The relationships between adult juvenile idiopathic arthritis and employment




The chronicity of juvenile idiopathic arthritis (JIA) into adulthood and attendant potential disability may adversely influence educational attainment and the ability to secure and maintain gainful employment. We undertook this study to investigate the effects of patient- and disease-specific factors on education and employment outcomes in a group of adult patients with JIA.


We performed a cross-sectional study of 103 consecutive adults attending a JIA continuity clinic, and patients who consented completed questionnaires relating to educational achievement, employment status, and functional disability (the Health Assessment Questionnaire disability index [HAQ DI]), and, for those who were employed, the rheumatoid arthritis Work Instability Scale. We used the structural equation modeling technique to study key patient and disease variables for employment in adults with JIA.


The median age of patients was 24 years (range 17–71 years) with median disease duration of 19 years (range 7–67 years). Functional disability (the mean HAQ DI score) was significantly lower in patients who were employed (P = 0.03) and in those with oligoarticular JIA (t = 2.29, P = 0.02). Educational achievement was not influenced by JIA subtype (F = 1.18, P = 0.33). Educational achievement measured by General Certificate of Secondary Education (GCSE) grades had a positive effect on the type of job achieved in later life (F = 11.63, P = 0.001), with greater success leading to more professional or managerial posts. In the complex structural equation model, job stability was influenced positively by educational achievement measured by GCSE grades and negatively by the HAQ DI score (t = 10.94, P = 6.36 × 10−16).


Educational attainment is key to successful employability and is influenced by functional disability rather than by JIA subtype. These findings have implications for choice of occupation and delivery of career advice to aid decision making by young people with JIA.

Juvenile idiopathic arthritis (JIA) is a heterogeneous disease with several subtypes of unknown etiology, clinically and genetically distinct from chronic arthritis in adults (1), with many patients having ongoing active disease requiring long-term treatment into adulthood (2). Medical management has markedly changed in the last 2 decades with the early, often aggressive use of disease-modifying antirheumatic drugs such as methotrexate, the advent of biologic therapies (3), and multidisciplinary specialist care. It is hoped that increasingly effective treatments will improve long-term outcomes, although collaboration with specialist orthopedic teams is still likely to be required for those patients with severe disease.

For many adults with JIA in current clinical care, medical management predated the routine use of methotrexate or biologic agents early in their disease course; many have entered their adult years with considerable joint damage, functional disability, and impaired quality of life (1, 4). In addition, outcome studies confirm the deleterious influence of JIA on employability despite often excellent academic attainment (1, 5–7). It is of great concern that for many individuals, ongoing disability may, in spite of high academic success, adversely affect durability and participation in the workplace. Indeed, Gerhardt et al (8) have shown that the deleterious effect of JIA is not evident early in the course of disease during the transition from adolescence to adulthood in comparison to a control group of age-matched healthy peers. Beyond adolescence and into adulthood there is a consistent and significantly higher reported rate of unemployment in this patient group compared to an age-matched control population (6). The reasons for such observations remain poorly understood and, in particular, the serial dependency between employment and educational achievement. This higher rate of unemployment may be secondary to cohort effects (i.e., younger adults with JIA may have had fewer employment experiences than much older adults) or result from the fact that problems with employment are more evident as youth age into middle and later adulthood (but not during early, entry-level jobs).

Simple univariate analyses of risk factors for poor employment do not consider potential interactions and dependencies between educational levels, disability, and employment. Analyses should also consider the educational level achieved (which will determine employability over and above any disability). The influence of disease type, magnitude of disability, and educational achievement on employability remains ill defined. It is also not clear how individuals with JIA perform at work in term of barriers at work (work instability), type of employment, and duration of employment.

We postulate that employment outcomes in adults with JIA depend not only on their type and level of disability but also on their educational achievements prior to becoming employed, with the former directly affecting their ability to work and indirectly affecting the latter. Thus, educational achievement at age 16 years will depend on JIA type and level of disability, and both in turn will affect educational achievement at age 18 years, with educational qualifications gained at age 18 years influencing the type of employment that can be undertaken (Figure 1). Thus, we posit the need for models that include indirect as well as direct effects. Hence, we used a 2-staged modeling approach. First, we used generalized linear modeling to investigate the effects of JIA and disability on education and subsequent employment. Second, we investigated both direct and indirect dependency between JIA, disability, and education and employment outcomes such as work instability, type of employment, and duration of employment. Our analyses were based on data from a prospective observational cohort study in a tertiary specialist clinic for which measures of known risk factors were also available. Our objective was to identify the role of different factors in determining employability for this patient group with a view to directing strategies to improve ability to work.

