To compare the burden of disease and its development over time in patients referred to an early arthritis cohort who were diagnosed either as having arthralgias without synovitis or as having rheumatoid arthritis (RA).
To compare the burden of disease and its development over time in patients referred to an early arthritis cohort who were diagnosed either as having arthralgias without synovitis or as having rheumatoid arthritis (RA).
Patients diagnosed as having arthralgias without synovitis or RA were selected from the Rotterdam Early Arthritis Cohort. Data on clinical and psychological characteristics, demographics, pain scores (Rheumatoid Arthritis Disease Activity Index), functional ability (Health Assessment Questionnaire), health-related quality of life (HRQOL; Short Form 36), fatigue (visual analog scale and Fatigue Assessment Scale), and health care utilization (HCU) were collected at baseline and at 6 and 12 months of followup. The burden of disease measures (pain, functional ability, fatigue, and HRQOL) and HCU levels were plotted over time for both groups. A Poisson regression model for repeated data was used to identify determinants of HCU for both groups.
At baseline, 330 patients with arthralgias without synovitis (nonsynovitis [NS] group) and 244 RA patients (RA group) were included. Overall, the burden of disease measures and HCU levels were very similar between groups. Both groups showed improvement over time with respect to pain scores, functional ability, HRQOL, and HCU levels. Independent predictors of high HCU were identified as more pain, worse physical health, and external locus of control in the NS group and as shorter duration of symptoms, low chance locus of control, and worse physical functioning in the RA group.
Despite the absence of an inflammatory diagnosis, patients with arthralgias without synovitis experienced a similar burden of disease compared with RA patients.
Over the past decade, early arthritis clinics have been established around the world (). The main goal of these clinics is the early detection and treatment of rheumatoid arthritis (RA), which has been shown to result in a more favorable course of disease ([2-4]). Although many of the patients referred to early arthritis clinics are diagnosed as having early RA and benefit from early treatment of their condition, other referred patients do not show any physical signs of an underlying inflammatory disease and are unlikely to develop RA in the future. The patients in this second group are referred to rheumatologists with symptoms that greatly resemble early arthritis symptoms, but no signs of arthritis can be found in these patients at physical examination. However, despite the absence of an underlying inflammatory condition, a previous study has suggested that these patients experience a burden of disease similar to early RA patients or patients with other inflammatory joint conditions (). Formal diagnoses, as well as a clear understanding of the processes underlying the symptoms, are usually lacking in this patient group. As a result, the consequences on functional ability, health, and general well-being, as well as the progression of health symptoms over time, are unknown. Furthermore, despite the high burden of disease, standardized treatment regimens are usually absent for this category of patients. It is unclear what strategies these patients adopt to deal with their symptoms after visiting their rheumatologist and being diagnosed as not having synovitis.
The aim of the present study was to compare the burden of disease and its development over time in patients referred to an early arthritis clinic who were diagnosed either as having arthralgias without synovitis or as having RA. To accomplish this, we set out to answer the following questions: 1) what are the quality of life (QOL), perceived pain, perceived fatigue, functional ability, and health care consumption levels in patients with arthralgias without clinical synovitis compared with those of RA patients and how do these domains evolve over time and 2) which factors are associated with higher health care consumption levels in patients with arthralgias without clinical synovitis and in patients with RA?
The patients in this study were selected from the Rotterdam Early Arthritis Cohort (REACH) (). All patients participating before June 2009 who gave permission to send questionnaires were included. Patients were excluded if they were diagnosed with non-RA arthritis at any moment during the 12 months of followup (Figure 1).
REACH is a prospective inception cohort that was set up in the greater Rotterdam area in July 2004. Patients were recruited at their first consultation either via the general practitioner or via the outpatient rheumatology clinics of 5 hospitals.
