Recently, the use of patient self-reporting instruments instead of clinical, objective measurements to assess rheumatoid arthritis (RA) patients was proposed. This assumes a constant association between disease activity and the self-reporting instruments. The objective was to explore the association (in time) between disease activity and patient perception of general health, disease activity, pain, and functional disability in patients with RA.
Data of 624 newly diagnosed RA patients who completed 3 years of followup were analyzed. Cross-sectional linear regression models and longitudinal regression models were estimated, with a visual analog scale (VAS) measuring general health (VAS-GH; 0 = best, 100 = worst) as a dependent variable and the Disease Activity Score (DAS28) without the VAS-GH as an independent variable. Other dependent variables were VAS disease activity, pain, and the Health Assessment Questionnaire.
The DAS28 and VAS-GH were significantly associated in RA patients (P < 0.001). However, the explained variance was low (6.7%). From diagnosis to 3 years after the diagnosis, the intercept decreased given the same regression coefficient. The longitudinal regression model showed that the VAS-GH improved during disease course independent of a change in DAS28. Analyses on the other outcome parameters showed similar results.
Patients' perception of health can be different with equal disease activity, depending on the moment in the disease course. Furthermore, our results indicate that self-reported measures on functionality, disease activity, and general health cannot substitute for objective measures of disease activity in RA in longitudinal studies; subsequently, both need to be measured.
Rheumatoid arthritis (RA) is a chronic, inflammatory, systemic disease, with an unpredictable course varying from mild to very severe. Patients with RA mostly experience pain and swelling in affected joints and generalized stiffness, especially in the morning. The disease has great influence on functional ability and quality of life. The goal of medical treatment of RA is to suppress disease activity and, consequently, to prevent joint damage and enhance the patient's quality of life.
Patients with RA are assessed in clinical trials and clinical practice using a core set of measurements consisting of a swollen joint count and tender joint count, assessment of acute phase reactants, physician's global assessment of disease activity, patient's global assessment of disease activity, assessment of general health and pain, and the assessment of functional ability (1). An index of disease activity for patients with RA, the Disease Activity Score in 28 joints (DAS28), can be calculated on the basis of the 28-joint count for swelling, 28-joint count for tenderness, erythrocyte sedimentation rate (ESR), and a visual analog scale (VAS) for general health (2). Response to treatment can be assessed by the American College of Rheumatology (ACR) 20% response criteria (3) or the European League Against Rheumatism (EULAR) response criteria (4).
Recently, it was demonstrated that a combination of the Health Assessment Questionnaire (HAQ), VAS pain, and patient global assessment (VAS or Likert scale) appeared to be as equally informative as the ACR20 response criteria or EULAR response criteria in distinguishing between active treatment and placebo treatment (5, 6). Subsequently, the use of only self-reporting instruments instead of physician-assessed disease activity instruments was suggested to evaluate results of drug therapy in clinical trials, observational studies, and clinical practice (7). Besides the ability to discriminate between responders and nonresponders, self-reporting instruments are very easy to administer. Patients can complete the instruments while waiting for the physician and the physician can scan them in approximately <30 seconds (5).
In all fields of medicine, clinical variables are closely related to the pathophysiologic process (i.e., blood pressure in hypertension, blood glucose concentration in diabetes, and disease activity in RA). Implicitly, clinicians assume that when they improve those clinical parameters, they will also improve the patient's quality of life and overall well-being. When using measures of functional disability and overall well-being to evaluate treatment in RA, as suggested recently, it is assumed that a change in inflammatory-related parameters is reflected in a change in overall well-being parameters. However, how valid is such an assumption?
It is well known that standards by which individuals assess their own health status can change over time (8–10), and therefore one should be cautious when using subjective, self-reporting instruments to evaluate treatment. Changes in health are accompanied by processes of readjustment, reassessment, and coping (11). When a patient is newly diagnosed with a disease, coping and, consequently, adaptation to the disease are probably barely present. In that case, the perception of health is likely to be associated with the objectively measured clinical status. Later in the disease course, the symptoms can still be present, but the patient is better able to cope with them. In this case, the health perception will often be better than one would expect based on the patient's objectively measured clinical status (9).
The objective of this study was to explore the association between objectively measured clinical status (as measured by the DAS28 excluding the VAS) and subjective perception of health in patients with RA. Further, we tested the hypothesis that later in the disease course the perception of health is better than one would expect based on the disease activity of the patient, compared with shortly after diagnosis.
PATIENTS AND METHODS
Data from 2 observational cohort studies of patients with RA, 1 from the Department of Rheumatology of the Radboud University Nijmegen Medical Centre (UMCN) and 1 from the Sint Maartenskliniek in Nijmegen (SMK), were used. Both cohorts are ongoing and continuously include all newly diagnosed patients with RA (disease duration <1 year and no prior use of disease-modifying antirheumatic drugs) since 1985 in the UMCN and since 1992 in the SMK. The assessment protocols are equal for both cohorts. Patients are treated according to the judgement of their rheumatologists and are assessed every 3 months by a trained research nurse. A more detailed description of the UMCN cohort is given elsewhere (12). For the present analysis, data from the first 3 years after diagnosis were used because it was thought that increasing joint damage later in the disease course would affect clinical status apart from present disease activity and thus would become a potential confounding factor within our analyses (13).
