Article Text
PDF
Abstract
Objectives The aim of this prospective study was to examine the role of physical workload, leisure-time physical activity, obesity and smoking in predicting the occurrence and course of multisite musculoskeletal pain (MSP).
Methods Data on physical and psychosocial workload, lifestyle factors and MSP were based on questionnaire surveys of 385 Finnish female kitchen workers. MSP (defined as pain at three or more of seven sites) during the past 3 months was measured repeatedly at 3-month intervals over 2 years. Four different patterns (trajectories) in the course of MSP were identified. The authors analysed whether the determinants at baseline predicted the occurrence of MSP (1) at the 2-year follow-up and (2) over the total of nine measurements during the 2 years by exploiting the MSP trajectories. Logistic regression was used.
Results High physical workload at baseline was an independent predictor of MSP at the 2-year follow-up (OR 3.8, 95% CI 1.7 to 8.5) in a model allowing for age, psychosocial factors at work and lifestyle. High physical workload (OR 2.0, 95% CI 1.0 to 4.0) and moderate (OR 2.4, 95% CI 1.2 to 4.9) or low (OR 2.3, 95% CI 1.1 to 4.7) physical activity predicted persistent MSP. Obesity (OR 2.8, 95% CI 1.0 to 7.8) predicted an increased, and not being obese (OR 3.7, 95% CI 1.1 to 12.7) a decreased, prevalence of MSP in models similarly including all covariates. Smoking had no effect.
Conclusion The results emphasise the importance of high physical workload, low to moderate physical activity and obesity as potential modifiable risk factors for the occurrence and course of MSP over time.
- Physical workload
- health-related lifestyle
- multisite pain
- longitudinal
- repeated measures
- musculoskeletal
- ergonomics
- epidemiology
- longitudinal studies
- intervention studies
- back disorders
- rehabilitation
- exposure assessment
- repetitive strain injury
- cardiovascular
- disability
Statistics from Altmetric.com
- Physical workload
- health-related lifestyle
- multisite pain
- longitudinal
- repeated measures
- musculoskeletal
- ergonomics
- epidemiology
- longitudinal studies
- intervention studies
- back disorders
- rehabilitation
- exposure assessment
- repetitive strain injury
- cardiovascular
- disability
What this paper adds
Multisite MSP is common but its risk factors are largely unknown. We used two prospective analysis strategies to identify predictors of MSP in a 2-year follow-up study among female municipal kitchen workers.
High physical workload had an important role in predicting the overall occurrence of MSP at the 2-year follow-up.
Making use of nine repeated measurements of MSP that identified latent groups within the study material, we found that high physical workload and low/moderate leisure-time physical activity at baseline predicted a persistently high prevalence of MSP. Obesity predicted an increased prevalence and not being obese a decreased prevalence of MSP over the 2 years.
Attention to physical workload, exercising during leisure and obesity may be useful in the screening of high-risk groups of MSP and in allocation of preventive interventions.
Introduction
Recent population-based studies in several European countries have shown that multisite musculoskeletal pain (MSP) is more frequent than single-site pain.1–6 MSP is common also in various occupational groups, such as scaffolders,7 manual workers,8 dentists,9 industrial workers,10 kitchen workers,11 nurses, office workers and postal clerks.12 Women report more MSP than men2 4 6 and the prevalence increases with age.4 11 Consequences of MSP seem to be more severe compared with single-site pain with respect to short- and long-term sickness absence,13 14 work disability,3 15–17 functional ability18 and frequency of consultations in primary care.19
The majority of epidemiological studies to date have examined pain at a specific anatomical site.20 The current understanding is that both work-related physical and psychosocial load21 and lifestyle factors, such as smoking22 and overweight or obesity,23 play a role in the multifactorial aetiology of MSP. Evidence on the role of physical exercise is less consistent.24 There is an expressed need of longitudinal studies with repeated measurements to better understand the course of pain and to identify subgroups of individuals with a similar development, as well as the determinants of the course of pain.25–27
The importance of finding risk factors of multisite pain, thus far largely unknown, has recently been highlighted.28 29 In some cross-sectional studies of occupational samples, physical workload has been associated with the occurrence of MSP.9 12 Low physical activity during leisure-time, high body mass index (BMI) and smoking were associated with MSP in the general Norwegian population,2 but in a 14-year follow-up of the study material, lifestyle had no independent effect.30 In a previous study of kitchen workers, we found that adverse psychosocial factors at work and MSP were strongly linked together.31 Kitchen work is very strenuous mentally and physically.32 33 Based on theoretical models on multifactorial pain causation,21 our aim in the current study was to examine whether physical workload, leisure-time physical activity (LTPA), obesity and smoking predict the occurrence of MSP at the 2-year follow-up or the course of MSP during the 2 years.
