Abstract
Abstract
Objectives
This study aimed to identify distinct trajectories of long-term sickness absence (LTSA, >10 consecutive working days) among young and early midlife Finnish employees who experienced pain at baseline. It also aimed to determine the pain characteristics and occupational and lifestyle factors associated with these LTSA patterns.
Design
Longitudinal occupational cohort study with register linkage.
Setting
The largest municipal employer in Finland.
Participants
The study population comprised 19–39-year-old Finnish municipal employees (n=1685) who reported pain in 2017.
Outcome measures
Prospective register data on all-cause LTSA through March 2020 were obtained from the Social Insurance Institution of Finland. Group-based trajectory modelling was used to identify distinct all-cause LTSA trajectories. Multinomial logistic regression was used to examine associations of pain characteristics and work- and lifestyle-related factors with trajectory group membership.
Results
Three distinct LTSA-trajectory groups were identified: no LTSA (74%), decreasing (18%) and increasing (8%). The decreasing trajectory group had a higher prevalence of chronic or multisite pain, smoking (average marginal effects (AME) 6% points, 95% CI 2 to 11), obesity (AME 8% points, 95% CI 2 to 13), manual or routine non-manual occupation (AME 9% points, 95% CI 4 to 13) and high physical workload, after adjusting for age and gender. No predictor was identified for the increasing trajectory.
Conclusion
A majority of young and early midlife employees with pain had no LTSA during follow-up; however, chronic and multisite pain, smoking, overweight or obesity, lower occupational class and higher physical workload were associated with the decreasing LTSA trajectory. Interventions at workplaces and in occupational healthcare to prevent LTSA should aim at supporting employees who work with pain and have these risk factors.
Keywords: Chronic Pain, SOCIAL MEDICINE, EPIDEMIOLOGY
STRENGTHS AND LIMITATIONS OF THIS STUDY.
The study population is drawn from a large occupational cohort representing multiple occupations.
Sickness absence data have been obtained from a reliable Finnish national register.
The study design was observational and does not allow inference of causal relationships between predictors and long-term sickness absence trajectories.
Questionnaire data on lifestyle factors may be subject to recall or reporting bias.
Introduction
Long-term sickness absence (LTSA) reflects ill health, incurs significant costs and is linked to future disability retirement.1 Due to the potential long-term consequences, it is of particular importance to prevent LTSA among young and early midlife employees. Most LTSA among younger employees in Finland and other Nordic countries is related to either musculoskeletal diseases or common mental health disorders, and pain is often the main symptom or comorbidity.2,4 Pain conditions alone are estimated to account for over one-third of all LTSA periods among older Finnish employees while corresponding estimates for young and early midlife employees are lacking.5
In an earlier study, we found that over 40% of Finnish municipal employees under 40 years old reported current pain.6 These employees had higher rates of LTSA compared with those without pain.6 Despite the high prevalence of pain, many younger employees maintain good work ability, while others experience prolonged or recurring sickness absence spells. Since sickness absence patterns vary over time, and between individuals, it is important to identify subgroups of employees with pain who share similar patterns of LTSA. Identifying characteristics or risk factors of these subgroups may help target preventive efforts for LTSA.
Pain is a personal and subjective experience, and its impact on work ability may depend on multiple individual, work- and pain-related factors. Chronic and multisite pain are associated with more LTSA.5,7 Lifestyle factors, such as smoking, low physical activity and obesity, are associated with more pain and with LTSA.8 Employees with a lower socioeconomic position and manual jobs report more pain.9 10 Psychosocial working conditions are also associated with more pain and LTSA.11,13
Factors that predict sickness absence trajectories among employees with pain have been studied previously.14,17 Haukka et al examined the determinants of LTSA trajectories in a nationally representative subsample of 30–55 year-old Finnish employees with multisite pain.17 Obesity, sleep disturbances and adverse physical and psychosocial working conditions were associated with a trajectory with more LTSA. Rysstad et al studied predictors of LTSA trajectories among 18–67 years old Norwegians on LTSA due to a musculoskeletal disorder. Multisite pain, poor self-rated general health and lower self-assessed workability were associated with trajectories with more sickness absence.15
To our best knowledge, no previous studies have examined LTSA trajectories specifically among young and early midlife employees with pain, a group who are expected to remain occupationally active for decades more but already are at increased risk of work disability. If modifiable lifestyle- and work-related factors that may predict LTSA patterns can be identified, this information could serve as a basis for intervention studies and healthcare or workplace interventions aiming to promote work ability among these employees. Therefore, the objective of this study was to identify subgroups of employees with pain who follow distinct trajectories of LTSA over the follow-up period and to examine whether pain characteristics, lifestyle and occupational factors at baseline are associated with these trajectories. While the study focused on general pain, adjustments were made for the specific effects of chronic pain (≥3 months) and multisite pain (≥2 body locations), as these are linked to both lifestyle factors and poor work ability.
