Rheumatoid arthritis (RA) is a chronic disabling condition with, among other things, negative social and economic consequences for both the individual patient and society. Striking economic consequences are the diminished ability to perform the current job (i.e., the same occupation or the same number of hours) or even work disability. Costs directly ensuing from work disability account for a large part of the total RA-related costs (1). Over the past decades, a number of studies have been published reporting on work disability. In general, the percentage of patients that becomes work disabled during the course of the disease is high, although rates between studies differ. This may be attributable to differences in definition of work disability, study design, population, mean disease duration or length of followup, flexibility of labor market at the time of study, and maybe most importantly, differences in social security systems across the different countries.
Regardless of the variability in study designs, it is of interest to identify the risk factors found in the studies. Potentially modifiable risk factors might be influenced early in the course of disease.
This overview has 3 purposes: 1) to summarize percentages of work disability in patients with RA, as measured cross-sectionally or longitudinally, 2) to see which sociodemographic characteristics, clinical variables, or work-related factors are associated with work disability or are predictors of work disability; and 3) to compare rates of work disability in groups of RA patients with the country-specific rates of the general population.
The PubMed database was searched for relevant literature published from 1980 until May 2002 using the following key words: “rheumatoid arthritis,” “work disability,” and “job.” All abstracts identified in PubMed were read by 1 reviewer. Based on information provided in the abstract, full-length articles were retrieved and read and judged for potential relevance for this overview. To complete the search, articles found by cross-reference search were obtained and read. Papers had to be in English and had to mention at least the percentage of RA patients that became (partly) work disabled due to RA. In this overview, cross-sectional surveys as well as longitudinal studies were included.
Cross-sectional surveys mostly included all RA patients of working age and the percentage of work-disabled patients was determined at 1 point in time. Longitudinal studies, most often, described the (cumulative) percentage of work-disabled patients at different points in time since the onset of disease. These latter studies enabled us to determine the increase of work disability over the course of the disease, and to identify possible baseline predictors of work disability. For reasons of clarity, reported patient characteristics and percentage of work disability will be summarized separately for cross-sectional and longitudinal studies.
In total, 27 articles were selected for this overview. In 21 studies, patients fulfilled the American College of Rheumatology (formerly American Rheumatism Association) criteria for RA (2); in the other 6 studies, not all patients fulfilled these criteria or it was not mentioned in the article. Table 1 shows patient characteristics and percentages of work disability for the 22 cross-sectional surveys. Although 10 studies (3–12) were prospective cohort studies, percentage of work disability was examined only at 1 point in time during followup; we therefore included these studies in this category. If no exact data were available on disease duration at the end of the followup period, we estimated it by adding the duration of the followup to the known disease duration at baseline; the same procedure was used for age. The other 12 studies shown in Table 1 clearly were cross-sectional studies (13–24). In general, work disability rates were estimated for 1 of the 2 different populations, i.e., for patients who were working at disease onset or for the total RA population of working age. For each definition of population, studies are presented by increasing disease duration. Mean disease duration of all cross-sectional studies ranged from 6 months (13) to 15 years (3). All studies, except 1 (14), included both women and men. The percentage of women ranged from 41 (15) to 100 (14). Mean age ranged from 46 years (10) to 60 years (11). In 5 studies, the group of patients younger than 65 years was part of a larger study population (4, 7, 11, 16, 17), and exact data on age, disease duration, and sex of these patients were not available; characteristics of the total study population are shown as a substitute. Included in Table 1 is the most recent study on work disability by Young et al (11) instead of the 2000 study (25), in which the identical study population was described. Data on work disability had been assessed either by interview (3–12, 14, 16–20) or by means of self-reporting questionnaires (13, 15, 21–24).
Table 1. Patient characteristics and percentage of work disability with increasing disease duration among cross-sectional studies
First author, year (ref)
Age, mean, years
Disease duration, mean, years
Work disabled, %
Definition of work disability
Patient characteristics presented are values of larger study populations, because exact data on the group of patients of working age were not available.
Same study population (Sweden).
Same study population (US).
