Juvenile Idiopathic Arthritis, Earnings, and Work Loss: A Nationwide Matched Cohort Study

Abstract

Objective

This study compares trajectories of earnings and work loss in individuals with juvenile idiopathic arthritis (JIA) versus matched comparators from the general population.

Methods

Patients with JIA (n = 4,737) were identified in the Swedish National Patient Register (2001–2017) and individually matched to up to five general population comparators on birth year, sex, and residence county (n = 23,645). Earnings and work loss data were retrieved from nationwide registers from age 18 years. Differences between patients with JIA and general population comparators were estimated using linear regression adjusted for sex, age, age at identification, and calendar year as well as parental education, work loss, and earnings.

Results

During a median of 11 years’ follow‐up, patients with JIA had 5.5% lower earnings than matched comparators (mean annual difference −€736; 95% confidence interval [CI] −€1,026 to €445). The difference in earnings was larger before than after age 26 years. Beyond age 26 years, the difference in earnings was less than 4%. Patients with JIA had more work loss than matched comparators throughout follow‐up (mean difference 11; 95% CI 8–13 days/year). This difference was consistent throughout follow‐up, but significant effect modification with calendar period of entry (<2005 vs ≥2005) was found, with later entry associated with lower work loss.

Conclusion

Patients with JIA had lower mean annual earnings and higher work loss than matched general population comparators, but earnings differences diminished in magnitude with age and work loss diminished with calendar period of identification. In JIA, a minority of patients accounted for the majority of the negative impact on economic outcomes, which persists into adulthood.

INTRODUCTION

Juvenile idiopathic arthritis (JIA) is the most common childhood rheumatic disease. ¹ Treat‐to‐target strategies and new drug therapies have revolutionized the management of JIA over the last decades, leading to improved clinical outcomes. ¹ , ² , ³ It has been speculated that this will improve long‐term outcomes for patients. The broader effects of these advances on patients’ outcomes later in life, for example earnings and work loss, remain underexplored (Supplementary Table 1).

SIGNIFICANCE & INNOVATIONS.

• To the best of our knowledge, there are no previous population‐based studies on work‐related earnings or work loss in patients with juvenile idiopathic arthritis (JIA).

• In this nationwide study of almost 5,000 patients with JIA, patients had 5.5% lower earnings and almost twice as high work loss compared with the general population.

• Among patients, 75% had no work loss longer than 14 days versus 85% in matched comparators, indicating that the difference is concentrated to a limited group of patients with JIA who experience substantial work loss.

Some previous studies have demonstrated that patients with JIA are more likely to be unemployed than community comparators, but these studies are older and do not reflect current treatment. ⁴ , ⁵ , ⁶ In studying the economic impact of JIA, it is important to differentiate between earnings (income generated through employment reflecting productivity at work) and work loss (periods of sick leave and disability pension because of impaired health).

The aim of this study was to examine the development of objective measures of earnings and work loss of patients with JIA compared with general population comparators, overall, by sex, and by calendar period of identification. JIA is made up of a diverse range of subtypes. Systemic onset JIA (sJIA) is the most distinct because of its clinical features, treatment, ⁷ and increased morbidity and mortality. ⁸ We therefore performed subgroup analyses by sJIA versus other types.

PATIENTS AND METHODS

This population‐based matched cohort study was performed using a linkage connecting nationwide Swedish registers using the unique personal identification number of each Swedish resident. Ethical approval was granted by the Swedish Ethical Review Authority (Regional Ethics Committee, Stockholm, Sweden. 2023‐03804‐01).

Patient identification

We identified patients with JIA as having at least two diagnoses (primary or supplementary) in the National Patient Register (inpatient/outpatient care) from 2001 to 2017 at age <16 years (International Classification of Diseases, Tenth Revision [ICD‐10]: M05, M06, M07, M08, and M09). We required that patients were born in 2001 or earlier (to ensure that they could reach age 18 years before the end of follow‐up) and lived until age 18 years (start of follow‐up; Supplementary Figure 1). Among patients with JIA, patients with sJIA were identified as having at least one registration listing of ICD‐10 code M08.2. The reasons for investigating this subgroup were that they have historically had more severe disease, and we have previously reported worse outcomes regarding marriage and childbearing. ⁹

The identification date for patients may not coincide with their first diagnosis date because the outpatient component of the National Patient Register began in 2001, resulting in not only incident but also some prevalent patients being identified in the early years (Supplementary Figure 2). Because patients could have been up to age 16 years at that time, their first diagnoses might have occurred before the outpatient data became available in 2001.