Disease activity was used as an objective measure of patients' clinical status and was measured by the DAS28. The DAS28 yields a score between 0 and 10, where a higher score indicates higher activity. The DAS28 is a statistically derived index combining the following variables: the 28-joint count for swelling, the 28-joint count for tenderness, ESR, and a VAS measuring general health (2). The DAS28 can also be calculated without the VAS, which is how it was used in the present analysis.
Several self-reporting measures were available in the data set. The primary outcome parameter was a VAS measuring general health as perceived by the patient (VAS-GH). Secondary outcome measures were a VAS measuring pain and a VAS measuring disease activity as perceived by the patient (VAS-DA). All VAS ranged from 0 to 100, where 0 was the most favorable state (best imaginable health or no pain at all) and 100 was the worst state. The HAQ was also included as an outcome measure. The HAQ is a 20-item questionnaire measuring functional disability, which results in a score between 0 (no disability) and 3 (severe disability). The study cohort provided more data on the VAS-GH and the VAS pain than on the VAS-DA and the HAQ.
Only patients who completed followup were included in the analyses. Means, standard deviations, and percentages were calculated for demographic characteristics of the patients at baseline. To explore the relationship between disease activity and the subjective perception of general health, linear regression models were estimated at baseline, 6, 12, 18, 24, 30, and 36 months with the VAS-GH as an outcome variable and the DAS28 as a predictor.
To investigate the hypothesis that the observed VAS-GH might differ from the predicted VAS-GH on the basis of the DAS28, the estimated regression model at baseline was used to predict the VAS-GH for 3, 6, 9, 12, 15, 18, 24, 30, and 36 months based on the observed DAS28 at these time points. This predicted VAS-GH was plotted with the observed VAS-GH at the time points concerned.
To test the hypothesis that VAS-GH scores will improve later in the disease course, independent of a change in disease activity (DAS28), a longitudinal regression model (random coefficient model) was used to determine the relationship between the VAS-GH and DAS28 over the course of the disease. A random intercept was chosen to account for variation in baseline values. Functional disability (HAQ), sex, and age at time of diagnosis were added to the model to test for confounding or effect modification. All analyses were repeated for the other outcome parameters and were performed using the statistical software package SAS version 8.2 (SAS Institute, Cary, NC).
In August 2004, 908 patients were included in the cohort, of which 624 completed followup (3 years). Of the 284 patients with <3 years of followup, 154 patients were censored patients (i.e., included <3 years ago) and 130 were lost to followup. Reasons for being lost to followup were as follows: deceased (n = 18), treatment at another hospital (n = 13), stopped involuntarily (e.g., severe comorbidities; n = 16), stopped voluntarily (e.g., no longer interested in participating, low disease activity; n = 53), and other reasons (n = 30).
At baseline, the majority of patients were women (65.5%) and rheumatoid factor positive (76.8%). The mean ± SD age was 54.3 ± 13.7 years. Mean ± SD baseline DAS28, VAS-GH, VAS pain, VAS-DA, and HAQ scores were 5.2 ± 1.3, 45.2 ± 23.4, 45.3 ± 22.1, 46.1 ± 23.6, and 0.77 ± 0.6, respectively.
Results from analysis.
The results of the cross-sectional regression analyses at baseline and at 36 months are shown in Figure 1. The association between VAS-GH and DAS28 was statistically significant (P < 0.0001) at all time points, but the explained variance was low (6.7% for both baseline and 36 months). The β coefficients (slopes of the regression) were found to be equal, but the value of the intercept was lower in the model at 36 months, which means that patients reported a better general health given the same disease activity later in the disease course.
The observed and the predicted VAS-GH scores are shown in Figure 2. Figure 2 also shows that the observed VAS-GH scores reflect a better general health than the predicted VAS-GH.
The longitudinal regression model (Table 1) demonstrated that the DAS28 and time were associated with the VAS-GH. Time was negatively associated with the VAS-GH, indicating that general health improves during the disease course independent of the DAS28. The significant quadratic term of time indicates that the effect of time on VAS-GH scores decreased during the disease course (i.e., after an initial improvement, the decrease in VAS-GH score stabilizes, as is also shown in Figure 2).
Table 1. Results of the longitudinal regression model with the visual analog scale measuring general health as the dependent variable*
DAS28 = Disease Activity Score based on 28-joint counts; HAQ = Health Assessment Questionnaire.
A second longitudinal regression model including covariates such as age and sex showed that these covariates did not confound the relationship between time and VAS-GH (Table 1). A third model demonstrated that functional disability (HAQ) was a confounding factor for time as a predictor (Table 1). However, time still had a substantial independent effect on VAS-GH scores. Results of all 3 analyses pointed in the same direction, namely, that general health perception improves after diagnosis, independent of disease activity in the first 3 years.