Patients in REACH were included if they had either or both of 1) synovitis in at least 1 joint on clinical examination and 2) pain, stiffness, or loss of function in at least 2 joints accompanied by at least 2 of the following criteria: morning stiffness >1 hour, unable to clench a fist in the morning, pain when shaking someone's hand, pins and needles in the fingers, difficulties wearing rings, difficulties wearing shoes, a family history of RA, and unexplained fatigue lasting <1 year. Patients were excluded if their symptoms existed for >12 months or if their symptoms resulted from trauma or overuse. This study was approved by the ethics committees of the 5 participating hospitals. All patients gave written informed consent.
Pain was assessed using the Rheumatoid Arthritis Disease Activity Index (RADAI) questionnaire (). The RADAI questionnaire was modified to exclude 2 questions regarding the activity of joint inflammation. Sum scores were calculated on a scale of 0–10, where higher values indicate more symptomatic disease.
Functional ability was assessed by the Health Assessment Questionnaire (HAQ) (). The HAQ ranges from 0–3, where higher scores indicate more disability.
HRQOL was assessed by the Medical Outcomes Study Short Form 36 (SF-36) health survey. The SF-36 is a generic 36-item questionnaire covering the following 8 dimensions: physical functioning, physical role functioning, bodily pain, general health, vitality, social functioning, emotional role functioning, and mental health ([8, 9]). To provide a global measure of physical and mental functioning, component summary scores (physical component scale [PCS] and mental component scale [MCS], respectively) were calculated from the 8 separate dimensions of the SF-36. The 8 dimensions and 2 summary scores may range from 0–100, where a higher score indicates a better HRQOL.
Fatigue was measured by a questionnaire using 2 different scales: a visual analog scale (VAS) and the Fatigue Assessment Scale (FAS). The VAS is a continuous scale ranging from 0–100. Total FAS scores can range from 10–50 (). On both the VAS and FAS, higher scores indicate higher levels of fatigue.
Health care utilization (HCU) was assessed by a questionnaire at baseline and at 6 and 12 months of followup. At each point, the patients were asked to report the number of visits to general practitioners, medical specialists, physiotherapists, nurse specialists, occupational physicians, and other health care providers for joint symptoms (to be specified by the patient). If any health care provider was consulted, patients were asked to report the number of visits.
To evaluate the overall use of care, a combined HCU measure was constructed as follows: visits to general practitioner + visits to a medical specialist + visits to a physiotherapist divided by 5 + visits to alternative health care providers. Because multiple sessions for physiotherapy are generally prescribed, we chose a correction factor of 5 (based on the distribution of our data, 15% had >5 visits to a physiotherapist at baseline) to be applied for its contribution to the combined measure.
A trained research nurse took a standardized history and conducted a physical examination at baseline, after 6 months, and after 12 months. Tender and swollen joint counts were computed evaluating 53 joints and 44 joints, respectively (as required to calculate the original Disease Activity Score) (). The presence or absence of synovitis was confirmed by the treating rheumatologist. The diagnoses were obtained directly from the treating rheumatologists or chart reviews.
The patients were asked about their age, sex, and ethnicity. Education level was categorized as low (primary school, lower and intermediate secondary schooling, or intermediate vocational training), intermediate (higher secondary schooling or intermediate vocational training), and high (higher vocational training or university). Employment status was defined as having paid employment (yes/no). Living status was ascertained and patients were classified as living alone or with others. Body mass index was categorized into obese (≥30 kg/m2) or nonobese (<30 kg/m2).
Coping style was assessed using the Coping of Rheumatic Stressors (CORS) questionnaire ([12-14]). The questionnaire consists of 2 scales: decreasing activities to cope with pain and pacing to cope with limitations. Sum scores were computed; a higher sum score indicates more frequent use of the coping strategy.
Perceived control over health outcomes was measured by the Multidimensional Health Locus of Control questionnaire (MHLC). The MHLC assesses 3 different dimensions of perceived health control by means of 3 scales: internal, external, and chance ([15, 16]). The internal scale reflects the belief that people are personally responsible for their own health. The external scale reflects that a physician is responsible for one's health. The chance scale reflects the belief that health depends on chance, luck, or fate. The subscale scores range from 6–36, where a higher score indicates that a patient's belief is stronger in that particular health locus of control. The scales are not opposite ends of the same spectrum; therefore, it is possible to have, for example, both internal and external beliefs about health status simultaneously.
