Over the last decade, major changes have occurred in the provision of health care for patients with rheumatoid arthritis (RA). Growing attention has been paid to improved management strategies with early and more aggressive treatment,1 which reflects important advances in the treatment along with the access to more effective and specific drugs for patients with RA. Some knowledge is available regarding the global change of health status in RA with time, and some studies have suggested that recent advances in treatment are also reflected in improved health status.2 3

Questionnaire instruments are useful practical tools for measuring the true burden that RA imposes on the patient in terms of outcomes and quality of life.4 Knowledge on patients’ quality of life is of interest both on the individual level in the clinical setting and on a population level for the purpose of management strategies.

Changes over time are typically studied in a setting of longitudinal observational studies of a patient cohort,5 and patient registers provide a setting with a high level of internal validity.6 The Oslo RA register (ORAR) provides data that are representative for the entire population in this geographic area. This setting makes comparison of cross-sectional cohorts possible with regard to changes in burden of health over time. The aim of this study was to examine possible changes from 1994 to 2004 regarding health status and levels of self-reported health within the setting of ORAR. We further sought to examine whether after adjustment for age, gender and disease duration there was a secular change towards less disease burden.

PATIENTS AND METHODS

Study population

Patients with RA living in Oslo were included in the ORAR.7 The register is annually updated with respect to new and deceased cases and with residential addresses. The completeness has been evaluated and is assumed to be 85%,7 8 and the number of enrolled cases has been fairly constant since the first data collection in 1994. The analyses of this study relate to all living patients in the age range 20–79 years for which the register has been validated:7 1994, n = 1315; 1996, n = 1375; 2001, n = 1411; 2004, n = 1454.

Age was recorded in years when questionnaires were filled in, disease onset as the year when the patient fulfilled the classification criteria for RA,9 and disease duration was recorded as the year of examination minus year of disease onset.

Measures

Mailed surveys to the living register patients were performed in 1994, 1996, 2001, and again in 2004. The same core set of health status measures were included in all surveys, ie pain, fatigue and patient global assessment of disease activity on 100 mm visual analogue scales (VAS), the Modified Health Assessment Questionnaire (MHAQ),10 the MOS short form SF-3611 12 and the Arthritis Impact Measurement Scales 2 (AIMS2).13 MHAQ10 is an eight-item questionnaire measuring ability to perform daily activities (scale 1–4, 4 worst health). SF-3611 12 is a generic measure of eight health dimensions (physical functioning, physical role, bodily pain, general health, vitality, social functioning, emotional role, mental health) with scale from 0 to 100 (0 worst health). Physical (PCS) and mental component summary (MCS) scores were aggregated from SF-36.14 The utility measure (SF-6D) as a value-based measure for overall disease impact was derived from the responses to the SF-36 questionnaire based on an algorithm developed by Brazier et al.15 This utility score has a range from 0 to 1, where 0 denotes death and 1 perfect health. The lowest possible value for living subjects with SF-6D is 0.30. AIMS213 is a multidimensional disease-specific instrument capturing information in 12 areas of health (mobility level, walking and bending, hand and finger function, arm function, self-care tasks, household tasks, social activity, support from family and friends, pain, work, level of tension, mood), which can be aggregated into five major dimensions (physical, social interaction, pain, work and affect). The scores for AIMS2 range from 0 to 10 (10 worst health). The self-efficacy scales16 for pain, function and symptoms were also used in 1994, 2001 and 2004 with scores ranging from 10 (low self-efficacy) to 100 (high self-efficacy).

The current use of drugs was self-reported for disease-modifying drug (DMARD) and glucocorticosteroids at all time points and for symptom-modifying drugs and analgesics at time points 1994, 2001 and 2004 on a check-list. Co-morbidities were assessed from question 60 in AIMS2 and categorised as present (at least one co-morbidity) or absent.

