- Top of page
- SUBJECTS AND METHODS
- AUTHOR CONTRIBUTIONS
Considerable evidence links arthritis to loss of employment and identifies clinical, personal, and environmental factors associated with changes in work status (1–13), but until recently, few studies have focused on the employment experiences of people with arthritis. This is changing, however, with increased emphasis on the diverse experiences of individuals with arthritis at work (14–21) and research being done to measure work place disability and productivity and the indirect costs of arthritis (22–34).
The shift in research to include a better understanding of working with arthritis has highlighted underemployment and the work transitions that individuals make in an effort to remain employed. Underemployment captures a wide range of adverse employment situations that result in reduced job satisfaction and earnings (35). These include unemployment, involuntary part-time work, and a mismatch between training and job requirements (e.g., being overqualified). Work transitions, a less evaluative term, describes labor force changes that may signal problems resulting in an individual leaving the workforce or may reflect positive changes enabling individuals to remain employed (22, 36–40). To our knowledge, there are few studies of work transitions in arthritis. The objective of our study was to prospectively examine arthritis-related work transitions, the relationships among different types of work transitions, and the factors associated with them.
Although there is little arthritis research examining work transitions, the benefits of mapping work trajectories for evaluating health services, productivity, and costs have been noted (22). Moreover, by examining whether occasional productivity losses and work changes are related to one another and to subsequently leaving employment, we are in keeping with research that conceptualizes participation in employment as multifaceted rather than a simple dichotomy of working versus not working (41).
In this study, we examined 3 categories of arthritis-related work transitions: productivity losses, namely absenteeism and job disruptions (e.g., not being able to take on extra work); work changes (reduced work hours, changing jobs); and leaving the labor force (Figure 1). Productivity losses are likely to be reported more frequently than work changes or leaving employment. Moreover, relationships among work transitions are expected to be complex. For example, if absenteeism occurs on an occasional basis, it may not create job disruptions or result in forgoing employment. However, if job disruptions and absenteeism become chronic, individuals may be more likely to make work changes or leave employment. Arthritis-related work changes such as reducing hours or changing jobs are also expected strategies for avoiding employment loss. However, it is unclear whether these strategies enable individuals to remain employed or whether they are associated with subsequently leaving employment (8, 21).
We also investigated whether demographic, health, work place, and psychological variables used in previous studies can help develop a more complete understanding of work transitions (10). For example, aging research finds that many older workers desire to gradually reduce work hours prior to retiring (42, 43). Older workers viewed reduced hours as a positive work transition because of the flexibility and the opportunity to remain working. This suggests that age is important in understanding work transitions. Chronic illness and disease factors such as pain and disability are related to greater absenteeism (30, 44–46). These are hypothesized to be especially important to productivity losses, which, in turn, are likely to relate to work changes and leaving employment. To our knowledge, few studies have examined job change, but a study of work place fatigue found that job disruptions, including problems with supervisors and reports of greater physical and emotional strain, preceded employees changing jobs (40). Other research has found that psychological variables such as depression and mastery are related to absenteeism (30, 46). Building on these studies, we expected that perceptions of a lack of job control, depression, and role conflict would be related to increased work transition (18, 30, 40, 46).
- Top of page
- SUBJECTS AND METHODS
- AUTHOR CONTRIBUTIONS
T1 and T4 sample characteristics are presented in Table 1. Seventy-one percent of participants were retained in the sample over 4.5 years: T1 n = 490, T2 n = 413, T3 n = 372, and T4 n = 349. Demographic and occupational characteristics remained similar over time. Approximately three-quarters of the sample were women. At T1, the respondents' age was ≥23 years, with a mean age of 51.1 years. Participants were relatively well educated and 60.6% were married. Approximately two-thirds of respondents worked in business/administration or health/sciences. At T1, 56.7% of participants reported OA and 33.3% reported inflammatory arthritis. Of these, 79% had RA. Fewer respondents reported having both OA and inflammatory arthritis at T1 than at T4 (10% versus 18.3%) (P < 0.001). Not surprisingly given the duration of the study, frequency of severe pain and fatigue were more variable than other characteristics. At T4, participants reported somewhat lower frequency of severe pain and fatigue (P < 0.05).
