- Split View
-
Views
-
Cite
Cite
C Deeney, L W O’Sullivan, Effects of cognitive loading and force on upper trapezius fatigue, Occupational Medicine, Volume 67, Issue 9, December 2017, Pages 678–683, https://doi.org/10.1093/occmed/kqx157
- Share Icon Share
Abstract
Musculoskeletal disorders (MSDs) are particularly common in the shoulder/neck region for some tasks that involve low force exertions, for example computer work. It has been demonstrated that muscle activity may be stimulated by cognitively demanding tasks. There is a lack of studies on the relationship between qualitative levels of cognitive loading, level of muscle activity, and muscle fatigue.
To investigate the effects of quantitative levels of cognitive loading on conduction velocity changes and isometric endurance times for the upper trapezius.
Participants performed a combination of three levels of a cognitively demanding computer task (0, 1.59 and 3 Bits) while simultaneously performing either of two isometric endurance tests (20 and 40% MVC) for the upper trapezius.
Information load had a significant effect (P < 0.05) on normalized conduction velocity slopes for the 40% but not for the 20% maximum voluntary contraction conditions. Information load had a highly significant effect on endurance times for both exertion levels (P < 0.01).
This study found that performing a high-difficulty cognitive task while simultaneously performing increasingly higher levels of static isometric shoulder abduction exertions, fatigued the trapezius muscle quicker than performing an equivalent exertion with low cognitive load. This is particularly relevant to workplaces with increasing levels of automation that require high levels of decision making and computer use.
Introduction
Musculoskeletal disorders (MSDs) remain the most prevalent occupational diseases in the European Union (EU) [1]. These include a wide range of inflammatory and degenerative conditions affecting the muscles, tendons, ligaments, joints, peripheral nerves and supporting blood vessels [2]. There has been a growing recognition that psychosocial risk factors in the workplace are associated with the development of work-related MSDs. A number of theories have been proposed to describe the mediation of stress and illness [3], but these have not identified specific pathways linking psychosocial risks to MSDs.
The fifth European Working Conditions Survey [4] found that MSDs directly related to physical working conditions are in decline whereas MSDs related to stressful working environments are increasing. It is clear that psychosocial risks contribute to risk of MSDs and that cognitive loading increases muscle tension, especially for the upper trapezius [5]. Although there are numerous psychosocial risk evaluation methods available [3], these are survey methods of workers’ perceptions of workload demands. They do not quantify the information loads involved in the work. The authors were unable to find studies that quantified cognitive load during psychophysiological studies of upper trapezius muscle activity and cognitive loading. It remains unclear how much cognitive loading at varying levels adds to indices of fatigue, especially when combined with fatigue induced by isometric contractions. This is practically important in today’s industrial world where there is continued adoption of automation. The recent automation and information and communications technology (ICT) initiative Industry 4.0 [6] embodies this wave of change, resulting in a change in workers roles away from manual work to more human robot interaction and supervision [7]. Essentially, there is less physical work and more computer-based monitoring and decision making.
The majority of studies examining the influence of cognitive loading on muscle activity, especially in computer related tasks, tend to focus on the upper trapezius [8]. This is a logical choice considering the fact that researchers have identified the neck/shoulder region as the most prevalent site for symptoms resulting from computer work. Exposure to a combination of physical and psychosocial workplace risk factors is said to be a major contributor to shoulder MSDs [9] but remains relatively under studied.
The purpose of this study was to quantify the effect of quantitative levels of cognitive loading and upper trapezius exertion level on shoulder (upper trapezius) muscle fatigue (muscle fibre conduction velocity) and also endurance time.
Methods
The experimental procedure was approved by the University of Limerick Research Ethics Committee. The participants were recruited through the use of advertisements on the university campus.
All treatments involved simultaneously performing an isometric shoulder abduction endurance test while also performing a cognitive task. Two levels of shoulder abduction exertion (20% and 40% maximum voluntary contraction (MVC)) and three levels of information load (no cognitive load, 1.59 Bits and 3.00 Bits) were tested in a full factorial design (Table 1). The levels of MVC were chosen to cover up to the higher levels of shoulder muscle effort observed in industrial tasks [10], whereas the levels of cognitive loading were based on the work of Shannon and Weaver [11].