Patient characteristics presented are from 91 patients who were either working full time or were receiving a work disability pension.
Because no continuous data on disease duration was available, we obtained data from a previous study including this study population (24).
Only patients with a paid job before disease onset
Work disabled (regardless of receiving disability payments)
Definitions of work disability, whether entirely or partly due to RA, differed between studies. Several studies classified patients as work disabled based on work disability certificates (3, 5, 7, 9, 11, 12, 15, 16). In other studies, work disability was defined as “stopped working due to RA” (4, 6, 8, 10, 14, 18, 19, 20, 24) or as “(partly) work disabled due to RA” (13, 21–23).
Figure 1 presents the percentages of work disability in cross-sectional studies with increasing disease duration for 19 studies, with the exception of the study by Reisine et al (14), in which rates were determined for the RA population with a paid job at onset of the disease. Looking at the influence of disease duration on work disability across studies, there is a trend toward an increasing percentage of work disability with increasing disease duration, especially among studies from the US. However, it must be stressed that most data shown are based on average disease duration and that ranges of disease duration differed between studies. Work disability percentages observed in European countries were in general higher than those seen in the US, especially in early RA. Overall, the percentage of patients that was identified as work disabled among cross-sectional studies ranged from 13% after a mean disease duration of 6 months (13) to 67% after an average disease duration of 15 years (3).
Table 2 summarizes patient characteristics and percentages of work disability as observed in longitudinal studies. In all studies, interviews either by telephone or during a visit were carried out to obtain data on employment status. In the 5 studies, 3 different methods were used to estimate work disability rates: life table analysis, cumulative work disability rates over time, or work disability rates at different time points during followup. Both disease duration at onset of the study and followup duration since study start are mentioned. Studies are presented in ascending order of publication year. Duration of followup ranged from 1 year (26–28) to 30 years in studies using survival analysis (27). As shown in Table 2, definitions of work disability also varied among longitudinal studies.
Table 2. Patient characteristics and percentage of work disability in longitudinal studies
Percentages were obtained using the life table method (L), in which disease duration rather than followup time was used, or cumulative work disability over years (C).
The last 2 studies obtained work disability rates at different time points during followup. Because both studies included RA patients with disease duration <1 year, we decided to define followup time similar to disease duration (29, 30).
Because 2 of the 5 longitudinal studies (27, 29) used survival analyses to determine percentage of work disability, we present work disability with increasing disease duration instead of the followup time of these 2 studies (Figure 2). By using the life table method, Yelin et al (27) estimated that the prevalence of work disability would be 10% after 1 year of RA and 90% after a disease duration of 30 years. As was also evident among cross-sectional surveys, work disability rates obtained from the 3 European studies (26, 28, 30) were higher, especially during the first years of RA, than rates obtained in the US studies.
Assessment of variables associated with work disability due to RA using univariate analyses.
A number of sociodemographic variables, clinical variables, and work-related factors have been found to be associated with work disability or not working due to RA in both cross-sectional and longitudinal studies (Table 3). The following most-frequently mentioned variables were significantly associated with work disability.
Table 3. Assessment of variables associated with work disability due to rheumatoid arthritis using univariate analyses*
First author (ref)
Long disease duration
More functional disability
High joint count
More radiographic damage
High pain score
Low general well being
More disease severity
Physically demanding job
ESR = erythrocyte sedimentation rate; RF = rheumatoid factor; + = statistically significant different between patients with work disability due to rheumatoid arthritis compared with the working population; − = variable included in analysis; however, no statistically significant association was found; b = baseline values.
No. (%) of studies finding a significant association
Patients who were work disabled or who stopped working due to RA were generally older (3, 5, 6, 11, 15, 20–22, 30, 31), less educated (3, 15, 20, 21, 29–31), and had lower incomes prior to onset of RA (13, 18, 19). Marital status was a risk factor in some studies (18, 19, 30), although De Roos and Callahan (23) found married men less likely to become work disabled or nonworking. Work disability supposedly was dependent on the working status of the spouse. Race was not associated with work disability (18, 19, 21), but being female in 1 study was (31).
