|
A. A REVIEW OF BIOMECHANICAL AND PSYCHOPHYSICAL RESEARCH ON RISK FACTORS ASSOCIATED WITH LOW-BACK PAIN A.1 Introduction Data collected through the U.S. Department of Labor's Annual Survey of Occupational Injuries and Illnesses (ASOII) demonstrate the high morbidity of work-related overexertion injuries and disorders in the United States. OSHA's estimates of the MSD injuries and illnesses based on 1996 data gathered by BLS or summarized as follows:
The true cost of work-related overexertion injuries and disorders in the United States is not known. Conservative estimates of annual expenditures, based on workers compensation payments (indemnity and medical services) and other direct costs, range between $13 to 20 billion (Bernard and Fine, 1997). The total cost to society is believed to be substantially higher due to various indirect costs (e.g., lost productivity, costs of hiring and training replacement workers, overtime, administrative costs, and miscellaneous transfer payments) that are not included in the conservative estimates. The total annual societal cost has been estimated to be as high as $100 billion (Bernard and Fine, 1997). Epidemiological and laboratory-based research methods have both been used to evaluate the significance of various risk factors associated with work-related musculoskeletal disorders (MSDs). Epidemiological studies are designed to look for significant associations between exposure to ergonomic risk factors (e.g., force, repetition, posture) and selected health outcomes (ranging from medically diagnosed disease entities to subjective reports of pain or discomfort) in selected populations of workers. NIOSH (Bernard and Fine, 1997) performed a comprehensive review of over 600 epidemiological studies of occupational MSDs. This study concluded that there was either "strong evidence" or "evidence" of a causal relationship between workplace exposures to forceful exertions, repetition, and awkward posture and MSDs of the neck, upper extremity and low back. This review also found "strong evidence" of a causal relationship between low-back pain and whole-body vibration, and between segmental vibration and hand-arm vibration syndrome. Epidemiological studies provide important insights into understanding the causes of MSDs. However, these studies are sometimes criticized due to their inability to precisely measure exposures to risk factors and the associated biomechanical and/or physiological responses to these exposures. Biomechanical models and laboratory studies do not replace epidemiological studies. However, these approaches provide important complementary information in the quest toward understanding the complex process of how exposures to ergonomic risk factors result in physiological responses that may ultimately lead to work-related injuries and illnesses. This paper presents a review of laboratory studies and biomechanical models of work factors associated with increased risk of low-back injuries and disorders. Section B of this Appendix addresses laboratory studies and biomechanical models of work factors associated with increased risk of upper-extremity injuries and disorders. Laboratory studies are controlled scientific investigations of how humans respond when exposed to specific ergonomic risk factors (e.g., forceful exertions, awkward work postures, high repetition, etc.) during simulated work activities. Responses include both objective biomechanical/physiological measurements, such as the electromyographic (EMG) activity of a working muscle, and subjective psychophysical measurements, such as ratings of perceived exertion. Most of the studies cited in this Appendix were performed in true laboratory settings. A few studies were performed in operational workplaces modified as necessary to collect data under carefully controlled conditions. Because of ethical issues related to the protection and safety of human subjects, laboratory studies are designed to keep exposures to risk factors at levels below the threshold of injury. As a result, these studies are generally incapable of "proving" a relationship between exposure and injury. Despite this limitation, laboratory studies provide important scientific insights as to how the body responds to ergonomic stresses. Combined with pathophysiological models of musculoskeletal injury mechanisms and epidemiological findings of positive relationships between exposure to ergonomic risk factors and musculoskeletal injury, laboratory studies are an essential element in understanding the causes and prevention of work-related overexertion injuries. This section also reviews biomechanical models that simulate and/or predict how the musculoskeletal system responds to work factors such as external loads placed on the hands, work posture, and movement dynamics. These models can be used to estimate musculoskeletal stresses in the absence of a human experiment. A.1.a Biomechanical Approaches To Studying Ergonomic Risk Factors The discipline of occupational biomechanics is concerned with measuring and/or modeling the "internal" mechanical responses of body tissues to the "external" physical demands of a work activity. These external demands include: 1) the magnitude and direction of force(s) exerted while working (e.g., weights lifted during manual handling tasks, exertions required to operate tools and equipment); 2) the location(s) where the external force acts on the body; 3) the posture(s) required to perform the job; and 4) movement dynamics (velocity and acceleration). A variety of methods have been developed to measure or predict internal responses to external demands, including:
Biomechanical methods have been used extensively in both laboratory and field settings. In some instances, direct and indirect measurement methods have been used to confirm the predictions of models. A.1.b Psychophysical Approaches to Ergonomics Research Psychophysics is a discipline dating back to the 19th century that bridges psychology and physics by examining the relationships between physical stimuli in the environment and the resulting sensations perceived by the humans who are exposed to the stimuli. Various quantitative relationships have been proposed, culminating in the general model of Stevens (1960): S = kIn where: S is the intensity of the perceived sensation The exponent n describes the strength of the relationship between the intensity of the stimulus and the intensity of sensation. Values of n have been determined empirically for stimuli commonly encountered by humans, such as 3.5 for electric shock, 0.6 for loudness, and 1.6 for muscular effort (Eisler, 1962; Snook, 1996; Borg, 1998). In the field of ergonomics, psychophysical methods have been used to empirically determine acceptable levels of work intensity by asking subjects to adjust their work load (the physical stimulus) so that the resulting discomfort/fatigue (the perceived sensation) is acceptable. Approximately 35 years ago, researchers at the Liberty Mutual Research Center (Snook and Irvine, 1967) initiated a series of psychophysical experiments to evaluate human responses to common manual materials-handling tasks and to develop guidelines for job design. The initial Liberty Mutual study focused on lifting, with maximum acceptable weight (MAW) being the dependent variable of primary interest. Subsequent studies at Liberty Mutual and other research labs have examined materials-handling activities such as carrying, pushing, pulling, and lowering (Snook et al., 1970; Snook and Ciriello, 1974, 1991; Snook, 1978; Mital, 1984a, 1984b; Mital and Fard, 1986; Smith, Ayoub, and McDaniel, 1992), and more recently, repetitive hand motions (Snook et al., 1995; 1997; Klein and Fernandez, 1997). The psychophysical approach has also been used to evaluate the stressfulness of work by having subjects in a laboratory or workers on the job rate their perceptions of the intensity of work effort and/or their perceptions of discomfort. Borg (1998) has demonstrated a strong correlation between heart rate and perceived exertion rate using 10-point and 20-point categorical rating scales. Derivations of the Borg methodology have been used by numerous ergonomic researchers to obtain perceptions of exertion and discomfort in a variety of work and simulated-work situations (Corlett and Bishop, 1976; Armstrong, Punnett, and Ketner, 1989; Ulin et al., 1990, 1992; Kihlberg, Kjellberg, and Lindbeck, 1995). Results from these studies have been used to recommend modifications to existing workplaces and to develop guidelines for job design. Table II-1 summarizes the key articles reviewed for this paper. Risk factors associated with an increased risk of work-related low-back pain are listed in the left-most column. For each risk factor, the results of relevant laboratory experiments and biomechanical model predictions are briefly described. A more detailed discussion of the literature is presented in the following sections. A.2 Biomechanical Factors In Work-Related Low-Back Pain Low-back pain symptoms are caused by a variety of injuries and disorders. Although the underlying cause of back pain cannot be determined definitively in up to 90% of patients, work-related cases are believed to result from the following mechanisms: muscle or ligamentous injury; herniation of the intervertebral disc with irritation of adjacent nerve roots; and degenerative changes in the intervertebral discs (Deyo, Rainville, and Kent, 1992). Significant biomechanical research has been devoted to understanding how the musculoskeletal tissues of the lower back are affected by the parameters of job demands, such as the postures required to perform a job and/or the forces exerted during manual materials-handling tasks. These studies have focused on evaluating how work requirements challenge the strength capabilities of muscles and connective tissues, and the load-bearing capacities of the spinal motion segments. In these studies, the human body is treated as a mechanical system, made up of rigid links (the bones), which are connected at joints. Forces and mechanical moments (torques) imposed on the system during work activities are estimated by static and dynamic biomechanical models and then compared with the strength capabilities and biomechanical tolerance limits of the affected tissues. The job is considered to be potentially hazardous if the imposed forces or moments exceed the strength or biomechanical tolerance limits of either an individual or an agreed-upon percentage of the population (NIOSH, 1981; Waters et al., 1991; Genaidy at al., 1993). Table II-1: Risk factors associated with low back pain -- a summary of literature describing laboratory experiments and biomechanical models.