Table 1. Sample characteristics of all respondents at time 1 and time 4*
|Variables||Time 1 (n = 490)||Time 4 (n = 349)|
|Demographic|| || || || |
| Age, mean ± SD years||487||51.1 ± 9.3||346||56.6 ± 9.0|
| Sex|| || || || |
| Education|| || || || |
| Elementary & secondary||85||17.4||59||17.0|
| Some postsecondary||112||22.9||74||21.3|
| Marital status|| || || || |
| Married/living as married||297||60.6||215||61.6|
| Never married||76||15.5||47||13.5|
| Lives alone||490||25.3||349||30.9|
|Illness related|| || || || |
| Arthritis type|| || || || |
| Inflammatory arthritis†||163||33.3||108||30.9|
| Both OA & inflammatory arthritis||49||10.0||64||18.3|
| Duration at wave 1, mean ± SD years||488||9.17 ± 8.7|| || |
| Pain severity in last month|| || || || |
| No days||86||17.6||74||21.2|
| A few days||120||24.5||113||32.4|
| Some days||117||23.9||80||22.9|
| Most days||109||22.2||51||14.6|
| All days||54||11.0||31||8.9|
| Fatigue in previous month|| || || || |
| No days||76||15.5||58||16.6|
| A few days||72||14.7||84||24.1|
| Some days||95||19.4||76||21.8|
| Most days||121||24.7||74||21.2|
| All days||123||25.1||57||16.3|
|Employment|| || || || |
| Job sector|| || || || |
| Business, finance, administration||162||33.1||88||34.5|
| Health, science, art||175||35.8||93||36.5|
| Sales, services||102||20.9||51||20.0|
| Trades, transportation, equipment operator||50||10.2||23||9.0|
| Psychosocial, mean ± SD score|| || || || |
| AWS (observed range 6–30)||464||17.9 ± 5.9||252||16.4 ± 5.3|
| CES-D (observed range 0–48)||484||10.9 ± 10.0||344||9.3 ± 8.5|
| WALS (observed range 0–24)||474||6.4 ± 4.4||255||7.3 ± 4.6|
Changes in employment status among participants for whom data was available across ≥2 time points during followup (n = 363) are summarized in Table 2. Sixty-three percent of respondents remained employed throughout the study, 29.8% left the labor force at some point during followup and did not return, and 7.2% left employment and returned at a later time.
Table 2. Employment status changes during followup (n = 363)
|Employment status||N (%)|
|Employed at all time points||229 (63.1)|
|Left labor force and did not return during followup||108 (29.7)|
| Within the first 18 months||42 (11.6)|
| Between 18 and 36 months||34 (9.4)|
| Between 36 and 54 months||32 (8.8)|
|Left labor force and returned at a later time||26 (7.2)|
| By 36th month of followup||13 (3.6)|
| By 54th month of followup||13 (3.6)|
Means and frequencies of work transitions at T2–T4 are presented in Table 3. These data represent changes to employment made after the onset of the study, allowing us to examine work transitions within a specific timeframe. Overall, 76.5% of participants reported ≥1 work transition during followup. Approximately half of the sample reported ≥1 job disruption at each time. The most common were arthritis-related work interruptions of ≥20 minutes (∼30% of all job disruptions at each time point), being unable to take on extra projects (15–20%), and arriving late or leaving early (10%). Problems with supervisors or coworkers (5–9% of job disruptions) and being unable to attend business trips (<5%) were least common. Arthritis-related absenteeism varied, with 27.9% and 20.1% of respondents reporting absenteeism at T2 and T4, respectively, and a total of 47.1% of participants reporting arthritis-related absenteeism during followup. At T2, 17.9% of respondents reported having reduced their work hours since their previous interview. This dropped to 12.4% and 12.9% at T3 and T4, respectively, for a total of 36.1% of respondents reporting arthritis-related reduced work hours. The percentage of respondents who reported changing jobs was 11.4%, 6.2%, and 5.2% at T2, T3, and T4, respectively, with a total of 20.9% changing jobs. At T2, 14.8% of respondents reported having left the labor force. At T3 and T4, 11.6% and 8.6% of respondents reported leaving employment, respectively. Overall, 36.9% of participants reported leaving employment during followup.