Figure 1.

Hypothetical pathway model of the interactions between juvenile idiopathic arthritis (JIA) subtype and other patient characteristics and their effects on educational and employment outcome. GCSE = General Certificate of Secondary Education; HAQ = Health Assessment Questionnaire; A-level = Advanced Level (examination); RA WIS = rheumatoid arthritis Work Instability Scale. Color figure can be viewed in the online issue, which is available at



A clinic for adults with JIA was set up in 1995 within the Musculoskeletal Unit, a regional service at the Freeman Hospital, Newcastle Hospitals NHS Foundation Trust, Newcastle-upon-Tyne, UK. The principal goal of the clinic was to provide continuity of care for young adult patients with JIA transferring from the children's arthritis service (patients age ≤16 years) within the same NHS Trust as well as for all adult patients with JIA, many of whom were hitherto under the care of adult rheumatologists or orthopedic surgeons within the same unit. Patient care was managed by 1 consultant (HEF), and all patients were classified using the International League of Associations for Rheumatology criteria for JIA, the term for the disease which has replaced “juvenile rheumatoid arthritis,” “juvenile chronic arthritis,” and others (2).

A total of 103 patients participated in the study (22 males and 81 females). Their mean age was 29.2 years (median 24 years, range 17–71 years), with a mean disease duration of 20.5 years (median 19 years, range 7–67 years). The JIA subtypes included oligoarticular (n = 40), polyarticular rheumatoid factor (RF) positive (n = 23), polyarticular RF negative (n = 17), systemic (n = 10), and other (n = 13), including psoriatic arthritis and enthesitis-related arthritis.


A cross-sectional study was performed to allow for collation of information on current functional status, history of employment, and educational achievement. This study was approved by the hospital research and development department, and ethical approval was waived since the study was considered an audit-of-service evaluation. All patients attending the clinic were invited independently to complete a series of validated questionnaires. These included questions on educational achievement and employment status. Also included were a validated Health Assessment Questionnaire (HAQ) and, for those who were employed, the rheumatoid arthritis Work Instability Scale (RA WIS) (9–11). No patient refused to participate in the study. All data were pseudoanonymized and entered into a hospital database in accordance with Caldicott confidentiality guidelines.

The Stanford HAQ 20-item disability index (DI) is a self-reported disability measure that has been validated and widely used for clinical trials in adult RA and other rheumatic diseases (12, 13). The construct validity, predictive validity, and sensitivity to change of the HAQ DI have also been established in numerous observational studies and clinical trials, although not in JIA. The HAQ DI (score of 0–3, with 3 being the worst functional ability) assesses functional disability and is predictive of long-term patient outcome and survival of joint replacement in patients with adult-onset RA, with evidence of its predictive value also reported in a cohort of adults with JIA requiring knee replacement (14). The RA WIS is a 23-item, simple, validated tool for screening work instability (the consequences of a mismatch between an individual's functional ability and job demands) (9). This self-administered measure enables the monitoring of work disability in RA patients. The RA WIS is scored in 3 bands indicating low (score <10), medium (score 10–17), and high (score >17) risk of work disability. Employment was categorized independently as job instability (as measured by the RA WIS score) or duration of employment and separately as the type of employment (Standard Occupational Classification 2000 [Volume 1] for the Office for National Statistics) (15).

Modeling analyses.

We used generalized linear modeling to investigate the differences in education and employment outcomes among patients with different subtypes of JIA. We used a range of patient-derived data to create covariates for analysis. HAQ data were used to create a score for disability. Educational achievement was recorded from grades in the General Certificate of Secondary Education (GCSE), an academic qualification awarded as the result of a UK public examination taken at age 16 years. We used the number of GCSE A–C grades achieved as the metric of success. Methotrexate was introduced as a routine treatment for JIA in 1994–1995 in the adult JIA clinic, so we created a covariate to compare educational and employment achievement for those who had or had not received the drug prior to completing their school education. However, the effects of this variable were confounded with those of age and, more specifically, the effects of “grade inflation” recorded in GCSEs in wider society (16); therefore, this variable was removed from subsequent analysis.