Table II-1: Risk factors associated with low back pain -- a summary of literature describing laboratory experiments and biomechanical models (continued)
Table II-1: Risk factors associated with low back pain -- a summary of literature describing laboratory experiments and biomechanical models (continued)
A.2.a Static Whole-Body Kinematic Models Chaffin and associates at the University of Michigan developed a static three-dimensional kinematic model of the musculoskeletal system that can be used to evaluate biomechanical responses to whole-body exertions such as lifting, pushing and pulling (Chaffin and Baker, 1970; Garg and Chaffin, 1975; Chaffin and Andersson, 1991). This model has generally been used for two purposes: to compare the strength demands of a task to the strength capabilities of the workforce to estimate the percentage of adult males and females who are capable of performing the task, and to predict compressive forces acting at the L5/S1 spinal disc during static exertions. The model has been used extensively to evaluate whole-body tasks that are performed at normal (non-jerky) movement speeds on an infrequent basis (typically less than once every 5 minutes). Because the model does not consider the effects of fatigue, it is generally not appropriate for highly repetitive tasks (psychophysical or metabolic job analysis tools are preferred) or highly dynamic motions (dynamic models are preferred). Despite these limitations, the model has been used to predict biomechanical responses to strenuous exertions associated with common manual handling tasks. Static Biomechanical Analysis of Strength Demands To use the Michigan model, it is necessary to describe the worker's anthropometry (height and body weight), working posture (angles at the ankles, knees, hip, trunk, shoulders, and elbows), and the vector (magnitude and direction) of the external load acting on the hands. The model uses this information to compute the strength required at the ankles, knees, hip, trunk, shoulders, and elbows to maintain the system in static equilibrium. Individual and task parameters such as body weight, posture, and hand force create resultant forces and mechanical moments at each joint. To maintain equilibrium, each joint must exert an equal and opposite reactive force and the muscles at each joint must have sufficient strength to create an equal and opposite reactive moment. For this reason, strength is characterized as the ability to create a mechanical moment. The Michigan model has been used to demonstrate how posture affects the strength demands of a job. For example, if a person with average male height and weight stands upright (trunk vertical) with the arms flexed at a 45 degree angle below the horizontal, the required lower back strength to counteract the resultant mechanical moment is only 14.6 Newton meters (Nm). If this person bends forward at the hip so that the trunk is at a 45-degree flexion angle, the required strength increases to 121 Nm (Chaffin and Andersson, 1991). The posture change has displaced the weight of the trunk, head, and upper extremities forward, increasing the moment arm and the required strength at the low back. Placing a load in the hand would increase the strength demands at the low back even further, due to the increased resultant moment created by the external load. The extensor muscles of the lower back must work harder during trunk flexion to counter the increased mechanical moment. This has been confirmed in numerous laboratory studies using EMG to measure the activity of back muscles during static trunk flexion (Morris, Lucas, and Bresler, 1961; Schultz et al., 1982; Seroussi and Pope, 1987). Andersson, Ortengren, and Nachemson (1977) found that the EMG activity in the lumbar back muscles increased linearly from a mean of approximately 25 microvolts at 10 degrees of forward flexion to approximately 60 microvolts at 60 degrees of flexion. As discussed above, the Michigan model computes the strength demands imposed by a task on major body joints. The ability (or lack of ability) to perform the task is a function of the strength capacities (i.e., the capability to produce the required reactive mechanical moments) at the joints (Andersson and Schultz, 1979). When a person attempts to lift, carry, push or pull, the resultant moments created at each joint due to the load in the hands and the body weight must be less than or equal to the strength at each joint. After the model has computed the strength required at each joint to perform a specific task, this value is compared to statistical distributions of the strength capabilities of U.S. adults to estimate the percentage of males and females who are capable of performing the task (Chaffin and Andersson, 1991). A task with a high percent capability prediction can be performed by most workers, while a task with a low percent capability prediction exceeds the ability of many workers. There is limited evidence that strength capability may be related to the risk of overexertion injuries. Chaffin and Park (1973) compared maximum isometric lifting strength to job demands in a study of 411 workers in an electronics manufacturing facility. The strength tests were designed to replicate job demands. During a 1-year follow-up, the rate of low-back injuries was three times greater in workers who did not demonstrate strength equal to or above job requirements; however, the sample size was not sufficient for statistical significance. In a similar study of aluminum and rubber workers, Keyserling, Herrin, and Chaffin (1980) found that workers whose maximum isometric strength matched job demands had fewer injuries than an unmatched group; however, the results were not statistically significant. A study by Battie et al. (1989) found no relationship between isometric strength and back injuries; however, the strength tests in this study were not designed to simulate job tasks. The injury studies cited in the previous paragraph compared the strength demands of a job to an individual's strength as determined by an isometric test. Herrin, Jaraiedi, and Anderson (1986) used the Michigan model to evaluate the relationship between injuries and model-predicted strength demands on the jobs of 6,912 workers in five different industries. For jobs where strength demands exceeded the model-predicted strength capability for the weakest 10% of the population, a significant (p < .05) increase was observed in reported back pain (sprains, strains, degenerative disc disease, and non-specific pain). The NIOSH Lifting Equation (NIOSH, 1981; Waters et al., 1993) considers population strength capability when evaluating the safety of a lifting task. For a job to be classified "acceptable," the strength requirements must be within the strength capability of at least 75% of the female working population. (Note: NIOSH used a combination of biomechanical and psychophysical strength considerations in developing the computational formula for the Lifting Equation.) Static Biomechanical Analysis of Compression Forces on the Spinal Discs From a biomechanical perspective, the fact that large resultant moments are induced in the lumbar spine during lifting and/or forward bending raises the question of the nature of internal forces that must be exerted to create the required reactive moments to stabilize the spine. An early biomechanical model proposed by Morris, Lucas, and Bresler (1961) assumed that two types of internal forces act to resist the resultant flexion moment. The largest contributor to the stabilizing reactive moment is the erector spinae muscles, located approximately 5 cm posterior to the center of rotation of the spinal discs. Through forceful contractions, these muscles create an extension moment to maintain stability. The second stabilizing force is created by intra-abdominal pressure, which pushes upward on the diaphragm in front of the spinal column, creating a small extension moment to resist the external load. Recent studies by McGill and Norman (1987) and Marras and Mirka (1992) have questioned the significance of the contribution of intra-abdominal pressure in creating the extension moment, so it is frequently not considered in contemporary models. Another recent study (Kumar, 1988) has shown that the moment arm of the erector spinae muscles is displaced approximately 5.5 cm to 7.0 cm from the center of rotation of the spinal discs, a slightly larger value than assumed in the original Morris model. Because the erector spinae muscles act through a small moment arm, they must exert high forces to counteract flexion moments during lifting and/or forward bending (Chaffin and Andersson, 1991). These exertions induce high compressive forces on the spinal motion segments. (Note: A motion segment consists of two adjacent vertebral bodies, the intervertebral disc, and the connecting ligamentous structures. The motion segment has been studied extensively because it is the smallest segment of the spine that exhibits biomechanical characteristics of the entire spine.) Figure II-1 summarizes compression forces acting at the L5/S1 spinal disc as a function of the load held in the hands and the horizontal reach distance. Under certain conditions of force and horizontal distance, compression forces may exceed limits recommended by NIOSH (1981) and others. Brinckmann, Biggeman, and Hilweg (1989) argue that epidemiologic and clinical evidence shows that some low-back problems are caused by primary mechanical destruction of tissues in the lumbar spine. Bone, cartilage, the intervertebral discs, and ligaments may fracture or rupture due to mechanical overload caused by excessive compression forces. Proof of this hypothesis is difficult to obtain, however, because it is unethical to perform in vivo experiments that expose the human body to mechanical loads at levels high enough to cause injury. Because of this dilemma, research has proceeded in two directions. The task-imposed load on the lumbar disc is either measured directly (see below) or calculated using biomechanical models. The load-bearing capacity of spinal structures (bones, discs, and ligaments) must be determined in vitro using tissue obtained from cadavers. By comparing task-induced loads to load-bearing capacity, it is possible to determine whether certain work activities exceed the mechanical limits of the spinal tissues.
Although most efforts to estimate spinal loads have used biomechanical models, there have been a few attempts to measure load in vivo. Nachemson and Elfstrom (1970) developed a small pressure transducer that fits inside a hypodermic needle. The needle can be inserted into a spinal disc, allowing the transducer to measure the hydrostatic pressure inside the nucleus pulposis while a person assumes different postures and performs various tasks. In vitro experiments using this system with cadaver spinal segments showed a linear relationship between internal disc pressure and the external compressive load applied to the segment (Berkson, Nachemson, and Schultz, 1979). In vivo disc pressure measurements have been used successfully to validate biomechanical model predictions of compressive loads (Schultz at al., 1982). Due to the highly invasive nature of this technique, it is best suited for slow, well-controlled movements (similar to those considered by static biomechanical models) and can only be used under strict laboratory conditions. As a result, in vivo studies of intra-discal pressure have been few in number. The tolerance of spinal motion segments to compressive loading has been studied extensively by biomechanical engineers and ergonomists in vitro using cadaver specimens. Genaidy et al. (1993) reviewed 12 of the larger studies with the objective of combining the data sources to develop a statistical model of spinal compression tolerance limits. Their resulting equation (R2 = .83) can be used to estimate lumbar compression strength limits for various percentiles of the population:
where: CS is the load tolerance limit (Newtons) Using this equation for a younger male with a strong lumbar spine (25 years old, 75th percentile), the resulting compression limit is 9,170 N. For an older male with a weaker lumbar spine (55 years old, 25th percentile), the compression limit drops to 3,176 N. For an older female with a weaker lumbar spine (55 years old, 25th percentile), the compression limit drops to 1,897 N. One of the limitations of cadaver studies is the uncertainty about whether compression damage to spinal segments in vitro is a reliable predictor of the risk of injury associated with in vivo compressive forces during the performance of work tasks. In spite of this limitation, NIOSH has utilized epidemiological and biomechanical evidence to establish 3,400 N as the maximum recommended L5/S1 disc compression force as one of the criteria used to develop the NIOSH Lifting Equation (NIOSH, 1981; Waters et al., 1993). Citing the cadaver study of Brinckmann, Biggemann, and Hilweg (1988), NIOSH acknowledged the large variability in compression forces associated with disc failure. NIOSH estimated that 21% of spinal segment specimens would fail at 3,400 N and that this limit may not protect the entire working population (Waters et al., 1993). The 3,400 N criterion has been incorporated into several biomechanical models. Figure II-1 above presents L5/S1 compression forces predicted by the University of Michigan model for various combinations of weight and horizontal distance. Compression forces increase with weight and distance, and the NIOSH 3,400 N limit can be exceeded when handling relatively light loads at an extended reach distance. (Note: Disc compression is the limiting criterion in the NIOSH Lifting Equation for infrequent lifting where fatigue is not a factor. For repetitive lifting, psychophysical and/or physiological factors become the limiting criterion due to fatigue (NIOSH, 1981; Waters et al., 1993).) The studies summarized by Genaidy et al. (1993) utilized experimental procedures where the compressive load on the spinal segment was increased to the point of mechanical failure in a single trial. A small number of studies have looked at the effects of repeated submaximal compressive loads on mechanical failure of the spinal segment (Brinckmann et al., 1987; Brinckmann, Biggemann, and Hilweg, 1988; Hansson, Keller, and Spengler, 1987). In these studies, loads were applied at a frequency of 0.25 Hz (15 cycles per minute). At compressive loads set at 55% of the single-trial failure load, mechanical failures were observed in 92% of the specimens after 5,000 cycles. At a 65% load, failures occurred in 91 percent of the specimens after only 500 cycles. At a 75% load, some specimens failed after only 10 cycles. These numbers must be interpreted cautiously because cadaver tissue may respond differently to repeated loads than live tissue. However, these studies present limited evidence supporting frequency as a risk factor for some back injuries. Summary of Static Biomechanical Models Static kinematic models of the human musculoskeletal system have been used to evaluate strength capability and compressive forces acting on the lumbar spine during common whole-body manual materials-handling tasks, such as lifting, pushing, and pulling. Large mechanical moments can be created in the lower back region by lifting heavy, compact loads close to the body or by lifting light-to-moderate loads at extended reach distances in front of the body. Forward bending also increases the mechanical moment due to the effects of body weight above the lower back. In order to counteract the