Methods
Study design and study population
This study was based on survey data collected as part of the Helsinki Health Study and on prospective register-based LTSA data obtained from the Social Insurance Institution of Finland. Questionnaire data were collected in autumn 2017 by online questionnaires (58%), mailed questionnaires (29%) and telephone interviews (13%) from employees of the City of Helsinki who were 19–39 years and had a working contract of at least 50% for at least 4 months before the survey (n=11 459).18 Online surveys were sent to all those who had an email address, and the surveys were mailed to the rest. All non-respondents after a reminder received a mailed survey with login details included in the cover letter, to choose their preferred participation method. Postage was covered. Telephone interviews were conducted among all non-respondents after the mailed reminders if they had a phone number available. The response rate was 51% (n=5 898), and 79% of the respondents were women, which largely corresponds to the proportion of women in the Finnish municipal sector (figure 1).19 The telephone interview was designed to collect only the most important survey items that are not available in registers, such as pain, physical workload, smoking and body mass index (BMI). Using data from all survey participants produced a more representative sample for the analysis of pain, as men and manual workers were somewhat over-represented among the phone-interviewed, balancing their somewhat lower participation in the full online and mailed surveys.18 Consent to register linkage (81%, n=4864) and reporting current pain (41%, n=2435) were required for inclusion in the study population. Of the 2009 respondents fulfilling these two criteria, those who did not work full- or part-time at the time of the survey (such as respondents who were on family leave, were studying, had been on sick leave for more than the 6 months prior to the survey, were on disability rehabilitation, unemployed, missed or reported ‘other’ employment status) were excluded (n=230). Respondents with missing information on covariates were also excluded (n=94). The final analytical sample included 1685 employees, of whom 81% were women.
Figure 1. Flow chart with exclusion criteria applied to the target population for obtaining the study population (n=1685).
Pain
The current study included respondents who answered ‘yes’ to the question: ‘Are you suffering from any pains or aches right now?’ at baseline. This defined them as having ‘current pain’. They were also asked ‘When did the pain begin?’ and given two options: up to 3 months ago or more than 3 months ago. Based on their answer, they were categorised as having chronic pain (more than 3 months) or acute/subacute pain (up to 3 months, hereafter referred to as acute pain).20 Additionally, they were asked ‘Where do you feel the pain?’ and given six choices: neck and shoulders, lower back, upper extremities, lower extremities, face or head or elsewhere. Each choice was considered as one pain site. A missing response was interpreted as no pain at that body site. A total score was calculated and analysed as single-site pain or multisite pain (≥2 pain sites).
Sickness absence
In Finland, the employer generally compensates the first ten working days of a sickness absence spell. The Social Insurance Institution of Finland pays sickness allowance for compensation of loss of income due to work disability that exceeds the first ten working days for up to 300 days, which may be consecutive or divided over multiple LTSA spells. Days on sickness allowance during the last 2 years count towards the maximum of 300 days. Sickness allowance is also paid to unemployed people after a ten-day waiting period. LTSA data were obtained from the Social Insurance Institution of Finland’s national register of sickness allowance payments (start date and end date) and did not depend on the employer at the time of data collection. LTSA days were defined as any sickness absence days that sickness allowance had been paid for. The data covered the period from 1 January 2018 to 31 March 2020. This end date was chosen to avoid any effect of the COVID-19 pandemic on the sickness absence rates. The follow-up was split into six observation periods, with cut-off points set at 4.5 months interval starting from 1 January 2018. The start and end dates of the sickness allowance payments were used to create a dichotomous variable for each observation period, indicating whether the participant had experienced any LTSA during that time.
Lifestyle factors
The participants provided self-reported data regarding their frequency of alcohol use, smoking habits, vegetable consumption, height and weight. Smoking was classified as either current use (daily or occasional) or non-use (never or stopped). The frequency of alcohol consumption was assessed and classified as either ‘twice weekly or more’ and ‘once weekly or less’. Vegetable consumption was classified as either ‘daily’ or ‘less often’. BMI was calculated based on height and weight, and participants were classified into three categories: < 25.0 kg/m2, 25.0–29.9 kg/m2 and ≥30.0 kg/m2.21
Occupational factors
The participants were categorised into three occupational classes: professionals (such as teachers and physicians), semi-professionals (such as registered nurses) and routine non-manual or manual workers (such as practical nurses, child minders, construction workers and cleaners). The classification was based on job titles listed in the employees’ register of the City of Helsinki. Routine non-manual and manual occupations are generally characterised by lower educational levels, and employees in these categories reported similar mean daily time in physically heavy work. This and the number of manual workers was low motivated collapsing these two groups into one category. Perceived physical and mental workload were assessed through the question ‘What is your job like?’ (physically/mentally). Participants provided responses on a five-point Likert scale, ranging from very light to very strenuous. A high physical workload was defined as reporting work that was physically quite or very strenuous, while a high mental workload was defined as reporting work that was mentally very strenuous, following previous procedures with similar data.22
Covariates
Covariates included age and gender. Gender was self-reported and categorised as either man or woman. Age was calculated based on the year of birth and dichotomised as <30 years and ≥30 years for descriptive purposes. However, in regression models, age was treated as a continuous variable to avoid residual confounding.
Statistical methods
A trajectory refers to the longitudinal pattern of development of a particular outcome.23 Group-based trajectory modelling (GBTM) is a statistical method used to identify latent trajectories of an outcome by employing maximum-likelihood estimation.24 In our study, we employed the traj package for STATA version 16 to model groups of employees who followed similar trajectories of LTSA.25 The total follow-up period of 27 months was split into six observation periods, and the number of LTSA days within each observation period was calculated. The proportion of employees who experienced LTSA during each observation period varied from 5.3% to 7.5%. The count of LTSA days exhibited zero inflation and dispersion among respondents who had experienced LTSA. To ensure a statistically valid approach that captures the temporal development of LTSA, we applied a logistic model with the outcome being either the occurrence or absence of LTSA within an observation period (yes/no).