Anxiety and depression were ascertained using the Hospital Anxiety and Depression Scale (HADS) ([17, 18]). The HADS was originally developed to identify anxiety disorders and depression among patients in nonpsychiatric hospital clinics. Both the anxiety and depression subscales range from 0–21, where higher scores indicate more anxiety or depression.
Characteristics of the study population were described using simple descriptive analysis techniques. The baseline differences between groups among continuous variables were tested with the unpaired t-test or with Wilcoxon's rank sum test if data were not normally distributed. Categorical variables were tested using Pearson's chi-square test.
Determinants of HCU were evaluated using Poisson regression analyses for repeated data. A 6-month time-lag model was chosen, implying that the measurement of a risk factor was related to the outcome measured 6 months later. For variables that measured overlapping constructs (HADS and SF-36 MCS, HAQ and SF-36 PCS, and tender joint count and RADAI) or were highly correlated (coping with pain and coping with limitations), one variable was selected for inclusion into the model. Covariates that were collected solely at baseline were included in the model as time-independent covariates. All univariate analyses were performed taking into account the evolution of HCU over time. For the multivariate models, variables were first selected based on their level of significance in the univariate analysis (P ≤ 0.20), then backward stepwise selection was performed, while the covariates month, age, and sex were included by default. Missing covariates were imputed with their corresponding individual value at 6 or 12 months of followup. If neither of these values was available, the group mean was imputed. Subsequently, missing values at 6 months of followup were imputed with their corresponding baseline values and missing values at 12 months of followup were imputed with their corresponding values at 6 months. All statistical analyses were performed with the statistical package STATA, version 12.0 SE, using a P value of less than or equal to 0.05 as the level of statistical significance.
At baseline, 330 patients with inflammatory arthralgias without synovitis (nonsynovitis [NS] group) and 244 RA patients (RA group) were included. Of the 244 patients in the RA group, 166 patients (68.0%) fulfilled the American College of Rheumatology (ACR) 1987 RA criteria and 171 patients (70.1%) fulfilled the ACR 2010 RA criteria. Furthermore, 124 patients (50.8%) fulfilled both criteria, 42 patients (17.2%) fulfilled the ACR 1987 RA criteria only, 47 patients (19.3%) fulfilled the ACR 2010 RA criteria only, and 31 patients (12.7%) fulfilled neither set of criteria ([19, 20]).
On average, patients in the RA group were older (mean age 54.0 years versus 45.0 years), were more often men, had lower education, were more often unemployed, had a shorter duration of symptoms (103 days versus 136 days), had more tender joints (9 versus 4), had higher pain scores (RADAI: 3.3 versus 2.5), and had higher functional disability scores (HAQ: 1 versus 0.6) than patients in the NS group (Table 1).