Matched comparators

For each patient, up to five comparators from the general population were identified in the Total Population Register ¹⁰ matched on birth year, sex, and residence county. Comparators were excluded if they had a JIA diagnosis before inclusion or died before age 18 years.

Outcomes

Taxable earnings

Data on taxable earnings and work loss (sum of sick leave and disability pension days) were retrieved from 2003 to 2019 from the Swedish Longitudinal Integration Database for Health Insurance and Labor Market Studies (LISA), ¹¹ covering the Swedish population from age 16 years. The earnings data originate from employer reports to the Swedish Tax Agency listing the annual total taxable earned gross income of employees. The earnings data were adjusted for inflation using the Swedish consumer price index and converted to 2019 euros using the exchange rate from July 1, 2019 (€1 = 10.54 Swedish kronor). Changes in taxable earnings can reflect several underlying changes in work ability, such as job changes within or among companies, shifts from permanent to temporary positions, shifts from full‐time to part‐time work, or, more generally, changes in earnings because of changes in the wage rate and/or hours worked.

Work loss

The Swedish welfare system provides complete or partial compensation for sick leave and disability pension. Sick leave is reimbursed by the employer from day 2 to day 14, and episodes >14 days are recorded by the Social Insurance Agency, which reimburses the employee for any lost income from day 15 onwards. An individual with a ≥25% reduced work ability (evaluated by a physician) expected to last at least 1 year may receive a disability pension. Our measure of work loss is the sum of sick leave and disability pension registered in Statistics Sweden's LISA database. Combining these two benefits gives a unified measure of work loss comparable over time and insensitive to institutional changes that may move individuals among benefit systems.

Parental data

Parents of patients with JIA and comparators were identified through the Multigeneration Register (biologic parents). Parents’ educational level for the year 2000 (before the start of follow‐up) was retrieved from the LISA Register ¹¹ at Statistics Sweden and categorized into ≤9, 10 to 12, and >12 years. The data on the education of each parent were merged into one variable recording the highest education among the two parents. The total earnings and work loss of parents (summing across each parent to obtain one measure for both parents) for the year 2000 was classified into two variables with five categories (missing and first to fourth quartile for parental earnings: missing, 0 days, 1–25 days, 26–100 days, and >100 days for parental work loss).

Statistical analysis

Results are presented as annual mean taxable earnings and work loss for patients with JIA and comparators by age and adjusted mean annual differences between patients with JIA and matched comparators with 95% confidence intervals (CIs) using linear regression and adjusting for sex, age, age squared, age at identification, year of observation, parental education (highest), total parental work loss, and total parental earnings. We also pooled observations across time and estimated the adjusted mean annual difference between groups from age 18 years and onwards. SEs were clustered at the individual level in all analyses using several observations for a patient with JIA or a matched comparator. In sensitivity analysis, we winsorized earnings at the first and 99th percentiles to reduce the impact of extreme values.

Outcome distribution

To examine the distribution of earnings by age, we compared the earnings of patients with JIA and comparators at the 25th, 50th, and 75th percentiles of the distribution. We also examined the fraction of employed patients and comparators (earnings >0) and the earnings and work loss of patients and comparators conditional on being employed. To examine the distribution of work loss by age, we divided patients with JIA and comparators into four categories: annual work loss of (1) 0 months, (2) >0 and ≤6 months, (3) >6 and <12 months, and (4) 12 months and plotted the distribution of individuals across categories.

Subgroup analyses

We conducted subgroup analyses by sex (women vs men), diagnosis subtype (sJIA vs no sJIA), and two periods of identification (identified 2001–2004 vs identified 2005–2017). We tested for subgroup differences with an interaction term between exposure and subgroup membership in the linear regression. In further exploratory analysis, we conducted subgroup analysis by four quartile periods of identification (2001, 2002–2004, 2005–2008, or 2009–2017).

Data were analyzed using SAS (version 9.4; SAS Institute) and Stata (version 13.1). All tests were two‐sided and P values <0.05 were considered statistically significant.