We used success on another public examination (the Advanced Level [A-level] examination) as a measure of achievement at age 18 years. There were too few cases to explore relationships with higher education. Employment outcomes were captured in 3 variables; jobs and careers were classified according to the categories of the Standard Occupational Classification 2000 (Volume 1) for the Office for National Statistics (15), an occupation-based classification that categorizes employment into 1 of 3 classifications. We adopted a classification with 9 categories ranging from class 1 (managers and senior officials) to class 9 (manual laborers), and we added “unemployed” as a final category. There were no patients in class 5. We used duration of employment and the RA WIS score as indicators of employment instability. Models were adjusted for age and sex as appropriate. We then analyzed indirect and direct effects of JIA, disability, and educational achievement on employment outcome using structural equation modeling, a technique used extensively in social science research (17).

Hypotheses tested.

This work was predicated on 3 principal hypotheses: first, JIA subtype influences HAQ DI score; second, JIA subtype influences educational achievement either at age 16 years (through the GCSE grade) or at age 18 years (through the A-level examination results); third, employment outcomes depend on the indirect effect of JIA subtype as mediated by the effects of subtypes on educational achievement at both ages.

Structural equation modeling.

If JIA were to result in disability that affects educational achievement, it is likely that there would be ongoing consequences for employment. Therefore, JIA might have both direct and indirect effects on employment. Simple linear regression modeling approaches do not account for complex interactions of the type posited in our third hypothesis. Therefore, we have used structural equation modeling to investigate the relative effects of different patient and disease types on educational achievement and their ultimate combined effects on employment.

Our hypothetical model (Figure 1) assumed that the different subtypes of JIA had differential effects on disability, which would affect educational achievement and ultimate employment. Since we know that the type of employment is likely to depend on age (older individuals are likely to have higher job status than young individuals who have just left school) and that educational achievement depends on age, we included age as an additional covariate. Structural equation modeling is usually based on analysis of a covariance matrix, expressing how each of the drivers and outcomes covary in the system of interest. Since our data were a mixture of categorical variables, we used polychoric correlations to create a correlation matrix for analysis in the structural equation modeling. Polychoric correlations allow estimates of correlation between ordinal and (when extended to tetrachoric correlations) categorical variables. These correlations are rarely bivariate normal, so this means that standard errors of parameter estimates derived from a structural equation modeling analysis are likely to be biased. We therefore used a bootstrapping approach to estimate means and standard errors for the structural equation modeling parameter estimates (18). Bootstrapping was undertaken by running the model 100 times with 100 separate sets of data comprising random samples of 70% of the total data set. The rationale was that analyzing the means and standard errors of the 100 random subsets of the data would provide unbiased structural equation modeling parameter estimates.

Our modeling procedure was first to fit a full model with all hypothesized pathways and then to remove all nonsignificant paths to create the simplest model containing only significant pathways. This was determined by assessing the root mean square error (RMSE) of association statistics and the standard errors of the parameter estimates from the bootstrapped 95% confidence limits for each coefficient. All modeling was undertaken using libraries in the R statistical package (19). We had 3 separate employment outcomes. These were the RA WIS score, which is a measure of employment stability; the job classification status achieved; and the total time of employment up to the census point of data collection. The full model is shown in Figure 1.


Of the 103 patients, 66 (64.1%) were employed either full time or part time; the remainder were unemployed, attending school, or homemakers. The HAQ DI score of those who were employed (mean ± SD 0.83 ± 0.85) was significantly lower (i.e., indicating less functional impairment) than the HAQ DI score of those who were not (mean ± SD 1.25 ± 0.92) (P = 0.028).

Influence of disease subtype on HAQ DI score.

The HAQ DI scores were not related to the JIA subtypes collated together (F = 1.18, P = 0.33). Those with polyarticular JIA (both RF-positive and RF-negative patients) had higher HAQ DI scores (i.e., greater functional disability) than those with oligoarticular JIA (t = 2.29, P = 0.02). In a separate analysis of oligoarticular and polyarticular JIA (RF positive and RF negative), patients with either polyarticular type had higher disability scores than patients with the corresponding oligoarticular type (for RF-positive patients, t = 1.81, P = 0.07; for RF-negative patients, t = 2.31, P = 0.02).

Influence of disease subtype on educational achievement.