Methods
Study design and sample
This 2-year prospective study is based on the data gathered in a cluster randomised controlled trial carried out during 2002–2005 in four large cities in Finland with the aim to find out the efficacy of an ergonomics intervention in preventing musculoskeletal disorders (MSDs) among 504 workers of 119 municipal kitchens. The Ethics Committee of the Finnish Institute of Occupational Health approved the study proposal.32 33
During the intervention or during the 1-year follow-up period, no systematic differences on MSDs, MSP or perceived physical workload between the intervention and control groups were found.31 32 Thus, for the further analyses regarding MSP, we pooled the data to be analysed prospectively over 2 years. Only 17 of the workers were men, and the analyses were restricted to women (n=487). Based on our inclusion criterion of trajectory analysis (see below), 385 women constituted the final sample.
Measures and data collection
All measures were based on self-reports collected by questionnaire. The researchers distributed the questionnaires at workplaces at baseline and thereafter, repeatedly at 3-month intervals, over 2 years. Data were thus collected altogether at nine time points. For those on vacation or sick leave, the questionnaires were mailed with the instruction to complete them within 1 week after returning to work. Non-responders were reminded by phone after 2 weeks. The response rates at each time point exceeded 90%.
Multisite MSP
The questionnaire included the following questions on MSP in the neck, shoulders, forearms/ hands, low back, hips, knees and ankles/feet: “Have you had…pain during the past three months (no/yes)?”. A manikin illustrated the body sites. Pain in the left and right shoulder was combined for analysis and similarly pain in the forearm and hand (if one of the answers was ‘no’ and the other missing, the response was defined as missing; if either response was ‘yes’, the combined variable was defined as ‘yes’). A sum index (0= ‘no pain’,…, 7= ‘pain in seven sites’) was calculated to assess MSP. A response to each item was required for the sum to be calculated. The sum score, dichotomised to 0–2 pain sites (no MSP) and ≥3 pain sites (MSP), was used as the outcome measure.
MSP trajectories
Figure 1 shows four MSP trajectories identified by trajectory analysis in this data set (nine time points).31 The trajectory labelled as Low was composed of the workers (n=129, 33%) among whom prevalence of MSP was constantly low, and the trajectory labelled as High of those (n=146, 37%) among whom the prevalence of MSP was constantly high over the 2-year follow-up. The Ascending trajectory (n=44, 11%) had a low initial prevalence of MSP that increased considerably towards the end of the follow-up. The Descending trajectory consisted of 66 workers (19%) and followed a declining pattern. In the current study, we examined whether the determinants at baseline predicted belonging to the High and Ascending MSP trajectories, with the Low MSP trajectory as the reference, and belonging to the Descending MSP trajectory, with the High MSP trajectory as the reference.
Multisite musculoskeletal pain trajectories (MSP, three or more pain sites) among 385 female kitchen workers. 1=Low (low prevalence of MSP), 2=Descending (decrease in prevalence of MSP), 3=Ascending (increase in prevalence of MSP) and 4=High (high prevalence of MSP).
Determinants
Perceived physical workload
The workers were asked to rate the physical strenuousness of seven main work tasks (preparation, cooking and baking, distribution and serving of food, packing food to be delivered to clients, dishwashing, cleaning and maintenance of room and equipment, receiving and storing of raw material) during the previous week. The scale ranged from 1 (‘not strenuous at all’) to 7 (‘very strenuous’). The mean of the seven work tasks was used to describe overall physical workload. Subjects were classified into tertiles with the following cut points: <3.3 (low), 3.3–4.2 (moderate) and >4.2 (high).
LTPA, BMI and smoking
The question regarding LTPA was as follows: ‘During the past 12 months, how many times a week did you exercise at least 20 min per session, to the extent to cause perspiration (‘not at all, less than once a week, once a week, 2–3 times, 4–5 times, 6–7 times’)?’ The exercise frequency was classified using tertiles into three classes: once or less a week (low), two to three times per week (moderate) and four or more times per week (high). BMI (kg/m2) was calculated based on self-reported height and weight and classified into three groups: ≤24.9 (normal), 25–29.9 (overweight) and ≥30(obese).34 Current smoking (yes/no) was asked by a question “Do you smoke regularly (daily/almost daily)?”
