Joint association of physical work demands and leg pain intensity for work limitations due to pain in senior workers: cross-sectional study

The present cross-sectional study regarding work-limiting pain and physical work demands bases upon data from the first wave of the SeniorWorkingLife study [42], which has been registered as a cohort study in ClinicalTrials.​gov (Identifier: NCT03634410). SeniorWorkingLife is a comprehensive questionnaire survey assessing work environment and health among senior workers in Denmark (≥50 years) [42]. The SeniorWorkingLife baseline questionnaire survey was sent out between July and October 2018. Specific questions used in the present analysis are specified below. The reporting of the study followed the guidelines for the reporting of observational studies in epidemiology (STROBE) [43].

Statistics Denmark drew a probability sample of a total of 30,000 Danes ≥50 years (18,000 in employment, 7000 unemployed, 3000 on voluntary early retirement, 2000 on disability pension), who were invited to participate in the questionnaire survey by a personal link via a web-based digital mailbox linked to each worker’s social security number (‘E-boks’) [42]. Subsequently, the survey data were merged with high-quality national registers provided by Statistics Denmark. We only included senior workers who confirmed on the questionnaire that they were employed at the time of the reply. The response rate for completing all questions in the questionnaire was 56% among currently employed workers, but in the present analyses, we also included those who did not fill out all the questions in the questionnaire (n = 11,786). Granted that not all participants filled in all survey questions, the exact number of participants for each analysis varies. Data from the SeniorWorkingLife study regarding the joint association of low-back pain and physical work demands with work limitations have previously been published [44].

Danish law permits scientific usage of questionnaire- and register data without collecting informed consent or applying for approval by ethical and scientific committees [45]. Statistics Denmark were responsible for depersonalizing and storing all data on their servers from where the researchers performed the analyses through remote access.

We used the following modified version of the Standardised Nordic Questionnaire for Musculoskeletal Symptoms [48] to assess work-limiting musculoskeletal pain: ‘To which degree did the pain limit you in your work during the last 3 months?’. The response options were 1) to a very high degree, 2) to a high degree, 3) to some degree, 4) to a small degree and 5) not at all. As done in our previous publications [16, 44], these response options were subsequently dichotomized to 1) not at all to a small degree and 2) some degree to a very high degree.

We controlled for the following potential covariates: age, sex, BMI, smoking status, educational level, psychosocial work factors as well as musculoskeletal pain in the other body regions. Age (years), body mass index (BMI) (kg/m²), psychosocial work environment (0–100, specified below), and musculoskeletal pain during the past three months in the low-back, neck/shoulders, and arms (0–10, see above) were assessed on continuous scales, whereas sex (‘male’ or ‘female’), educational level (see below), smoking status (‘No, never’, ‘Ex-smoker’, ‘Yes, but not every day’ and ‘Yes, every day’), and physical activity during leisure (see definitions below) were categorical variables.

Leisure-time physical activity level was assessed by replying to the following question: ‘How would you describe your physical activity level during leisure for the last 12 months?’. Respondents were given the following response options 1) ‘Mostly sedentary’, 2) ‘Light exercise at least 4 h a week’, 3) ‘Sports or heavy physical activity at least 4 h per week’ and 4) ‘Training and competing regularly and several times a week’ [49]. Psychosocial work factors, i.e. influence at work and recognition from colleagues, were assessed on a continuous scale from 0 to 100 (with 100 being best) by specific questions from the comprehensive and validated Copenhagen Psychosocial Questionnaire [50]. Thus, the greater the score, the higher/better the collegial recognition and influence at work.

Highest attained educational level was drawn from the linked register: 1) Primary school or unknown, 2) High school, 3) Short-term higher education, 4) Medium-term higher education, and 5) Long-term higher education.

We controlled for these variables because previous studies have shown associations between work ability and sickness absence and these occupational and lifestyle factors, i.e., age [28, 29], sex [28, 29], smoking [28], overweight [51], leisure-time physical activity [49], MSD [7‐9] and psychosocial work environment [29].

The statistical analyses were carried out and controlled for potential covariates in the SAS statistical software for Windows (Proc Glimmix, SAS version 9.4, SAS Institute, Cary, NC). We applied two different statistical models. Statistical model 1 was adjusted for age, sex, and pain in the other body regions, whereas the fully adjusted model 2 was additionally adjusted for smoking status, body mass index (BMI), psychosocial work factors, educational level, and physical activity level during leisure. In order to make estimates representative for the population of senior workers, model-assisted statistical weights were applied in all analyses. Weights included sex, age, origin, highest completed education, occupational industry, as well as family income and type.

We performed logistic regression analyses of the associations between work-limiting pain (dichotomous outcome variable) and physical work demands (exposure variables), leg pain intensity, and the interaction between physical work demands and leg pain intensity. Due to the presence of a statistically significant interaction, the results are reported stratified by leg pain intensity. Results are presented as odds ratios (OR) and 95% confidence intervals (CI) unless otherwise stated, with alpha levels below 0.05 considered statistically significant differences.