The number of GCSE A–C grades attained did not differ significantly among JIA subtypes (t = −0.37, P = 0.71), suggesting that type of disease did not affect educational achievement measured by GCSE grades. A similar finding was observed for achievement on A-level examinations (t = −1.67, P = 0.10). Achievement on A-level examinations was strongly correlated with achievement measured by GCSE grades (t = 7.28, P = 2.30 × 10−10), and older patients had lower grades than younger patients (t = −2.84, P = 0.005). Educational achievement measured by GCSE grades was found to influence the class of job achieved in later life (F = 11.63, P = 0.001) (Figure 2). This was equally the case for achievement on A-level examinations (F = 13.171, P = 0.0006), although educational achievement measured by GCSE grades and A-level examinations did not influence job stability (t = −0.37, P = 0.71 and t = −1.67, P = 0.10, respectively). There was a relationship between patient age and number of GCSE grades achieved (t = −2.84, P = 0.005), with older patients having fewer grades above C. This was believed to most likely reflect the annual 1.1% increase in the percentage of candidates obtaining A–C grades in the general UK population since 1988 (42% in 1988 compared to 69% in 2011) (16) and indicates that assessment has to be adjusted for age in the present context.

Figure 2.

Numbers of General Certificate of Secondary Education (GCSE) A–C grades held by individuals in each of 10 Standard Occupational Classification 2000 (SOC2000) employment classes (1 = managers and senior officials; 2 = professional occupations; 3 = associate professional and technical occupations; 4 = administrative and secretarial occupations; 5 = skilled trades occupations; 6 = personal service occupations; 7 = sales and customer service occupations; 8 = process, plant, and machine operatives; 9 = elementary occupation [manual laborer]). We added a tenth class to include “unemployed.” There were no patients in class 5. Data are presented as box plots, where the boxes represent the 25th to 75th percentiles, the dots within the boxes represent the median, and the whiskers outside the boxes represent the 10th and 90th percentiles.

Influence of disability on employment.

Employment instability as measured by the RA WIS score was strongly dependent on the HAQ DI score (t = 10.94, P = 6.36 × 10−16). While the effects of JIA subtype on educational achievement and subsequent employment are complex, the results of the structural equation modeling analyses indicated that many of the variables interact and have an effect on employment for the individual, emphasizing the need for an analytic approach that encompasses indirect and direct pathways to outcomes. Length of employment was not related to JIA subtypes or to any other covariate, with the exception of age as an expected finding. The pattern of employment types across the JIA subtypes was highly variable (Figure 3), with a spectrum of job classes although with too few cases for further analysis.

Figure 3.

Distribution of Standard Occupational Classification 2000 (SOC2000) employment classes (see Figure 2) obtained by patients with different types of juvenile idiopathic arthritis. Values are the number of patients in each job type. RF+ve = rheumatoid factor positive; RF−ve = RF negative. Color figure can be viewed in the online issue, which is available at

The simplest structural equation modeling pathways investigating the effects of the covariates on the job stability score and the job type achieved are shown in Figures 4 and 5. Job stability (Table 1) appeared to be driven by the GCSE grades and the HAQ DI score, which in turn were affected by the age of the patient. The mean ± SD RMSE of association score for this model for the 100 runs of the model was 0.03 ± 0.05, indicating almost perfect fit. The HAQ DI score had a 7-fold greater effect on job stability (0.85 SD change per unit SD change in HAQ DI score) than did educational achievement measured by GCSE grades (0.11 SD change per unit SD change in number of A–C grades obtained).

Figure 4.

Simplified model of significant pathways of effects of disease and other patient covariates on the job stability of adults with JIA. Values within arrows represent the effects of unit change in SD of the individual variable in the text node preceding the arrow on the outcome at the end of the arrow. Values in the text nodes represent residual variation. Model estimates are derived from 100 runs of 70% random samples from the full data set. The mean ± SD root mean square error of association score for this model for the 100 runs of the model was 0.03 ± 0.05. See Figure 1 for definitions. Color figure can be viewed in the online issue, which is available at

Figure 5.