Analysis

Descriptive statistics present continuous variables with mean (SD), and categorical values as counts or proportions, providing 95% confidence intervals (CI) as appropriate. As a conservative estimate we assumed that an important/significant change (cohort effect) had occurred if the 95% CI intervals of a measure in 1994 and at subsequent years of examination did not overlap.

To study the longitudinal relationship between time of disease onset or time of examination and indicators of disease severity we grouped years of disease onset for the periods 1980–4 (n = 312), 1985–9 (n = 476) and 1990–4 (n = 534). Indicators of disease severity were used as dependent variables and the four time points of examination as explanatory variables. A repeated measures analysis of variance (ANOVA) using a mixed models approach was applied in this approach. The main advantages of the mixed model approach are that patients with missing data at single time points are retained in the analyses and that unequal time intervals are handled appropriately. Sex, age and co-morbidity were kept as a factor or a covariate in all analyses, while disease duration in years and current DMARD use were kept as a covariate in the analysis if they contributed statistically significantly (p<0.05) in the model with a particular health status outcome.

Statistical Package for Social Sciences (SPSS Inc, Chicago, Illinois, USA) version 12.0.1 and the Statistical Analysis System (SAS version 9.1.3) were used for analyses. p⩽0.05 was considered significant. No further correction for multiple testing was carried out.

RESULTS

The respondents in the examined cohorts in at all four time points during the 10-year period did not differ for important demographic and disease variables (table 1). Of 931 respondents in the age range 20–79 years in 1994, 310 (33%) provided data for all four time points of data collection.

Health status in RA was from 1994 to 2004 statistically or numerically improved across all dimensions of health (table 2). The most consistent and pronounced improvement across instruments was seen for the physical dimension and global health measures (table 2), and statistically significant improvement was also observed for all three pain measures. The mean value for SF-6D derived utility was 0.616 in 1994 and 0.647 in 2004.

DMARD (including biological agents) and non-steroidal anti-inflammatory drugs as well as cyclo-oxygenase II inhibitors were, in 2001 and 2004, used by more than half of the patients. At least one co-morbidity was at the four time points reported by between 60.6% and 64.5% of patients. Use of biological medication increased from 3.1 to 11.8% over the 3-year period 2001–4 (table 3).

Figure 1 presents mean health status and utility values in patients with RA examined in the repeated cross-sectional analyses from 1994 to 2004 stratified to periods of disease onset (1980–4, 1985–9, 1990–4) applying mixed model analyses with adjustments for age, sex, co-morbidity and time point of examination. DMARD treatment and number of years with disease duration were only adjusted for if these variables contributed statistically significantly to the model with the specified outcome measure. The period for disease onset had effects on longitudinal outcome for SF-36 PCS (p<0.001), pain (VAS pain p<0.001, SF-36 pain p<0.001) and SF-6D utility (p<0.05) (fig 1A,D,G,H). Further, effects of the time point of examination were seen in fatigue VAS (p<0.001), pain VAS (p<0.001), patient global (p<0.001), SF-6D utility (p<0.05) and SF36 pain (p<0.05) (fig 1C–E,G,H). Both the period of disease onset and the time point of examination were associated mainly with outcomes of pain and utility, but not consistently with other health dimensions.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 (A–H) Health status and utility in patients with RA examined cross-sectionally in 1994–2004 according to periods with disease onset (1980–4, 1985–9, 1990–4) with adjusted mean values applying mixed model analyses. MCS, mental component summary (scores); MHAQ, Modified Health Assessment Questionnaire; PCS, physical component summary (score); VAS, visual analogue scale. ‡Significant effect of disease onset. †Significant effect of examination time point.