Work disabled patients had longer disease duration than currently employed RA patients (3, 15, 18, 19, 21, 31), higher erythrocyte sedimentation rates (ESR) (11, 20, 22, 29), higher joint counts (swollen joint count or tender joint count) (5, 11, 15, 20, 29), more radiographic damage (5, 11, 15, 19, 22, 26, 30), higher pain scores (11, 15, 21, 29, 30), less general well being (15, 29, 30), more frequently presence of rheumatoid factor (29), and more disease severity (e.g., Disease Activity Score) (11, 18, 29–31). In all studies in which functional disability (e.g., Health Assessment Questionnaire) was included in the analysis, more functional disability appeared to be associated with work disability (3, 6, 10, 11, 13, 15, 20–22, 26, 28–31). In addition, patients who underwent joint surgery or patients who received more disease-modifying antirheumatic drugs (DMARDs) or used a glucocorticoid, probably reflecting higher disease activity, were also more frequently work disabled (11, 18, 20).
Patients who stopped working due to RA more often had blue-collar jobs (6, 10, 11, 15, 21, 31) and more physically demanding jobs (3, 5, 21, 29, 30) compared with currently employed patients. However, Yelin et al (18) and Meenan et al (19) did not find significant differences regarding type of job, but they found that both greater self-employment and more determination to keep one's own pace at work were strongly associated with lower probability of work loss.
Assessment of predictors of work disability due to RA using multivariate regression analyses.
Table 4 shows the (baseline) demographic, clinical, and job-related predictors of work disability. Predictors most frequently found were older age (6, 8, 9–11, 14, 15, 23, 27, 30), less education (10, 20, 23, 29), longer disease duration (9, 20, 23), more functional disability (6, 7, 9–11, 14, 15, 20, 23, 26–29), higher ESR (11, 20, 29), higher joint count (swollen joint count or tender joint count) (8, 9, 27, 30), blue-collar jobs (6, 11, 15), and physically more-demanding jobs (7, 14, 27, 29, 30). Other predictors were more DMARDs or glucocorticoid use (30), high body mass index (29), more radiographic damage (9, 11), more deformed joints (8), less manual dexterity (8), and lower desire to remain working (8).
Table 4. Assessment of predictors of work disability due to rheumatoid arthritis using multivariate regression analyses*
First author (ref)
Long disease duration
More functional disability
High joint count
More radiographic damage
High pain score
Less general well being
More disease severity
Physically demanding job
ESR = erythrocyte sedimentation rate; RF = rheumatoid factor; + = significant predictors of work disability corrected for all other variables; − = included in analysis; however, no significant predictor; b = baseline values; m = mean values from baseline until last visit.
No. (%) of studies finding a significant association
Comparison of percentages of work disability among RA patients with country-specific percentage of the general population.
In a Dutch cross-sectional study by van Jaarsveld et al (22), the percentage of patients not working among both men and women aged 45–64 with RA was significantly higher compared with the same age group of the general Dutch population. Full or partial work disability for people younger than 65 years in the early RA group was 37% compared with 9% in the Dutch population. A second Dutch cross-sectional study (16) showed that 27% of all patients younger than 65 years were officially recognized as (partly) work disabled. The adjusted work disability percentage was 4 times higher for men and 15 times higher for women compared with the general population. These results are in contrast with another study also performed in the Netherlands, in which the labor force participation of patients with RA with a mean disease duration of 11 years was not significantly different from that of the general population after adjusting for age, sex, and education (24); 61% and 66%, respectively (not significant). It must be noted that in the latter study, labor force participation rather than work disability was assessed and compared with the general population.
Mau et al (9) observed that the percentage of permanent work disability for RA patients (37%) in their study was higher than they had expected from the incidence of 6 cases per 1,000 subjects receiving permanent work disability compensation in Germany per year. Furthermore, work disability was found to be 14% higher for women with RA and 11% higher for men with RA when compared with the German population (32).
As far as we know, there is only 1 case-control study in which the employment status of early RA patients has been repeatedly compared during followup with age- and sex-matched controls (28). In this UK study, RA patients were 32 times more likely to stop their job due to health reasons than their healthy controls.