Among the respondents included in the analysis, 74.5% (n=1256) did not have any LTSA periods, while 25.5% (n=429) had at least one LTSA period during the follow-up. Respondents without LTSA throughout the follow-up period were classified into a single trajectory. A logistic group-based trajectory model was used to identify latent trajectories of LTSA based on the occurrence or non-occurrence of LTSA within each of the six observation periods. We systematically fitted trajectory models with varying numbers of trajectory groups (ranging from 1 to 3 groups) and polynomial orders (cubic, quadratic, linear, intercept). Ultimately, a two-trajectory model with polynomial orders (1 2) was selected based on the simultaneous consideration of several model fit indices, as previously recommended by Nagin and Odgers.24 The chosen model exhibited Bayesian and Akaike’s information criteria closer to zero, and the average posterior probability for each group in the model exceeded the recommended threshold of 0.7, whereas the entropy value of 0.55 was relatively low. However, the model maintained sufficiently large group sizes for further analyses. Model fit statistics for all tested models are displayed in online supplemental table 3. A spaghetti chart displaying individual trajectories is presented in online supplemental figure 1. Finally, the two-trajectory model was combined with the trajectory of respondents without LTSA during the follow-up period.
Descriptive characteristics of the three trajectory groups were initially examined through cross-tabulation and the X2 test. Moreover, we assessed multicollinearity among the predictors (see online supplemental table 2) by calculating variance influence factors (VIF). A VIF value higher than one indicates multicollinearity, with a VIF of >5, suggesting significant multicollinearity. The highest observed VIF, between occupation class and physical workload, was measured at 1.26, indicating low multicollinearity (mean VIF 1.08). We observed a Spearman correlation coefficient of 0.42 between occupational class and physical workload; otherwise, the correlation between variables was low. Correlation coefficients are presented in online supplemental table 2. To examine the association between pain characteristics, lifestyle and occupational factors and trajectory membership, we performed multinomial regression analyses using two models. Model 1 was adjusted for age and gender, while Model 2 was further adjusted for pain chronicity and number of pain sites. The link between illness and work disability is mediated by work demands, which was why occupation was not adjusted for in our analyses, to prevent over-adjustment. The estimates are presented as average marginal effects (AME) with their corresponding 95% CIs. The AME is presented in percentage points and interpreted as the average difference in predicted probability, compared with the reference group, of belonging to the trajectory group in question.26 AME coefficients are derived based on the multinomial regression models (STATA command margins) (online supplemental table 4). The corresponding ORs are presented in online supplemental table 5.
Ethical considerations
As part of the Helsinki Health Study, the observational study in the submitted manuscript did not require a separate ethical assessment; however, ethical aspects of the study have been considered in line with the Declaration of Helsinki. The Helsinki Health study’s plan received a favourable statement from the ethical board of the Faculty of Medicine at the University of Helsinki (decision nr. 01/2017) and was approved by the City of Helsinki (updated approval in 2017, decision nr. HEL 2017–0 04 490). The respondents have by participation in the Helsinki Health study agreed to the questionnaire data being used for scientific research purposes and that the data are being handled in line with the data protection statement of the Helsinki Health Study, which can be accessed via the research group’s website. Only respondents who have given their informed consent to register linkage have been included in the current study. Respondents have had the possibility to withdraw from the Helsinki Health Study at any point.
Patient and public involvement
Neither patients nor the public have been actively involved in the design, conduct, reporting or dissemination plans of the research.
Results
Study population characteristics
Approximately half of the participants had pain that was chronic, and almost two-thirds had multisite pain (table 1). Alcohol use twice weekly or more frequently was reported by 12% and smoking daily or sporadically was reported by 28%. One-third did not consume vegetables daily, and almost half of the participants were overweight or obese. The proportions of professional, semi-professional and manual or routine non-manual employees in the study population were similar.
Table 1. Descriptive characteristics of the study population and by trajectory group (n=1685).