The weighted prevalences of work-limiting pain by physical work demands and leg pain intensity among the whole sample are provided in Fig. 1, while the associations between physical work demands and work-limiting pain stratified by sex and leg pain intensity are presented in Table 2.

Overall, we observed a significant dose-response association (p < 0.0001) between leg pain intensity and work-limiting pain among senior workers. We also observed a significant exposure-response association between physical work demands and work-limiting pain (p < 0.0001). In addition, a significant interaction (p < 0.001) between pain intensity and physical work demands for work-limiting pain existed, and the results are therefore stratified by leg pain intensity.

As it appears, the weighted prevalences of work-limiting pain generally increase with physical work demands and leg pain intensity, with a weighted prevalence of approximately 68% among senior workers with very high leg pain intensity (≥7/10) in occupations characterized by ‘heavy or fast work that is physically demanding’ (Fig. 1). As an example based on statistical model 1, the odds of work-limiting pain was significantly increased among senior workers with ‘mostly standing and walking work that otherwise is not physically demanding’ and moderate leg pain (pain intensity: 3–4/10) compared to senior workers having ‘mostly sedentary work that is not physically demanding’ (OR: 1.66, 95% CI: 1.58–1.76), and even higher odds among senior workers with ‘heavy or fast work that is physically demanding’ and moderate leg pain (OR: 3.82, 95% CI: 3.58–4.08). While the sex-stratified odds ratios follow the same aforementioned pattern (increased leg pain intensity and physical work demands being associated with increased odds of work-limiting pain) between sexes, women seem at higher odds for work-limitations due to pain compared to men, especially women with heavy or fast work.

Low-back and neck-shoulder pain remain highly prevalent and debilitating pain regions globally [1] and have accordingly received much scientific and work environmental rehabilitative attention [3]. However, in our study among senior workers, we found equally high prevalences of musculoskeletal pain in the low-back, neck-shoulders, and legs of 35, 37, and 36%, respectively. Of the entire sample, 11% reported doctor-diagnosed osteoarthritis, which may explain some of the high leg pain prevalence, although not all cases. Overall, this underscores that the societal impact of lower body leg pain – specifically among senior workers – should not be underrated, especially not considering that lower body knee pain, specifically, has been associated with long-term sickness absence at a proportionately lower pain intensity threshold than low-back and neck-shoulder pain [14].

For the entire study sample of senior workers, we found a significant dose-response association (p < 0.0001) between leg pain intensity and work-limiting pain. Our results thus elaborate on previous findings demonstrating the importance of not just the presence of pain but also the severity of pain and functional limitation for long-term sickness absence and the ability to work [8, 9, 14, 40]. For instance, we have previously demonstrated strong dose-response associations of increasing pain intensities in the low-back, neck-shoulders, and knees with increased risk of long-term sickness absence in healthcare workers [14]. In accordance, Hallman and colleagues reported that sick leave increased and work ability decreased across all longitudinal trajectory classes of neck-shoulder pain (i.e. from low, moderate, strong fluctuating to severe persistent pain) [9]. From a rehabilitation perspective, it seems that even a small decrease in leg pain intensity could lead to a parallel decrease in work limitation due to pain. This indicates that senior workers do not have to be pain-free in order to reduce work limitations due to pain. Previous research has shown that strength training at the workplace is effective for reducing pain intensity [3, 52]. Even though strength training does not completely eliminate the pain, it reduces pain intensity with moderate to large effect sizes, making it a potent tool for workplace rehabilitation of senior workers with work limitations due to pain.

In the present study, we also demonstrated a significant exposure-response association between physical work demands and work-limiting pain among senior workers (p < 0.0001). This finding is in accordance with previous studies reporting a higher risk of low work ability [10, 29, 37] as well as sickness absence [33‐36, 53] and disability pension [54] among workers with MSD and high physical work demands, which further emphasizes the practical consequences of MSD.

However, and importantly, our analysis demonstrated a strong significant interaction between physical work demands (p < 0.001) and pain intensity, indicating a combined and interacting effect of physical work demands and leg pain intensity on work-limiting pain.

Several studies have reported compromised work ability and productivity (presenteeism) among workers with MSD having manual jobs characterized by higher physical work demands as compared to workers with MSD in more sedentary jobs [10, 29, 37‐40]. A Dutch study reported that high physical workload (including high exposures of heavy lifting, awkward and static postures) compared to low physical workload was associated with low work ability among workers with physician-diagnosed MSD [37], whereas Pensola and coworkers demonstrated that having jobs with predominantly ‘strenuous’ work tasks was associated with lower work ability compared to jobs with ‘moderately heavy’ or ‘light’ work tasks among workers with multi-site pain (including lower extremity pain) [29]. More specifically, semi-manual and manual labour was found to be significant risk factors for work-limitations (presenteeism) among workers with chronic knee pain [39]. In opposition, Wilkie and colleagues did not report manual labour as a risk factor for developing work restriction among older workers (50–59 years) with lower body MSD [40].