Similar results of the longitudinal models for all other outcomes are shown in Table 2. All other analyses performed with the VAS pain and the VAS-DA showed the same results as the VAS-GH (data not shown). The HAQ, however, showed conflicting results, namely, that functional disability worsened in time despite a decrease in disease activity. Based on the relationship between the DAS28 and the HAQ at baseline, HAQ scores were expected to decrease (predicted HAQ) (Figure 3). Instead, the observed HAQ scores increased.
Table 2. Results of the longitudinal regression model with the VAS pain, the VAS measuring disease activity, and the HAQ as the dependent variable*
VAS = visual analog scale; HAQ = Health Assessment Questionnaire; DAS28 = Disease Activity Score based on 28-joint counts; M Sharp = modified Sharp score for measuring joint destruction.
Recently, the use of patient self-reporting instruments instead of clinical, objective measurements to assess patients with RA in clinical trials, observational studies, and clinical practice was proposed (7). The use of subjective, self-reporting instruments assumes at least a moderate and constant association between objectively measured disease activity and the data of self-reporting instruments. Analyses showed that there is an association, although weak, between subjective patient self-reporting instruments (VAS-GH) and objective clinical measures (DAS28). More importantly, it was demonstrated that the patient's perception of general health is better, given the same disease activity (DAS28) later in the disease course as compared with shortly after diagnosis. Similar results were seen from analyses with the VAS pain and the VAS-DA as outcome measures.
This incongruence over time between objectively measured and self-perceived health has implications for interpreting subjective, self-reporting instrument data in the evaluation of treatments. It emphasizes the fact that disease activity and the patient's perception of health, disease activity, and pain should be measured complementary to each other.
An important goal in managing RA is to prevent or control joint damage and loss of function and decrease pain (14). Recently, it was demonstrated that a therapeutic strategy intended for tight control of disease activity (measured with the original DAS) improved disease activity, radiographic disease progression, physical functioning, and quality of life compared with routine outpatient care (15). It is unknown what the effect will be of a treatment strategy intended for improving patients' perception of general health or functional ability on joint damage or disease activity later in the disease course. There is broad agreement on the fact that the goal of medical treatment should be to suppress disease activity and prevent joint damage. However, patients' self-reported well-being, pain, and disease activity are of major concern in making treatment decisions.
The incongruence over time between objectively measured and self-reported health might be associated with response shift effects (10). Possible causes of such a response shift might be adaptation and coping strategies, social comparison (16), information bias in time (learning processes) (17), or a change in expectations (18). At presentation of a disease (such as RA), there will be great anxiety about the future prospects and the unknown course of the disease, which will be reflected in measures of well-being. After a few months, patients have learned about the treatment options and their prospects, and they start to cope with their disease and subconsciously compare themselves with other patients with RA. Standards by which individuals assess their own health status are dynamic, and therefore response shift processes as described here can occur with every life event experienced by an individual patient.
Further, general health is an encompassing construct, of which disease activity is only one aspect. Other factors like social functioning, fatigue, and depression play a part in general health and are not captured by the DAS28. This also applies more or less to self-reported disease activity, pain, and functionality. Recent randomized controlled trials in rheumatology including all core set measures as well as measures of overall well-being demonstrated a smaller effect on health-related quality of life measures than on disease activity measures (19, 20). Together, this is why self-reported measures of functionality, disease activity, and general health cannot substitute objective measures of disease activity in RA; subsequently, both need to be measured.
Analyses with the HAQ as an outcome parameter showed conflicting results. One explanation might be that the HAQ is a more objective self-reporting questionnaire and is therefore less susceptible for possible response shift effects. Second, joint destruction could affect functional disability besides disease activity. Analyses showed that the effect of joint destruction on the HAQ almost reached statistical significance (P = 0.07). A last possible explanation for the increase in HAQ scores, despite a decrease in disease activity, could be that HAQ scores increase with increasing age in the RA population as well as in the general population (21).
Some limitations of this study should be mentioned. First, the DAS28 was used as the only objective variable associated with the patient's subjective general health assessments. The DAS28 was chosen because the purpose of this study was to explore the relationship between objective and subjective measures in RA, and in our opinion the DAS28 is the most objective measure in this field. Further, it is possible that other variables are associated with the general health perception of patients with RA as well; however, longitudinal regression analysis including variables such as sex, age, and functional disability did not change the independent association between time and VAS in either magnitude or direction. Finally, there were less data available on the VAS-DA and the HAQ, therefore results of analyses with these parameters are less robust.
It can be concluded that the patient's perception of health can be different with equal disease activity, depending on the moment in the disease course. Furthermore, our results indicate that self-reported measures of functionality, disease activity, and general health cannot substitute objective measures of disease activity in RA in longitudinal studies; subsequently, both need to be measured.