RESULTS

Participant characteristics

In total, 4,737 patients with JIA were identified in the National Patient Register from 2001 to 2017, together with 23,645 comparators from the general population (Table 1; Supplementary Figures 1 and 2). The median follow‐up time for patients with JIA and comparators was 11 years. Among the patients with JIA and their matched comparators, 63% were women, and the mean ± SD age at register‐based identification was 10.8 ± 3.7 years. Relative to parents of the comparators, parents of patients with JIA had similar educational attainment and earnings (€43,400 vs €41,900) but more work loss (52 vs 39 days; mean standardized difference 12.7%).

Table 1.

Characteristics of patients with JIA and matched general population comparators^*

Characteristics	JIA (n = 4,737)	Comparators a (n = 23,645)	Mean standardized difference, %
Women, n (%)	3,004 (63.4)	14,998 (63.4)	0.0
Age at identification, b mean (SD), y	10.8 (3.7)	10.8 (3.7)	−0.3
Age 0–5 y, n (%)	617 (13.0)	3,101 (13.1)	−0.3
Age 6–10 y, n (%)	1,502 (31.7)	7,471 (31.6)	0.2
Age 11–16 y, n (%)	2,618 (55.3)	13,073 (55.3)	0.0
Year of identification, mean (SD)	2005 (4)	2005 (4)	0.0
Year of identification, median (IQR)	2004 (2001–2008)	2004 (2001–2008)	–
sJIA, n (%)	434 (9.2)	–	NA
Parental education level, n (%)
Primary school	190 (4.0)	1,380 (5.8)	−8.0
High school	2,463 (52.0)	11,583 (49.0)	6.0
University	1,955 (41.3)	9,474 (40.1)	2.5
Education missing	129 (2.7)	1,208 (5.1)	−11.3
Parental workplace variables, mean (SD)
Annual work loss, days	52 (114)	39 (102)	12.7
Annual earnings	43,400 (48,300)	41,900 (38,700)	3.7

^*IQR, interquartile range; JIA, juvenile idiopathic arthritis; NA, not applicable; sJIA, systemic onset juvenile idiopathic arthritis.

^a Matched by birthyear, sex, and residence county.

^b The identification date does not necessarily match the first diagnosis date in all patients because the outpatient care component of the National Patient Register from which patients were identified was established in 2001 and patients were allowed to be younger than age 16 years at that time point and may, therefore, have had their initial diagnosis earlier.

Earnings and employment

Throughout the follow‐up period from age 18 years onward, patients with JIA had 5.5% lower earnings than matched general population comparators (mean difference −€736 [95% CI −€1,026 to −€445]) (Figure 1; Tables 2 and 3). After winsorization, the corresponding reduction in earnings was 5.6% (Supplementary Table 2). The most pronounced difference in earnings between patients and comparators occurred before age 26 years. Beyond age 26 years, patients with JIA and comparators had a <4% difference in annual earnings. At age 30 years, patients with JIA had 3.9% lower annual earnings (mean difference −€977 [95% CI −€2,255 to €300]). The lower earnings for patients with JIA were consistent across different earning percentiles (Supplementary Figure 3), but the difference was numerically larger at the lower end of the distribution. Patients with JIA had lower employment levels than matched comparators, overall and in both sexes, from age 18 years and onward (overall mean difference −2.8% [95% CI −3.6% to −1.9%]; Supplementary Figures 4 and 5). Beyond age 26 years, differences in employment levels were smaller. Conditional on being employed, patients with JIA had 2.5% lower earnings than comparators (mean difference −€388 [95% CI −€657 to −€119]) (Supplementary Table 3).

Figure 1.

Earnings and work loss for patients with juvenile idiopathic arthritis (left) and matched general population comparators (right).

Table 2.

Estimated association between JIA and earnings observations for those aged 18 years and older^*

Subgroup	Estimate of change in earnings (95% CI) a	P value	Percentage change b	P for interaction
Overall	−736 (−1,026 to −445)	<0.001	−5.5	NA
Men	−709 (−1,306 to −112)	0.020	−4.7	0.90
Women	−744 (−1,051 to −437)	<0.001	−6.0
Systemic onset JIA	−690 (−1,709 to 328)	0.18	−5.1	0.94
No systemic onset JIA	−737 (−1,040 to −434)	<0.001	−5.6
Identified 2001–2004 c	−864 (−1,235 to −493)	<0.001	−6.1	0.14
Identified 2005–2017	−430 (−856 to −3)	0.048	−3.9

^*CI, confidence interval; JIA, juvenile idiopathic arthritis; NA, not available.