moments created by loads in the hand and body weight, the extensor muscles of the lower back must exert high forces, creating a compression load on the lumbar spine. Based on biomechanical analysis, the critical task factors associated with lifting are the amount of weight lifted, the location of the load (horizontal distance from the lower back), and body posture (forward bending of the trunk increased the load on the lower back). The predictions of static biomechanical models are consistent with direct and indirect measurements of strain on musculoskeletal tissues. EMG studies have shown that EMG activity in the erector spinae muscles increases with increased load in the hands and/or forward bent postures. Intra-discal pressure measurements have shown that hydrostatic pressure in the nucleus pulposis of the disc increases with increased load in the hands and/or forward bent postures. Cadaver studies of spinal motion segments have demonstrated mechanical failure of spinal tissues under compressive loads smaller than those predicted by biomechanical models for many lifting tasks. Furthermore, a small number of cadaver studies have demonstrated that cyclical loading reduces the mechanical tolerance limits of the lumbar spine, indicating that lifting frequency may be a factor in some back injuries. However, cadaver studies have been questioned because it is not known whether the behavior of cadaver tissue is similar to living tissue. Although static whole-body biomechanical models are a useful tool for evaluating stresses associated with common materials-handling activities, they should not be used in all situations. The data base of muscle strength capability in these models was collected using isometric tests of maximum strength capability. Because muscles cannot perform at maximum levels for extended periods of time or on a highly repetitive basis, the static biomechanical models tend to overestimate strength capability on jobs that require repeated exertions. For this reason, psychophysical and/or physiological-based methodologies are a preferred alternative for evaluating repetitive whole-body exertions (Ayoub, 1992). Because static models do not consider forces and moments imposed on the musculoskeletal system from the acceleration/deceleration of external loads and body segment masses during highly dynamic movements, they may underestimate strain in work activities that involve rapid body motions (Marras et al., 1993). Finally, the spine is an extremely complex mechanical structure. Without making many assumptions (such as using single equivalent muscles to represent multiple muscles) to simplify the system, solutions become statically indeterminate. As a result of these simplifications, the models are imprecise in evaluating certain stressors, such as shear and torsional loads, and the effects of co-activation of multiple muscle groups. Despite these limitations, static biomechanical models have provided good insight into some of the risk factors associated with manual handling activities. A.2.b Dynamic Whole-Body Biomechanical Models Dynamic biomechanical models for evaluating whole-body materials-handling activities have been developed by several investigators (Leskinen et al., 1983; McGill and Norman, 1985; Jager and Luttman, 1989; Marras and Sommerich, 1991a, 1991b). These models are inherently more complex than static models. In addition to considering external forces acting on the body (the load applied to the hands and effects of body weight) and posture, these models must also consider the effects of motion dynamics, including velocity and acceleration. Due to inertia, acceleration or deceleration of body segments and any load in the hands requires the application of additional force as stated by Newton's second law (F = ma). Because the body is composed of multiple links and multiple joint centers, dynamic models require high-frequency measurements of many reference points in order to determine instantaneous locations, velocities, and accelerations of model components. For this reason, dynamic models are often restricted to laboratory environments where accelerometers, goniometers, and/or motion analysis equipment can be used to collect reliable data. In spite of this limitation, dynamic models provide important insights into the additional biomechanical strain imposed by rapid motions. Leskinen (1985) used a simple two-link model (upper limbs + trunk above the L5/S1) to compute compression forces at the L5/S1 spinal disc under static and dynamic assumptions. Input for the model was collected by obtaining posture and inertial data from 20 males who lifted a 15 kg box from a height of 10 cm to knuckle height. Model-predicted peak L5/S1 compression was 33% to 60% higher (depending on subject and lift technique) when the dynamic inertial load was added to the static model. Bush-Joseph et al. (1988) used a dynamic model to compute peak moments at the L5/S1 spinal segment using data collected from 10 male subjects who lifted a 150 N box from floor level to a height of 1m at slow, medium, and fast speeds. Peak moments increased linearly with the lift speed. The effects of movement dynamics and other task factors on selected indices of biomechanical strain on the lower back were evaluated by Jager and Luttmann (1989) using a whole-body model (the Dortmunder model) developed at the Universitat Dortmund in Germany. The basic representation of the skeletal system in the Dortmunder model is similar to the Michigan model described above; however, the lumbar spine is depicted as a system of five joints representing the five lumbar intervertebral discs (the Michigan model uses only one joint [the L5/S1] in the lower back region). The Dortmunder model was used to estimate the moment at the L5/S1 joint, compressive forces at the L5/S1 disc, and shear forces at the L5/S1 disc under various static and dynamic task conditions using 50th percentile male anthropometry during symmetric, sagittal plane lifting. The static analysis produced results consistent with the Michigan model. All three outcome measures increased monotonically as the weight in the hands increased from a no-load condition to a 50 kg load. All three outcomes increased monotonically as the horizontal location of the load in front of the L5/S1 increased. With the hands empty, increased trunk flexion angle (forward bending) caused monotonic increases in L5/S1 moment, compression, and shear. The dynamic analysis with the Dortmunder model demonstrated the significance of velocity and acceleration. The task of raising objects of various weights (no load, 20, and 40 kg) from floor to elbow height was simulated under three conditions: slow (where the lift was completed in 2 seconds); medium (where the lift was completed in 1.5 seconds); and fast (where the lift was completed in 1 second). Under the slow condition, compressive forces at the L5/S1 were similar to static conditions. At medium speed, peak compressive forces were about 20% higher than under static conditions. At high speed, peak compressive forces were roughly 50% higher than under static conditions. Jager and Luttman (1989) also simulated a jerking motion, assuming that all of the upward acceleration was completed in 0.1 second. Under these conditions, the peak L5/S1 compression for lifting a 20 kg load from the floor was approximately 8,000 N, more than double the peak load under static conditions and considerably higher than compressive loads shown to cause mechanical damage in cadaver tissues. Shear forces were not reported for the dynamic analyses. Marras and Sommerich (1991a, 1991b) of Ohio State University developed three-dimensional dynamic biomechanical model for evaluating loads on the lumbar spine during lifting. Instead of using a single equivalent trunk extensor muscle, this model extended earlier work by Schultz and Andersson (1981) by including 10 functional muscle groups in the lower back. By considering multiple muscle groups, Marras and Sommerich were able to evaluate the effects of co-contraction and asymmetric postures on spinal stresses during lifting. The Ohio State model was used to evaluate the effects of trunk velocity (10, 20, and 30 degrees/second), trunk torque output (27.1 and 54.2 Nm), and trunk posture symmetry (symmetric, 30 degrees clockwise rotation) on biomechanical loads on the spine. Eleven subjects participated in this test by performing isokinetic trunk extensions. During these exertions, EMG measured muscle activity in 10 trunk muscles and was used as input to the model along with subject anthropometry and trunk kinetics. The model calculated compression, shear, and torsion loading in the lumbar spine. Muscle activity levels in the erector spinae were balanced between the left and right sides during symmetric lifting; however, during the asymmetric lift (truck twisted clockwise), activity on the left side was dominant. Under symmetric conditions, peak compression at the L5/S1 increased with velocity (approximately 100 N for each increase of 10 degrees/second) and trunk torque output. Compression forces approached 3,900 N at the 30 degrees/second - 54.3 Nm condition. Under asymmetric conditions, peak compression was level (approximately 3,000 N) over the range of velocities tested and approximately 25% lower than under similar symmetric conditions. Peak anterior/posterior shear forces were greater under symmetric conditions and increased with the magnitude of trunk torque output. Left/right shear was approximately 40 N under asymmetric conditions compared to about 10 N under equivalent symmetric conditions In a follow-up cross-sectional epidemiological study, Marras et al. (1993, 1995) used historical medical records of low-back injuries to classify over 400 cyclical jobs as either high risk or low risk. Dynamic trunk motions were measured for each job using a triaxial goniometer system (called the Lumbar Motion Monitor (LMM)) to document three-dimensional angular position, velocity, and acceleration of the lumbar spine while workers performed their jobs. In addition, basic biomechanical variables (weight lifted, lift frequency, posture, etc.) were determined for each job. Logistic regression was used to identify the following five risk factors that distinguished between high- and low-risk jobs: lifting frequency, load moment, trunk lateral velocity, trunk twisting velocity, and the trunk sagittal angle (odds ratio for five variables combined was 10.7). The Ohio State group (Allread, Marras, and Parnianpour, 1996) used the lumbar motion monitor in a laboratory study to evaluate the effects of lift symmetry and to compare one- versus two-handed lift technique. Twenty-four subjects (all male) wore the LMM while performing one- and two-handed lifts ranging between 0 degrees (in the sagittal plane) and 135 degrees. Trunk motion characteristics associated with increased risk of back injury (Marras et al., 1993b, see above) were all higher with one-handed lifts, and velocities and accelerations increased substantially with the angle of asymmetry. Kim and Chung (1995) conducted a laboratory study of the EMG activity of the lower back during dynamic lifting. Eight healthy males participated in four 2-hour trials, lifting and lowering weights between floor and knuckle height. Independent variables were lift type (frequent-lifting, defined as lifting a weight normalized to 10 percent of the subject's strength six times/minute vs. heavy-lifting, defined as lifting a weight normalized to 20% of the subject's strength three times/minute) and posture (symmetric vs. 90 degree offset). Muscle activity (normalized EMG) was significantly higher during heavy lifting and during asymmetric lifting (p < .001). Muscle fatigue during the 2-hour trial (measured by a decrease in the mean power frequency of the EMG signal) was significantly greater during asymmetric and frequent lifting (p < .001). Summary of Dynamic Biomechanical Models and Laboratory Experiments Dynamic biomechanical models have been used by several investigators to overcome some of the limitations of static models discussed previously. These models allow investigators to consider the effects of inertia and acceleration when estimating biomechanical stresses on the lower back during lifting activities. Under slow, controlled sagittal plane lifts (approximately 2 seconds from floor to elbow height), the Dortmunder model shows that compressive forces acting at the L5/S1 spinal disc are similar to compressive forces under static conditions. When the lifting speed doubled, compressive forces increased by 50%. Studies performed at Ohio State University showed similar results for sagittal lifts; back compression increased with lifting speed. For asymmetric lifts, the Ohio model found unbalanced muscle activity between erector spinae muscles on the left and right side, and increased lateral shear forces. Based on dynamic biomechanical analyses and limited epidemiological studies, the critical risk factors associated with lifting are: 1) load moment about the spine (weight x horizontal distance for the hand-held load and the weight of body segments above the L5/S1); 2) velocity of lift; 3) frequency of lift; 4) lift asymmetry (lateral and twisting velocities), and 5) the trunk sagittal flexion angle. At faster lifting speeds, relatively light loads in the hand can result in back compression forces that exceed the 3,400 N threshold level established by NIOSH. Dynamic models are not appropriate for all lifting situations. For highly repetitive lifting, dynamic models do not consider the effects of fatigue; therefore, psychophysical and/or physiological methods are preferred. Due to the increased complexity of the data collection and analysis when using dynamic models, static approaches may prove to be more practical if lifts are performed using slow, controlled motions. A.3 Psychophysical Studies of Manual Materials Handling Psychophysical studies of lifting and related manual materials-handling activities have focused on workers' perceptions of physical strain, discomfort, and fatigue associated with work. There are several distinct differences between the biomechanical methods discussed above and the psychophysical methods discussed in this section. First, psychophysical methods are used to measure subjective responses to work (discomfort, fatigue, etc.) while biomechanical methods focus on objective responses (EMG activity, disc compression, etc.) Second, biomechanical approaches are primarily concerned with predicting how body tissues react during a single exertion, whereas psychophysical methods can be used to assess how workers respond to work demands that are distributed over a shift of 8 or more hours. A.3.a The Liberty Mutual Studies (Two-Handed Tasks Performed Over an 8-Hour Shift) The Liberty Mutual experiments (see summary in Snook and Ciriello, 1991) were all performed in a laboratory setting using subjects recruited from local industries (evening shift workers) near Hopkinton, Massachusetts. Subjects performed a variety of common materials-handling tasks, such as lifting, carrying, pushing, and pulling. They were instructed to perform as if they were being paid on an incentive basis, working as hard as they could without becoming unusually tired, weakened, overheated, or out of breath. Instead of measuring maximum strength in a single exertion, this approach measured exertions that could be performed on a repeated basis over an extended period without excessive fatigue or discomfort. Subjects were given control over one task variable, the weight (or resistance force for pushing and pulling tasks) of the object being handled. They could increase or decrease this weight at will. All other variables, such as task frequency, initial height of load, distance lifted/lowered/carried, etc. were controlled by the experimenter. For each condition tested, the subject adjusted the weight to the maximum amount they would be willing to handle if the task were performed throughout an 8-hour work shift. The final weight (following adjustments) was recorded as the MAW. All tasks were performed with two hands symmetric to the sagittal plane. Liberty Mutual has published extensive tables of MAWs for various task conditions (Snook and Ciriello, 1991). Separate tables exist for males and females, listing MAWs for various percentiles of the working population. The Liberty Mutual studies can be summarized by the following points:
The Liberty Mutual tables show a large variance in population capability. Within gender, MAWs and MAFs demonstrated by the strongest 10% of the population were roughly double those demonstrated by the weakest 10%. While males as a group were stronger than females, there was considerable overlap between the genders. Relationship Between Liberty Mutual Results and Workplace Injuries Snook, Campanelli, and Hart (1978) performed a retrospective study of 191 compensable work-related low-back injuries from 32 states. Physical demands on the jobs where injuries occurred were evaluated and compared to the Liberty Mutual psychophysical database to estimate the percentage of workers who would find the demands acceptable. Physical demands on jobs where no injuries occurred were analyzed in a similar manner. Workers assigned to jobs where physical demands exceeded the level deemed acceptable by 75% of the population were three times as likely to experience a back injury when compared to workers on jobs where physical demands were below the level acceptable to 75% of the population. Based on the distribution of workers in the injury vs. no-injury jobs, the authors concluded that up to one-third of compensable back injuries could be prevented by designing jobs to fit at least 75% of the population. Herrin, Jaraiedi, and Hart (1986) used the Liberty Mutual psychophysical tables (Snook, 1978) to evaluate the physical demands on the jobs of 6,912 workers in five different industries. A significant negative correlation (p <.05) was found between overexertion injury incidence rates and psychophysical percent capable rating for the most stressful task on these jobs (i.e., as "percent capable" increased, injury rates decreased). Similar to the Snook and Ciriello (1978) study described in the previous paragraph, there was a background level of overexertion injuries that could not be attributed to physical demands. The Snook and Ciriello (1978) and Herrin, Jaraiedi, and Hart (1986) studies provide evidence supporting the utility of using psychophysical percent capable scores for identifying manual materials-handling activities that place workers at increased risk of overexertion injury. NIOSH included psychophysical criteria (along with biomechanical and physiological criteria) in the development of a quantitative tool for evaluating stresses associated with manual lifting. The original (1981) and revised (1993) NIOSH Lifting Equations establish weight limits that are acceptable to 75% of adult females and 99% of adult males (NIOSH, 1981, Waters et al. (1993). A.3.b Studies of Two-Handed Tasks During 12-Hour Work Shifts Mital (1984a, 1984b) replicated a subset of the Liberty Mutual experiments to investigate the effects of an extended work shift (12 hours) on MAW and energy expenditure. Industrial workers (37 males and 37 females) participated as subjects in this University of Cincinnati laboratory study. Independent variables were lift frequency, height of lift (floor-to-knuckle, knuckle-to-shoulder, and shoulder-to-reach), box size, and shift length. For a given condition of frequency, height, and box size, subjects selected MAWs first assuming an 8-hour shift and then assuming a 12-hour shift. Similar to the Liberty Mutual studies, the following results were significant (p < .05):
Over all conditions tested, the MAW for males decreased by an average 22% when going from an 8- to 12-hour shift (p < .05). Although the MAW decreased by an average of 12% for females, the difference was not statistically significant. Energy expenditure rates, based on measurement of oxygen uptake and computed as a percentage of aerobic capacity, decreased with the longer work shift. For males, average energy expenditure decreased from 29% to 23% of aerobic capacity. For females, the corresponding change was from 28% to 24% of aerobic capacity. Neither change was statistically significant. A.3.c Studies of Asymmetric Lifting In studies at the University of Wisconsin, Garg and Badger (1986) used the psychophysical method to compare MAW for two-handed floor-to-table (81 cm) lifts in the sagittal plane versus two-handed floor-to-table lifts at asymmetry angles of 30, 60, and 90 degrees in a laboratory study of 18 male college students. MAW decreased by 7%, 15% and 22% as the asymmetry angle deviation from the sagittal plane increased (p < .05). MAW also decreased with larger box sizes (p < .01). In a follow-up experiment, Garg and Banaag (1988) studied the effects of lift symmetry, frequency, and height of lift on MAW in a laboratory study of eight male college students. MAW at angles of 30, 60, and 90 degrees was compared to MAW in the sagittal plane with mean observed decreases of 9%, 14%, and 21% (p < .01). In addition, MAW decreased with increasing lift frequency (p < .01). At the University of Cincinnati, Mital and Fard (1986) studied 18 male students to determine the effects of lift symmetry, load symmetry, load size, frequency, and height of lift on MAW. Heart rate and oxygen uptake were also monitored during the experiment. Major findings included the following:
Neither lift symmetry nor load symmetry had a significant effect on heart rate or oxygen consumption. A.3.d Studies of the Effects of Handles and Container Shape Garg and Saxena (1980) used the psychophysical method to determine the effects of handles, container shape, and container dimensions on MAW in a laboratory study of 10 college students at the University of Wisconsin. Subjects lifted six different-sized tote boxes and three different-sized mail bags from the floor to a 76 cm bench. Each of the tote boxes had two configurations: with and without handles. The lack of handles on boxes decreased MAW by 7.2% averaged across all box sizes and subjects (p < .01). MAW for the smallest box (38 x 51 cm) was on average 10% greater than MAW for the largest box (64 x 64 cm) for both handle and no-handle conditions (p < .01). MAW averaged across all mailbags was greater than the average MAW for no-handle boxes but less than the average MAW for handle boxes (p < .01). Ciriello, Snook, and Hughes (1993) evaluated the effects of handles vs. no handles on MAW using six male industrial workers in a laboratory study performed at Liberty Mutual. When workers lifted boxes without handles, MAW was consistently lower (median reduction of 16%) compared to lifting similar boxes equipped with handles. A.3.e Studies of Maximum Acceptable Weights in Restricted Work Postures Gallagher (1991) used the psychophysical method to determine MAW under conditions of restricted headroom where it is impossible for a worker to stand fully erect (such as low-ceiling coal mines). Eight experienced coal miners served as subjects in a laboratory study to evaluate the effects of posture (kneel vs. stoop), lift symmetry, and lift distance (35 cm vs. 60 cm) on MAW. All tests were performed under a 1.22 m ceiling to prevent the subjects from standing. Mean MAW was reduced by 11% in the kneeling posture compared to stooping (p < .05). MAW was significantly greater under asymmetric conditions (p < .01) and at the smaller lift distance (p < .05); however, the relative differences were small (less than 5%). Smith, Ayoub, and McDaniel (1992) used the psychophysical method to determine MAW when lifting, lowering, or carrying in non-standard postures such as twisting, lying down, kneeling, squatting, and work with restricted ceiling heights. One hundred subjects (50 male, 50 female) recruited from a college-age population participated in this laboratory study at Texas Tech University. Although this study did not include measurements of MAW in normal lifting (two-handed, symmetric sagittal plane) to use as a basis of comparison, the following trends were observed:
A.3.f Comparison of Psychophysical Findings to Other Criteria for Job Design Several studies have been performed to compare MAWs established using the psychophysical approach against energy expenditure and biomechanical criteria for designing manual materials-handling tasks. Investigators have compared energy expenditure when working at psychophysically determined MAWs to the NIOSH (1981) recommendation of 3.5 kcal/minute to avoid excessive physiologic fatigue. These studies (Karwowski and Yates, 1986; Ciriello and Snook, 1983; Ciriello et al., 1990) have shown that at rapid lifting frequencies (4.3 lifts per minute or faster), psychophysically determined MAWs exceed the NIOSH-recommended energy expenditure levels. For low-to-moderate lifting frequencies, energy expenditure when working at the MAW was below the 3.5 kcal/min criterion. At the other extreme, MAWs determined for very low lifting frequencies (once every 5 minutes) have been used to compute spinal compression forces using biomechanical models. A recent analysis by Chaffin and Page (1994) for floor-to-knuckle-height lifts found that the "once per 5-minute lift" MAW acceptable to a relatively large proportion of the working population (90% of adult males based on Snook and Ciriello [1991] tables) resulted in L5/S1 compression forces that exceed the 3,400 N limit recommended by NIOSH (Waters et al., 1993). Questions have also been raised regarding whether MAWs established during a relatively short experimental session (typically on the order of 30 minutes) are valid for extended work periods. Mital (1983), for example, found decreases in MAW at the end of an 8-hour experimental session when workers lifted at frequencies greater than six lifts/minute. Ciriello et al. (1990) found stability in MAWs during 4-hour sessions as long as the lifting frequency was slower than 4.3 lifts/minute. A.3.g Summary of Psychophysical Studies of Manual Materials Handling The psychophysical method has been used in a series of laboratory studies to determine MAWs (for lifting and lowering tasks) and MAFs (for pushing and pulling tasks). Using this approach, subjects adjust the level of weight they are willing to lift (or the level of force they are willing to exert) as task requirements change. These studies have demonstrated that the following task factors are significant:
There is limited evidence (Snook, Campanelli, and Hart, 1978; Herrin, Jaraiedi, and Anderson, 1986) supporting the use of psychophysical guidelines for designing manual materials-handling tasks. Potentially up to one-third of compensable back injuries could be prevented by designing jobs to accommodate at least 75% of the working population. A.4 Summary of Biomechanical and Psychophysical Studies of Risk Factors Related to Low-Back Pain This paper summarizes recent biomechanical and psychophysical research on workplace factors associated with low-back pain. The biomechanical studies have examined the relationship between selected work parameters (e.g., weight lifted, reach distance, posture) and selected strain responses of body tissue (e.g., EGM activity of muscles, intradiscal pressure, job strength requirements vs. worker strength capabilities). The psychophysical studies have examined the relationship between selected work parameters and the amount of weight that people are willing to handle without excessive fatigue. Table II-2 presents a summary of task characteristics found to be significantly relevant to one or more biomechanical or psychophysical measures. Considerable additional research is needed to understand the quantitative relationships between exposure to these factors and the incidence and severity of work-related overexertion injuries and disorders of the lower back. Nonetheless, these task attributes should be considered during job evaluation and job design procedures to reduce exposures to those factors proven to cause increased biomechanical and/or psychophysical strain. Table II-2. Summary of job and task factors significantly related to biomechanical and/or psychophysical measures of strain.
A.5 References
B A REVIEW OF BIOMECHANICAL AND PSYCHOPHYSICAL RESEARCH ON RISK FACTORS ASSOCIATED WITH UPPER-EXTREMITY DISORDERS B.1 Introduction Section A reviewed biomechanical and psychophysical research on risk factors associated with low-back pain. Section A.1 presented general information on the morbidity of MSDs in the United States, a brief overview of biomechanical and psychophysical research methods, and a brief discussion of how laboratory and epidemiological studies complement each other in understanding the causes and prevention of overexertion injuries and disorders. For reasons of brevity, those topics are not repeated here. Pain and discomfort in the upper extremities are associated with a variety of underlying disorders of the muscles, tendons, nerves and/or blood vessels in the neck, shoulder, elbow, wrist, hand, and/or fingers. Common overuse disorders affecting these tissues include myalgia, myofacitis, tendinitis, synovitis, tenosynovitis, de Quervain's disease, epicondylitis, carpal tunnel syndrome, cubital tunnel syndrome, and vibration-induced white finger syndrome. The etiology of these disorders is quite complex; epidemiological studies have identified significant relationships between these outcomes and a spectrum of both personal and occupational risk factors. Personal risk factors include age, gender, previous acute trauma, rheumatoid arthritis, diabetes mellitus, hormonal factors, wrist size/shape, and vitamin deficiency. Occupational factors include performance of jobs that involve repeated/sustained exertions, forceful exertions, awkward postures, mechanical contact stresses, exposure to vibration, exposure to low temperatures, and work organization (Armstrong et al., 1986; Armstrong et al., 1993; Hagberg et al., 1995). It has been shown that the prevalence of MSDs increases with exposure to certain risk factors; however, it is not known at what level the risk becomes significantly elevated for a single factor or combination of work factors. A significant biomechanical research effort has been devoted to understanding how the muscles, tendons, and nerves of the upper extremity are affected by the parameters of job demands and tool design in hand-intensive work. Psychophysical methods have also been used extensively to evaluate workers' perceptions of hand-intensive work and to determine acceptable levels of work intensity. Table II-3 provides a summary of the key biomechanical and psychophysical articles reviewed here. Work factors associated with increased risk of upper-extremity disorders are listed in the left column. For each risk factor, the results or relevant laboratory studies and biomechanical model predictions are briefly described. A more detailed discussion of these papers is presented below. Table II-3: Risk factors associated with upper-extremity disorders -- a summary of literature describing laboratory experiments and biomechanical models.