| Total | No LTSA | Decreasing | Increasing | P value* | |
| N (%) | N (%) | N (%) | N (%) | ||
| 1685 (100) | 1256 (74) | 299 (18) | 130 (8) | ||
| Gender | 0.007 | ||||
| Woman | 1367 (81) | 997 (79) | 258 (86) | 112 (86) | |
| Man | 318 (19) | 259 (21) | 41 (14) | 18 (14) | |
| Age | 0.205 | ||||
| <30 | 532 (32) | 383 (30) | 107 (36) | 42 (32) | |
| ≥30 | 1153 (68) | 873 (70) | 192 (64) | 88 (68) | |
| Pain chronicity | 0.103 | ||||
| Acute | 900 (53) | 689 (55) | 144 (48) | 67 (52) | |
| Chronic | 785 (47) | 567 (45) | 155 (52) | 63 (48) | |
| Pain sites | <0.001 | ||||
| Single site | 599 (36) | 480 (38) | 75 (25) | 44 (34) | |
| Multisite | 1086 (64) | 776 (62) | 224 (75) | 86 (66) | |
| Alcohol use | 0.352 | ||||
| <twice weekly or missing (n=51) | 1484 (88) | 1098 (87) | 270 (90) | 116 (89) | |
| ≥twice weekly | 201 (12) | 158 (13) | 29 (10) | 14 (11) | |
| Smoking | 0.005 | ||||
| Never or stopped | 1218 (72) | 932 (74) | 194 (65) | 92 (71) | |
| Daily or sporadically | 467 (28) | 324 (26) | 105 (35) | 38 (29) | |
| Vegetable consumption | 0.182 | ||||
| Daily | 1171 (69) | 888 (71) | 198 (66) | 85 (65) | |
| Less than daily | 514 (31) | 368 (29) | 101 (34) | 45 (35) | |
| BMI (kg/m2) | 0.028 | ||||
| <25.0 | 925 (55) | 713 (57) | 140 (47) | 72 (55) | |
| 25.0–29.9 | 470 (28) | 342 (27) | 95 (32) | 33 (25) | |
| ≥30.0 | 290 (17) | 201 (16) | 64 (21) | 25 (19) | |
| Physical workload | 0.006 | ||||
| Light | 1002 (59) | 770 (61) | 153 (51) | 79 (61) | |
| Heavy | 683 (41) | 486 (39) | 146 (49) | 51 (39) | |
| Mental workload | 0.351 | ||||
| Light | 1347 (80) | 1013 (81) | 230 (77) | 104 (80) | |
| Heavy | 338 (20) | 243 (19) | 69 (23) | 26 (20) | |
| Occupational class | 0.002 | ||||
| Professional | 430 (26) | 345 (27) | 53 (18) | 32 (25) | |
| Semi-professional | 647 (38) | 466 (37) | 121 (40) | 60 (46) | |
| Manual and routine non-manual workers | 608 (36) | 445 (35) | 125 (42) | 38 (29) | |
| Observation period specific mean number of LTSA days and percentage of participants on LTSA by trajectory group | |||||
| Observation period 1 | 2.2 (6.9) | 0 (0.0) | 12.6 (38.1) | † | <0.001‡ |
| Observation period 2 | 1.7 (6.7) | 0 (0.0) | 8.8 (35.5) | † | <0.001‡ |
| Observation period 3 | 2.5 (7.5) | 0 (0.0) | 13.5 (39.1) | † | <0.001‡ |
| Observation period 4 | 2.0 (6.2) | 0 (0.0) | 13.5 (28.4) | 4.1 (15.4) | 0.004‡ |
| Observation period 5 | 1.5 (5.3) | 0 (0.0) | 2.1 (6.0) | 15.1 (55.4) | <0.001‡ |
| Observation period 6 | 1.9 (6.0) | 0 (0.0) | 0.9 (4.0) | 21.9 (68.5) | <0.001‡ |
Pearson’s Chiχ2 squared.
aA cell contains less than ten individuals.
Refers to percentage of participants on LTSA.
BMI, body mass index; LTSA, long-term sickness absence
Sickness absence trajectories
The best-fitting trajectory model produced three distinct LTSA trajectories, of which two trajectories were obtained by GBTM (decreasing (17.7%) and increasing (7.7%), and one trajectory was based on the observed data (no LTSA (74.5%). This model was selected for further analyses. Model fit estimates are presented in online supplemental table 3, and a graphical illustration of the LTSA trajectories throughout the follow-up period is presented in figure 2. In the decreasing trajectory, the share of employees with LTSA was initially high but decreased over time. In the increasing trajectory, the share of employees with LTSA was initially low but increased over time. The no LTSA trajectory comprised employees who had no LTSA days registered during the follow-up. The percentage of women was higher in the decreasing and increasing LTSA trajectories, as compared with in the no LTSA trajectory (table 1). There was also a higher percentage of employees with manual and routine-non manual occupations in the decreasing LTSA trajectory. Pearsons X2 test showed that gender, number of pain sites, smoking, BMI, physical workload and occupational class were associated with trajectory group membership (p<0.05) (table 1). Detailed information on sociodemographic factors, pain characteristics, lifestyle- and occupational factors for each LTSA trajectory are presented in table 1. Participant characteristics by gender is presented in online supplemental table 1.
Figure 2. Long-term sickness absence (LTSA) trajectories among employees with pain in 2017. (n=1685). The Y-axis reflects the proportion of participants who had LTSA days (yes/no) during the 4.5 months that preceded each observation point.
Associations of health-related and occupational factors with LTSA trajectories
After adjusting for age and gender, employees with chronic pain had five percentage points lower predicted probability of belonging to the No LTSA trajectory and four percentage points higher likelihood of belonging to the decreasing trajectory, as compared with those with acute pain (figure 3). Employees with multisite pain had nine percentage points lower predicted probability of belonging to the no LTSA trajectory and nine percentage points higher predicted probability of belonging to the decreasing trajectory, as compared with those with single site pain.
Figure 3. Associations between pain characteristics, lifestyle factors, occupational factors and long-term sickness absence trajectories (n=1685). Average marginal effects and their 95% CIs from multinomial logistic regression models. Model 1: adjusted for age and gender. Model 2: adjusted for age, gender, pain chronicity and pain sites.