^a The estimate represents the adjusted mean annual difference in earnings between patients with JIA and comparators obtained from linear regression adjusting for sex, age, age squared, age at identification, year of observation, and parental characteristics, with SEs clustered at the individual level.

^b The percentage change is defined as the difference between patients with JIA and general population comparators estimated by linear regression divided by the mean level of the outcome variable for the general population comparators during the follow‐up period.

^c The identification date does not necessarily match the first diagnosis date in all patients because the outpatient care component of the National Patient Register from which patients were identified was established in 2001 and patients were allowed to be younger than age 16 years at that time point and may, therefore, have had their initial diagnosis earlier.

Table 3.

Estimated association between JIA and work loss observations for those aged 18 years and older^*

Subgroup	Estimate of change in days of work loss (95% CI) a	P value	Percentage change b	P for interaction
Overall	10.6 (8.43–12.8)	<0.001	92	NA
Men	9.4 (5.9–12.8)	<0.001	91	0.33
Women	11.3 (8.5–14.0)	<0.001	92
Systemic onset JIA	17.4 (9.8–24.9)	<0.001	184	0.08
No systemic onset JIA	9.9 (7.6–12.1)	<0.001	84
Identified 2001–2004 c	12.1 (9.3–14.9)	<0.001	99	0.02
Identified 2005–2017	7.0 (4.2–9.9)	<0.001	70

^*CI, confidence interval; JIA, juvenile idiopathic arthritis; NA, not available.

^a The estimate represents the adjusted mean annual difference in work loss between patients with JIA and comparators obtained from linear regression adjusting for sex, age, age squared, age at identification, year of observation, and parental characteristics, with SEs clustered at the individual level.

^b The percentage change is defined as the difference between patients with JIA and general population comparators estimated by linear regression divided by the mean level of the outcome variable for the general population comparators during the follow‐up period.

^c The identification date does not necessarily match the first diagnosis date in all patients because the outpatient care component of the National Patient Register from which patients were identified was established in 2001 and patients were allowed to be younger than age 16 years at that time point and may, therefore, have had their initial diagnosis earlier.

Work loss

Patients with JIA had more work loss days than matched comparators throughout the follow‐up period (mean difference 11 [95% CI 8–13] days/year; Figure 1). This higher level of work loss for patients with JIA versus comparators was more pronounced in patients with the highest number of work loss days, at the 95th percentile, than at the 75th or 90th percentiles (Supplementary Figure 3). Patients ≥95th percentile for work loss at age 20 years differed from the total sample in having a greater percentage with sJIA (14% vs 9%) and having parents with lower education level (9% vs 4% with primary school only), more work loss (99 vs 52 days per year), and lower earnings (€36,500 vs €43,400) (Supplementary Table 4). The distribution of work loss was skewed, with 81% of patients with JIA and 89% of general population comparators not having any compensated work loss days at age 25 years, whereas approximately 7% versus 3% had ≥6 months of compensated work loss at that age (P < 0.001; Supplementary Figure 4).

Subgroup analysis

Patients with and without sJIA had a similar earnings differential versus matched comparators (mean difference vs comparators −€690 vs −€737) but higher work loss (mean difference versus comparators 17 vs 10 days) (Supplementary Figure 6; Supplementary Table 5). No statistically significant effect modification by diagnosis subtype (sJIA vs no sJIA) was found (Tables 2 and 3). There was no statistically significant effect modification by sex or by identification period (identified 2001–2004 vs 2005–2017) for earnings (Tables 2 and 3; Supplementary Figures 7 and 8). Regarding work loss and compared with the general population, patients identified 2001 to 2004 (mean difference 12 [95% CI 9–15] days/year) had more work loss compared with those diagnosed 2005 to 2017 (mean difference 7 [95% CI 4–10] days/year; P _interaction = 0.02) (Tables 2 and 3). Exploratory analysis by four quartile periods of identification showed the highest mean difference in work loss for patients identified in 2001 (relative to comparators) (Supplementary Figures 9 and 10). That subgroup consisted of patients diagnosed primarily in the 1990s with little or no exposure to biologics before adulthood (age 18 years).