With regards to sickness absence, manual labour has been associated with an increased risk of sickness absence among workers with MSD [33‐36]. Haukka demonstrated that having a physically light job with no exposure to lifting, carrying, kneeling, squatting and other physical work demands was associated with a lower risk of sickness absence among Finns with multi-site MSD [33]. More specifically, lower body pain due to hip and especially knee osteoarthritis has been associated with an increased risk of sickness absence in manual workers compared to non-manual workers [35, 36]. Thus, these studies together with our data could indicate larger practical consequences of having MSD among manual workers than among sedentary workers, which should call for differentiated action between working groups.

While our sex-stratified analyses show similar patterns between sexes, women generally seem at higher odds of work-limiting pain. At a similar pain intensity, women with high physical work demands (heavy or fast work) seem at particularly high risk of work-limitiations due to pain (ORs of 5.49, 3.20, and 5.63 in women with moderate, high, and very high pain, respectively) compared to men with equally high physical work demands (ORs of 2.64, 2.96, 2.97, respectively). High physical work demands (by job title) have previously been associated with increased risk of sick leave and disability pension among both women and men with knee osteoarthritis [35]. As pertains to sex, arthritis-attributable work limitations have been reported to be more prevalent among women than men [55]. Likewise, sick leave and disability pension are also more prevelant among women with knee osteoarthritis than men with knee osteoarthritis [56]. Further, in a previous study pertaining work factors facilitating working beyond state pension age, women tended to be more affected by high physical work demands [57]. In opposition, the female gender (as compared to male) has been associated with good work ability among workers with multi-site pain [29], whereas other studies have reported no gender differences in the odds for work restrictions, absenteeism and presenteeism as well as risk of sick leave among workers with lower limb joint pain, chronic knee pain and knee and osteoarthritis, respectively [36, 39, 40].

Thus, both reducing the physical work demands and pain intensity may be targets that are worth pursuing by work environment professionals. As mentioned above, workplace-based exercise in terms of strength training is a well-documented beneficial means to reduce MSD pain intensity [2, 45] that may be particularly effective in terms of reducing work-limitation among senior workers with manual jobs.

Another viable strategy may be tailoring the physical work demands (e.g. reducing) to the capacities and age of the worker with MSD [58]. As example, Haukka reported that the possibility to sufficiently adjust the length of the working day was a protective factor for sickness absence in workers with MSD [33]. In the Nordic countries, however, it currently seems that older manual workers experience similar or even higher occupational physical demands compared to younger workers [58, 59]. In addition, we have previously reported a strong labour market inequality in opportunities offered by the workplace for supporting a long and healthy worklife, whereby senior workers in manual labour are offered fewer opportunities such as exercise training during work compared to sedentary senior workers [60]. This poses a further threat to prolonging worklife among senior manual workers [47] and needs to be further addressed in future studies.

The present study contains several strengths and limitations. It is a clear strength of the study that the invitation of potential participants was based on a probability sample drawn by Statistics Denmark among all eligible Danish residents ≥50 years. Further, all analyses were performed by employing statistical weights based on national registers, which ensured that the data was representative of senior workers in Denmark, and reduced the potential influence of non-response bias. The large and representative sample strengthens the statistical power and reduces the chances of statistical type II errors. A limitation of the study is the cross-sectional design, which does not allow for causal inferences and entails the risk of reverse causation.

The step-wise adjustment for potential covariates in the statistical models only changed the odds-estimates to a very small extent. Thus, it seems that educational attainment, physical exercise, smoking, BMI, and psychosocial work factors are less important factors for the interplay between pain intensity and work-limiting pain. However, other studies show that both overweight and smoking may be work-limiting in terms of increasing the risk of sickness absence [61].

The presented results may also be biased by ‘the healthy worker effect’ as we only included senior workers still capable of working and excluded those that have left the labor market prematurely or changed to a less physically demanding job due to health problems such as MSD. Thus, it cannot be ruled out that individuals with the longest history of physically and mentally demanding work could already be outside the labour market or in another job when the SeniorWorkingLife study was initiated. Hence, we consider our odds estimates to be relatively conservative and may thus underestimate the actual odds estimates. We used self-reports to determine physical activity during work, which may have introduced self-report bias and inaccuracy [62], and hence represent a limitation to the study. However, we have previously demonstrated strong agreement between the specific research item used regarding physical work characteristics and grouping based on ISCO (International Standard Classification of Occupations 1), where respondents are stratified into occupational groups based on high-quality national registers [47]. Additionally, self-reporting can result in common method variance where the answers may be influenced by e.g. the respondents mood, health status and socioeconomic status [63]. Furthermore, different time frames between variables (leg pain intensity, physical activity in leisure, current physical work characteristics etc) also constitute a limitation of the study. Future studies investigating the joint association of physical work demands and pain intensity with work-limiting pain may employ other methods such as technical measurements of physical activity during a working day or using job titles.