Treatment Number . | Weight (MVC) . | Information Loads (bits) . |
---|---|---|
1–3 | 20% | 0.00, 1.59, and 3.00 |
4–6 | 40% | 0.00, 1.59, and 3.00 |
Treatment Number . | Weight (MVC) . | Information Loads (bits) . |
---|---|---|
1–3 | 20% | 0.00, 1.59, and 3.00 |
4–6 | 40% | 0.00, 1.59, and 3.00 |
Treatment Number . | Weight (MVC) . | Information Loads (bits) . |
---|---|---|
1–3 | 20% | 0.00, 1.59, and 3.00 |
4–6 | 40% | 0.00, 1.59, and 3.00 |
Treatment Number . | Weight (MVC) . | Information Loads (bits) . |
---|---|---|
1–3 | 20% | 0.00, 1.59, and 3.00 |
4–6 | 40% | 0.00, 1.59, and 3.00 |
The information load processing task involved a choice reaction test on a computer. Task difficulty increases with the information content, i.e. the number of alternatives to choose from. This was quantified using the Shannon and Weaver (1949) [11] Information Theory, where three alternatives comprise 1.59 Bits and eight alternatives comprise 3.00 Bits. The zero bit condition involved the participant only performing the isometric endurance test. Software developed in house using Visual Basic 5.0 presented a random number (1 to 10) and the participant selected that value from the number of alternatives (3 or 8 depending on the Information Load condition) using the mouse with the dominant hand. Participants were instructed to perform the cognitive task as quickly and as accurately as possible. The total duration of the experiment was 2 hours (including set-up of electromyogram (EMG) electrodes).
EMG signals were recorded for the duration of the isometric endurance tests. A 128 channel EMG-USB amplifier system (OT Bioelecttronica, Turin, Italy) was used to detect multi-channel surface EMG signals from an adhesive linear electrode array, consisting of eight electrodes equally spaced with a 5-mm inter-electrode distance. The cavities of each electrode were filled with 20 µl of conductive gel using a pipette (Multipette Plus, Eppendorf AG, Hamburg, Germany). A dry silver bar array electrode was used to determine the muscle’s Innervation Zone (IZ). The IZ was determined by visual inspection of the EMG signal on the computer (12).
Raw EMG signals were amplified, band pass filtered (3 dB bandwidth, 10–500 Hz, roll-off of 40 dB/decade), sampled at 2048 Hz, and then stored on the computer using a 12 bit A/D converter. Specialized software (EMG-GV v1.1.3, LiSin, Turin, Italy) was used to analyse the EMG signals by calculating the conduction velocities (CVs) over a minimum of four channels for 1 second epochs as per the guidance of Farina et al. [13].
The velocity of propagation of an action potential is indicative of the motor unit contractile properties. Muscle fibre CV is a basic physiological parameter and is related to the type and diameter of muscle fibres, ion concentration, pH, muscle temperature and motor unit firing rate [14]. During muscle fatigue, there is a depletion of calcium ions and a co-commitment reduction in CV [15]. Hence, a negative slope of a regression line through CV data recorded continuously throughout a constant exertion level isometric contraction is indicative of the level of fatigue [16]. The slope of CV values, normalized as a percentage of the values at the start of the treatment, indicates the rate of change as a percentage per unit time (i.e. percentage per second) [17].
For the MVC and for the isometric endurance tests, a handle with adjusting chain was attached to a Mecmesin AFG force meter, which was in turn fixed to a resisting load on the floor. For the isometric endurance tests, forces were displayed in real-time on the computer interface with the exertion target for each treatment (20 or 40% MVC). The non-dominant hand was used to perform the isometric shoulder abduction endurance contractions while the dominant hand simultaneously performed the choice reaction task on the computer. Hence, the exercises were performed and the EMG data collected on the non-dominant.
Statistical analysis was performed in SPSS version 14.0. The CV and Endurance Time (ET) data violated the assumptions for parametric statistics; normality of data (Shapiro-Wilk Test, P < 0.05) and homogeneity of variance (Levene’s Test, P < 0.05). As a two way ANOVA was not possible, the Friedman Test, the non-parametric equivalent to a one way ANOVA, was used to test for the effect of information load on normalized CV slopes and endurance times (ET) for the 20% MVC and 40% MVC conditions separately.