DISCUSSION

This study provides evidence that RA today is a milder disease with improvement of especially physical function during the decade 1994–2004. These findings of improved RA health coincide with major advances in treatment and health care structure in the field of rheumatology.1 17 Health status over the 10-year period in cross-sectional assessments consistently improved across all dimensions of health also applying different instruments within the same dimensions (table 2). Using multivariate analyses, patients with a more recent disease onset had more beneficial longitudinal health outcomes in key outcome variables, especially with respect to physical function, pain and utility. These observations correspond to the perception among experienced clinicians that RA presently is characterised by a reduced disease burden compared with some decades ago. This evidence for an improvement in the health of patients with RA and in utility during the last years is relevant, taking into account individual suffering and considerable costs to society resulting from this disease.18

Importantly, we also examined whether improvement in health status over time was dependent on the time of RA onset. It has specifically been suggested to examine the possible effect of time of disease onset longitudinal outcomes in the ORAR.6 Our analyses demonstrated that patients with more recent disease onset had more favourable effects on longitudinal health outcomes (fig 1), with better physical function, less pain and higher utility.

We have previously reported improved health status among patients with RA in Oslo over a 7-year period,3 and the present report demonstrates further improvement of health status in patients with RA. To address the question whether RA presents as a milder disease at the beginning of the new millennium, one needs to examine data from successive incidence cohorts where health status measurements have been performed consistently over the years. Even though a milder disease has been suspected earlier,19 such evidence has not been easy to document.2 As in our study, Krishnan and Fries2 found reduced disability in patients with RA examined over the time period 1977–98, applying the HAQ twice yearly and estimating the average disability for each patient. Comparing subsequent calendar years and adjusting for demographic and response factors as well as disease duration, the average disability declined 2% per year.2 Swedish patients with RA had lower disease activity in 1995 compared with 1978.20

In the study by Welsing et al,21 patients from more recent inception cohorts had lower disease activity at baseline and 5-year follow-up, whereas in contrast physical function assessed with HAQ showed a worsening trend in the latest cohort. Thus, the trend of less disease activity in early RA over time was not evident in patient self-report measures. The trend coincided with a tendency toward a shorter duration of symptoms at the time of diagnosis and more aggressive use of DMARD treatment during the disease course.21

Time trends in the disease status of patients with RA at first presentation, radiographic damage and mortality are also indicators for changes in disease severity. The disease characteristics of patients presenting for the first time to rheumatology care may be an indirect indicator of a milder disease per se, but may also reflect other factors such as access to health care or awareness for symptoms. Jansen et al22 found that patient referral to early arthritis clinics changed over a 6-year period towards fewer patients who fulfilled the diagnosis of RA and towards a lower erythrocyte sedimentation rate, whereas the number of swollen joints and the duration of complaints remained unchanged. The health status of patients with RA presenting at an outpatient clinic was found to be better in 2000 than in 1985, which was expressed in the number of swollen and tender joints, global assessments or disease activity score, and better physical function.23

Such observations of improved health status at referral may be confounded by a changed referral pattern if patients with milder disease gain access to rheumatology care. Furthermore, a beneficial health outcome in RA over time may not only apply to patients with RA but may correspond to a better overall health in the population.

Less radiographic damage in more recent cohorts with early RA in Finland was reported by Sokka et al.24 Differences over time were mainly due to an almost complete absence of radiographic deterioration in the latest cohorts with early RA. As possible reasons, the authors discuss improved treatment with the more extensive use of DMARD, milder disease and patient selection. More active treatment could be considered as the most likely reason in this study.25 This finding of reduced RA damage is in line with another study with patients from Wichita Kansas, which compared patients with disease onset in the 1970s, 1980s and 1990s with respect to radiographic progression of joint damage.26 While there was a trend in unadjusted analyses towards a lower progression in recent decades, these differences disappeared after adjustment for use of DMARD, glucocorticoids and baseline predictors, lending support to a milder disease attributable to effective treatment.26 An association between physical function and radiographic damage, independent of disease activity adds to the importance of radiographic damage for disease burden.27 The observed stable rates for hip and knee joint replacement in Finland between 1986 and 2003 in patients with RA compared with an apparent increase in controls suggest improved outcomes in RA.28