Covariates
Age (continuous) and MSP at baseline were included as covariates. Due to the experimental study design, the study arm (intervention/control group) was also considered as a covariate in all analyses. The intervention showed some adverse effects on psychosocial factors at work especially in two of the participating cities where organisational reforms in the foodservice took place concurrently with the intervention.33 Thus, all the models were adjusted for the occurrence of organisational reforms (yes/no).
Psychosocial factors at work were also allowed for, based on the literature and our recent results among this sample showing that psychosocial factors and MSP were strongly linked together. Eight questions from a validated questionnaire were used in the assessment31: job control (“At work, can you influence matters concerning you?”), skill discretion (“Can you use your knowledge and skills in your work?”) and supervisor support (“Does your supervisor provide support and help when needed”?) were classified similarly (1= ‘very much’,…, 5= ‘not at all’). The other items were: job satisfaction (“How satisfied are you with your present work?” 1= ‘very satisfied,…, 5= ‘very dissatisfied’), co-worker relationships (“How do workmates get along at your workplace?” 1= ‘very well’,…, 5= ‘badly’), hurry at work (“Do you have to hurry to get your work done?” 1= ‘never’,…, 5= ‘constantly’), mental strenuousness of work (“Is your work mentally strenuous?” 1= ‘not at all’,…, 5= ‘very strenuous’) and difficult work phases (“Does your work have phases that are too difficult?” 1= ‘never’,…, 5= ‘constantly’). A sum score was constructed (Cronbach's α=0.76) to describe adverse psychosocial work environment and categorised into tertiles: ≤17 (low), 18–20 (intermediate) and ≥21 (high).
Statistical analyses
We used two different approaches to examine the relationships of physical workload, LTPA, BMI and smoking with MSP.
First, we studied whether these determinants at baseline predict the overall occurrence of MSP at the 2-year follow-up (at single time point). Since there were missing values in the outcome (n=96, 25%), this analysis was carried out among 289 workers. The effects of physical workload and lifestyle factors were studied separately and adjusted for age (years) and the baseline level of MSP. In the final model, in addition to the former, also psychosocial factors at work were included.
Second, the measurements at nine time points over 2 years enabled us to study the role of the determinants in predicting the course of MSP over the 2-year follow-up using trajectory analysis (see below). By contrasting the MSP trajectories (figure 1), we studied whether the determinants at baseline predicted the persistence (High vs Low), an increase (Ascending vs Low) and a decrease (Descending vs High) of MSP prevalence.
Logistic regression models with ORs and 95% CIs were used to estimate the effects. In the analyses, all models were adjusted also for study arm (intervention/control) and organisational reforms (yes/no). Statistical analyses were performed using V.18.0 of the SPSS software.
The MSP trajectories (figure 1) were identified by using a semiparametric group-based modelling strategy by PROC TRAJ in SAS version 9.2.35 36 This method identifies different latent groups (trajectories) which tend to have a similar profile over time. All available data points during a given period are used. Trajectory analysis accommodates missing data, but for individuals with very incomplete histories exclusion from analysis is a practical necessity.36 We set as a criterion for inclusion that data from at least four time points of the nine were available. Based on this, our final sample comprised of 385 women.
Selection of the optimal model and number of trajectories and their shape is based on the Bayesian information criterion. According to posterior membership probability (ideally near 1, at least ≥0.70), subjects are assigned to the group to which they have the highest probability to belong.36 In the analysis, the best fit was a four-group model including two trajectories with only the intercept (no change over time) and two with a linear shape (change over time, figure 1). The mean trajectory group assignment probabilities varied between 0.69 and 0.92, indicating that there was relatively little misclassification. A sensitivity analysis using four or more pain sites as the cut-off point for MSP yielded similar results regarding the number and shape of the trajectories.31
Results
At baseline, single-site pain during the past 3 months was reported by 14%, pain in two body sites by 19% and pain in at least three sites by 53% of the workers. Table 1 describes the distribution of the studied determinants and covariates in the total study sample at baseline and in the MSP trajectories. Among the total sample, 30% of the workers reported that work was physically highly strenuous. About half of the workers were overweight or obese, and a quarter were regular smokers. Only 22% reported exercising at least four times per week.
Distribution of perceived physical workload, leisure-time physical activity (LTPA), body mass index (BMI, kg/m2), smoking and covariates in the total sample of female kitchen workers at baseline and in multisite musculoskeletal pain (MSP, three or more pain sites) trajectories
The workers in the Low MSP trajectory were a little younger, perceived their work physically lighter and were more satisfied with their psychosocial work environment compared with the workers in the other MSP trajectories.