Table 3. Arthritis-related work transitions during followup*
| ||Time 2, 18 months (n = 413)||Time 3, 36 months (n = 372)||Time 4, 54 months (n = 349)||Total (n = 363)|
|Job disruptions, %|
|Job disruptions, mean ± SD||1.28 ± 1.7||1.07 ± 1.6||1.04 ± 1.6||1.15 ± 1.6|
|Absenteeism||115 (27.8)||85 (22.8)||70 (20.1)||171 (47.1)|
|Reduced work hours||74 (17.9)||46 (12.4)||45 (12.9)||131 (36.1)|
|Changed jobs||47 (11.4)||23 (6.2)||18 (5.2)||76 (20.9)|
|Left labor force†||61 (14.8)||43 (11.6)||30 (8.6)||134 (36.9)|
|Total arthritis-related work transitions||271 (65.6)||192 (51.6)||186 (53.3)||316 (76.5)|
The coefficient estimates for the multivariate longitudinal analyses of each work transition using GEEs are presented in Table 4. Analyses included the independent variables as well as work transitions from preceding waves. Men reported more job disruptions than women (P < 0.05). Workplace activity limitations predicted later job disruptions (P < 0.05), as did perceptions of arthritis-work spillover (P < 0.001), previous reports of arthritis-related absenteeism (P < 0.01), and changing jobs (P < 0.05). Absenteeism was reported less often in the health/science sector compared with business/administration (P < 0.05). Perceptions of arthritis-work spillover (P < 0.05), depressive symptoms (P < 0.01), and reports of previous job disruptions (P < 0.001) were also associated with later absenteeism. Older adults (P < 0.05), those reporting greater arthritis-work spillover (P < 0.05), or those who previously changed jobs (P < 0.01) were more likely to report reduced work hours. Individuals with postgraduate education were significantly more likely to change jobs (P < 0.01). Changing jobs was also related to previously reducing work hours (P < 0.05). Finally, leaving the labor force was associated with being older (P < 0.001), a lower education level (P < 0.01), having less control over one's work schedule (P < 0.05), being employed in the health/sciences sector (P < 0.05), and reporting previous job disruptions (P < 0.05) and reductions to work hours (P < 0.01).
Table 4. Longitudinal analyses of the association of demographic, illness, work context, psychological, and previous work transitions with subsequent arthritis-related work transitions using generalized estimation equations*
|Variables||Productivity losses||Work changes||Left labor force|
|Job disruptions||Absenteeism||Reduced hours||Changed jobs|
|Demographic|| || || || || |
| Age|| || ||0.03 (0.01, 0.06)†|| ||0.06 (0.03, 0.09)‡|
| Female||−0.31 (−0.57, −0.06)†|| || || || |
| Living alone|| || || || || |
| Education§|| || || || || |
| Some postsecondary|| || || || || |
| Postsecondary|| || || || || |
| Postgraduate|| || || ||1.60 (0.43, 2.80)¶||−1.41 (−2.20, −0.61)¶|
|Illness related|| || || || || |
| Arthritis duration|| || || || || |
| OA diagnosis#|| || || || || |
| Previous fatigue|| || || || || |
| Previous pain|| || || || || |
|Work context|| || || || || |
| Previous schedule control|| || || || ||−0.19 (−0.33, −0.04)†|
| Previous WALS||0.03 (0.00, 0.05)†|| || || || |
| Occupation**|| || || || || |
| Health, science, art, sports|| ||−0.48 (−0.94, −0.03)†|| || ||0.70 (0.17, 1.24)†|
| Sales and services|| || || || || |
| Trades, transportation, equipment|| || || || || |
|Psychological|| || || || || |
| Previous AWS||0.07 (0.05, 0.10)‡||0.04 (0.00, 0.09)†||0.06 (0.01, 0.11)†|| || |
| Previous CES-D|| ||0.03 (0.01, 0.05)¶|| || || |
|Previous work transitions|| || || || || |
| Previous job disruptions||–||0.22 (0.10, 0.35)‡|| || ||0.16 (0.02, 0.31)†|
| Previous absenteeism||0.26 (0.08, 0.45)¶||–|| || || |
| Previous reduced hours|| || ||–||0.75 (0.18, 1.36)†||0.74 (0.25, 1.23)¶|
| Previous change of job||0.26 (0.04, 0.47)†|| ||0.74 (0.26, 1.23)¶||–|| |
- Top of page
- SUBJECTS AND METHODS
- AUTHOR CONTRIBUTIONS
This research examined arthritis-related work transitions reported over 4.5 years. The findings extend previous studies by capturing dimensions of participation in paid work beyond the categories of employed or not employed. Although >60% of respondents remained working throughout the study, work transitions were common, being reported by three-quarters of participants. Moreover, these changes were often related to subsequently making other work transitions, including leaving employment. As such, this research sheds light on a process of diverse employment changes that may occur in the working lives of many people with arthritis.