Participants were seated in an upright position on an adjustable height chair with the feet flat on the floor and the knees at 90 degrees for the duration of the experiment. Prior to placement of the electrodes, the skin was cleaned with water and lightly abraded with gel and the innervation zones were detected using a bar electrode. The linear array electrode was placed on the straight line between the spine of the seventh cervical vertebra (C7) and the lateral edge of the acromion. The electrode was attached outside of the innervation zone. The electrode was then connected to an EMG amplifier and a grounding wrist strap attached to the participant.
MVC was recorded at the start of the experiment by way of shoulder abduction isometric contraction. The force meter was adjusted such that when the participant abducted the upper arm at 90 degrees when applying an abduction force on the force meter. The participant was asked to pull upwards against the force meter and directed to generate the force over 2–3 seconds, holding their maximum contraction for a further 2–3 seconds. The procedure was repeated three times with a 3-minute break between each measurement. The highest value across the trials was recorded as the MVC.
The experiment tasks comprised shoulder abduction isometric endurance tests for the non-dominant limb. Participants simultaneously performed the choice reaction cognitive task on the computer with a mouse using the dominant hand. Participants were instructed to abduct their arm at a 90 degrees angle in the scapula plane (i.e. parallel to the floor) and to lift the force meter vertically. During the isometric endurance tests the participant was requested to hold the force exerted (as displayed and controlled on the computer) for as long as they could. The task was terminated when participants were no longer able to sustain the required posture/exertion or when the participant requested to stop. Participants rested for at minimum 10 minutes between the tasks. Presentation of the cognitive load conditions were balanced using Latin square orders.
Results
In total, 12 participants performed the experiment (12 males; age range 20–26 years, mean ± SD: 23.67 ± 1.87 years). All were right handed and healthy, and reported no previous history of musculoskeletal disorders.
A summary of the statistical analyses are presented in Table 2 with summary descriptive statistics presented in Table 3. The results found information load did not have a significant effect on CV slope for the 20% MVC conditions but did have a significant effect for the 40% conditions (P < 0.05). Information load had highly a significant effect on endurance times for both exertion levels (both P < 0.01). Figure 1 is a plot of the average normalized CV slopes with 95% confidence intervals.
Dependent . | Exertion . | Significance . |
---|---|---|
CV slope | 20% MVC | NS |
40% MVC | P < 0.05 | |
ET | 20% MVC | P < 0.01 |
40% MVC | P < 0.01 |
Dependent . | Exertion . | Significance . |
---|---|---|
CV slope | 20% MVC | NS |
40% MVC | P < 0.05 | |
ET | 20% MVC | P < 0.01 |
40% MVC | P < 0.01 |
CV slope = Conduction Velocity normalized slopes; ET = Endurance Time seconds; NS = not significant.
Dependent . | Exertion . | Significance . |
---|---|---|
CV slope | 20% MVC | NS |
40% MVC | P < 0.05 | |
ET | 20% MVC | P < 0.01 |
40% MVC | P < 0.01 |
Dependent . | Exertion . | Significance . |
---|---|---|
CV slope | 20% MVC | NS |
40% MVC | P < 0.05 | |
ET | 20% MVC | P < 0.01 |
40% MVC | P < 0.01 |
CV slope = Conduction Velocity normalized slopes; ET = Endurance Time seconds; NS = not significant.