Simplified model of significant pathways of effects of disease and other patient covariates on the type of employment undertaken by adults with JIA. Values within arrows represent the effects of unit change in SD of the individual variable in the text node preceding the arrow on the outcome at the end of the arrow. Values in the text nodes represent residual variation. Model estimates are derived from 100 runs of 70% random samples from the full data set. The mean ± SD root mean square error of association score for this model for the 100 runs of the model was 0.08 ± 0.05. See Figure 1 for definitions. Color figure can be viewed in the online issue, which is available at

Table 1. Parameter estimates and 95% CIs for final structural equation modeling of the path models relating employment outcomes to disease severity and other patient characteristics*
FromToMean coefficient (95% CI)
  • *

    Outputs were derived from 100 bootstrapped model runs with 70% random samples of the patient data set for each model run. “From” and “to” represent the interaction between various correlates in influencing stability and type of employment. Note that coefficients for covariances (self-self) are effectively residual variances in each variable not explained by the model. 95% CIs = 95% confidence intervals; GCSE = General Certificate of Secondary Education; RA WIS = rheumatoid arthritis Work Instability Scale; HAQ DI = Health Assessment Questionnaire disability index; JIA = juvenile idiopathic arthritis; A-level = Advanced Level (examination).

Stability of employment  
 GCSE gradesJob instability (RA WIS)0.115 (0.022, 0.185)
 GCSE gradesAge−0.229 (−0.338, −0.131)
 Job instability (RA WIS)HAQ DI score0.851 (0.809, 0.899)
 HAQ DI scoreAge0.436 (0.272, 0.566)
 GCSE gradesGCSE grades0.943 (0.885, 0.982)
 Job instability (RA WIS)Job instability (RA WIS)0.291 (0.219, 0.345)
 HAQ DI scoreHAQ DI score0.802 (0.659, 0.921)
Type of employment  
 JIA typeGCSE grades−0.191 (−0.375, −0.021)
 GCSE gradesA-level scores0.735 (0.683, 0.788)
 GCSE gradesEmployment class−0.411 (−0.542, −0.283)
 GCSE gradesAge−0.389 (−0.518, −0.261)
 Employment classHAQ DI score0.285 (0.160, 0.432)
 Employment classAge−0.222 (−0.395, −0.067)
 HAQ DI scoreAge0.398 (0.257, 0.511)
 JIA typeJIA type1.000
 GCSE gradesGCSE grades0.831 (0.736, 0.924)
 A-level scoresA-level scores0.457 (0.374, 0.531)
 Employment classEmployment class0.745 (0.617, 0.843)
 HAQ DI scoreHAQ DI score0.836 (0.735, 0.925)

The structural equation modeling analysis for the job type achieved in employment showed a more complex relationship with the JIA subtype and other covariates, with education, age, and disability affecting the type of job achieved (Table 1). The mean ± SD RMSE of association score for this model for the 100 runs of the model was 0.08 ± 0.05, suggesting a reasonable fit of model to data. In this model the effect of JIA subtypes was indirect and through an effect on educational achievement, a relationship that was not apparent in the univariate modeling analysis. The HAQ DI score contributed to the job type achieved both through its indirect effect on age as well as through a direct effect on job type, giving a combined effect of 0.37 change in SD of employment type per unit change in SD of HAQ DI score (i.e., [−0.19 for JIA type × −0.43 for GCSE grades] + 0.29 for HAQ DI score). The contribution of JIA type to employment type was comparatively small at 0.08 SD change in job type per unit SD change in JIA type (i.e., −0.19 × −0.43). Both structural equation modeling analyses clearly emphasize the role of disability rather than the type of disease in determining employment stability and job type.


In this cohort of adults with JIA, educational achievement measured by GCSE grades and A-level examination were found to be independent of disease subtype. This observation is similar to an earlier reported finding in this cohort in 2003 (1), that educational achievement in adults with JIA was comparable, if not better, than that of local controls irrespective of the level of functional disability. There is also general consensus that educational achievement in patients with JIA is similar to that in the general population (1, 5, 7, 20–23). The effects of JIA on subsequent employment are less clear-cut, and previous work has found conflicting information on the employment rates (1, 4–8, 20–23). The unemployment rate in the adult JIA patients in the present study was 3-fold higher than that of the local controls (1), and higher unemployment rates have been noted among adults with JIA in relation to local averages (1, 20, 21) and comparison groups (5, 7). Our current study goes further in suggesting that there are indirect effects of disability on employment type and stability that are more subtle and mediated by education.

We have shown in this study that oligoarticular JIA was associated with a lower HAQ DI score as compared with polyarticular disease. Previous work has shown that the need for orthopedic intervention in oligoarticular disease is lower than in “nonoligoarticular” disease (24), and that systemic- and polyarticular-onset JIA subtypes yield reduced functional outcome and have prolonged duration (5, 25–28). While our results indicate that JIA subtype had an effect on levels of disability (the HAQ DI score), they also suggest that level of disability was a more important determinant of job stability (RA WIS) and employment type than JIA type per se. This may simply reflect the fact that the HAQ DI score is a better descriptor of employment capability than is the JIA classification, since it is recognized that JIA is a heterogeneous disease. The increased complexity may also indicate that job type may be a more subtle indicator of patients' abilities within jobs, relative to their subtype of JIA.