Table II-3: Risk factors associated with upper-extremity disorders -- a summary of literature describing laboratory experiments and biomechanical models (continued).
Table II-3: Risk factors associated with upper-extremity disorders -- a summary of literature describing laboratory experiments and biomechanical models (continued).
Table II-3: Risk factors associated with upper-extremity disorders -- a summary of literature describing laboratory experiments and biomechanical models (continued).
Table II-3: Risk factors associated with upper-extremity disorders -- a summary of literature describing laboratory experiments and biomechanical models (continued).
Table II-3: Risk factors associated with upper-extremity disorders -- a summary of literature describing laboratory experiments and biomechanical models (continued).
B.2 Biomechanical Studies of Upper-Extremity Disorders Prehensile activities, such as gripping and pinching, involve exertion of the extrinsic finger flexor muscles (the flexor digitorum profundus, flexor digitorum superficialis, and pollicus longus), which are located in the forearm. These muscles are attached to the fingers by long tendons that pass through the carpal tunnel, a rigid anatomical structure formed by the seven carpal bones on the dorsal side of the wrist and the flexor retinaculum on the palmar side of the wrist. These tendons are surrounded by synovial sheaths that reduce sliding friction at points where the tendon passes through a confined area or over bony trochleas. The median nerve also passes through the carpal tunnel, and is situated between the flexor tendons and the flexor retinaculum. A sketch depicting these anatomical structures is presented in Figure II-2. B.2.a Static Biomechanical Models of the Wrist During pinching and grasping, the finger flexor muscles contract, creating a tension force in the tendon that pulls the tip of the finger in the palmar direction. Static biomechanical models developed at the Mayo Clinic (Chao, Opgrande, and Axmear, 1976) and the University of Michigan (Armstrong and Chaffin, 1979) have been used to predict the amount of tension in the finger flexor tendons during various work tasks. When a normal (i.e., perpendicular) load is applied to the palmar side of the fingers during pressing, pinching, or grasping actions, mechanical moments are created at the finger and wrist joints. Depending on the grasp type, the tensile force in the muscles and tendons required to oppose these moments is approximately 2.8 to 4.3 times the normal force acting on the fingers (Armstrong, 1976). The multiplier is higher when the normal force is exerted only with the fingertips as when pressing or pinching with the pulps of the distal phalanx. The load is lower when the fingers can wrap around the object using a power grip. When the wrist is straight, the tendons are only subjected to a tensile load. However, if the wrist is bent, the tendons are exposed to compressive and frictional forces as the tendon slides across the internal boundaries of the carpal tunnel. The angle of the wrist and the tensile force are therefore important factors in determining the total force on the tendon. Exertions of the hand performed with a deviated (i.e., bent) wrist create greater force on the tendons than exertions with a straight wrist (Armstrong and Chaffin, 1979). In a cadaver study, Goldstein et al. (1987) found that wrist deviation (either flexion or extension) increased the shear traction forces exerted by the carpal bones and flexor retinaculum on the flexor tendons and their sheaths. Traction forces also increased as a function of the tensile load on the tendon. In another aspect of this cadaver study, cyclical loading similar to work/rest cycles on repetitive jobs produced viscoelastic creep in the finger flexor tendons.
 Moore, Wells, and Ranney (1991) used the Armstrong model to estimate several parameters of strain on the extrinsic flexor muscles and tendons during a simulated caulking task performed in a laboratory. Eight task conditions were tested: four combinations of trigger force and exertion frequency (low force/low repetition, low force/high repetition, high force/low repetition, high force/high repetition) were performed using a normal (straight) and flexed wrist. Six subjects performed each condition for approximately 5 minutes. Using the low-force condition as a basis for comparison, peak tensile force in the flexor tendons was approximately doubled under the high-force conditions (p < .001). Using a straight wrist as basis for comparison, computed normal forces on the tendons were approximately tripled when the subjects worked with a bent wrist. Exertions with a bent wrist also create mechanical stresses on the median nerve. Flexion causes the nerve to be compressed between the flexor tendons and the flexor retinaculum. Extension causes the nerve to be stretched over the tendons and the head of the radius (Phalen, 1966; Smith et al., 1977; Gelberman et al., 1981). The flexed wrist posture (Phalen's test) has been used by clinicians to compress the median nerve and induce symptoms associated with carpal tunnel syndrome (CTS) (Phalen, 1972). B.2.b Dynamic Biomechanical Models of the Wrist Schoenmarklin and Marras (1990) at Ohio State University enhanced the Armstrong static model by adding the dynamic component of acceleration. Because wrist motions are controlled by the extrinsic flexor and extensor muscles in the forearms, dynamic activities create additional tension in the tendons as they pass through the wrist. When the wrist is accelerated while in an extreme flexed or extended posture, the combined effects of overcoming inertia and the deviated posture create both tensile and normal loads in the tendons that are significantly higher than those required to simply maintain a static grasp. Furthermore, the normal load on the tendon increases compression on the median nerve as it passes between the flexor tendons and the flexor retinaculum when the wrist is accelerated during a flexed posture. In a pilot study that compared biomechanical attributes of jobs with high and low rates of upper-extremity cumulative trauma disorders (CTDs), Marras and co-workers measured wrist position, velocity, and acceleration on 40 workers assigned to 20 different jobs (10 jobs with high rates of CTDs, 10 jobs with low rates). Velocity and acceleration were significantly higher in the high injury-rate jobs. Peak wrist acceleration was found to be the best discriminator between and high- and low-risk jobs (Marras and Schoenmarklin, 1993; Schoenmarklin and Marras, 1994). B.2.c Biomechanical Studies of Intra-Carpal Pressure Cadaver Studies Exertions with non-neutral wrist postures, such as extreme flexion or extension, increase hydrostatic pressure within the carpal tunnel, augmenting compression of the median nerve. Tanzer (1959) measured carpal tunnel pressure (CTP) in six patients diagnosed with CTS and in six cadavers. Similar trends were found in both sets of wrists -- flexion and extension postures produced increased CTP in the proximal portion of the carpal tunnel while only extension increased CTP in the distal portion of the tunnel. In another cadaver study, Smith, Sonstegard, and Anderson (1977) replaced the median nerve with a water-filled cylindrical tube and found the pressure inside the tube increased when the wrist was flexed to an extreme angle. Pressures also increased when the finger flexor tendons were tensed while the wrist was flexed. Clinical Studies Gelberman et al. (1981) measured CTP in 15 patients with diagnosed CTS and 13 controls. Wrist flexion and extension resulted in increased CTP in both patients and controls. In a more recent study, Weiss et al. (1995) examined CTP in four CTS patients and 20 controls. In both patients and controls, CTP followed a parabolic arc as a function of wrist position in the flexion/extension and radial/ulnar directions. In flexion-extension, mean CTP was approximately 100 mm Hg at 60 degrees extension, approximately 5 mm Hg in the neutral position, and approximately 80 mm Hg at 60 degrees flexion. In radial/ulnar deviation, mean CTP was approximately 100 mm Hg at 40 degrees ulnar deviation, approximately 10 mm Hg in the neutral position, and approximately 90 mm Hg at 40 degrees radial deviation. Rempel et al. (1994) measured CTP in healthy subjects under four test conditions: resting with and without a wrist splint and performing a repetitive task with and without a wrist splint. The repetitive task involved handling 0.45 kg cans 20 times/minute for a period of 5 minutes. Under unsplinted conditions, CTP rose from a baseline median 8 mm Hg at rest to 18 mm Hg during the task (p < .001). With the splint, CTP rose from a baseline median of 13 mm Hg at rest to 21 mm Hg during the task (p < .001). Differences between splinted and unsplinted conditions were not significant. Mean wrist positions (flexion/extension and radial/ulnar deviations) were similar under splinted and unsplinted conditions; however, the splint significantly reduced the range of motion (ROM) during the 5-minute work period (p < .001). The authors concluded that the median nerve was exposed to elevated hydrostatic pressure within the carpal tunnel during repetitive work. The application of a wrist splint did not reduce CTP during work. CTP increased at rest with the splint, probably as a result of direct external pressure on the carpal canal. In a follow-up study, Rempel et al. (1997) measured CTP as a function of force exerted by the tip of the index finger and wrist angle. Fifteen subjects with no history or symptoms of CTS participated in this laboratory study. CTP was significantly affected by both fingertip force (p < .001) and wrist posture (p < .01); however, the interaction was not significant. In pairwise analyses, CTP under any finger load was higher than the no-load condition (p < .05). CTP was lowest in the neutral and near-neutral postures and was highest at 45 degrees extension. Werner, Bir, and Armstrong (1994) measured CTP in five healthy subjects in four different wrist postures: neutral, Phalen's posture (flexion), modified Phalen's posture (flexion combined with pinching), and reverse Phalen's posture (wrist extension). Using the neutral wrist posture as a baseline, CTP increased slightly (5 to 8/mm Hg) for the two flexion postures. CTP increased substantially (30 to 40/mm Hg) in the reverse Phalen's extension posture. In a follow-up study, Werner et al. (1997) measured CTP as a function of hand position (closed grip, three-point chuck pinch, relaxed with fingers slightly flexed, and straight with fingers extended); forearm position (supinated, semi-pronated, and pronated); and wrist position (neutral, flexed, extended, ulnar deviation, and radial deviation). CTP was lowest when the wrist was in a neutral position, the hand was relaxed, and the forearm was slightly pronated. Wrist extension caused the greatest increase in CTP followed by wrist flexion, forearm pronation, and forearm supination. Radial and ulnar deviation produced slight increases in CTP. The authors concluded that the wrist and forearm should be maintained in neutral postures during work activities in order to reduce CTP. Seradge, Jia, and Owens (1995) measured CTP in a laboratory study of 102 hands, 81 with diagnosed CTS and 21 controls. CTP was measured with the hand at rest with a neutral wrist posture, with the wrist in flexion, with the wrist in extension, while the large finger performed a pressing action, while making a fist, and while holding an object in a power grip. Using the resting neutral position as a basis for comparison, CTP increased significantly (p < .05) in both patients and controls for all hand actions. The largest increases were observed during wrist flexion and while making a fist. CTP was also measured following 15 minutes of rest after a 1-minute exercise of repeated wrist and finger flexion and extension. Compared to the initial resting measurement, there was a significant decrease in post-exercise CTP. The authors concluded that carpal tunnel patients should avoid hand activities that cause CTP to increase, and suggest that brief, infrequent hand exercise may be beneficial in reducing CTP. B.2.d Summary of Biomechanical Models and Intra-Carpal Pressure Studies Static and dynamic biomechanical models of the wrist have been used to estimate tensile, normal, and frictional forces acting on the finger flexor tendons during static and dynamic hand tasks. These models show that common work activities such as pinching and gripping produce significant tension in the finger flexor tendons. The magnitude of tensile strain in the tendons increases as a function of the force exerted during the pinching or gripping activity. Due to the effect of longer moment arms, forces exerted with the tips of the fingers (pinches and presses) result in higher tensile forces than power grip exertions of a similar magnitude. Static biomechanical models predict that normal and frictional forces on the finger flexor tendons increase