|Nonsynovitis (n = 330)||Rheumatoid arthritis (n = 244)||P|
|Clinical and demographic characteristics|
|Age, mean ± SD years (no. of patients)||45.0 ± 12.4 (330)||54.0 ± 13.7 (244)||< 0.001a|
|Women, no./total (%)||282/330 (85)||165/244 (68)||< 0.001b|
|Dutch ethnicity, no./total (%)||251/309 (81)||188/229 (77)||0.823c|
|Education, no./total (%)||0.008b|
|Low||154/310 (50)||145/230 (59)|
|Intermediate||96/310 (31)||54/230 (22)|
|High||60/310 (19)||31/230 (13)|
|Living alone, no./total (%)||40/312 (13)||38/231 (16)||0.266c|
|Paid employment, no./total (%)||204/313 (65)||127/231 (52)||0.017b|
|BMI, mean ± SD kg/m2 (no. of patients)||26.8 ± 5.1 (321)||26.4 ± 4.6 (219)||0.376d|
|Duration of symptoms, median (range)/no. of patients||136 (7–380)/329||103 (7–373)/244||< 0.001e|
|Tender joint count, median (range)/no. of patients||4 (0–26)/328||9 (0–45)/241||< 0.001e|
|RADAI (range 0–10), median (range)/no. of patients||2.5 (0–8.4)/304||3.3 (0–9.5)/225||0.004e|
|HAQ (range 0–3), median (range)/no. of patients||0.6 (0–2.3)/310||1 (0–2.9)/229||< 0.001e|
|Fatigue (FAS), median (range)/no. of patients||22 (10–48)/307||21 (10–47)/225||0.177f|
|Diagnosis, no./total (%)||N/A|
|Joint symptoms not specified otherwise||201/330 (61)|
|Comorbid conditions, no./total (%)||0.382c|
|None||83/320 (26)||76/232 (31)|
|1||124/320 (39)||80/232 (33)|
|2||66/320 (21)||44/232 (18)|
|≥3||47/320 (15)||32/232 (13)|
|Coping with pain (range 8–32), median (range)/no. of patients||14 (8–29)/310||15 (8–30)/229||< 0.001e|
|Coping with limitations (range 8–40), median (range)/no. of patients||20 (10–37)/310||23 (10–40)/229||< 0.001e|
|Locus of control, median (range)/no. of patients|
|Internal (range 6–36)||21 (7–32)/307||21 (6–34)/231||0.541f|
|External (range 6–36)||17 (6–34)/307||20 (9–35)/231||< 0.001e|
|Chance (range 6–36)||19 (6–35)/307||20 (6–36)/231||0.585f|
|HADS anxiety (range 0–21), median (range)/no. of patients||6 (0–21)/309||5 (0–18)/231||0.151f|
|HADS depression (range 0–21), median (range)/no. of patients||3 (0–18)/309||4 (0–18)/229||0.067f|
|PCS (range 0–100), median (range)/no. of patients||39 (12–59)/306||32 (9–58)/225||< 0.001e|
|MCS (range 0–100), median (range)/no. of patients||54 (20–70)/306||54 (21–74)/225||0.343f|
Psychosocial characteristics were very similar between groups (Table 1). Although significant differences were found on coping with pain and limitations (CORS), physical health (SF-36 PCS), and external locus of control, the differences were too small to have clinical relevance. Pain scores (RADAI) decreased over time in both groups (Figure 2A) from 2.9 (95% confidence interval [95% CI] 2.7–3.1) to 2.1 (95% CI 1.8–2.3) and from 3.4 (95% CI 3.1–3.7) to 1.8 (95% CI 1.6–2.0) in the NS and RA groups, respectively. A trend was seen for higher pain scores at baseline and lower pain scores after 6 and 12 months in the RA group. Baseline functional ability (HAQ) scores were worse in the RA group but decreased over time (Figure 2B). HAQ scores remained more or less constant over time in the NS group. Fatigue scores, as measured with the FAS, were similar in both groups and remained constant over time (Figure 2C). Fatigue scores measured on the VAS were similar in both groups and decreased from 55 (95% CI 52–57) and 51 (95% CI 48–54) at baseline to 47 (95% CI 44–51) and 42 (95% CI 38–46) after 12 months in the NS and RA groups, respectively (Figure 2D). HRQOL (SF-36 subscales) was comparable or somewhat lower for the RA group, improving over time for both groups (Figure 3). However, HRQOL remained considerably lower than the population average in both groups.