Clearly, there are other unmeasured factors that affect job stability in the workplace, and this necessitates more detailed investigation (1, 5, 7, 21). There is evidence to suggest that a key influence on gainful employment is a successful transition from adolescence to adulthood (8), and this is an ultimate aim of transitional care programs (29). We accept that interpretation of these disparate employment rates in isolation is difficult, and we have not included the influence of employers' opinions on the employability of such individuals and the effects of psychosocial factors, including family and patient expectations and aspirations. However, to our knowledge, the current study is the first exploration of maximal job achievement and sustainability in the workplace in the context of functional disability, JIA subtype, and other patient factors.

The current study has extended our understanding of the complexity of job stability in this patient group. Previous work has confirmed the higher rate of unemployment without evaluating the barriers faced at work, and subjective questioning found that the majority of patients without work attributed their unemployment to the disabling effects of their disease (21). In a review of 242 patients with chronic illness and disability in the adolescent age group (72% of whom had JIA), White and Shear found that specific prevocational assistance improved the employment rate (30).

We accept that our study is biased toward more severe JIA subtypes followed up in a tertiary unit, and that the patients were not an incidence cohort. However, all patients are being followed up regularly, including those with inactive disease, who are reviewed annually. The unemployment rate noted in our patient group may not reflect the true picture, and therefore we have not attempted to compare it with unemployment rates of control groups or historic cohorts. Moreover, the distribution of JIA subtypes is not typical of that observed in a regional pediatric rheumatology service, since we have a relatively low proportion of patients with oligoarticular-onset JIA and a higher proportion of patients with more severe subtypes, namely, systemic-onset and RF-positive polyarticular JIA; however, this distribution of subtypes is likely representative of adult JIA patients in current adult rheumatology practice. In addition, we acknowledge that the measures used, namely, the HAQ DI and the RA WIS, have been validated for the adult population of RA patients and not specifically for adults with JIA. Nonetheless, we have explored the direct and indirect pathways between potentially important drivers of disability and education and their effects on the stability of employment outcomes and the job classification status achieved.

This work has demonstrated that educational achievement at ages 16 and 18 years predicts job type for patients with JIA in later life. Furthermore, the degree of disability has a direct bearing on the employment outcome. This information is important and useful to professionals, both in health care and in educational environments, who provide career advice to young people with JIA. Our work emphasizes the importance of informed career counseling to young people through schools and colleges; furthermore, such dialogue needs to include health care providers, as JIA is complex and the functional outcomes are variable. We acknowledge that our study includes older patients, many of whom were not exposed to treatments that are now routinely used. Increasingly, with modern therapies, functional outcomes are expected to improve, and it is important that career advisors and, indeed, prospective employers are aware that for many young people with JIA, the functional outcomes are often excellent. For those patients who do have functional limitations, employability may be optimized by often relatively simple interventions, such as flexible scheduling or ergonomic adaptation at the work station. We acknowledge that the outcomes in this study were primarily patient-reported; further work is needed to explore the experiences of patients who reported difficulty in sustaining employment as well as the views of employers and, indeed, career counselors.

In summary, we have found that the HAQ DI score was significantly lower in patients who were employed and in those with oligoarticular JIA. Educational achievement was not influenced by JIA subtype. Educational achievement measured by GCSE grades had an effect on the class of job achieved in later life, with posts in managerial and professional occupations associated with better GCSE A–C grades. Job stability as measured by the RA WIS score appeared to be driven both by the level of educational achievement measured by GCSE grades and by the HAQ DI score, which themselves were affected by the age of the patient. Length of employment was not related to JIA subtypes or to any other covariate with the exception of age. We have measured the effect of the disease on various employment outcomes. This has implications for the choice of occupation and forms the basis for further research to guide career decisions for these patients.


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. Mr. Malviya 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. Malviya, Foster, Ferris, Muthumayandi, Deehan.

Acquisition of data. Foster, Ferris, Hanson, Muthumayandi, Deehan.

Analysis and interpretation of data. Malviya, Rushton, Foster, Ferris, Muthumayandi, Deehan.