. | . | 20 MVC . | 40 MVC . | ||||
---|---|---|---|---|---|---|---|
. | . | 0 Bits . | 1.59 Bits . | 3 Bits . | 0 Bits . | 1.59 Bits . | 3 Bits . |
UT CV slopes (%/sec) | Mean | −0.013 | −0.036 | −0.044 | −0.043 | −0.067 | −0.087 |
SD | 0.05 | 0.08 | 0.08 | 0.05 | 0.06 | 0.04 | |
ET (seconds) | Mean | 107.4 | 116.6 | 82.8 | 78.8 | 51.2 | 47.5 |
SD | 23.9 | 49.4 | 25.1 | 26.7 | 10.5 | 23.2 |
. | . | 20 MVC . | 40 MVC . | ||||
---|---|---|---|---|---|---|---|
. | . | 0 Bits . | 1.59 Bits . | 3 Bits . | 0 Bits . | 1.59 Bits . | 3 Bits . |
UT CV slopes (%/sec) | Mean | −0.013 | −0.036 | −0.044 | −0.043 | −0.067 | −0.087 |
SD | 0.05 | 0.08 | 0.08 | 0.05 | 0.06 | 0.04 | |
ET (seconds) | Mean | 107.4 | 116.6 | 82.8 | 78.8 | 51.2 | 47.5 |
SD | 23.9 | 49.4 | 25.1 | 26.7 | 10.5 | 23.2 |
UT = Upper Trapezius; ET = Endurance Time
. | . | 20 MVC . | 40 MVC . | ||||
---|---|---|---|---|---|---|---|
. | . | 0 Bits . | 1.59 Bits . | 3 Bits . | 0 Bits . | 1.59 Bits . | 3 Bits . |
UT CV slopes (%/sec) | Mean | −0.013 | −0.036 | −0.044 | −0.043 | −0.067 | −0.087 |
SD | 0.05 | 0.08 | 0.08 | 0.05 | 0.06 | 0.04 | |
ET (seconds) | Mean | 107.4 | 116.6 | 82.8 | 78.8 | 51.2 | 47.5 |
SD | 23.9 | 49.4 | 25.1 | 26.7 | 10.5 | 23.2 |
. | . | 20 MVC . | 40 MVC . | ||||
---|---|---|---|---|---|---|---|
. | . | 0 Bits . | 1.59 Bits . | 3 Bits . | 0 Bits . | 1.59 Bits . | 3 Bits . |
UT CV slopes (%/sec) | Mean | −0.013 | −0.036 | −0.044 | −0.043 | −0.067 | −0.087 |
SD | 0.05 | 0.08 | 0.08 | 0.05 | 0.06 | 0.04 | |
ET (seconds) | Mean | 107.4 | 116.6 | 82.8 | 78.8 | 51.2 | 47.5 |
SD | 23.9 | 49.4 | 25.1 | 26.7 | 10.5 | 23.2 |
UT = Upper Trapezius; ET = Endurance Time
CV slopes (Figure 1) decreased with increasing exertion level and increasing information load. As expected, the CV slopes were greater (more negative) for each of the 40% MVC levels than for 20% MVC.
Endurance time decreased with increasing information level and exertion level (Figure 3) with the exception of 20% MVC where a small increase was observed between 0 Bits ( 107 sec) and 1.59 Bits ( 116.6 sec). Endurance time averaged across the force conditions reduced from 93.1 sec for zero Bits to 83.9 sec for 1.56 Bits to 65.1 sec for the 3 Bit condition.
Endurance times were longer for each of the 20% MVC isometric endurance exertion levels than the corresponding information loads for 40% MVC. For zero Bits, the average endurance time was 107 seconds for 20% MVC and this reduced to 79.8 seconds for 40% MVC, a reduction of 26%. The difference was 56% for the 1.59 bit condition and 42% for the 3 Bit condition.
Multiple regression analysis was used to develop equations predicting both normalized CV slope and endurance time (separately) based on MVC and Information Level. Both models were a good fit to the data. The equation predicting normalized CV slope had a R2 value of 0.97 (P < 0.01) and the equation predicting Endurance Time had a R2 values 0.89 (P < 0.05).
Discussion
The principal finding from this study was that shoulder muscle fatigue increased with increases in the level of mental demand. The experimental approach permitted the modelling of the effects of information level and exertion level on CV slopes and endurance time using regression analysis. The high R2 values indicate that the effect of information level on normalized CV slopes and endurance times was close to linear. Information load had a significant effect on CV slopes for the 40% MVC condition but not for the 20% MVC condition. There was high variability in the data set (as discussed below) and it is unclear if this affected the 20% MVC condition. More research is needed to study information load effects for lower exertion levels.