Employees who smoked had seven percentage points lower predicted probability of belonging to the no LTSA-trajectory and six percentage points higher predicted probability of belonging to the decreasing trajectory, as compared with non-smokers, while employees who consumed vegetables less than daily had six percentage points lower predicted probability of belonging to the no LTSA trajectory, as compared with those who consumed vegetables daily. Employees with BMI of ≥30 had eight percentage points lower predicted probability of belonging to the no LTSA trajectory and eight percentage points higher predicted probability of belonging to the decreasing trajectory, as compared with employees with a BMI of <25.
Employees experiencing a heavy physical workload had five percentage points lower predicted probability of belonging to the no LTSA trajectory and six percentage points higher predicted probability of belonging to the decreasing trajectory, as compared with those with light physical workload. Those with a manual or routine non-manual occupation had eight percentage points lower predicted probability of belonging to the no LTSA trajectory and nine percentage points higher predicted probability of belonging to the decreasing LTSA trajectory, as compared with professionals.
Adjusting lifestyle and occupational predictors for pain characteristics slightly attenuated the associations, and the association between vegetable consumption and trajectory membership became non-significant. Perceived mental workload and frequency of alcohol use were not associated with trajectory membership. None of the variables examined in this study showed a significant relationship with the increasing LTSA trajectory.
Discussion
Main findings
This study sought to identify trajectories of LTSA among young and early midlife employees with pain and factors that predict trajectory membership. Most employees who reported pain had no LTSA during follow-up, and three distinct LTSA trajectories were identified: no LTSA, decreasing and increasing. Employees in the no LTSA trajectory were less likely, and employees in the decreasing LTSA trajectory were more likely, to have chronic or multisite pain, to smoke, to be overweight or obese, to have a manual or routine non-manual occupation and report heavy physical workload. Mental workload and the frequency of alcohol use did not predict trajectory membership in this study population.
Interpretation
To our knowledge, this is the first study to examine LTSA trajectories among young and midlife employees with pain, which is a large group of individuals who have many potential years of active working life ahead but who already are at increased risk of later work disability. The point prevalence of pain in this cohort has previously been estimated to be more than 40%, and self-reported pain in this age group is associated with higher sickness absence rates during the following year, as it is among older employees.5 6 10 14 The high prevalence of pain in this cohort and the proportions of chronic (47%) and multisite (64%) pain, in relation to the proportion of those experiencing LTSA during the following 27 months (25%), indicates that working despite a chronic or widespread pain condition is common among employees in this age group.
Among young and midlife employees who had LTSA during the follow-up, we identified a Decreasing and an Increasing LTSA-trajectory. The Decreasing LTSA-trajectory comprised employees who were more likely in a manual or routine non-manual occupation, perceived their work as physically strenuous, smoked, and had overweight or obesity. The reasons for the decline in LTSA in the Decreasing trajectory cannot be determined using this study design, but several factors may play a role. Finnish employers are obliged to provide occupational healthcare, which becomes involved if an employee is work disabled for a longer period.27 A medical assessment of work ability and need for rehabilitation is required by the Social Insurance Institution for continued payment of sickness allowance.28 Thus, one explanation for declining LTSA could be related to regained work ability after recovery, successful rehabilitation, workplace adjustments or modifications, or changing of workplace. Sickness allowance in Finland is generally paid for a maximum of 300 days, after which it is possible to claim disability pension. Information about the proportion of employees who transferred from LTSA to disability pension during follow-up was unavailable to the researchers and this may account for some of the decreased LTSA although work disability continued. Disability pension is, however, relatively uncommon in this age group, so that this is unlikely to account for most of the effect seen.
The Increasing trajectory was identified as a group with initially low LTSA, which increased during the latter part of the follow-up. However, none of the examined lifestyle or occupational factors were found associated with this trajectory. Descriptive analyses revealed that this group included somewhat more semi-professional employees, of which many are known to work within healthcare or education. During the follow-up period, there was an increase in sickness absence spells exceeding 1 week among Finnish public sector employees, particularly within the educational and healthcare sectors.29 Depression and anxiety-related sickness absence among younger employees increased during the same period, particularly among women.4 This increase in mental health related sickness absences could explain the LTSA pattern in the Increasing trajectory. Alternatively, there may be factors outside of work that contribute to the increasing LTSA. This group provides a useful focus for future research, particularly given their young age.
Considering the lack of predictors for the Increasing LTSA trajectory in the main analyses and trends in sickness absences in Finland during the study period, additional sensitivity analyses were performed to examine the association between trajectory membership and mental symptoms, which were measured by two items from the RAND-36 role limitations due to emotional distress-subscale and reflect role limitations due to affective symptoms at baseline.429,31 These items were negatively associated with the No-LTSA trajectory, and positively associated with the Decreasing and Increasing trajectories, after controlling for age and gender (online supplemental table 6). These questionnaire responses can only be regarded as proxy measures of mental health, and the results were as expected, given that mental health related LTSA dominates in this age group.4 However, this finding also indicates that a poorer role functioning due to affective symptoms among young and midlife employees with pain is associated with LTSA both short and long-term. This further suggests that detection and treatment of mental health impacts associated with pain in this group of workers could improve work outcomes.
