Introduction
To tackle work-related musculoskeletal problems due to lifting, the Danish Work environment Authority currently uses a model for supervision that primarily focuses on the weight of the load and the perpendicular distance from the center of gravity. The model also takes several other factors into account such as duration, frequency and the shape of the load. While the model in many respects is purposeful shortcomings are also present. One shortcoming is that the model does not take into account individual variations in physical capacity. Obviously, all else equal, a strong person will, in relative terms, use less muscle force to manually lift 20 kg and experience less physical exertion while performing the lift than a weaker person. Accordingly, perceived physical exertion seems to reflect the balance between physical work demands and physical capacity of the individual. Perceived exertion during work is a risk factor for development of musculoskeletal disorders (Andersen et al.
2012a) and sickness absence in certain occupations (Andersen et al.
2012b).
Studies performed in controlled laboratory settings have demonstrated a close relationship between perceived physical exertion and work demands expressed as a percentage of the individual physical capacity. This has been observed both in terms of objectively assessed cardiovascular (Borg
1982; Scherr et al.
2012) and muscular work load (Andersen et al.
2010; Fontes et al.
2010). These observations are interesting because they imply that ratings of perceived exertion could be used to further qualify the existing lifting model. However, only few studies have examined the associations between perceived physical exertion and relative physical workload in workplace settings (Balogh et al.
2004; Village et al.
2005). These workplace studies showed show weak associations between physical exertion and relative physical workload, showing that results from controlled laboratory settings may not necessarily be transferred to the workplace.
There are potentially several reasons for the observed discrepancies between laboratory studies and workplace studies. For example, heart rate monitoring of cardiovascular load provide information on whole body energy expenditure, whereas muscular workload measured by electromyography is associated with local muscular workload. Accordingly, a strenuous workday with separated bursts of high-intensity muscle contractions may have little effect on the average cardiovascular load, but still have significant impact on the perceived physical exertion due to local muscular exertion. Furthermore, hours of repetitive low-to-moderate intensity muscle contractions may affect the average cardiovascular load as well as the perceived physical exertion. Hence, the relative load and the resulting perceived exertion seems highly dependent on the type, intensity and duration of the work performed. As a consequence, the duration and differential tasks of a workday, as opposed to the controlled laboratory tasks, may compromise the association between the objectively assessed workload (cardiovascular and muscular) and the perceived exertion.
Furthermore, as direct measurements are very resource demanding the number of measurements is often reduced to a minimum that may compromise the validity of the measurement. As opposed to direct measurements self-reports of perceived exertion can be applied for a low-cost making it a relevant tool for large workplace surveys. In any event, if perceived exertion in relation to lifting is to be used as a general practical tool in ergonomic supervision situations there is a need to improve the knowledge base by collecting and analyzing individual reports of perceived physical exertion and objectively assessed physical workload (muscular and cardiovascular) during actual work. For this reason, the present study was designed to investigate the association between perceived exertion and objectively assessed muscular and cardiovascular load during a full working day among a broad category of blue-collar workers with lifting tasks.
Discussion
The present study investigated the association between perceived exertion and objectively assessed muscular and cardiovascular load during a full working day in jobs characterized by lifting tasks. Logistic and linear multiple regression analysis demonstrated significant associations between perceived exertion and objectively assessed cardiovascular and muscular load. Particularly, muscular load measured as the percentage of the working day with high neck/shoulder muscle activity was associated with perceived exertion. The logistic regression analysis indicates that a threshold exists when experiencing high (Borg ≥ 4) perceived exertion.
Our observations show that many participants reported a moderate perceived exertion during work. These reported levels of perceived exertion are in fair agreement with the reported levels of perceived exertion among Danish blue-collar workers that were engaged in heavy construction work (Persson et al.
2006). Accordingly, the participants seem to be fairly good representatives for people in strenuous work. However, it is also clear that the perceived exertion ratings are skewed to the lower end (cf. Figure
2).
When dividing the participants into three perceived exertion groups that grouped the answers in relation to the meaning of the underlying response scale (“low”, “moderate” and “high”), we observed significant age and gender-adjusted group differences in objectively assessed cardiovascular and muscular load. Accordingly, the “high” perceived exertion group demonstrated longer periods with moderate and high muscle activity (% of work time > 40 and 60 % max EMG, respectively) and greater cardiovascular load compared to the “low” and “moderate” group. Furthermore, participants experiencing “low” perceived exertion demonstrated longer periods with inactivity in their leg muscles compared with the “high” perceived exertion group. In support of this, the multiple linear regression analysis demonstrated significant, but weak associations between muscle activity (high activity in the neck/shoulder and inactivity in the legs) and perceived exertion and cardiovascular load and perceived exertion assessed during the workday. However, in the fully adjusted linear model including both muscle activity and cardiovascular load, only the muscle activity of the neck/shoulder muscles remained significantly related to perceived exertion. Correspondingly, when dichotomizing the Borg scale into “low or moderate” (Borg < 4) and “high” (Borg ≥ 4) perceived exertion, the logistic regression analysis demonstrated strong association between perceived exertion and high neck/shoulder muscle activity. Notably, this indicates that the associations are non-linear and a threshold exists when experiencing high (Borg ≥ 4) perceived exertion.
