = 0.57, p<0.001). A similar relationship was seen between “Avg-Dscore” and “Final J-HAQ score” (r = 0.47, p<0.001).

Relationship of RA disease activity with long-term functional disability by multiple linear regression analysis

The analysis was performed using six multiple linear regression models to examine the relationship of RA disease activity to “Final J-HAQ score” (table 3). For prediction of “Final J-HAQ score”, after adjusting for all the covariates it was apparent that that “Avg-Dscore” and “Avg-DAS28” were the most significant factors contributing to “Final J-HAQ score”, indicating a positive correlation.

For prediction of “Change in J-HAQ score”, multiple linear regression analysis after adjusting for all the covariates also revealed that “Avg-Dscore” and “Avg-DAS28” were the factors most significantly contributing to “Change in J-HAQ score”, indicating a positive correlation (table 4).

Relationship of RA disease activity with long-term functional disability individually in the three groups that were classified based on “Avg-Das28”

Of the 2775 patients with RA, the number of patients in the “Under 2.6”, “Between 2.6 and 3.2” and “Above 3.2” groups were 346, 461 and 1968 respectively. The DAS28 values were significantly decreased during the follow-up period in all the three groups. The J-HAQ scores were significantly decreased only in the “Under 2.6” group. The difference in J-HAQ scores was not significant in “Between 2.6 and 3.2” group. However, the J-HAQ scores were significantly increased in the “Above 3.2” group.

DISCUSSION

Using a prospective observational large cohort in Japan, we clearly demonstrated that a tight control of RA disease activity has significant association with lower functional disability level after an average 3.6 years of follow-up when a DAS28 value of <2.6 was considered as a tight control.

In the cohort in this study, (table 2), there was a statistically significant correlation between “Avg-DAS28” or “Avg-Dscore” and “Final J-HAQ score”, indicating a significant relationship between tighter disease control and lower disability levels. Combe et al found that 3- and 5- year HAQ disability could be predicted mainly by baseline HAQ score in 191 patients with early RA.27 Other previous studies focusing on the predictive factors of HAQ disability have also identified baseline HAQ score as the best prognostic factor.16–18 23 25 Hence, we also made multiple linear regression models (models 2, 5, 8 and 11) that were adjusted for age, gender, RA disease duration, time of enrolment, seasonal effect, and baseline J-HAQ score as an explanatory variable (tables 3 and 4). These multiple linear regression models also confirmed the strong relationship between “Avg-DAS28” or “Avg-Dscore” and “Final J-HAQ score” or “Change in J-HAQ score” after adjusting for all covariates including baseline J-HAQ score and baseline DAS28 (tables 3 and 4), indicating managing disease activity efficiently by maintaining DAS28 score values under 2.6 has significant association with reduced long-term disability.

Some prospective studies reported the relationship between functional capacity and disease activity of RA. Drossaers-Bakker et al found that functional capacity was strongly influenced by disease activity throughout the course of RA in 132 female patients with early RA over 12 years of cohort follow-up.11 By contrast, Welsing et al reported that the effect of disease activity and joint destruction on functional capacity changed over the course of the disease in 378 patients with early RA with 9 years of cohort follow-up.12 They concluded that the loss of functional capacity was caused mainly by disease activity in early stages of the disease, while joint destruction was the main determinant for functional capacity later in the disease course. In our study, since mean RA duration of our cohort sample was 9.5 years at entry, most patients were categorised as having established RA. Although patient characteristics at entry of this study were different from the previous two reports in which their cohort was based on patients with early RA,11 12 our findings are in concordance with the cohort study by Drossaers-Bakker et al, in which disease activity was the most important factor in determining the loss of functional capacity throughout the course of RA.11

Various definitions of “remission” in RA have been proposed, such as ACR remission criteria,45 radiographic remission34 46 and remission assessed by DAS,29 DAS28,33 simplified disease activity index (SDAI)47 48 and clinical disease activity index (CDAI).49 50 Although DAS28 remission at a cut-off level of 2.6 is often used in clinical practice and clinical trials32^–34 and we use this cut-off point in this study, many researchers have chosen their own favourable DAS28 cut-off points such as 2.66,32 2.81,51 2.32,48 and 2.4.31 48 Based on our analysis in the three groups that were classified by “Avg-DAS28”, the disease activity was significantly reduced in all three groups. However, functional capability was significantly improved only in the “Under 2.6” group that had average DAS28 values under 2.6. The change in functional capability in the “Between 2.6 and 3.2” group with average DAS28 values between 2.6 and 3.2 was not significant and in the “Above 3.2” group with average DAS28 values above 3.2, functional capability significantly deteriorated. Hence, it can be suggested that we have to control the disease activity of patients with RA more strictly. A targeted treatment strategy should be undertaken in the management of RA in daily practice, and based on our results, the target DAS28 should be <2.6 to prevent progression of disability. The threshold DAS28 level of 2.6 may be useful in developing guidelines for targeted treatment strategies in patients with RA from a viewpoint of predicting patient functional disability. The definition of “remission” generally is the total absence of signs and symptoms of the disease: (1) no tender and swollen joints, (2) no increase in joint damage, and (3) full functioning (HAQ = 0). However, we have to acknowledge that a definition of remission according to DAS28 at a cut-off level of 2.6 may include many patients with a considerable number of tender and/or swollen joints,31 48 and that a DAS28 cut-off level of 2.6 has to be used with caution in clinical practice and clinical trials of RA treatment.

In RA, disability consists of two major components, reversible components such as joint pain and swelling due to joint inflammation, and irreversible components such as joint destruction, deformity, and muscle weakness. Aletaha et al reported that average HAQ scores despite clinical remission increased progressively with duration of RA and the reversibility of HAQ scores decreased with the duration of RA in 295 patients in whom clinical remission was achieved in several clinical trials.52 Hence, when we assess remission using DAS28 score, the cut-off point might not be limited to a single value for the whole population of patients with RA, and this may be considered separately for subgroups of patients with different disease duration. Further longer-term observational studies are required to examine a cut-off level of DAS28 remission.

The low continuation rate of our study and selection bias of patients with RA may be possible limitations. In fact, among 7511 patients with RA enrolled in IORRA from October 2000 to April 2005, only 2775 patients with RA (36.9%) had continuous enrolment for at least 3 years. Since our institute is located in the centre of Tokyo, many patients with RA may travel long distances. The purpose of their visit may be to obtain a second opinion for therapeutic strategy, and thus they may not keep visiting us thereafter. However, it is our principle to enrol all patients with RA into this cohort and thus we may lose many patients during follow-up. Additionally, we sometimes refer patients with RA to their family doctors, who may lose them during follow-up. However, the baseline features of 2775 patients with RA included in this study sample were quite similar to the 4736 excluded patients with RA. Hence, we believe selection bias of patients with RA may be minimal.

In conclusion, efficient management of disease in patients with RA, by maintaining the DAS28 value at <2.6, has significant association with improved functional capability as assessed by J-HAQ over a long-term period. The threshold DAS28 level of 2.6 may be useful in developing guidelines for targeted treatment in patients with RA.