HCU decreased over time in both groups and was similar in both groups, except for more visits to medical specialists in the RA group, which one would expect (Figure 4). In the multivariate Poisson analysis, high levels of HCU were associated in the NS group with increased pain (incidence rate ratio [IRR] 1.10, 95% CI 1.02–1.17), worse physical health (IRR 0.98, 95% CI 0.96–0.99), and external locus of control (IRR 1.04, 95% CI 1.01–1.07) (Table 2). In the RA group, worse physical functioning (IRR 0.98, 95% CI 0.97–0.99) was associated with high levels of HCU, while a longer duration of symptoms (IRR 0.65, 95% CI 0.50–0.85) and high change locus of control (IRR 0.97, 95% CI 0.95–0.99) were associated with low levels of HCU (Table 2).
|Nonsynovitis, multivariate model||Rheumatoid arthritis, multivariate model|
|IRR (95% CI)||P||IRR (95% CI)||P|
|Month||0.944 (0.922–0.967)a||0.000a||0.962 (0.942–0.982)a||0.000a|
|Ageb||1.001 (0.990–1.013)||0.858||1.000 (0.992–1.008)||0.992|
|Female sex||1.110 (0.755–1.630)||0.597||1.085 (0.872–1.348)||0.465|
|Duration of symptoms|
|90–180 days||0.922 (0.737–1.153)||0.476|
|>180 days||0.653 (0.500–0.852)a||0.002a|
|Pain (RADAI)||1.095 (1.023–1.172)a||0.009a|
|External locus of controlb||1.036 (1.006–1.066)a||0.018a|
|Chance locus of controlb||0.972 (0.953–0.991)a||0.004a|
|SF-36 PCS||0.975 (0.963–0.987)a||0.000a||0.980 (0.971–0.988)a||0.000a|
Overall, we observed that the burden of disease (as measured by pain, disability, fatigue, and QOL) and health care consumption levels were very similar between patients with arthralgias without synovitis compared to patients with RA. Our results indicate that, despite the absence of an underlying inflammatory process, patients with arthralgias without synovitis experienced a burden of disease that was similar to that of patients diagnosed with early RA, at least during the first year of followup. Pain scores, functional ability, HRQOL, and health care consumption levels improved over time in both groups. Fatigue scores improved only marginally (VAS fatigue) or not at all (FAS) in both groups. The groups differed most with respect to baseline disability scores (HAQ), which were higher in the RA group. Small differences were observed with respect to health care consumption levels, which were somewhat higher in the RA group, and HRQOL scores (SF-36 subscales), which were either similar or somewhat worse in the RA group. The finding that HCU levels in the NS group were quite similar to those observed in the RA group is surprising. The RA patients were actively being treated for their disease and had regularly scheduled visits with their rheumatologists, which accounted for a large part of the health care consumption levels observed. Patients in the NS group, conversely, probably sought care for different reasons, possibly searching for a diagnosis and relief for their symptoms.
Studies on the burden of disease measures and health care consumption in RA patients are sparse and are absent in patients with arthralgias without synovitis. In a prospective cohort study of 183 patients with early RA included between 1985 and 1989, Lindqvist et al found that median HAQ scores increased from 0.8 at baseline to 1 after 1 year of followup (). In contrast, we found that median HAQ scores decreased in the RA group from 1.0 at baseline to 0.6 after 12 months. These differences are probably a consequence of the current availability of more effective treatment regimens. In accordance with our findings, a study comparing 4 cohorts of patients with early RA found that pain scores (VAS) decreased in all cohorts during the first year of followup (). In this Dutch cohort, pain scores decreased from 3.9 at baseline to 2.6 at 1 year of followup ().
Despite substantial improvements in pain, functional ability, and HRQOL scores, we only observed a modest improvement in fatigue scores as measured on the VAS, while no improvement in fatigue scores was observed on the FAS. It is possible that the FAS is less sensitive than the VAS for the detection of subtle changes in fatigue. We did not find any studies investigating levels of fatigue in NS patients published in the current literature. However, in line with our findings, studies investigating the effect of treatment with conventional or biologic disease-modifying antirheumatic drugs on VAS fatigue scores in RA patients found improvements, but with small effect sizes as well ([23, 24]).