Employees in the Decreasing trajectory were more likely to report chronic and multisite pain conditions, which are more likely to have long-term consequences for work ability.6 Pain chronicity and number of pain sites explained only some of the associations between lifestyle and occupational factors and trajectory membership, suggesting that smoking, overweight or obesity and physically heavy and manual work may indicate an elevated risk of future LTSA irrespective of whether the pain was acute or chronic, single site or multisite at baseline. Smoking and overweight are linked to a range of pain related and musculoskeletal disorders, and these modifiable lifestyle factors predicted a trajectory with more LTSA. Considering this, prevention of smoking and overweight could be targets for reducing LTSA among younger employees with pain, not forgetting the socioeconomic differences in smoking and overweight.32,34
Manual work including factors such as repetitive movements or awkward working positions, is associated with pain and manual workers report more disabling pain.10 35 Their LTSA rates are also higher compared with other occupational groups.36 37 This may partly be attributable to the higher physical workload, but also psychosocial work-related factors, limited opportunity to modify physically demanding work tasks, and poorer general health could contribute to more sickness absence among younger employees in manual occupations who have a pain condition.36 38 Psychosocial working conditions are previously known to be associated with LTSA, and the lack of association between perceived mental workload and trajectory membership was unexpected.12 Sensitivity analyses revealed that high mental workload in our study population was predominantly reported by professionals, which is a group with lower general morbidity, less physical work that pain could interfere with, and generally lower LTSA rates. One single item may have been inadequate to capture mental workload. Multiple factors are thus likely contributing to the lack of association in our data and this finding should be interpreted cautiously.
Methodological considerations
The study was based on data from a large cohort of young and early midlife employees, among whom studies on sickness absence trajectories are scarce, as previous studies have mostly focused on ageing employees or employees of all ages.15,17 Assessing pain in a survey enables capturing of all pain conditions and not only those that have led to seeking healthcare or been medically diagnosed. In the current study, having current pain was the only pain-related inclusion criterion, so that the types of pain being experienced were likely heterogenous, ranging from temporary minor aches to longstanding or recurring disabling pain conditions. By including phone respondents, we obtained a more representative sample with more men and manual workers, but this restricted the variables to the selection enquired about in the phone interviews. Validated measures of mental health and physical activity were not available for all participants, as this was not enquired in detail in the phone interview. However, we were able to perform sensitivity analyses of the association between two items from the RAND-36 role limitations due to emotional problems-subscale and LTSA-trajectories, with obtained results demonstrating an association with LTSA-trajectories among younger employees with pain. The data on pain, lifestyle- and occupational factors were cross-sectional survey data, while earlier studies have shown that repeatedly reported pain has an even stronger association with work disability.39 40 Information on LTSA was based on reliable national register data from the Finnish the Social Insurance Institution. The data included information on LTSA spells compensated by the Social Insurance Institution during specific time periods and was independent from the employer, thus also including LTSA spells occurring during employment for other employers than the City of Helsinki. Gender was reported and analysed as either man or woman, without separately considering non-binary genders. This dichotomous approach to gender could contribute to misclassification bias. The registered data could not be controlled for death, emigration or transfer to disability pension; however, death and disability retirement are relatively rare in this age group.
Conclusions
We found different LTSA patterns among young and early midlife employees with pain, of whom most did not have LTSA during the follow-up period. Chronic and multisite pain, lifestyle factors, occupation and higher physical workload were associated with a pattern with more LTSA. This knowledge could help design interventions to prevent LTSA by providing early support to young and midlife workers with pain, particularly those with chronic and multisite pain, poor lifestyle habits and physically demanding jobs.
supplementary material
Acknowledgements
A warm thanks to Jatta Salmela, PhD at the Department of Public Health, University of Helsinki, for statistical advice.
Footnotes
Funding: PF was supported by Stiftelsen Dorothea Olivia, Karl Walter och Jarl Walter Perkléns Minne. PF and TL were supported by the Social Insurance Institution of Finland (Kela) (grant 29/26/2020). OR was supported by The Juho Vainio Foundation (grant 202300041).
Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2024-085011).
Provenance and peer review: Not commissioned; externally peer reviewed.
Patient consent for publication: Not applicable.
Ethics approval: This study involves human participants and was approved by As part of the Helsinki Health Study, the observational study in the submitted manuscript did not require a separate ethical assessment, however, ethical aspects of the study have been considered in line with the Declaration of Helsinki. The Helsinki Health study’s study plan received a favorable statement from the ethical board of the Faculty of Medicine at the University of Helsinki (decision nr. 01/2017) and was approved by the City of Helsinki (updated approval in 2017, decision nr. HEL 2017-004490). The respondents have by participation in the Helsinki Health study agreed to the questionnaire data being used for scientific research purpose, and that the data is being handled in line with the data protection statement of the Helsinki Health Study, which can be accessed via the research group’s website. Only respondents who have given their informed consent to register linkage have been included in the current study. Respondents have had the possibility to withdraw from the Helsinki Health Study at any point. Participants gave informed consent to participate in the study before taking part.
Data availability free text: Data are available upon reasonable request. The Helsinki Health Study survey data cannot be made publicly available due to strict data protection laws and regulations. The data can only be used for scientific research and to the research group’s cooperation partners with a reasonable request and study plan. More information on the availability of the survey data can be inquired from the Helsinki Health Study research group (kttl-hhs@helsinki.fi). Register data cannot be shared.
Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Contributor Information
Pi Fagerlund, Email: pi.fagerlund@helsinki.fi.
Rahman Shiri, Email: rahman.shiri@ttl.fi.
Karen Walker-Bone, Email: karen.walker-bone@monash.edu.
Ossi Rahkonen, Email: ossi.rahkonen@helsinki.fi.
Tea Lallukka, Email: tea.lallukka@helsinki.fi.
Data availability statement
Data are available upon reasonable request.
References
- 1.Salonen L, Blomgren J, Laaksonen M. From long-term sickness absence to disability retirement: diagnostic and occupational class differences within the working-age Finnish population. BMC Public Health. 2020;20:1078. doi: 10.1186/s12889-020-09158-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Moberg LL. Frafall fra arbeidslivet etter langtidssykefravær. 2022. https://www.nav.no/no/nav-og-samfunn/statistikk/sykefravar-statistikk/infografikk Available.
- 3.Försäkringskassan Social insurance i figures 2022. 2022. https://www.forsakringskassan.se/download/18.405631581831c9acbfb1b0/1664347127777/social-insurance-in-figures-2022.pdf Available.
- 4.Blomgren J, Perhoniemi R. Increase in sickness absence due to mental disorders in Finland: trends by gender, age and diagnostic group in 2005-2019. Scand J Public Health. 2022;50:318–22. doi: 10.1177/1403494821993705. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Saastamoinen P, Laaksonen M, Lahelma E, et al. The effect of pain on sickness absence among middle-aged municipal employees. Occup Environ Med. 2009;66:131–6. doi: 10.1136/oem.2008.040733. [DOI] [PubMed] [Google Scholar]
- 6.Fagerlund P, Shiri R, Suur-Uski J, et al. Separate and joint associations of chronic pain, multisite pain and mental health with sickness absence among younger employees: a register based longitudinal study. Arch Public Health. 2023;81:97. doi: 10.1186/s13690-023-01115-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Fernandes R de CP, Burdorf A. Associations of multisite pain with healthcare utilization, sickness absence and restrictions at work. Int Arch Occup Environ Health. 2016;89:1039–46. doi: 10.1007/s00420-016-1141-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Virtanen M, Ervasti J, Head J, et al. Lifestyle factors and risk of sickness absence from work: a multicohort study. Lancet Public Health. 2018;3:e545–54. doi: 10.1016/S2468-2667(18)30201-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Saastamoinen P, Leino-Arjas P, Laaksonen M, et al. Socio-economic differences in the prevalence of acute, chronic and disabling chronic pain among ageing employees. Pain. 2005;114:364–71. doi: 10.1016/j.pain.2004.12.033. [DOI] [PubMed] [Google Scholar]
- 10.Fagerlund P, Salmela J, Pietiläinen O, et al. Life-course socioeconomic circumstances in acute, chronic and disabling pain among young employees: a double suffering. Scand J Public Health. 2023;51:257–67. doi: 10.1177/14034948211062314. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Sundstrup E, Andersen LL. Joint association of physical and psychosocial working conditions with risk of long-term sickness absence: Prospective cohort study with register follow-up. Scand J Public Health. 2021;49:132–40. doi: 10.1177/1403494820936423. [DOI] [PubMed] [Google Scholar]
- 12.Goorts K, Boets I, Decuman S, et al. Psychosocial determinants predicting long-term sickness absence: a register-based cohort study. J Epidemiol Community Health. 2020;74:913–8. doi: 10.1136/jech-2020-214181. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Madsen IEH, Gupta N, Budtz-Jørgensen E, et al. Physical work demands and psychosocial working conditions as predictors of musculoskeletal pain: a cohort study comparing self-reported and job exposure matrix measurements. Occup Environ Med. 2018;75:752–8. doi: 10.1136/oemed-2018-105151. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Hiilamo A, Butterworth P, Shiri R, et al. Within-individual analysis of pain and sickness absence among employees from low and high occupational classes: a record linkage study. BMJ Open. 2019;9:e026994. doi: 10.1136/bmjopen-2018-026994. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Rysstad T, Grotle M, Aasdahl L, et al. Identification and Characterisation of Trajectories of Sickness Absence Due to Musculoskeletal Pain: A 1-Year Population-based Study. J Occup Rehabil. 2023;33:277–87. doi: 10.1007/s10926-022-10070-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Hallman DM, Holtermann A, Björklund M, et al. Sick leave due to musculoskeletal pain: determinants of distinct trajectories over 1 year. Int Arch Occup Environ Health. 2019;92:1099–108. doi: 10.1007/s00420-019-01447-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Haukka E, Kaila-Kangas L, Ojajärvi A, et al. Pain in multiple sites and sickness absence trajectories: a prospective study among Finns. Pain. 2013;154:306–12. doi: 10.1016/j.pain.2012.11.003. [DOI] [PubMed] [Google Scholar]
- 18.Lallukka T, Pietiläinen O, Jäppinen S, et al. Factors associated with health survey response among young employees: a register-based study using online, mailed and telephone interview data collection methods. BMC Public Health. 2020;20:184. doi: 10.1186/s12889-020-8241-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Local Governments and County Employers Municipal sector and personnel. https://www.kt.fi/en/municipal-sector-and-personnel n.d. Available.
