|
Note: This section is almost a word for word repeat of Appendix B of "Musculoskeletal Disorders and Workplace Factors: A Critical Review of Epidemiologic Evidence for Work-Related Musculoskeletal Disorders of the Neck, Upper Extremity, and Low Back" (NIOSH, 1997), that has been augmented with current literature. A.1 Introduction A worker's ability to respond to work factors may be modified by his or her own capacity. The level, duration, and frequency of the loads imposed on tissues, as well as the adequacy of recovery time, are critical components in whether increased tolerance (a training or conditioning effect) occurs, or whether reduced capacity occurs and increases susceptibility to musculoskeletal disorders (MSDs). The capacity to perform work varies with gender and age, among workers, and for any individual over time. The relationship between these factors and the resulting risk of injury to the worker is complex and not fully quantified. Certain epidemiological studies have used statistical methods to take individual factors (e.g., gender, age, body mass index) into account, that is, to control for their confounding or modifying effects when looking at the strength of work-related factors. Studies that fail to control for the influence of individual factors may either mask or amplify the effects of work-related factors. The multifactoral nature of MSDs requires a discussion of individual factors that have been studied to determine their association with or influence on the incidence and prevalence of work-related MSDs. These factors include age (Guo et al., 1995; Biering-Sorensen et al., 1983; English et al., 1995; Ohlsson et al., 1994); gender (Hales et al., 1994; Johansson, 1994; Chiang et al., 1993; Armstrong et al., 1987a); anthropometry (Werner et al., 1994; Nathan et al., 1993; Heliovaara, 1987); physical activity (Holmstrom, Lindell, and Moritz, 1992; Baron et al., 1991; Craig et al., 1998); strength (Chaffin and Park, 1973; Chaffin et al., 1977; Troup, Martin, and Lloyd, 1981); cigarette smoking (Finkelstein, 1995; Owen and Damron, 1984; Svensson and Andersson, 1983; Kelsey, Golden, and Mundt, 1990; Hildebrandt, 1987); and alcohol, caffeine, and vitamins (Nathan et al., 1996, Keiston et al., 1997). In addition, psychosocial factors have been associated with upper-extremity and back disorders. A.2 Age The prevalence of MSDs increases as people enter their working years. By the age of 35, most people have had their first episode of back pain (Guo et al., 1995; Chaffin et al., 1979). Once people are in their working years (ages 25 to 65), however, the prevalence is relatively consistent (Guo et al., 1995; Biering-Sorensen et al., 1983). Musculoskeletal impairments are among the most prevalent and symptomatic health problems of middle and old age (Buckwalter et al., 1993). Decreases in musculoskeletal function due to the development of age-related degenerative disorders and loss of tissue strength with age may increase the probability or severity of soft tissue damage from a given insult. Nonetheless, age groups with the highest rates of compensable back pain and strains are the 20-to-24 age group for men and the 30-to-34 age group for women. Another problem is that advancing age and increasing number of years on the job are usually highly correlated. Age is a true confounder with years of employment, so these factors must be adjusted when determining the relationship to work. Many of the epidemiologic studies that looked at populations with a wide age variance have controlled for age by statistical methods. Several studies found age to be an important factor associated with MSDs (Guo et al., 1995; Biering-Sorensen et al., 1983; English et al., 1995; Ohlsson et al., 1994; Riihimaki et al., 1989; Toomingas et al., 1991), whereas others have not (Herberts et al., 1981; Punnett et al., 1985). Age may not equate to less strength. Mathiowetz et al. (1985) demonstrated that hand strength did not decline with aging; average hand pinch and grip scores remained relatively stable in their study's population with a range of 29 to 59 years. Torell, Sanden, and Jarvholm (1988) found no correlation between age and the prevalence of MSDs in a population of shipyard workers. They found a strong relationship between workload (categorized as low, medium, or heavy) and symptoms or diagnosis of MSDs. Other studies have also reported a lack of increased risk associated with aging. For example, Wilson and Wilson (1957) reported that the age and gender distribution of 88 patients with tenosynovitis from an ironworks closely corresponded to that of the general population of that plant. Similarly, Wisseman and Badger (1976) reported that the median age of workers with chronic hand and wrist injuries in their study was 23 years, while the median age of the unaffected workers was 24 years. Riihimaki et al. (1989) found a significant relationship between sciatica and age in machine operators, carpenters, and sedentary workers. Age was also a strong risk factor for neck and shoulder symptoms in carpenters, machine operators, and sedentary workers (Riihimaki et al., 1989). Some authors may have incorrectly attributed age as the sole cause of their findings, though data presented suggested a relationship with work (Schottland et al., 1991). An explanation for the lack of an observed relationship between an increased risk for MSDs and aging may be "survivor bias." If workers who have health problems leave their jobs, or change to jobs with less exposure, the remaining population includes only those workers whose health has not been adversely affected by their jobs. As an example, in a study of female plastics assembly workers, Ohlsson, Attewell, and Skerfving (1989) reported that the degree of increase in the odds of neck and shoulder pain with the duration of employment depended on the age of the worker. For the younger subjects, the odds increased significantly as the duration of employment increased (p = .01), but for the older ones no statistical change was found with length of employment. The older women who had been employed for shorter periods of time had more reported symptoms than the younger ones, while older workers with longer employment times reported fewer symptoms than younger workers. Ohlsson, Attewell, and Skerfving (1989) interviewed 76 former assembly