On average, the neck muscles were exposed to longer periods of high muscle activation compared to leg and lower back muscles during the workday. As implied by the types of jobs investigated (cf. Table
2), this indicates that the majority of tasks were performed either seated or standing fully erect at an assembly line or table lifting boxes, handling meat, etc. Accordingly, this supports the strong association between high neck muscle activity and perceived exertion and the lack of association between lower back and leg muscle activity and perceived exertion. On the other hand, several of the jobs investigated may have had higher relative loading on muscles that were not examined, i.e., hand- and arm muscles (i.e., in meat cutters and machine operators). As a consequence, this may have weakened the relationship between perceived exertion and the muscular load of the muscles examined. Nevertheless, the current muscles represent large muscle groups where one or several groups to some extent are involved in almost all tasks performed in the present job types. Accordingly, when, i.e., cutting meat using hand and arm muscles, the neck/shoulder muscles are also highly active (Arvidsson et al.
2012) and would inherently contribute to the association between trapezius muscle load and perceived exertion.
In line with the present findings, Balogh et al.
2004 observed comparable low associations between perceived exertion and cardiovascular load in day measurements in cleaners and office workers (Balogh et al.
2004). It may therefore be suggested, that the relative cardiovascular load induced by the investigated jobs was not high enough to have significant impact on perceived exertion.
It may in work situations, be difficult for the individual to distinguish between the lowest levels on the Borg scale, for example very weak, weak and moderate exertion (Borg 1–4), which can increase the risk of misclassifications and thereby bias the findings of our linear correlation analysis. As a consequence, the Borg scale was divided into a “low” (<4) and a “high” (≥4) perceived exertion part in the logistic regression analysis. The odds ratio of 18 observed between perceived exertion and neck/shoulder muscle activity means that the likelihood of experiencing “high” exertion is increased by a factor of 18 for every percentage point (1 percentage point is 4½ min of a 7½ h shift) increase in the total duration of the workday with high muscle activity in the neck/shoulders. Translated into practical guidelines, this roughly means that for every 15 heavy lifts of 1 s with high muscle activation the likelihood of experiencing “high” perceived exertion is doubled. In view of this, the workers in the group that rated “high” perceived exertion (≥4 Borg) worked on average 0.23 % of the day (~7½ h shift) with high neck/shoulder muscle activity (>60 % of max EMG). This corresponds to a total exposure of ~1 min of high muscle activity or in more practical terms roughly 65 heavy lifts at an estimated lift duration of 1 s. Accordingly, as high physical exertion is a risk factor for development of musculoskeletal disorders (Andersen et al.
2012a) long-term sickness absence (Andersen et al.
2012b), the present results may help in the development of better lifting-guidelines for preventing work-related poor health in blue-collar workers.
The incidence of musculoskeletal complaints has been associated with misclassification of self-rated exposure. Several studies have shown that people with musculoskeletal disorders rate their exposure higher than healthy people (Hansson et al.
2001; Balogh et al.
2004). Furthermore, Viikari-Juntura et al. (
1996) observed that the relationship between self-reports and hand, neck and shoulder movements was lower for workers with low-back pain than those without pain. Accordingly, it seems that pain has a greater effect on perceived exertion than on the balance between physical capacity and workload that may bias the association between self-report and direct measurement. However, adjusting for self-reported muscle pain in a model with self-reported physical exertion is meaningless, because muscle pain may be both the cause and effect of physical exertion. Thus, if physical exertion leads to muscle pain, adjusting for muscle pain will underestimate the association between perceived exertion and the objectively assessed variables. Another method is to stratify analyses, however, at the risk of hampering generalizability of the findings. On an exploratory basis, when excluding participants with moderate and high pain (VAS > 3 in either body region) only the differences in moderate-to-high and high neck/shoulder muscle activity remained significant between the “high” and moderate and “low” perceived exertion group. Thus, regardless of pain, this highlights the strong association between neck/shoulder muscle activity and perceived exertion in manual strenuous jobs.
As previously indicated, the large variability in job types, tasks and physical capacity of the worker may cause a large variation in perceived exertion. However, this only strengthens the generalizability of our findings. Nevertheless, possible further explanations that may have compromised the association between perceived exertion and the objectively assessed workload may be addressed the complexity and frequency of the question asked (Spielholz et al.
2001; Stock et al.
2005). The participants were asked to rate their average perceived exertion of the past 2 h at midday and after 7½ h of work on a Borg CR10 scale. Although the question seems rather simple, it may be difficult to provide a valid estimate of the average exertion experienced during the past 2 h, indeed if the subject had performed several different tasks. It may be speculated that the association between perceived exertion and objectively assessed workload would have been higher if the frequency of questions were higher during the day or alternatively if the workers could rate their perceived exertion after every task performed. However, because only minimal data collection, e.g., a single question posed, is feasible in a real-world workplace setting, the aim of this study was to investigate whether a minimum of perceived exertion reports obtained during the day could provide valuable evidence of a relationship between objectively assessed physical workload and perceived exertion of a full working day.
There are both strengths and limitations to our study. Because maximal isometric muscle activity and maximal heart rate capacity may be reduced in participants with musculoskeletal disorders (Andersen et al.
2008c,
2009) the present type of EMG and ECG normalization may be a limitations. On the other hand, this procedure has been widely used (Andersen et al.
2006; Jakobsen et al.
2011) and has shown to generate greater test–retest reliability scores compared with non-normalized data (Wilk et al.
1996; Kellis and Baltzopoulos
1996; Rutherford et al.
2001; Alkjaer et al.
2003). Additionally, performing maximal exertion-tests (MVC and
VO
2max) after a strenuous day of work may potentially skew the results. Nevertheless, normalizing EMG with maximal EMG values obtained from separate days is not advisable and performing the MVC and
VO
2max test on two separate days was not possible for practical reasons.
Furthermore, an exploratory analysis excluding employees with musculoskeletal pain did not change the results. A strength is the large population size and variability among participants and job types, highlighting the generalizability of the study results.