Research has assessed productivity losses by taking measure of absenteeism (25, 26, 44–46) and more recently by examining presenteeism (i.e., at-work productivity losses) (22, 24–26). Job disruptions such as being unable to work at particular times and difficulties with coworkers have also been characterized as potential productivity problems, although these have been assessed less frequently (40). In the current research, approximately half of the participants reported ≥1 job disruption, most frequently in the form of work interruptions, being unable to take on extra projects, and arriving late or leaving early. A difficulty facing researchers measuring this type of productivity loss is that job disruptions are difficult to quantify in terms of their frequency, duration, and cost. Also, it is unclear whether some job disruptions or constellations of problems are more important than others in predicting future work transitions. Additional research measuring the productivity impact of job disruptions is needed to assess their full impact.
Over the duration of the study, nearly half of the respondents reported arthritis-related absenteeism, more than one third had reduced their work hours, and approximately 1 in 5 had changed jobs. At each successive wave, fewer respondents reported these work transitions. It is likely that this is because some work transitions constitute relatively rare events; having reduced their hours or changed jobs, respondents are less likely to do so again within a short time. Overall, however, the ongoing impact of arthritis on employment is notable. Assessing a range of arthritis-related work place outcomes can enhance our understanding of the impact of the disease on employment and deserves increased attention in future research.
Similar to other research, older workers in the current study were more likely to reduce work hours or leave employment than younger workers (42, 43). Financial resources make these work transitions more likely with age. It is not clear whether the changes were desired. Research with older workers found reduced hours were a preferred option prior to retirement (42). In this study, reducing hours was associated with perceptions of spillover (i.e., difficulty balancing the demands of work and caring for arthritis). Additional research is needed to examine whether reduced hours is an adverse job status among some groups or a work transition that facilitates employment and other lifestyle goals. This also has implications for studies estimating the indirect costs of arthritis because they have typically included reduced hours as an undesired productivity loss.
Men reported more arthritis-related job disruptions than women. There were no other sex differences in work transitions. As noted, we lacked comprehensive data examining job disruptions. However, in a large Canadian study of arthritis disability, men were more likely to report work place limitations, whereas women were more likely to leave the labor force (52). It may be that financial considerations, such as being the main or sole income earner, result in men remaining employed despite job disruptions. Differences may also relate to the types of jobs men and women occupy.
Similar to previous studies, education acted as a resource (10). Individuals with greater education were more likely to change jobs and less likely to leave employment. This suggests that arthritis work place interventions need to consider skill training and education that may enable people with arthritis to remain working in a different job or occupation, as well as work place accommodations and arthritis therapies to treat the disease.
Illness variables, including diagnosis of OA or inflammatory arthritis, were not related to later work transitions. However, it is inappropriate to conclude that disease factors have no relationship to work transitions. Instead, the findings point to the need for investigation of the mechanisms of the impact of arthritis on work transitions. Specifically, although direct effects of chronic, severe pain and fatigue on work transitions were not found, symptoms may act indirectly by increasing the likelihood of work place activity limitations or by changing perceptions of work, such as increased role conflict or spillover. Activity limitations and changed perceptions may in turn result in decisions to make work transitions. Some evidence for this exists elsewhere. The effects of symptoms on participation restrictions have been found to be mediated by activity limitations and depression (53), and fatigue and activity limitations were important predictors of perceived spillover in other research (18). Studies testing complex models of the moderating and mediating effects of illness, disability, and psychosocial variables are needed.