US and Canada.
In addition to the studies discussed above, several population-based studies have been published in which work disability in RA patients was compared with that of the general population. The 1978 US social security survey of disability and work (33) demonstrated that the labor force participation among women with symmetric polyarthritis was 31% compared with 58% for the total US female population, and 42% versus 69% for men. Similar differences were found in the study by Yelin (34) in which labor force participation in the US among men with arthritis was ∼20% lower than among men without arthritis and 25% lower among women with arthritis compared with women without arthritis. For Canada, data from the 1990 Ontario Health Survey showed that 19% of men and 36% of women with arthritis were not in the labor force compared with 7% of men and 23% of women in the general population (35).
Despite differences in definition of work disability and of study populations, the percentages of work disability reported across all studies were quite high. Among the cross-sectional studies, the percentages of work disability in RA patients with a paid job before disease onset varied from 13% after a mean disease duration of 6 months (13) to 67% after a mean disease duration of 15 years (3). In longitudinal studies, work disability ranged from 10% after 1 year to 90% after 30 years of followup using life table methods (29). Percentages of work disability as presented in this overview increased steeply during the first years of disease and subsided somewhat later in the course of disease. Interestingly, the percentage work disability observed by Chorus et al (24) was relatively low after a mean disease duration >10 years when compared with the 3 other Dutch studies (16, 20, 22). However, the same author reported that the presented 33% of patients with a paid job at disease onset stopped working due to RA after a mean disease duration of 4.8 years (31), which is more in accordance with the work disability rates found among the 3 other Dutch studies. In a review of ankylosing spondylitis, it was found that work disability ranged from 3% after 18 years to 50% after 45 years of disease duration (36). These percentages are lower when compared with the observed percentages among patients with RA as presented in this overview, probably due to differences in age and sex distribution.
The discrepancies in the observed work disability rates due to RA across all studies might be attributed to methodologic problems and differences. Actual comparison of the results would require using 1 uniform definition of work disability in the future. In concordance with Allaire (37), we would recommend defining work disability as “work cessation (at least in part) caused by RA prior to the normal age of retirement.”
Exact estimates of subjects who become work disabled due to RA in cross-sectional studies can be influenced by recall bias, unless figures are obtained from governmental work disability databases. Because only baseline characteristics were reported in a number of studies, we had to add followup duration to baseline values. For longitudinal studies, it is recommended that disease duration rather than followup duration be depicted. The study population under survey can alter the number of patients who become work disabled (e.g., population of working age or not, paid job at onset or all patients irrespective employment history). For the population under survey, investigators should therefore clearly state disease duration (mean and range), mean and range of age, and number or proportion of women.
Notwithstanding methodologic differences, the average percentage work disability was already high (47%) during the first years of the disease as observed in 2 European studies (16, 20) and continued to rise with increasing disease duration. Even higher work disability rates due to RA might be expected in the future, because labor force participation among women seems to be increasing, and RA predominately affects women. Given that, in general, women tend to hold different jobs than men, the percentage of work disability might also differ between women and men. Only a few studies provided work disability rates for women and men separately (16, 20); in these studies work disability rates between women and men did not differ very much (∼2%). Using almost similar definitions of work disability, the percentage of work disability in a population of only women (14) was somewhat lower than that found in an RA population comprising both women and men (18). When ordering work disability rates by study years, no possible influence of treatment on work disability rates was observed. Because the methods differed between studies and social security policies have changed over the past years, the direct influence of treatment changes in these studies would be difficult to observe.
Figures 1 and 2 in this overview illustrate that work disability rates in general are higher in European countries than in the US. These differences may partly be explained by differences in social security systems. The primary criterion for approval of disability claims is very similar across countries; “incapability of earning” (Germany, the Netherlands), “incapability of work” (Austria, Finland, United Kingdom, and Sweden), and “incapability of pursuing any substantially gainful occupation” (Canada and US) (38). Nevertheless, the disability pension coverage differs greatly between countries and depends on the insurance program (universal or earnings-related social insurance), work history, previous contribution to the insurance, and range of earning loss percentages (e.g., 100% loss for disability in Canada or 15% loss for a minimum pension and a 80% loss for a maximum pension in the Netherlands) (38).