A weakness of the study is the large variations in the response variables, especially the normalized CV slopes. The raw CV values were within normal ranges (Figure 2) but the variation in values was higher for low information than for high information levels (Figure 1). Ultimately, the variation can be attributed to the main effects, but also between subject difference, and variation introduced by the experiment control (or lack thereof). The Friedman test is a within subject test so differences in that respect are largely controlled for. An explanation for the high variation in CV slopes could be due to spatial muscle activation. Kleine et al. [18] found variation in the typography (RMS and median frequency) of the upper trapezius as the muscle fatigued. Troiano et al. [19] suggested that this is possibly due to the participants being able to adjust their posture and hence recruitment pattern of the muscle. Toriano et al. studied the activation typography of the upper trapezius using a shoulder shrug (rather than the arm elevated as in Kleine et al. [18]) and found no evidence of spatial muscle activation. They attribute this to the isolation of the muscle by the specific nature of the posture during the fatiguing contraction. Future studies of this nature might consider the shoulder shrug rather than shoulder abduction as in this study to help obviate this issue.
The results of this study give some evidence of increased fatigue which suggests that increased cognitive loading in the form of information level (Bits) increased muscle activity in the upper trapezius muscle. Bloemsaat et al. [20] studied the effects of cognitive loading on neck/shoulder muscles (upper trapezius and deltoid), two upper arm muscles (biceps brachii and triceps), and four forearm muscles (flexor digitorum superficialis, extensor digitorum, extensor carpi radialis longus and extensor carpi ulnaris). Only the upper trapezius muscle was significantly affected by cognitive loading (memory processing). The authors suggested that the muscles in the neck/shoulder region and upper arm are more susceptible to cognitive task demands than muscles in the forearm and hand, based on their findings.
From a psychosocial perspective, Lundberg et al. [21] investigated the effects of experimentally induced mental and physical stress on motor unit recruitment in the upper trapezius muscle. They demonstrated that the same motor units can be activated by mental stresses that are activated by physical stress. Wærsted et al. [22] found that during a 10-minute exposure to a cognitively demanding task the same motor units were continuously active. It was found that when participants were exposed to a similar set of cognitively demanding tasks as in the previously mentioned study, muscle activity was found to increase as mental workload increased [23]. The implications of these findings are very important, as they suggest that the same motor units can be activated either by physical exertion and/or some levels of mental activity.
High information level is proposed to be akin to high mental job demands, which is recognized as a key psychosocial risk in occupational settings. It is recognized that whilst psychosocial risks impact on MSDs, the exact relationship remains unclear. Although the data cannot be used to prove or disprove either of the theories, they provide additional insight into the effects of cognitive activity and muscle fatigue for the upper trapezius. The data verified that the psychophysiological relationship between information and both CV slope and endurance time could be accurately modelled with linear regression for the treatments studied.
These results further clarify the relationship between cognitive demands and risk of musculoskeletal disorders in work settings. Existing tools evaluate variation in aspects related to cognitive loading as part of work, notably job demands. So, tools are available to evaluate the exposures. Regarding impacting change through policymaking, Niedhammer [24] detailed wide variation in psychosocial work exposures in a study across 31 European countries. They detailed how the data could be useful to guide prevention policies at European level, but this is not easy, especially considering that a recent proposal for an EU directive on work-related MSDs was unsuccessful. There is continued research in the development of tools to assess psychosocial exposures, including in occupational disability [25]. The need to survey individual workers to assess exposures continues to be a barrier. An interesting approach to this is the use of company data to assess exposures [26]. At country level, there are individual initiatives. In Ireland, the Work Positive tool is a national policy initiative, based on work by the Health and Safety Authority (Ireland) and the Health and Safety Executive UK. The Work Positive tool included the key psychosocial risks and is noted to be reliable [27]. The challenge that remains is the uptake of the use of existing tools that include assessment of job demands [28]. These are considered to be risk assessment tools, but that application is generally not enforced such as for physical ergonomics risks in the workplace. This remains a significant barrier to the wider adoption of psychosocial risk evaluation in the workplace.
This study highlights how high cognitive demand tasks, for example in intensive ICT/automation supervision, can be expected to result in continued high prevalence of shoulder/neck MSDs in industry, despite an expected reduction in physical workload for these workers.