A second aim of this study was to identify the factors associated with high HCU levels in patients with arthralgias without synovitis and patients with RA. Our multivariate Poisson regression analyses showed that overall HCU was associated with worse physical health in both groups. Differences between the groups were found for the duration of symptoms at baseline (>6 months) and high chance locus of control, which were associated with lower HCU in the RA group, while higher pain scores and external locus of control were associated with higher HCU in the NS group. We could not find any previous studies in which HCU was assessed in an early arthritis cohort. In a cross-sectional study among 1,200 patients with established RA, Jacobi et al found that overall high use of care was associated with younger age, female sex, longer disease duration, and having ≥2 comorbidities (). We could not confirm a relationship for age, sex, and comorbidities. These discrepancies could be explained by the smaller sample size of our study, or the difference between patients in the 2 studies (longstanding RA in the study by Jacobi et al versus newly diagnosed RA in our study). Moreover, Jacobi et al found increased disease duration to be associated with high HCU, while we found a longer duration of symptoms to be associated with less HCU.
This study has several strengths and limitations. The strong points include that data were obtained prospectively and repeatedly within an early arthritis cohort. The determinants for HCU were analyzed using a Poisson model for repeated data, taking into account progression of the outcome over time. One possible limitation of this study is the relatively small sample sizes of the 2 patient groups; the possible determinants for HCU might not have been detected because of this. In addition, HCU was measured using questionnaires, introducing a potential for recall bias. Recall bias could have led to both an over- and an underestimation of HCU, although underreporting was found to be the more common problem when using questionnaire data (). In either case, because disease burden was very similar among the groups, we would expect that if recall bias indeed was present, it would have been similar among the groups. This would then lead to nondifferential misclassification of health care use, resulting in a dilution of effect sizes for determinants of HCU in our Poisson regression models. Therefore, the relatively weak effect sizes observed in the multivariate analyses for determinants of HCU might have been a consequence of nondifferential misclassification of HCU due to recall bias. Another limitation might be that some patients diagnosed with fibromyalgia (n = 10) were included in the NS group because fibromyalgia was not an a priori exclusion criterion. This could have biased the results for the group; however, excluding the 10 patients with fibromyalgia (3.8% of the total) from the analysis of determinants of HCU in the NS group did not substantially change the results as previously found (data not shown). Furthermore, the finding that HRQOL was similar between groups pertained only to the first year of followup. Further observation is required to prove if this was still true later on. Another limitation could be that patients in the NS group were selected on the basis that they did not develop any form of arthritis during the first 12 months of followup, while they might develop arthritis later on. Patients in our cohort were followed for 24 months; therefore, we checked whether patients in the NS group had developed any form of arthritis at 24 months of followup or beyond. At 24 months of followup, 4 patients in the NS group had developed arthritis; 2 patients were diagnosed with RA, 1 patient was diagnosed with polyarthtritis, and 1 patient was diagnosed with oligoarthritis. Beyond the 24 months of followup, 8 more patients developed a form of arthritis; 5 patients were diagnosed with RA, 2 patients were diagnosed with psoriatic arthritis, and 1 patient was diagnosed with synovitis due to osteoarthritis. However, excluding the 12 patients developing arthritis beyond 12 months of followup from the analysis did not significantly change the results (data not shown).
In conclusion, we found that the burden of disease as well as HCU in patients with arthralgias without synovitis were comparable to those of patients with RA. Although both groups showed improvements in pain scores, QOL, and reduction of HCU over time, only a modest improvement was seen for fatigue in both groups.
HRQOL in the NS group remained substantially lower compared with the general population. We therefore believe that, after synovitis is sufficiently ruled out by the rheumatologist, patients should be offered further support and monitoring by their general practitioner. Depending on the burden of health symptoms, the emphasis could lie on helping patients to better cope with their symptoms, for instance, through psychological interventions. However, if any joint swelling does occur in the future, patients should be referred back to a rheumatologist for further examination.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Kuijper had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Luime, Alves, Barendregt, Hazes.
Acquisition of data. Kuijper, Alves, Barendregt, van Zeben, Hazes.
Analysis and interpretation of data. Kuijper, Luime, Bindels, Hazes.