In general, disability payments in Germany, the Netherlands, and the Scandinavian countries are relatively easily accessible, whereas in Canada and the US, limited welfare facilities exist, which are in addition more difficult to enter. Benefits in the latter countries are not only lower, but also of more limited duration and are, in general, restricted to the neediest persons. Thus, one might expect that patients who become work disabled in 1 of the European countries are less likely to return to work because of the higher compensation for loss of earnings. However, social security benefits in western European countries have been decreasing since the end of the last millennium. So, the difference with the US may diminish.
Another issue specific for patients in the US is that patients with disabilities face a potential loss of employment-linked health insurance if they become disabled. Also, patients may run into difficulties finding new health insurance coverage with a new employer because of the preexisting conditions, or expensive insurance premiums due to increased risk. So, overall, it remains difficult to compare results from countries with totally different social security systems and describe the impact of the economy on work disability due to RA. Furthermore, the flexibility of the labor force market at time of the study might influence the number of patients that become work disabled.
To illustrate the consequences of RA in relation to work disability, it is advisable to compare work disability rates of the RA population with those of the general population, thus considering similar economic and legislative circumstances. All studies showed increased (risks of) work disability in patients with RA versus the general population. Because the distribution of age and sex in RA populations differs from the general population, it is recommended to adjust rates for age and sex. However, only 2 of the 4 presented cross-sectional studies provided adjusted percentages (16, 24).
Badley et al (35) concluded that the risk of being out of the labor force was slightly increased for men with chronic arthritis, but not for women. However, the odds ratio for not being in the labor force was significantly increased for both men and women with disability arthritis compared with persons without arthritis. The latter finding confirms the results seen in cross-sectional and longitudinal studies, i.e., that severe functional disability is a predictor of and is associated with work disability.
Identification of risk factors of work disability early in the course of the disease might enable rheumatologists or employment officers to influence them positively, thus postponing or even avoiding work disability. The most striking risk factors associated with or defined as predictors of work disability were more functional disability, physically demanding jobs, and older age. Because functional disability is associated with increased disease activity during the first years of the disease (39), reducing disease activity in early RA patients might lead to extended labor force participation. In addition, evaluation of the workplace and of job characteristics should start as soon as possible after RA diagnosis to see whether adaptations are needed. In contrast to older age, clinical variables and working conditions are considered as factors that might (partly) be influenced. For patients of older age, it is probably more difficult and less worthwhile, especially from the employer's point of view, to adapt their work environment or even to change to less physical jobs than for younger patients. However, due to aging of the population and changes in the currently applied social security systems, it might become interesting to also retain these older patients in the labor force.
We did not go into data on a possible association between depression or anxiety with work disability, but a significant association was found in a couple of studies (17, 29), but not in others (6, 10, 13). It is interesting to note that the observed predictors of work disability are almost the same as variables found to be associated with difficulties in returning to work among persons unemployed due to arthritis and musculoskeletal disorders (40). This implicates that these factors need special attention when evaluating the working conditions of RA patients. Therefore, in accordance with Sokka and Pincus (41), we conclude that, next to improved medical management of disease, reduction of work disability and reemployment are dependent on social, economic, and political issues.
Both cross-sectional and longitudinal studies reported substantial work disability rates among RA populations. Overall, a tendency toward higher work disability with increasing disease duration was observed, especially in the US studies. Because a number of methodologic differences and problems exists in studies on work disability due to RA, we suggest the following recommendations for future studies. We suggest defining work disability as “work cessation partly caused by RA prior to the normal age of retirement.” Furthermore, we recommend to 1) include only patients of working age and clearly characterize the RA population under survey (i.e., at least disease duration, age, and sex), 2) give separate work disability rates for women, men, and the total population, and 3) compare adjusted work disability rates with country-specific rates.
The authors wish to thank all participating rheumatologists and research nurses of the Utrecht Rheumatoid Arthritis Cohort Study Group (SRU) for their contribution to the investigation described in this article.