Determinants at baseline in relation to MSP at the 2-year follow-up
MSP had a high persistence rate with 84% of the workers with MSP at baseline reporting it also at the 2-year follow-up. MSP at baseline was thus a strong predictor for MSP at follow-up (OR 9.8, 95% CI 5.4 to 17.7) when adjusted for age, study arm and organisational reforms (table 2). In this model, those who perceived their physical workload as high at baseline had a clearly increased risk (OR 4.6, 95% CI 2.2 to 9.7) for MSP at the 2-year follow-up, and the independent effect remained in the final model (OR 3.8, 95% CI 1.7 to 8.5). The risk connected to low LTPA was increased but not statistically significant. Neither BMI nor smoking predicted the occurrence of MSP at follow-up.
Perceived physical workload, leisure-time physical activity (LTPA), body mass index (BMI, kg/m2) and smoking at baseline in relation to multisite musculoskeletal pain (MSP, three or more pain sites) at the 2-year follow-up among female kitchen workers (n=289)
Age and the intermediate (OR 2.9, 95% CI 1.4 to 6.0) and high (OR 2.2, 95% CI 1.0 to 4.5) levels of adverse psychosocial factors at work at baseline also increased the risk for MSP at follow-up.
Determinants at baseline in relation to MSP trajectories
With the Low MSP trajectory as reference, moderate or high physical workload (ORs 1.9 and 3.1) predicted belonging to the High trajectory of MSP when only age, study arm and organisational reforms were adjusted for (table 3). Low or moderate LTPA (ORs 2.2 and 2.3) predicted belonging to the High trajectory of MSP when age, BMI and smoking were included in the model (table 3). The effects of high physical workload (OR 2.0, 95% CI 1.0 to 4.0) and low or moderate LTPA (ORs 2.3 and 2.4) remained after adjustment for all covariates.
Perceived physical workload, leisure-time physical activity (LTPA), body mass index (BMI, kg/m2) and smoking in relation to trajectories of multisite musculoskeletal pain (MSP, three or more pain sites) among female kitchen workers (n=385)
When the Ascending trajectory was contrasted with the Low trajectory of MSP in age-adjusted analysis, we found that high physical workload (OR 2.5, 95% CI 1.1 to 6.0) predicted belonging to the former group (table 3). However, when also psychosocial factors at work and the lifestyle variables were included in the model, the estimate for high physical workload attenuated (OR 1.9, 95% CI 0.7 to 5.0). Obesity was a predictor of membership in the Ascending trajectory when age, LTPA and smoking were included in the model (OR 2.5, 95% CI 1.0 to 6.4). The relationship persisted when also physical workload and psychosocial factors at work were allowed for (OR 2.8, 95% CI 1.0 to 7.8).
In the last analyses, the Descending trajectory was contrasted with the High trajectory of MSP. Low physical workload predicted a decrease in MSP prevalence in age-adjusted analysis (OR 2.2, 95% CI 1.0 to 4.5), but the effect attenuated when psychosocial factors and lifestyle variables were considered (table 3). Not being obese predicted a decrease in MSP and this effect remained after adjustment for all covariates (OR 3.7, 95% CI 1.1 to 12.7 for overweight; OR 2.9, 95% CI 0.9 to 9.6 for normal weight, table 3).
Increasing age was predictive of MSP in all analyses, except in the last trajectory contrast (Descending vs High). Adverse psychosocial factors at work increased the risk of belonging to the High (ORs 2.0–4.0) and Ascending (ORs 2.9–3.1) trajectories of MSP when contrasted with the Low MSP trajectory.
Discussion
We found among female kitchen workers that high physical workload independently predicted the occurrence of MSP in a 2-year follow-up. High physical workload also predicted a persistent course of MSP over time. On the other hand, low and moderate LTPA at baseline was connected with a persistently high prevalence of MSP. Of the lifestyle factors, also obesity, but not smoking, had a role as a risk factor for MSP. Obesity predicted an increased prevalence and not being obese a decreased prevalence of MSP.