when the wrist deviates from a neutral position due to compression against adjacent anatomical structures. These predictions have been confirmed by cadaver studies of forces between tendons, tendon sheaths, ligaments, and bones. Static models also predict mechanical compression of the median nerve during work activities performed when the wrist is deviated from a neutral posture. Dynamic biomechanical models of the wrist predict that tensile and normal forces on the finger flexor tendons increase substantially during rapid wrist accelerations. These predictions have been supported by a limited cross-sectional epidemiological study that found wrist acceleration to be a significant factor on jobs with elevated rates of upper-extremity CTDs. Studies of intra-carpal pressures in cadavers and living subjects have confirmed the predictions of the biomechanical models. CTP has been shown to increases as a function of several task variables, including the magnitude of force exerted by the fingertips, the angular deviation of the wrist from a neutral position, and the performance of repetitive work. Increased carpal tunnel canal pressure may be a causal factor in the development of a mononeuropathy of the median nerve at the wrist. Increased interstitial fluid pressure causes the capillaries to collapse and interfere with the profusion of the median nerve. This condition may cause symptoms of numbness and tingling that are consistent with CTS (Lundborg, Gelberman, and Minter-Convery, 1962; Gelberman et al., 1981; Werner et al., 1997.) B.3 Psychophysical Studies of Generic Hand Activities B.3.a One-Handed Transfer Tasks Krawczyk, Armstrong, and Snook (1992) used the psychophysical method to determine preferred weights for one-handed horizontal transfer tasks (e.g., the motion used to move an object across a supermarket scanner) performed over an 8-hour shift. Sixteen experienced industrial workers (eight males and eight females) were evaluated in a laboratory setting. Independent variables were transfer distance (0.5 and 1.0 meters) and transfer frequency (10, 20, and 30 times/minute). Dependent variables were MAW and perceived exertion. Distance and frequency were significant (p < .05) determinants of both MAW and perceived exertion. The mean preferred weight decreased from 6.5 kg in the short-distance, low-frequency (0.5 m, 10 times/minute) condition to 3.2 kg in the long-distance high-frequency (1.0 m, 30 times/minute) condition. Task duration was also significant (p < .001) with MAW decreasing and perceived exertion increasing over the 8-hour session. B.3.b Maximum Acceptable Torque During Repetitive Wrist Motions In a Liberty Mutual study, Snook and Ciriello (1995) used the psychophysical method to determine maximum acceptable torque (MAT) during repeated flexion and extension of the wrist. Tactile sensitivity and symptoms were also measured. Sixteen females were evaluated in a laboratory setting, performing the task for 8-hours on a twice-a-week schedule. Independent variables were motion type (wrist flexion with power grip, wrist flexion with pinch grip, and wrist extension with power grip), and frequency (2, 5, 10, 15, and 20 motions/minute). Frequency was significant (p < .05) with MAT higher at low repetition rates. MAT also decreased (p < .05) over the course of the 8-hour session. MAT was higher (p < .05) in flexion motions compared to extension motions, but no significant differences were found between pinch and power-grip. Interactions were not significant. Contrary to expectations, tactile sensitivity improved over the course of the testing session and at higher repetition rates (p < .05). Symptoms of discomfort increased with repetition. Wrist extension produced more symptoms than flexion-pinch and flexion-power-grip motions. In a follow-up study, Snook and Ciriello (1997) measured MAT, symptoms, and tactile sensitivity during repeated ulnar deviation of the wrist. Sixteen females participated in this laboratory study. The primary independent variable was repetition rate: (15, 20, and 25 motions per minute). There was a slight decrease in MAT with increased task frequency and a slight increase in symptom; however, these changes were not statistically significant. B.3.c Maximum Pinch Force and Pinch Frequency In a University of Texas study, Imrhan (1991) evaluated the effects of pinch-grasp type and wrist posture on maximum voluntary pinch strength (MVPS) in a laboratory experiment using 30 male subjects. Pinch type was tested at four levels (lateral, chuck, pulp 2, and pulp 3) and wrist posture was tested at five levels (neutral, radial deviation, ulnar deviation, extension, and flexion) in a full factorial design. Overall, the lateral pinch MVPS was about 150% of the two pulp-pinch values (p < .001). MVPSs were highest for the neutral wrist and lowest for the flexed wrist, regardless of pinch type (p < .001). The mean neutral wrist MVPS was about 170% of the mean flexion MVPS. Dempsey and Ayoub (1996) performed a similar study at Texas Tech University with several extensions and methodological differences. Gender (eight males, eight females) and pinch width (1 cm, 3 cm, 5 cm, and 7 cm separation distances) were added as independent variables. Pinch width was a significant main effect (p < .01); maximum pinch strength occurred at the 5 cm separation. Gender was also significant (p < .01); the mean female strength was 63% of the mean male strength. Similar to Imrhan (1991), the lateral pinch was the strongest grasp type while the pulp pinches were the weakest (p < .01). The neutral wrist was the strongest posture while the flexed wrist was the weakest. Overall, Dempsey and Ayoub (1996) obtained strength values that were approximately 50% of those obtained by Imrhan (1991). This was attributed to methodological differences. Dempsey and Ayoub (1996) used a 3-second sustained exertion while Imrhan (1991) used an instantaneous peak exertion. At Wichita State University, Klein and Fernandez (1997) used the psychophysical method to determine MAF during repeated lateral pinches under varying conditions of force, pinch duration, and wrist posture. EMG activity of the forearm flexors and extensors, blood pressure, heart rate, perceived exertion, and discomfort were also measured. Task frequency was paced by a metronome; however, subjects (12 male college students) were instructed to adjust the metronome frequency to the level they considered to be reasonable for an 8-hour day. Perceived exertion and discomfort were rated after each trial. MAF decreased with increased force, increased duration, and wrist deviation (p < .05). The EMG analysis showed similar findings; EMG activity increased and median frequency decreased with increased force, duration, and wrist deviation. Ratings of perceived exertion were significantly higher with increased force, duration, and wrist deviation, as were body discomfort ratings. In the analysis of heart rate, force was the only significant factor; heart rate was lower at 15% (MVC) compared to the two other treatments (30% and 50% MVC). No significant effects were found in the analysis of blood pressure. B.3.d Summary -- Psychophysical Studies of Generic Hand Activities The above-cited studies considered different types of generic hand activities and therefore are not directly comparable. However, several consistent trends were observed:
B.4 Laboratory Studies of Specific Tasks and Tools Laboratory studies using both biomechanical and psychophysical approaches have been used to evaluate the design and operational features of workstations, tasks, and tools. The following sections present a brief summary of laboratory findings from studies of powered hand tools, keyboards, gloves, and workstation layouts. B.4.a Pneumatic-Powered Hand Tools -- Threaded Fasteners and Drills Biomechanical Studies At the University of Wisconsin, Radwin, VanBergeijk, and Armstrong (1989) used EMG to measure muscle activity in the finger flexor and extensor muscles in a laboratory study of right-angle nutrunners equipped with shut-off torque control. Five subjects (four males, one female) participated. Independent variables included torque output (four levels ranging between 30 Nm to 100 Nm) and joint stiffness (two levels: a "hard" joint requiring 0.5 seconds to build up to the shut-off torque and a "soft" joint with 2.0 seconds build-up time). Muscle activity was measured during four phases of a simulated task: 1) pre-start -- holding the tool prior to activation; 2) run down -- free spinning of the fastener, prior to the onset of torque build-up; 3) torque-reaction -- period of torque build-up prior to shut-off; and 4) post-shut-off -- holding the tool after air supply to the motor has been cut off. Finger flexor forces, estimated from EMG measurements, were four times higher during the torque reaction phase than during the pre-start and post-shut-off phases, and two times higher during the torque reaction phase than during the run-down phase. During the torque reaction phase, average finger flexor forces increased from 372 N when using the 30 Nm tool to 449 N when using the 100 Nm tool (p < .01). The authors interpreted this result as an indication that the larger handle length on the high-torque tool did not provide the necessary increased mechanical advantage to resist the higher torque. They suggested several interventions for reducing hand forces when using high-torque tools, including longer handles, torque-reaction bars, and torque-absorbing suspension systems. Average flexor forces increased from 390 N to 440 N when moving from the soft joint to the hard joint (p < .01). However, the impulse force, calculated by integrating flexor EMG activity over the full duration of the torque reaction phase, decreased from 839 N with the soft joint to 312 N with the hard joint (p < .01). In a recent University of Wisconsin study of right-angle nutrunners, Oh and Radwin (1997) used six subjects (four males, two females) to evaluate muscle activity and tool stability when using a right-angle nutrunner. Dependent variables included measures of tool stability: peak handle velocity (PHV) and peak handle displacement (PHD), work done on the hand-tool system, power involved in performing the work, and EMG activity. Independent variables were torque level (25 Nm and 50 Nm), joint hardness (35 ms and 900 ms build-up time), workstation orientation (vertical vs. horizontal), and operator reach distance (10 cm and 35 cm). Significant results included the following:
Freivalds and Eklund (1993) used EMG and subjective ratings of perceived exertion to evaluate six different nutrunners in a laboratory experiment. Four subjects (three males, one female) with varying experience using powered hand tools participated. Muscle activity in the finger flexors and trapezius musclesshowed a significant interaction (p < .05) between tool type and work surface orientation. EMG activity in both muscle groups was lowest when using in-line tools on horizontal surfaces and pistol-grip tools on vertical surfaces. EMG activity in the finger flexors increased with tool torque (p < .05); however, there was no significant relationship between trapezius EMG and torque. Ratings of perceived exertion were negatively correlated with tool impulse (torque integrated over time, measured in units of Nm/s). Trials where torque quickly built up to shut-off levels (because of a high revolutions-per-minute [RPM] tool and/or a stiff joint) resulted in lower exertion ratings than trials where the torque build-up occurred slowly. Subjects also preferred pulse-type tools based on ratings of perceived exertion. The authors attributed this preference to the fact that because pulse-type tools cycle quickly between torque-on and torque-off states, the inertia of the tool resists the applied torque, reducing the impulse sensed by the operator. Kilberg, Kjellberg, and Lindbeck (1995) used the psychophysical method to evaluate the acceptability of torque reaction when using a right-angle nutrunner. Four different nutrunners were tested: three operated at a torque of 50 Nm and utilized three different shut-off mechanisms: fast, slow, and delayed. The fourth tool operated at 75 Nm and used a delayed shut-off. Thirty-eight subjects (37 males, one female) participated in the experiment; all were experienced employees in a truck assembly plant. Subjects rated their discomfort using a 20-point scale (0 = no discomfort at all, 20 = almost unbearable). Subjects also made a yes/no determination to whether they would be willing to use the tool over the course of a full workday. Biomechanical measurements, including handle displacement, tool impulse (torque integrated over time as described in ISO Standard 6544), and ground reaction forces were taken on a subset of subjects due to time limitations and equipment malfunctions. Significant differences were found among all four tools in the mean discomfort rating (p < .001). The 75 Nm tool had the highest discomfort score (14 = very discomforting) while the 50 Nm fast shut-off tool had the lowest score (2 = slightly above the "no discomfort" anchor). The 50 Nm slow shut-off tool was rated 4 (somewhat discomforting) while the 50 Nm delayed shut-off tool was rated 9 (rather discomforting). Only the 50 Nm fast shut-off tool was considered to be acceptable for a full day of work by 100% of the subjects. Discomfort and acceptance ratings were strongly correlated with the biomechanical measures of hand tool displacement, ground reaction forces, and tool impulse. As these biomechanical measures increased, discomfort increased and acceptability for a full day's work decreased. Psychophysical Studies In a University of Michigan survey of autoworkers, Armstrong, Punnett, and Ketner (1989) used a psychophysical approach to evaluate tool attributes and workstation characteristics of pneumatic tools used in automobile assembly. Twenty-three autoworkers (13 females, 10 males) used a visual analog scale to evaluate the following characteristics of the workstations and tools used to perform their jobs: tool mass, required grip force, handle size, horizontal reach, and vertical reach. Significant (p < .05) findings included the following:
Gender was not a significant factor in this study. Although females responded to heavier tools with somewhat higher discomfort ratings than males, the differences were not statistically significant. In a University of Michigan laboratory study, Ulin et al. (1990) investigated the effects of working height on perceived exertion while using a pistol-shaped screwdriver. Thirty-six students (18 males, 18 females) served as subjects, driving screws into vertically mounted perforated sheet metal. The independent variable was working height; 25 screws were driven at each of seven levels, ranging between 38 and 191 cm. Dependent variables included perceived exertion (measured using the Borg 10-point ratio scale and two visual analog scales [VAS]), time required to drive each screw, and the number of cross-threaded screws. Working height was a significant factor (p < .01) in all three measures of perceived exertion. Exertion was lowest at 114 cm (approximately elbow height) regardless of the subject's body size. Exertion was greatest at 38 cm, the lowest height tested, and at 191 cm, the highest height tested. Subject anthropometry was a factor at 191 cm, with smaller subjects reporting significantly higher perceived exertion than the tallest subjects (p < .01). Working height was a significant factor in the time required to drive a screw. At 191 cm, approximately 16% more time was required than for the other six levels. The number of cross-thread screws was highest at the 165 cm and 191 cm levels; however, the differences were not significant from other levels. In a follow-up study, Ulin et al. (1992) evaluated three types of pneumatic screwdriver (in-line, pistol, and right-angle) on both horizontal and vertical surfaces. Independent variables included work surface orientation (vertical and horizontal), work location (seven heights on the vertical surface, four reach distances on the horizontal surface), and tool type. The dependent variable was the Borg 10-point ratio scale rating of perceived exertion. A working height of 114 cm was preferred for the vertical surface; the highest Borg ratings occurred at the lowest (38 cm) and highest (191 cm) heights. At the 114 cm level, the pistol-shaped tool resulted in the lowest Borg ratings. (Note: In a recent intervention study of furniture assembly where workers were allowed to introduce changes in work methods to reduce musculoskeletal load on the upper extremities and lower back, Hakkanen, Viikari-Juntura, and Takala [1997] found that pistol-shaped tools were substituted for in-line tools when drilling and driving screws into a vertical surface at elbow height.) Reach distances of 13 cm and 38 cm produced the lowest Borg ratings when working on the horizontal surface. Over all reach distances, the right-angle tool produced the lowest exertion scores while the pistol-shaped tool produced the highest scores. Anthropometry was a factor in high vertical reaches and long horizontal reaches. The shortest subjects reported higher scores than taller subjects at the 191 cm vertical height and the 88 cm horizontal reach. Fernandez and co-workers at Wichita State University have used the psychophysical method to study pneumatic drills. Although drilling operations are somewhat different from driving screws and nuts, the results of the Wichita studies are reported here due to similarities between pneumatic drills and drivers. Marley and Fernandez (1995) measured MAF during simulated use of a pistol-shaped drill under varying conditions of wrist posture. The task required subjects to exert a 5.4 kg force with a simulated drill bit for a period of 1 second. EMG activity of the forearm flexors and extensors, blood pressure, heart rate, and perceived exertion (using the Borg 20-point scale) were also measured. Twelve female college students with no history of upper-extremity CTDs served as subjects. Wrist posture was set at nine levels in a full-factorial design (flexion: neutral, one-third maximum ROM and two-thirds maximum ROM; ulnar: neutral, one-third maximum ROM and two-thirds maximum ROM). Task frequency was paced by a metronome; however, subjects were instructed to adjust the metronome frequency to the level they considered to be reasonable for an 8-hour day. Wrist flexion angle was a significant factor for several dependent variables. At one-third maximum flexion, MAF was 88% of MAF with a neutral wrist; at two-thirds maximum flexion, MAF was 73% of the neutral posture value (p < .05). Using the neutral wrist posture as a baseline, the following physiologic responses were observed with the wrist in the two-thirds flexion position: heart rate increased by 4%, systolic blood pressure increased by 5%, forearm flexor EMG root mean square (RMS) activity increased by 118%, and deltoid EMG RMS activity increased by 53% (all significant at 0.05 level). Ratings of perceived exertion increased as follows: wrist: + 16%, forearm: + 25%, shoulder: + 17%, whole body: + 24% (all significant at 0.05 level). No significant relationships were found between any of the outcome variables and ulnar deviation. In a related study, Kim and Fernandez (1993) used the psychophysical method to determine MAF during simulated use of a pistol-shaped drill under varying conditions of wrist posture and applied force. Wrist angle was tested at three levels (neutral, 10 degrees flexion, 20 degrees flexion) and force was tested at four levels (2.4 kg, 5.5 kg, 8.2 kg, and 10.9 kg). Fifteen female college students with no history of upper-extremity CTDs served as subjects. Force and angle were significant (p < .01) for the following variables: MAF, heart rate, and Borg range of perceived exertion (RPE) ratings in the upper extremity and whole body. MAF decreased with increased force and with increased deviation from neutral while heart rate and Borg ratings both increased. Increased force produced significant increases in EMG activity level in the flexor and deltoid muscles; however, the relationship between EMG and angle was not significant. Summary of Biomechanical and Psychophysical Studies of Pneumatic Power Tools Laboratory studies of pneumatic power tools demonstrate that several characteristics of tool and task design have a significant effect on both biomechanical load and psychophysical measures:
B.4.b Laboratory Studies of Keyboarding Tasks Studies of Keyboard Characteristics At the University of Michigan, Armstrong et al. (1994) examined relationships between keyboard characteristics and forces exerted during keyboard operation. Three different keyboards were tested, with varying "make" force (the minimum amount of force that must be exerted to register a key stroke), "break-away" force (the reduction in force or "click" felt after a keystroke is registered), "make" travel (the displacement required to register a keystroke), and "total" travel (the maximum displacement between the highest and lowest positions of the key). Ten subjects (four males, six females) participated in the study; all were experienced typists. Force was measured with strain gauges positioned under the keyboard. Forces exerted while keying were significantly related to several variables, including subject (p < .01), keyboard type (p < .01), keying rate (p < .01), and several first-order interactions. Maximum forces were 2.5 to 3.9 times the make forces, indicating that keys were depressed all the way to the mechanical stops (i.e., beyond the minimum required travel distance). Forces were lowest for the keyboard with the smallest make force. Forces tended to decrease as the keying rate increased. The authors drew no conclusions regarding relationships between keyboard characteristics and MSDs. In a University of California study, Rempel et al. (1994) used a piezoelectric load cell mounted in the "H" key of a computer keyboard to measure forces exerted on a single key during a keystroke. Four male subjects with varying degrees of typing skill participated. Three distinct force exertion phases were identified: 1) keyswitch compression -- force buildup required to overcome the make force; 2) finger impact -- force at the bottom of key travel; and 3) pulp compression -- deformation of the finger pulp while the key remains stationary, prior to key release. Force during the key compression phase was approximately equal to the keyswitch make force of 0.6 N. The maximum force occurred at finger impact and ranged between 1.6 N and 5.3 N, depending on subject. Force dropped during the pulp compression phase with peak values ranging between 1.4 N. to 2.5 N. The results of this study were consistent with the Armstrong et. al. study reported above -- in both cases forces exerted by the typists exceeded the make forces of the keyswitch. Fernstrom, Ericson, and Malker (1994) used EMG to measure muscle activity levels in the shoulders and forearms of eight subjects (one male, seven female) while performing a typing task on five different keyboards: a mechanical typewriter (make force = 1.8 N, key travel = 8 mm); electromechanical typewriter (make force = 0.6 N, key travel = 5 mm); electronic typewriter (make force = 0.35 N, key travel = 3 mm); standard IBM-XT keyboard (make force = 0.5 N, key travel = 4 mm); and modified IBM-XT keyboard with 20 degree slope to middle keys (make force = 0.5 N, key travel = 4 mm). EMG activity was normalized as a percentage of MVC for each muscle group within each subject. Muscle activity was less than 10% MVC for almost all muscle groups and keyboards; the only exception was that finger extensor activity exceeded 10% for the mechanical and electromechanical keyboards. Muscle activity in the finger flexor and extensors was significantly greater (p < .05) when using the mechanical keyboard compared to any of the more modern options. When a palm rest was used in a subset of treatments, there were no significant changes in the activity level of any muscle groups. Gerard et al. (1996) measured typing speed, EMG activity in finger flexor and extensor muscles, and keyboard reaction force in a laboratory study of six experienced typists. Subjects typed for two 2-hour sessions, using a keyboard with a 0.28 N make force and a keyboard with a 0.83 N make force. The subjects were instructed to type at a normal speed. Other than the difference in make force, the keyboards were identical. There was a small decrease in typing speed from a mean of 40 words per minute (wpm) to 38 wpm when using the 0.83 N keyboard; however, this change was not significant. Peak keyboard reaction forces were 54% higher on the 0.83N keyboard (p < .01), while flexor EMG activity was 34% higher (p < .05). There was no significant difference in extensor EMG activity. Change in electrical efficiency (peak force/EMG activity) in the finger flexors was used to characterize fatigue. During the 2-hour session, average electrical efficiency decreased by 12% with the 0.83 N keyboard (p < .05) compared to 2% with the 0.23 N keyboard (n.s.). Studies of Keyboard Characteristics and Typing Speed Sommerich, Marras, and Parnianpour (1996) evaluated the relationship between keyboard reaction force and typing speed in an actual work setting and in a laboratory. Keyboard reaction force was measured using load cells mounted below the keyboard. The 25 subjects in the work setting typed at their normal speeds. No significant relationship was found between typing speed and keyboard reaction force. A weak positive relationship (p < .05) between speed and cumulative force (integrated over time) was found; however, this may have been due to the greater number of keystrokes among the faster typists. Five subjects participated in the laboratory study and typed at three speeds: slower than preferred, preferred, and faster than preferred. Significant positive correlations (p < .05) were found between speed and keyboard reaction force in some subjects and between speed and cumulative force in all subjects when working at a "forced" pace. Effects of Posture and Arm Support Erdelyi et al. (1988) used EMG to evaluate the effects of arm supports during keyboard tasks. Twelve subjects, all women, served as subjects. Several of the subjects had a history of shoulder and neck pain prior to the study. EMG activity in the upper trapezius muscle was recorded using surface electrodes. EMG activity was higher among subjects with a past history of shoulder/neck pain (p < .05). Across all subjects, EMG activity in the trapezius was significantly lower when the arms were supported with either a fixed or hanging device (p < .05). Aarås et al. (1997) used EMG to evaluate muscle loading on the upper trapezius and the erector spinae at the L3 level during keying and mousing activities in a laboratory study. During keyboard use, the load on the right trapezius muscle increased significantly (p < .05) from a mean of 0.8% MVC when sitting with forearm support to 3.6% MVC when sitting without forearm support. The duration of task time when muscle activity was below 1% MVC decreased from a mean of 44% when using forearm support to 10% with no forearm support. During mouse work, forearm support resulted in significant (p < .05) reductions in both trapezius and erector spinae EMG activity. Effects of Task Duration on Discomfort During Keyboard Tasks Schleifer and Amick (1989) evaluated discomfort in the trunk/leg and arm/hand regions as part of a laboratory study of keyboard work. Forty-five professional typists, all female, served as subjects in a 4-day study. Each day, subjects typed for 150 minutes in the morning and afternoon, with a 45-minute lunch break. There was also a 10-minute break in the middle of the morning and afternoon sessions. Discomfort was rated at the beginning of each session, and then at 50-minute intervals. Reported discomfort followed a similar pattern in both the trunk/leg and arm/hand regions. Ratings increased monotonically during the morning and afternoon sessions. Although the lunch break provided some relief, the mean discomfort in both body regions was greater during the afternoon session than during the morning (p < .01). Summary of Biomechanical and Psychophysical Studies of Keyboard Work Laboratory studies of keyboard work demonstrate that several characteristics of the workstation and work organization have a significant effect on both biomechanical and psychophysical measures:
B.4.c Laboratory Studies of Gloves Gloves are routinely worn in many occupations to protect workers against thermal extremes and mechanical injuries (e.g., lacerations, abrasions, blisters). Several laboratory studies have been performed to evaluate how gloves affect human performance during hand-intensive work. Effects of Glove Type Sudhakar et al. (1988) evaluated the effects of glove type on grip strength in a laboratory study of 12 subjects (six male, six female) at Ohio State University. Maximum grip strength and EMG activity in the extrinsic flexors and extensors were measured under three conditions: bare-handed, wearing a rubber glove, and wearing a leather glove. Due to strength differences in subjects, all measurements were normalized by dividing by the subject's bare-handed strength. Grip strength was significantly reduced by wearing gloves (p < .05), but there was no significant difference between the types of gloves. EMG activity was not significantly different across the treatments. The authors concluded that muscle exertions were not affected by the presence of gloves, but that gloves reduced the efficiency of translating exertions into usable grip force. Hallbeck and McMullin (1993) performed a laboratory study at the University of Nebraska to measure changes in maximum power grasp and three-jaw "chuck pinch" force as a function of wrist angle, age, and glove type. Thirty subjects (15 male, 15 female) participated. Independent variables included:
Three-jaw chuck pinch forces were approximately one-fourth to one-third the magnitude of power grasp forces under equivalent conditions of glove type and wrist posture. Power grasp force was significantly affected by glove type. Using the bare hand condition as 100%, strength ranged from 80% in both of the two-glove treatments to 89% with the single cotton thermal glove. Three-jaw pinch force was not significantly affected by glove type. Power grasp strength was significantly (p < .01) affected by wrist position and gender. Using the neutral posture as 100%, power grasp strength ranged from 57% at 65 degrees flexion to 82% at 45 degrees extension. The mean female power grasp strength was 74% of the mean male strength. Three-jaw pinch was also significantly (p < .01) affected by wrist position and gender. Using the neutral posture as 100%, mean strength ranged from 75% at 65 degrees flexion to 92% at 45 degrees extension. The mean female pinch strength was 78% of the mean male strength. Age was not a significant factor in either power grasp or three-jaw pinch. Batra et al. (1994) evaluated the effects of glove thickness in a laboratory study of 52 males. Grip strength was measured bared-handed and while wearing six types of gloves (leather, asbestos, rubber, cotton, open-fingered, and surgical). The ratio of gloved strength/bare-hand strength was computed for each subject to normalize for strength differences. This ratio ranged from a mean of 82% with leather or asbestos gloves to 96% with surgical gloves (p < .05) Fleming, Jansen, and Hasson (1997) performed a laboratory study to determine the effect of cotton-lined leather gloves on time endurance to exhaustion when performing a sustained grip at 60% MVC. Twenty-one subjects (4 male, 17 female) participated in the study. EMG activity was measured throughout the trial and subjects verbally reported perceived exertion levels at 10-second intervals using the Borg 10-point scale. Independent variables included glove (yes, no) and the type of exertion (isometric vs. eccentric). Time to exhaustion was significantly affected by wearing gloves; the mean time of 76 seconds in the no-glove condition was reduced to 58 seconds when wearing gloves (p < .05). There were EMG changes over time (a decrease in mean power frequency) in all conditions; however, none of the independent variables was significant. Gloves and Tool Use In a University of Cincinnati study, Mital, Kuo, and Fard (1994) measured the effects of gloves on the ability to use non-powered wrenches and screwdrivers. Dependent variables included: EMG activity of the extrinsic flexors, EMG activity of the extrinsic extensors, and maximum torque exerted with the tool. The independent variable was glove type; nine different gloves were tested along with a bare-hand condition. Nineteen males participated as subjects. EMG activity was not significantly affected by glove type for either the flexor or extensor muscles. Torque output was influenced by the presence of gloves. Torques exerted while wearing gloves were significantly higher than torques exerted with the bare hand (p < .05). The only exception to this trend was torque exerted with a latex glove; which was not significantly different from the bare hand condition. Summary of Glove Studies Laboratory studies of gloves show that effective grip strength decreases as a result of glove use. Grip strength is approximately 95% of bare-hand values when wearing thin surgical gloves, dropping to approximately 80% of bare-handed levels when wearing thick leather gloves or combinations of gloves. Gloves also decrease the maximum endurance time for a sustained power grip. In at least one study, gloves have been shown to increase the ability to exert torque when using non-powered hand tools such as screw drivers. B.4.d Lab Studies of Shoulder and Neck Posture Studies of Shoulder Elevation Herberts, Kadefors, and Broman (1980) measured EMG activity as a function of static shoulder posture in a laboratory study using 10 male subjects. The primary independent variable was posture. Subjects held a 2-kg load in the hand at waist, shoulder, and overhead heights using different combinations of flexion and abduction at the shoulder. EMG activity was measured using wire electrodes in the anterior and posterior portions of the deltoid, the supraspinatus, the infraspinatus, and the upper portion of the trapezius. Localized fatigue (a shift in EMG mean power frequency [MPF]) was observed in all muscle groups during shoulder-level and overhead work (p < .05) during the 1-minute trials. Even at waist level, fatigue was observed when the upper arm was abducted at an angle of 30 degrees. Hagberg (1981) measured EMG activity and discomfort in the shoulder in a laboratory study of six female subjects. Surface electrodes recorded EMG activity in the descending trapezius, anterior deltoid, and biceps brachii while subjects performed repeated flexion of the shoulder every 4 seconds to an angle of 90 degrees for a period of 60 minutes. Heart rate and perceived exertion using Borg's scale was also recorded. Hand load was the independent variable: weights of 0.6 kg, 1.6 kg, and 3.1 kg were held in the hand (in addition to a no-load treatment). Heart rate and perceived increased over the course of the trial. Heart rate and perceived were greater when a load was held in the hands. EMG activity in the trapezius was closely correlated with the external moment at the shoulder joint. Oberg, Sandsjo, and Kadefors (1994) measured EMG activity and subjective discomfort in the shoulder-neck region in a laboratory study of 20 subjects (10 male, 10 female). Surface electrodes measured EMG activity in the right trapezius muscle while subjects abducted the arm to a 90 degree angle. Subjects reported fatigue using the Borg 10-point scale. Each subject was tested under two conditions: a 5-minute test with no load in the hand and a 2.5 minute test with a 2-kg load in the hand. At the no-load level, there was no change in EMG MPF over the course of the trial; however, subjective fatigue increased. With the 2-kg. load, there was a small linear decrease in MPF over the trial and there was a negative correlation between MPF and the Borg rating (r = 0.46). The authors concluded that MPF was not a good proxy for perceived fatigue during low-intensity static exertions of the shoulder. Studies of Neck Flexion Harms-Ringdahl and Ekholm (1986) measured EMG activity and discomfort in the shoulder-neck region in a laboratory study of 10 female subjects. Surface electrodes recorded EMG activity in the splenus, thoracic erector spinae-rhomboid, and descending portion of the trapezius while subjects were seated while maintaining the neck in a fully flexed posture during a trial of up to 60 minutes. Discomfort was recorded on a 10-cm VAS once per minute. All subjects reported discomfort within 15 minutes; only one subject was able to complete the entire 60-minute session. Following relaxation, the pain subsided within 15 minutes, but 9 of the 10 subjects experienced discomfort later in the day. EMG activity was low (less than 6% MVC), but increased over the period of the trial. Summary Of Shoulder and Neck Studies Laboratory studies of shoulder posture show that muscle activity and subjective fatigue in the shoulder region increases as a function of shoulder elevation angle and load moment at the shoulder joint. There is also evidence of localized muscle fatigue based on a shift in the MPF of the EMG spectrum. Prolonged periods of neck flexion cause increased levels of discomfort and increased EMG activity in the neck extensor muscles. B.5 Summary of Biomechanical and Psychophysical Studies of Risk Factors Related to Upper-Extremity Disorders This paper summarizes recent biomechanical and psychophysical research on factors associated with upper-extremity MSDs. The biomechanical studies have examined the relationship between selected work parameters (e.g., forces exerted during hand-intensive work, wrist postures, shoulder postures, repeated exertions, use of gloves) and selected strain responses of body tissue (e.g., EMG activity of muscles, intra-carpal tunnel pressure, compression of tendons and nerves). The psychophysical studies have examined the relationship between selected parameters and perceived discomfort as well as the relationship between selected parameters and performance levels that can be achieved without incurring excessive fatigue. Table II-4 presents a summary of task characteristics found to be significantly relevant to one or more biomechanical or psychophysical measures. Considerable additional research is needed to understand the quantitative relationships between exposure to these factors and the incidence and severity of work-related overexertion injuries and disorders of the upper extremity. Nonetheless, these task attributes should be considered during job evaluation and job design procedures in order to reduce exposures to those factors proven to cause increased biomechanical and/or psychophysical strain. Table II-4. Summary of job and task factors significantly related to biomechanical and/or psychophysical measures of strain.
B.6 References
|
