Patients

For this study, patients from the Leiden Early Arthritis Clinic (Leiden, The Netherlands)—a population-based inception cohort of patients with newly diagnosed early arthritis—were selected. This cohort presented in 1993 at the Department of Rheumatology of the Leiden University Medical Center, the only referral centre for rheumatology in a healthcare region of approximately 400 000 inhabitants in The Netherlands. General practitioners were encouraged to refer patients directly when arthritis was suspected; patients were included if physical examination revealed arthritis.¹⁸ In the study period (1993–9), 1009 patients with early arthritis were included. After 2 weeks of follow-up, 182 patients had fulfilled the 1987 American College of Rheumatology (ACR) criteria for RA¹⁹ and 326 patients had arthritis that could not be readily classified (undifferentiated arthritis (UA)). After 1 year of follow-up, a total of 285 patients fulfilled the 1987 ACR criteria for RA. In all, 103 of these patients had been classified as UA at 2 weeks of follow-up. From these patients with RA, two categories of patients with extreme disease courses were selected: those who entered sustained remission (remitting RA) and those who had progressed to the most destructive disease (severe RA). Remission was defined as satisfying the proposed American Rheumatism Association criteria for clinical remission²⁰ after having discontinued the use of disease-modifying antirheumatic drugs (DMARDs) for at least 1 year. After one additional year of follow-up, the presence of sustained remission was confirmed. 28 patients had achieved sustained remission after a mean disease duration of 3.7 years (range 210–3159 days, SD 852 days).²¹ The remitting patients are further referred to as “remitting RA”. Radiographs of the hands and feet were taken at baseline and yearly thereafter, and the level of radiological joint destruction was scored according to the Sharp–van der Heijde method.^22,23 The rheumatologist who scored the radiographs was unaware of the study question. The patients with the most destructive disease course were identified by selecting the 28 patients who had the highest Sharp–van der Heijde scores after 1 year of follow-up. This corresponded with a Sharp–van der Heijde score >34 (mean 59). These patients are further referred to as “most severe RA”.

Methods

At inclusion, the rheumatologist questioned the patient about the initial symptoms as well as the presence of morning stiffness and the symptom duration. Physical examination at baseline included a swollen joint count, scaling each joint on a 0–1 scale (0 denotes no swelling and 1 indicates the presence of swelling). Subsequently, the joints or joint groups were categorised (eg, all metacarpophalangeal joints on one side were counted as one joint). The following joints (groups) were studied: shoulders, elbows, wrists, metacarpophalangeal joints, proximal and distal interphalangeal joints, knees, ankles, and metatarsophalangeal joints. For the analysis, a dichotomous approach that indicated the presence or absence of swelling per joint group was used.

Baseline laboratory parameters included eosinophil sedimentation rate, CRP, IgM rheumatoid factor ELISA as described previously²⁴ and anti-CCP2 antibodies (ELISA, Immunoscan RA Mark 2, Euro-Diagnostica, Arnhem, The Netherlands). According to the manufacturer’s instruction, a level >25 arbitrary units was considered positive.

Statistical analysis

The baseline parameters of the patients with remitting and severe RA were compared using the Mann–Whitney U test for continuous variables and the χ² test for nominal variables. Subsequently, a logistic regression analysis with backward selection procedure (p>0.10 as removal criterion, using a likelihood ratio test) was performed, with the presence of remitting RA or most severe RA as the dependent variable and the individual joint groups that were significantly associated with the disease outcome in the univariate analysis and the total number of swollen joints as possible explanatory variables.

To validate the findings, the association between the joint groups and radiological joint destruction was assessed in the total group of patients with RA. The Sharp–van der Heijde scores at 1-year follow-up were entered as a dependent variable in a linear regression analysis (backward selection procedure, p>0.10 as removal criterion), with the total number of swollen joints and the joint count for the individual joint groups being possible explanatory variables. To assess the association between the distribution of swollen joints and the level of radiological joint destruction in relation to other clinical variables that might be associated with RA severity, another linear regression analysis (backward selection procedure, p>0.1 as removal criteria) was performed. In this analysis, the Sharp–van der Heijde score at 1-year follow-up was entered as a dependent variable, and the total number of swollen joints, the swollen joint count for the individual joint groups, gender, age, the presence of anti-CCP antibodies, RF, morning stiffness, the CRP level and symptom duration entered as possible explanatory variables.

To assess whether the findings from these analyses were specific for RA or were also applicable to arthritis of various causes (UA), a logistic regression analysis with backward selection procedure (p>0.10 as removal criterion, using likelihood ratio test) was performed, with the outcome RA after 1 year (RA yes or no) being the dependent variable and the baseline parameters and individual joint groups being possible explanatory variables in patients who were classified as UA at 2 weeks follow-up.

SPSS V.11.0 was used to analyse the data. A p value ⩽0.05 was considered to be significant.