A reduction in mortality from the 1960s to the 1990s in patients hospitalised with RA has been reported compared with the general population, which was possibly due to an increased access to specialised rheumatology care29 or after using methotrexate.30^–32

The more aggressive use of DMARDs may be a possible explanation1 for better RA health as well as increased access to new biological treatments.17 Improved outcomes in RA seem to apply to Western countries, whereas the majority of patients with RA in many countries, including Europe, still experience a high disease burden.33 34 We found that there was an increase in the use of medication, both symptom-modifying and disease-modifying, in 2004 compared with 1994. There is, however, no consensus on how to compare the use of longitudinal DMARD in clinical care,24 and it is difficult to estimate to what extent disease-modifying treatment contributed to a milder disease. The initiation of medication is subject to channelling bias, and in our analysis of health status in patients with RA we did not only adjust for demographic and time variables, but also for current DMARD treatment in those cases where treatment had statistically significant effects. While we may believe that RA is continuously improving due to better treatments, reasons for improved disease outcomes may include decreased tobacco smoking, increased overall living standards or increased vaccination.35 Epidemiological data over the last years have also indicated a decreased overall incidence of RA with a shift to a higher age at RA incidence.36 37 Similar age at all time points in our study indicates that the average age for patients with RA did not increase. Thus, besides a hypothesis that increasingly aggressive treatment has reduced disease severity in RA, our study cannot differentiate between different explanations, including possible changes in the natural history, generally improved quality of life also in the general population, and better access to care for patients with RA.

Generally, we have reported a high level of patient involvement in rheumatology care,38 and during the whole 10-year period our department has had overall responsibility for rheumatological care of the patients with RA described in this study. Improvement in health status could thus be related to structural changes in health care from 1994 to 2004, as during that time period the number of beds and the duration of hospital stays decreased, although similar numbers of patients were hospitalised, and waiting lists and surgical procedures were largely similar in our department in 1994 and 2004. Outpatient consultations increased considerably during this time period with potentially better access to early and aggressive treatment with new drugs.1

The strengths of this study include examination of representative patients with RA in Oslo over a 10-year period with application of RA classification criteria.9 The ORAR gives the opportunity to examine health status in a well defined and representative RA population at various time points, thus following real-life impact of RA on health status. The assumption that the ORAR is representative of the entire underlying RA population in the area is based on a validation demonstrating 85% completeness in the age group 20–29 years;8 nevertheless, the number of patients in the register has remained fairly constant since, and patients with RA at the different time points had similar demographics, which strengthens our findings. Health status was at all time points assessed in an identical way by postal questionnaire also to patients not regularly seen in the clinic. The repetitive application of a combination of generic and disease-specific instruments allowed for the cross-validation of the results. The setting of the ORAR allows examination of the evolution of RA itself over consecutive time periods.

As a limitation, the setting of a repeated cross-sectional study does not allow to distinguish the impact of therapeutic interventions from that of other effects. Further, non-response to postal questionnaires containing 20–30 pages may introduce potential response bias. In 2001, the response rate was lowest with 58.8% for the age group 20–79 years; however, looking at all four time points there were no systematic differences in demographics between respondents and non-respondents. Possible differences in censorship at the different time points can, however, not be completely excluded or eliminated through statistical adjustment. In order to increase patient participation in 2004 we used incentives—we entered people into a lottery to win vouchers that could be exchanged in a nearby travel agency—and also included those who were unwilling to provide health status information. This may have contributed to an increase in the response rate to 62.9% in 2004.

While some studies have suggested an improved health status in RA during the last years, our disease register is representative and in a setting with high external validity to address issues about changes in health status over time. This study demonstrates consistent improvement in RA health status in the population from 1994 to 2004, and also better health status outcomes in patients with more recent disease onset. This reduced burden of RA may be related to improved treatment, and our findings thus support the importance of improved management strategies in population settings where the treatment of RA has not yet been given a sufficiently high priority.