Preventing workplace injury

       The increase in workplace computer usage since 1980 has been accompanied by a corresponding increase in incidence rates of chronic musculoskeletal disorders reported in the United States (BLS, 2002).  From 1980 to 2001, the incidence rate across all industries increased from 3.6 to 23 new cases per 10,000 workers, with a high of 41 in 1994. Up to fifty percent of computer workers experienced musculoskeletal symptoms within the first year of employment (Gerr et al., 2002). 
        The estimated average cost of an upper extremity case in 1989 was $8,070 (Webster and Snook, 1994), and the total cost of care in the United States was $563 million for 1989. Considering the increased incidence rates, the total cost at the beginning of the 21st century could be well over one billion dollars.  These reported costs, however, account only for medical costs, and do not include costs associated with workers’ compensation, lost time, training costs, and loss of productivity associated with symptomatic workers. Conservative estimates of the cost to the U.S. economy for all musculoskeletal disorders, which include these indirect costs, range from $45 to $54 billion annually (NSR/IOM, 2001).  
        Several hypotheses for the specific biomechanical injury mechanisms underlying MSDs have been proposed (Kumar, 2001; Moore, 2002). Injuries are associated with several risk factors, including repetition, force, posture, direct pressure, and vibration (Armstrong et al., 1987; Hales and Bernard, 1996; Kurppa et al., 1991; Luopajarvi et al., 1979; Malchaire et al., 1996; Rempel et al., 1992; Silverstein et al., 1987; Sjogaard and Sjogaard, 351; Stock, 1991; Vikari-Juntura, 1998).  For example, over-exertions are correlated with the prevalence of cumulative trauma disorders (Moore and Garg, 1994; Silverstein et al., 1986), such as tendinitis (McCormack et al., 1990) and carpal tunnel syndrome (Falck and Aarnio, 1983).
        Computer work subjects users to many of these risk factors. Using keyboards and pointing devices (e.g. mouse) involves prolonged exposure to force, repetition, and non-neutral postures of the upper extremity (Armstrong et al., 1987; Hales et al., 1994; Marcus et al., 2002; Rempel et al., 1999; Sauter et al., 1991). Epidemiological studies have linked chronic musculoskeletal disorders (MSDs) to prolonged use of computer workstations. For example, an increase in the number of hours spent working on computer video display terminals is associated with an increased risk of upper extremity musculoskeletal disorders (Bergqvist et al., 1995; Faucett and Rempel, 1994; Matias et al., 1998). Forces exerted on the environment (e.g. keyboard) during computer work may be specifically associated with the development of MDSs (Feuerstein et al., 1997; Rempel et al., 1999.
        Characterizing external exposures to risk factors during interactions with the workplace environment is a necessary part of understanding and preventing MSDs. However, external forces and motions are not uniquely related to exposure of the internal tissues (e.g. muscles, tendons, bones) to stresses which directly lead to injury (Hagberg et al., 1997; Fig. 1). External risk factors cause internal exposure through interactions with musculoskeletal elements. However, the internal exposure resulting from external forces and motions can be complex. For example, anatomy can cause internal forces experienced in muscles and tendons to be greater than the external forces (Dennerlein et al., 1998a). Moreover, the dynamics of segmented systems are inherently non-linear, and the potential for differences in internal state (i.e. muscle activity) could result in differences in internal exposure even if the external exposure is the same (Valero-Cuevas et al., 1998). Consequently, external exposure is altered by the dynamic biomechanical system, resulting in internal exposure (Fig. 1). Computer simulations can represent the dynamics of the musculoskeletal system, and facilitate the prediction of specific internal exposures.

        Exposure of musculoskeletal tissues to force and repetitive motion can lead to injury. However, the type and extent of injuries that develop depend on the dose-response characteristics of the affected tissues (Moore, 2002; Winkel and Mathiassen, 1994). Resulting injuries can in turn cause changes to behavior or external exposure, and directly affect internal exposure, presenting the possibility of increased injury through a positive-feedback mechanism (Carter and Banister, 1994).
        Biomechanical models represent a critical link between measurable external exposures and resulting injuries. Anatomically-based models allow the prediction of internal exposure based on direct measurements of external forces and relevant anthropometrical and physiological parameters, and can suggest specific mechanisms of injury (Moore, 2002). Using biomechanical models to analyze and interpret experimental measurements can lead to better predictors of injury outcomes than consideration of external exposure alone (Norman et al., 1998). Both experimental measurements and modelling of the upper-extremity can generate hypotheses of specific injury mechanisms underlying computer work-related MSDs.
        Working with Jack Dennerlein at the Harvard School of Public Health, we conducted studies to characterize finger mechanics and motor control during tapping on a computer keyswitch, towards understanding the causes of upper-extremity musculoskeletal disorders and improved keyboard design. We tested several hypotheses, among them: 1) finger joints act similarly during tapping in terms of kinematics, torque production, and energy production; 2) A simple lumped-parameter model can describe finger impedance during tapping; and 3) Humans employ postures and generate horizontal forces against keyswitches that result in joint torque and energy transfer minimization. We collected simultaneous measurements of finger joint kinematics and endpoint force production during tapping on different types of computer keyswitches. These experiments show that finger joints act differently when tapping. The proximal (metacarpo-phalangeal) joint, flexes and produces the energy necessary to depress the keyswitch. However, during the contact phase of tapping, the distal joints extend and then flex, absorbing and releasing energy in a spring-like manner. A simple spring-damper model can describe the behavior of the distal joints. Humans do not appear to employ postures or patterns of force production that result in torque or energy transfer minimization during tapping on keyswitches. We used these findings to propose a new keyswitch design to reduce joint loading during typing.