The absence of pain and fatigue as predictors of work transitions may also relate to measurement. Similar to other studies, we asked participants about pain and fatigue in the previous month. However, individuals with arthritis reported considerable variability in symptoms, making it unlikely that symptoms at a particular point influenced work transitions assessed 18 months later. This creates difficulties for researchers interested in long-term changes in employment. Reviews of research examining work disability in RA also found inconsistent results for illness variables (10). More frequent assessment of disease factors or additional measures capturing symptom variability may be needed.
Work context factors were associated with work transitions. Greater work place activity limitations were related to later reports of job disruptions. Assessments of activity limitations may be ideal for capturing specific productivity problems (e.g., difficulties with the schedule of work) rather than broad work transitions (e.g., changing jobs). Similar to disease symptoms, however, activity limitations may be part of a complex causal pathway, indirectly affecting other variables. In this study, workplace activity limitations were related to job disruption, which in turn predicted later absenteeism and leaving employment.
Participants working in the health/sciences sector were less likely to report absenteeism, but more likely to leave employment than those in business/administration. The former group included health professionals (e.g., nurses) and teachers, and comprised the largest of the job sectors examined. These occupations can be physically demanding for individuals with arthritis (e.g., lifting, standing for long periods). Moreover, the specialized training and nature of these jobs may make work transitions such as reduced hours and changing jobs less viable. Less control over the work schedule was also associated with leaving employment. This highlights the need for researchers to work in concert with employers to develop work place practices that impart greater flexibility and control to workers.
Psychological perceptions, especially arthritis-work spillover, were important in understanding productivity losses such as job disruptions and absenteeism, as well as in reducing hours. To date, employment studies have emphasized the impact of arthritis on work. Arthritis-work spillover underscores the importance of perceived role stress and role balance. The relationship of arthritis-work spillover to work transitions suggests that it may be an important factor in the early identification of work place problems that may lead to later work transitions. As such, future research may want to examine its potential for screening individuals for inclusion in work place interventions.
Of interest are the interrelationships among work transitions. Measures of productivity loss (job disruptions, absenteeism) were related in the current study. Job disruptions were also associated with subsequently leaving employment. Work changes (i.e., reduced hours, changing jobs) were interrelated, and reduced hours were associated with leaving employment. These findings suggest conceptual links among the different types of work transitions, and point to the importance of examining patterns of work transitions and their accumulation over time. Work transitions were not independent of one another and may form a hierarchy of changes that occur as individuals with arthritis struggle to remain employed. It was noteworthy that absenteeism and changing jobs were not directly associated with leaving employment. Absenteeism may be an episodic problem that affects working insofar as it may create job disruptions. Changing jobs may actually enable individuals to remain employed. In this sample, changing jobs was not associated with negative psychological perceptions or work context factors, but occurred among those with greater education. At the same time, if changing jobs is associated with reducing hours, it may subsequently relate to leaving employment.
Several limitations in this research need to be addressed. Attention to measurement of work transitions such as job disruptions and changing jobs would increase understanding of work transitions and their interrelationships. Research into the measurement and short- and long-term impacts of arthritis disease factors is also important given the variability of symptoms experienced by working individuals. Although our recruitment strategy enabled us to generalize beyond a clinical sample and was comparable with other samples in terms of demographic and disease characteristics, the sample was purposive. Research comparing diverse subgroups of arthritis, as well as comparing those with arthritis with those with no chronic health problems, is needed to replicate and extend the findings. Finally, assessing perceptions of work transitions would allow examination of whether work transitions were viewed as adverse employment options or were proactive efforts to remain employed.
In conclusion, this study highlights the diverse work transitions made by individuals with arthritis and underscores insights to be gained from using a multifaceted approach to measurement of participation in employment. The study also emphasized the interrelationships among productivity, work changes, and employment loss, as well as the role of other factors in predicting work transitions. A greater understanding of work transitions may enable new intervention efforts to target work changes that signal impending difficulties with remaining employed.