- 20.Treede R-D, Rief W, Barke A, et al. Chronic pain as a symptom or a disease: the IASP Classification of Chronic Pain for the International Classification of Diseases (ICD-11) Pain. 2019;160:19–27. doi: 10.1097/j.pain.0000000000001384. [DOI] [PubMed] [Google Scholar]
- 21.World Health Organization A healthy lifestyle - WHO recommendations. 2010. https://www.who.int/europe/news-room/fact-sheets/item/a-healthy-lifestyle---who-recommendations Available.
- 22.Lahti J, Lallukka T, Harkko J, et al. Working conditions and antidepressant medication use: A prospective study among 18 to 39-year-old municipal employees. Psychiatry Res. 2021;305:S0165-1781(21)00509-6. doi: 10.1016/j.psychres.2021.114213. [DOI] [PubMed] [Google Scholar]
- 23.Nagin DS. Group-based trajectory modeling: an overview. Ann Nutr Metab. 2014;65:205–10. doi: 10.1159/000360229. [DOI] [PubMed] [Google Scholar]
- 24.Nagin DS, Odgers CL. Group-based trajectory modeling in clinical research. Annu Rev Clin Psychol. 2010;6:109–38. doi: 10.1146/annurev.clinpsy.121208.131413. [DOI] [PubMed] [Google Scholar]
- 25.Jones BL, Nagin DS. A Note on a Stata Plugin for Estimating Group-based Trajectory Models. Sociol Methods Res. 2013;42:608–13. doi: 10.1177/0049124113503141. [DOI] [Google Scholar]
- 26.Mood C. Logistic Regression: Why We Cannot Do What We Think We Can Do, and What We Can Do About It. Eur Sociol Rev. 2010;26:67–82. doi: 10.1093/esr/jcp006. [DOI] [Google Scholar]
- 27.Keskimaki I, Tynkkynen L-K, Reissell E, et al. Finland: Health System Review. Health Syst Transit. 2019;21:1–166. [PubMed] [Google Scholar]
- 28.Finnish Social Insurance Institution Sickness allowance 2023. 2023. https://www.kela.fi/sickness-allowance Available.
- 29.KEVA Sairauspoissaolot kunta-alalla vuosina 2019-2022. https://www.keva.fi/contentassets/387295a047784732aae5e013b6e80038/sairauspoissaolot-kunta_alalla-vuonna-2022.pdf Available.
- 30.Aalto A-M, Aro ATJ. RAND-36 terveyteen liittyvän elämänlaadun mittarina. 1999. https://www.julkari.fi/bitstream/handle/10024/76006/Tu101.pdf?sequence=1&isAllowed Available.
- 31.Hays RD, Sherbourne CD, Mazel RM. The RAND 36-Item Health Survey 1.0. Health Econ. 1993;2:217–27. doi: 10.1002/hec.4730020305. [DOI] [PubMed] [Google Scholar]
- 32.Salmela J, Lallukka T, Kanerva N, et al. Intergenerational social mobility and body mass index trajectories - A follow-up study from Finland. SSM Popul Health. 2021;13:100723. doi: 10.1016/j.ssmph.2020.100723. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Laaksonen M, Rahkonen O, Karvonen S, et al. Socioeconomic status and smoking: analysing inequalities with multiple indicators. Eur J Public Health. 2005;15:262–9. doi: 10.1093/eurpub/cki115. [DOI] [PubMed] [Google Scholar]
- 34.Paavola M, Vartiainen E, Haukkala A. Smoking from adolescence to adulthood. Eur J Public Health. 2004;14:417–21. doi: 10.1093/eurpub/14.4.417. [DOI] [PubMed] [Google Scholar]
- 35.Andersen LL, Vinstrup J, Sundstrup E, et al. Combined ergonomic exposures and development of musculoskeletal pain in the general working population: A prospective cohort study. Scand J Work Environ Health. 2021;47:287–95. doi: 10.5271/sjweh.3954. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Piha K, Martikainen P, Rahkonen O, et al. Trends in socioeconomic differences in sickness absence among Finnish municipal employees 1990-99. Scand J Public Health. 2007;35:348–55. doi: 10.1080/14034940601160706. [DOI] [PubMed] [Google Scholar]
- 37.Pekkala J, Blomgren J, Pietiläinen O, et al. Occupational class differences in diagnostic-specific sickness absence: a register-based study in the Finnish population, 2005-2014. BMC Public Health. 2017;17:670. doi: 10.1186/s12889-017-4674-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.van Zon SKR, Bültmann U, Mendes de Leon CF, et al. Absolute and Relative Socioeconomic Health Inequalities across Age Groups. PLoS ONE. 2015;10:e0145947. doi: 10.1371/journal.pone.0145947. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Lallukka T, Hiilamo A, Oakman J, et al. Recurrent pain and work disability: a record linkage study. Int Arch Occup Environ Health. 2020;93:421–32. doi: 10.1007/s00420-019-01494-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Lallukka T, Mänty M, Cooper C, et al. Recurrent back pain during working life and exit from paid employment: a 28-year follow-up of the Whitehall II Study. Occup Environ Med. 2018;75:786–91. doi: 10.1136/oemed-2018-105202. [DOI] [PMC free article] [PubMed] [Google Scholar]