MSP was very persistent over time. MSP at baseline was the strongest determinant of all for the occurrence of MSP at the 2-year follow-up. The trajectory with a constantly high MSP prevalence was the largest group. Similarly, in the Norwegian general population, the baseline MSP was the most important predictor in a 14-year follow-up study.30
We know of no previous study that has longitudinally examined the role of physical workload in relation to MSP. Our results are, however, in accordance with some earlier reports with a cross-sectional design. In a study of Greek dentists,9 high perceived physical workload was associated with MSP, ORs increasing from 2.5 for two and 3.1 for three to 4.4 for four pain complaints. Similarly, among Greek nurses, office workers and postal clerks,12 an increase in a self-reported physical load score was linearly associated with the number of pain sites affected. The score composed of heavy lifting, working with hands above the shoulder level, repeated bending and straightening of the elbow, repeated wrist–hand movements and kneeling, squatting or climbing stairs. These physical work demands are comparable to those in kitchen work, and equally, their sample consisted mostly of women.
In previous prospective studies,37–39 the above-mentioned physical risk factors and pushing/pulling heavy weights37 increased the risk of chronic widespread pain. Chronic widespread pain (present at least 3 months in the axial skeleton, above and below the waist, on the left and right side of body) has long dominated research, although its prevalence is lower than that of MSP.5 A less restrictive approach to pain in multiple sites has been encouraged.1 Especially the number of pain sites was recently observed to be a stronger predictor for functioning than location or chronicity of pain.18
Our subjects were mostly workers of kitchens of schools, nurseries and nursing homes, an average four workers working per kitchen. Regardless of the type of kitchen, the work is very demanding both physically and mentally. The workers work under time pressure and perform various parallel tasks, many of which include exposure to a combination of physical risk factors of MSDs. The high occurrence of MSP could be connected with the pattern of physical workload that is rather uniformly distributed on the musculoskeletal system.32 In our analyses, psychosocial factors at work showed even higher risk estimates than physical workload or lifestyle factors with regard to MSP. This has been often the case in previous studies on MSP12 or chronic widespread pain37–39 as well. It may be that in small work communities—such as the kitchens studied here—where work is done in close co-operation, there is an emphasis on the importance of good social relationships for well-being.31
Earlier findings regarding lifestyle factors in relation to MSP have been inconsistent. In the Norwegian general population, smoking, BMI and low LTPA were associated with MSP at baseline.2 In a 14-year follow-up of the study material, however, lifestyle factors had no independent effect, when the baseline MSP was included in the model.30 In a study of working-aged English adults,6 smoking was associated with the occurrence of pain at multiple sites. Among adolescents in the Northern Finland Birth Cohort, smoking and a high physical activity level in both genders and overweight, although not obesity in girls, were associated with pain in three to four sites.40 In a 2-year follow-up of a subsample of this birth cohort, persistent pain in at least two sites was predicted by smoking and high physical activity among girls.41 In our study, opposite to this, smoking had no effect on MSP, whereas less than high LTPA predicted persistent MSP. Our finding is in line with a large Norwegian population-based study where physical inactivity at baseline was associated with chronic widespread musculoskeletal complaints 11 years later.42
The relationship between LTPA and MSP is likely complex, since symptoms may be a cause as well as a consequence of inactivity. Our analyses found support for low and even moderate physical activity being an antecedent of a persistent course of MSP. This association persisted when physical workload was allowed for. It is intriguing that physical activity at work increased and physical activity during leisure-time decreased the probability of belonging to the group with a persistent course of MSP. This paradox has been pointed out before, for example, with regard to long-term sickness absence43 and physical functioning.44 It is conceivable that differences in the type, timing, duration, intensity and control of activity contribute to the different effects.
We found that obese workers had a higher probability of belonging to a group with an increasing prevalence of MSP compared with the constantly low-prevalence group. Moreover, not being obese predicted belonging to a group with a decreasing prevalence of MSP when compared with the constantly high-prevalence group. This supports the hypothesis that obesity could be causally linked with MSP. On the other hand, no relationship between obesity and the overall level of MSP in the 2-year follow-up was seen (table 2). The number of subjects belonging to the ascending/descending trajectories was small, however. The relationships revealed in the subgroup analyses were perhaps not strong enough to influence the overall result.
Two types of explanations for the development and persistence of MSP have been presented. First, risk factors shared by different pain sites, such as injury, mechanical overuse, low physical activity or obesity, may cumulatively lead to MSP. Second, MSP may develop as a result of a strong association of chronic MSP with psychological factors, like somatisation, depression and anxiety.29 40 Unfortunately, we lack for information regarding these specific individual psychological characteristics and were not able to consider these in the analyses.
However, adverse psychosocial factors at work, high physical workload, low LTPA and obesity had all some role in predicting MSP. These have been often mentioned as risk factors of site-specific pain conditions. Thus, our results can be seen as supporting the notion that single-site pain and MSP share risk factors.