Univariate comparison of patient characteristics at first presentation of patients with RA with extreme disease courses

The patients in the remitting and most severe RA group had no significant differences in the distribution of gender, age and morning stiffness. The patients with the most severe disease course harboured RF and anti-CCP antibodies more frequently, had higher levels of CRP and had higher Sharp–van der Heijde scores at inclusion compared with patients with remitting RA (table 1⇓). The patients with the most severe disease course also had a significantly higher number of swollen joints at first presentation than the patients with remitting RA. The distribution of swollen joints was different between the two groups: the patients with the most severe RA had, significantly more often, arthritis of the shoulders, elbows, proximal interphalangeal joints, knees and ankles (table 2⇓). There was no difference in the prevalence of swollen metacarpophalangeal and metatarsophalangeal joints between the two groups of patients with RA.

Regression analysis in patients with RA with extreme disease courses

To investigate which joints (groups) were independently associated with the disease outcome, a logistic regression analysis with backward selection procedure was performed with the presence of remitting RA or most severe RA as the dependent variable and the individual joint groups that were significantly associated with the disease outcome in the univariate analysis (shoulder, elbow, proximal interphalangeal, knee and ankle) and the total number of swollen joints as possible explanatory variables. In this analysis, only the presence of a swollen knee (OR 7, p = 0.004) was found to be significantly associated with the disease outcome (table 3⇓).

Regression analysis in all 285 patients with RA

To validate the findings, the association between the distribution of swollen joints and radiological joint destruction was assessed in the total group of patients with RA. The Sharp–van der Heijde score at 1-year follow-up was entered as a dependent variable in a linear regression analysis with a backward selection procedure and the total number of swollen joints and all evaluated joint groups as possible explanatory variables. This analysis showed that, in the total group of patients with RA, the total number of swollen joints (p = 0.004) and swelling of the knee (p = 0.03) were independently associated with the level of radiological joint destruction of hands and feet (table 4⇓). A similar analysis was performed with the Sharp–van der Heijde scores at 2- and 3-year follow-up as dependent variable (x rays were available of, respectively, 202 and 162 patients). In these analyses, only swelling of the knee was found to be associated independently with the level of radiological joint destruction (B = 7.2, SE = 3.9, p = 0.05 for the 2-year time point and B = 3.5, SE = 4.7, p = 0.005 for the 3-year time point). Figure 1⇓ displays the level of radiological destruction of the small joints of the hands and feet during 3 years of follow-up of the patients with RA who at first presentation did have or did not have arthritis of the knee (fig 1⇓).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

 The level of radiological destruction of the small joints of hands and feet during 3 years of follow-up of patients with rheumatoid arthritis with or without arthritis of the knee at first presentation. *p<0.05 (Mann–Whitney U test). At the 2- and 3-year time points, radiographs of hands and feet were available of 202 and 162 patients, respectively.

In conclusion, regression analyses in both the whole group of patients with RA and using the approach of the extremes of the disease courses yielded the total number of swollen joints and, particularly, the presence of arthritis of the knee as independently associated with the level of joint destruction of the small joints of hands and feet during follow-up.

Predictive ability of distribution of swollen joints in relation to other clinical characteristics for RA severity

Next, we investigated the predictive ability of the presence of arthritis of the knee and the total number of swollen joints in relation to other clinical variables that are observed or described to be associated with a severe disease outcome: CRP, RF, anti-CCP antibodies, morning stiffness and symptom duration (table 1⇑).^4,17 A linear regression analysis with a backward selection procedure revealed that the total number of swollen joints (B = 0.9, SE = 0.4, p = 0.03), the presence of anti-CCP antibodies (B = 8.4, SE = 2.4, p<0.001), the CRP level (B = 0.02, SE = 0.007, p<0.01) and the symptom duration (B = 0.2, SE = 0.04, p<0.001) were independently associated with the level of joint destruction after 1-year of follow-up. In this analysis, the presence of arthritis of the knee was not an independent predictor for disease severity. The fraction of explained variation of this model was 0.2. As, after correction for clinical variables in a regression analysis, the presence of arthritis of the knee was not found to be independently associated with RA severity (in contrast with the CRP level, the presence of anti-CCP antibodies and symptom duration), we hypothesised that these variables might be distributed differently among the patients with RA with and without arthritis of the knee. Arthritis of the knee was present in 106 (39%) patients at first presentation. The presence of anti-CCP antibodies and symptom duration, and also the variables age, gender, RF and morning stiffness, were not significantly different between the patients with RA with or without arthritis of the knee. However, there was a significant difference in the level of CRP: patients with RA with arthritis of the knee at first presentation had higher CRP levels (mean (SD) 48 (35) mg/l) compared with patients with RA without involvement of the knee (mean (SD) 22 (24) mg/l, p<0.001).