DISCUSSION

This study of 4,737 patients with JIA observed from age 18 years found patients to have 5.5% lower earnings compared with matched general population comparators as well as lower employment levels. Earnings were lower for patients with JIA, both because they were less likely to be employed and because they had lower earnings conditional on being employed. The numerical difference between the earnings of patients and comparators was larger at the bottom of the earnings distribution. Beyond age 26 years, the gaps between patients and comparators were generally smaller overall as well as for women, although it appeared to widen for men. This finding should be interpreted with caution as no statistically significant effect modification by sex was found, and the uncertainty regarding the estimates increased with age as the sample size diminished. Future studies with longer follow‐up periods are needed to establish whether there is a sex difference in the earnings trajectory.

Patients with JIA accumulated almost twice as many days of work loss, averaging 11 additional days annually versus matched comparators. The distribution of work loss was skewed, and the difference versus comparators was more pronounced at the 95th percentile of work loss, indicating that a subgroup of patients experienced a disproportionately high amount of work loss. These individuals may face considerable challenges in maintaining consistent employment because of frequent and prolonged work loss episodes.

The toll of JIA in terms of work loss was smaller in patients diagnosed after 2004 compared with those identified in 2001 to 2004 (7 vs 12 additional days annually vs comparators), with significant effect modification by calendar period. We cannot determine what caused this temporal change, but we speculate that the adoption of treat‐to‐target strategies and the increase in the number of effective drug treatments are contributing factors.

Despite patients with JIA having reportedly similar levels of education, ¹² previous studies have found their unemployment levels to be significantly higher than those in the general population. ⁴ , ⁵ , ⁶ , ¹³ To our knowledge, there are no previous studies on earnings or work loss in patients with JIA versus matched general population comparators.

This study did not investigate the main factors driving the differences in earnings and work loss versus the general population. Potential factors include pain, mobility limitations, time lost to regular health care visits and treatments, and loss of self‐esteem because of a childhood with chronic disease and failed treatments as well as painful procedures.

The main strengths of our analysis are that it is population‐based on a nationwide level, resulting in a large group of patients with JIA, long follow‐up of objective data on earnings and work loss, and individually matched comparators sampled from the general population to contrast outcomes in patients with JIA as well as discern any potential changes in the level of earnings or work loss over time caused by policy adjustments, business cycle effects, or something similar.

One limitation of our study is that we could not distinguish between patients in long‐term remission and those with active disease. Regarding work loss, we observed that a minority of patients disproportionately contributed to higher work loss relative to the general population, similar to findings in adults with rheumatoid arthritis. ¹⁴ , ¹⁵ At least 75% of patients with JIA did not exhibit any work loss >14 days in a year compared with 85% in matched comparators. The lack of information on whether patients with JIA worked part‐time could impact the interpretation of our results. Patients who work part‐time may experience fewer work loss days yet still experience earnings reductions because of their decreased working hours. However, we found lower earnings as well as greater work loss among patients with JIA versus matched general population comparators.

Patients with JIA had lower mean annual earnings and higher work loss than matched general population comparators, but earnings differences diminished in magnitude with age and work loss diminished with calendar period of identification. In JIA, a minority of patients accounts for most of the negative impact on economic outcomes that persists into adulthood. For patients with sJIA, the average loss in earnings versus the general population was similar to that of the one for patients without sJIA, but higher work loss was observed for patients with sJIA.

AUTHOR CONTRIBUTIONS

All authors contributed to at least one of the following manuscript preparation roles: conceptualization AND/OR methodology, software, investigation, formal analysis, data curation, visualization, and validation AND drafting or reviewing/editing the final draft. As corresponding author, Dr Bruze confirms that all authors have provided the final approval of the version to be published and takes responsibility for the affirmations regarding article submission (eg, not under consideration by another journal), the integrity of the data presented, and the statements regarding compliance with institutional review board/Declaration of Helsinki requirements.

Supporting information

Disclosure form.

Appendix S1: Supplementary Appendix