Methodological considerations
Among the strengths of our study were the longitudinal design and the use of two different analytical methods. We studied the effects of the determinants measured at baseline in predicting MSP at a single time point 2 years later. The other approach was to use repeated measurements that identified latent groups with a different course of MSP and to find out the role of the determinants on that course. The approaches led to different numbers of subjects in the groups to be compared. Previously, Dunn et al25 have used latent class analysis to characterise the course of low back pain and to identify trajectories of back pain and some other pain disorders.26
The use of only two time points may be inadequate in terms of capturing the often recurrent and fluctuating nature of pain. Trajectories better describe the course of pain and identify subgroups among the sample with divergent patterns.25 26 We found that 70% of the workers belonged to a trajectory with a constantly stable (high or low) prevalence of MSP but that also changes (increase or decrease) in the occurrence of MSP occurred. These findings emphasise the need to find out possible underlying factors for the different courses of MSP.
A limitation of our study was that we considered only women in a single occupational group. Hence, the generalisability of the results is limited and probably can only be extended to other occupations among women with a high physical and mental workload and small work communities. On the other hand, the homogeneity of the sample might have been an advantage for revealing associations. Although MSP is more common in women, it occurs also in men2 4 5 and the risk factors may differ according to gender. Analyses among men would be worth of further study.
Our sample was relatively small for the assessment of risk, increasing imprecision of the estimates specifically in the contrast of Ascending versus Low MSP trajectories. Due to missing values in the outcome at the 2-year follow-up, the number of subjects included in the two time point analysis was 289. However, that sample encompassed 75% of the original sample of 385 workers. The distribution of the determinants and covariates at baseline were similar in both of these samples (n=289 and n=385). We did the corresponding analyses also among the maximal sample of 487 workers and the results did not change. This indicates that selection bias did not threaten the validity of our results. The number of missing values in the determinants at baseline was small, at highest 2%.
The cut-off point for MSP is not well established. The mean number of pain sites was three. Thus, we dichotomised MSP and used three or more pain sites as a cut-off point. A sensitivity analysis with four or more pain sites as a cut-off point did not change the results with respect to the number and shape of MSP trajectories.
Pain and psychosocial factors at work were assessed by questions from validated questionnaires.31 The question regarding the physical workload was constructed specifically for kitchen work since no suitable published methods for the purpose were available. The use of seven categories was based on previous findings suggesting that the minimum number of categories should be between five to seven and that up to seven the reliability of the measure increases.45 Our measures of the exposures were based on self-reports due to the lack of more objective methods practicable in epidemiological studies taking also financial constraints into account. The common method may have produced some shared variance in the reporting of work exposures and pain, in this case possibly overestimation, since the presence of pain could have increased the workers' awareness of physical and psychosocial conditions at work.46 However, the prospective results on the relationship between physical workload and MSP are not likely to be substantially influenced by this. We examined the association of physical workload at baseline with MSP at follow-up, allowing for MSP at baseline and with trajectories of MSP. The results obtained by these two different types of analyses supported each other. We additionally checked whether MSP at baseline predicted the reporting of physical workload at the 2-year follow-up but found no association. This differs from the previously reported results on reciprocal associations between psychosocial factors and MSP.31 Thus, in our interpretation, high physical workload may be causally connected with MSP.
A limitation of the material is the crude measure of smoking as we lacked information to calculate, for example, pack-years. This may have diluted possible associations. Also BMI was based on self-reported weight and height. People tend to underestimate their weight and overestimate their height. However, it has been argued that self-reported height and weight are valid for identifying relationships in epidemiological studies.47
In conclusion, our results underline the importance of considering physical workload as a determinant of MSP. The study also adds to the rather sparse literature on lifestyle factors in relation to MSP in showing that the insufficient LTPA and obesity may have a role as predictors of the course of MSP. These modifiable factors should be considered when planning preventive interventions on MSP.
Acknowledgments
The authors thank other members of the ERGO research group who participated in execution of the cluster randomised controlled trial among municipal kitchen workers, to which the current study was based on. The Academy of Finland (Health Promotion Research Programme), the Finnish Work Environment Fund, the Ministry of Labour and the Local Government Pensions Institution financially supported that core study.
References
Footnotes
Funding This study was supported by the Finnish Cultural Foundation.
Competing interests None.
Patient consent Obtained.
Ethics approval This study was conducted with the approval of the Ethics Committee of the Finnish Institute of Occupational Health.
Provenance and peer review Not commissioned; externally peer reviewed.