workers and found that 26% reported pain as the cause of leaving work. This finding supports the likely role of a survivor bias in this study, the effect of which is to underestimate the true risk of developing MSDs, in this case in the older workers. A.3 Gender Some studies have found a higher prevalence of some MSDs in women (Bernard et al., 1994; Hales et al., 1994; Johansson, 1994; Chiang et al., 1993). A male-to-female ratio of 1:3 was described for carpal tunnel syndrome (CTS) in a population study in which occupation was not evaluated (Stevens et al., 1988). However, in the Silverstein et al. (1985) study of CTS among industrial workers, no gender difference could be seen after controlling for work exposure. Franklin et al. (1991) found no gender difference in workers' compensation claims for CTS. Burt, Hornung, and Fine (1990) found no gender difference in reporting of neck or upper-extremity MSD symptoms among newspaper employees using video display terminals (VDTs). Nathan et al. (1988, 1992) found no gender differences for CTS. In contrast, Hagberg and Wegman (1987) reported that neck and shoulder muscular pain is more common among females than males, both in the general population and among industrial workers. Whether the gender difference seen with some MSDs is due to physiological differences or differences in exposure is unclear. One laboratory study, Lindman et al. (1991), found that women have more type I muscle fibers in the trapezius muscle than men, and have hypothesized that myofascial pain originates in these type I muscle fibers. Ulin et al. (1993) noted that significant gender differences in work posture were related to stature and concluded that the lack of workplace accommodation to the range of workers' height and reach may, in part, account for the apparent gender differences. The reporting bias may exist because women may be more likely to report pain and seek medical treatment than men (Armstrong et al., 1993; Hales et al., 1994). The fact that more women are employed in hand-intensive jobs and industries may account for the greater number of reported work-related MSDs among women. Bystrom et al. (1995) reported that men were more likely to have de Quervain's disease than women; they attributed this to more frequent use of hand tools. Some studies have reported that workplace risk factors account for increased prevalence of MSDs among women more than personal factors (e.g., Armstrong et al., 1987; McCormack et al., 1990). In a recent evaluation of Ontario workers' compensation claims for repetitive strain injury (RSI), Asbury et al. (1995) reported a relative risk (RR) for female to male claims ranging from 1.3 to 1.6 across industries. Within five different broad occupational categories, females were approximately 2 to 5 times as likely to have a lost-time RSI claim. No information on gender differences in hand-intensive jobs was reported. Many researchers have noted that men and women tend to be employed in different jobs. More research is needed to separate the effect of work risk factors from potential effects that might be attributable to biological differences. The research must control for exposure to workplace risk factors, and researchers should study jobs that men and women perform "relatively equally". A.4 Anthropometry Weight, height, body mass index (BMI) (a ratio of weight to height squared), and obesity have all been identified in studies as potential risk factors for certain MSDs, especially CTS and lumbar disc herniation. Obesity seems to play a small but significant role in the occurrence of CTS (Section B.4.a). Anthropometric data are conflicting, but in general indicate that there is no strong correlation between stature, body weight, body build, and low-back pain (Section B.4.b). A.4.a Upper-Extremity MSDs Few studies examining anthropometric risk factors in relationship to CTS have been occupational epidemiologic studies; most have used hospital-based populations that may differ substantially from working populations. Nathan et al. (1988, 1992, 1994) have published several papers about a single industrial population and have reported an association between CTS and obesity; however, the methods employed in their studies have been questioned in a number of subsequent publications (Gerr and Letz, 1992; Mackinnon et al., 1997; Stock, 1991; Werner et al., 1994). Several investigators have reported that their industrial study subjects with CTS were shorter and heavier than the general population (Cannon et al., 1981; Dieck and Kelsey, 1985; Falck and Aarnio, 1983; Nathan et al., 1992; Werner et al., 1994; Wieslander et al., 1989). Werner et al. (1994) studied a clinical population requiring electrodiagnostic evaluation of the right upper extremity, patients classified as obese (BMI > 29) were 2.5 times more likely than slender patients (BMI < 20) to be diagnosed with CTS. These researchers developed a multiple linear-regression CTS model (with the difference between median and ulnar sensory latencies as the dependent variable). The regression highlighted BMI as the most influential variable, but still only accounted for 5% of the variance in the model. In Nathan's (1994) logistic model, BMI accounted for 8.6% of the total risk; however, this analysis used both hands from each study subject as separate observations, although they are not independent of each other. Falck and Aarnio (1983) found no difference in BMI among 17 butchers with (53%) and without (47%) CTS. Vessey, Villard-Mackintosh, and Yeates (1990) found that the risk for CTS among obese women was double that for slender women. Studies assessing BMI and CTS categorized BMI according to various criteria, including criteria from the National Center for Health Statistics (Werner et al., 1996) or Michigan Department of Public Health (Nordstrom et al., 1997). The NCHS criteria were slender (BMI < 20), normal (BMI 20-27), overweight (men: BMI 27.8-31.09; women: 27.3-32.29), obese (men: > 31.1; women: > 32.3). The Michigan Department of Health criteria were thin (BMI < 20), normal (BMI 20-25), heavy (BMI 25-29), and obese (BMI > 29). Nordstrom et al. (1997), in a study of risk factors for CTS in a general population, concluded that BMI is one factor that seems to have a causal relation to CTS. These researchers found that for each increase of one unit of BMI, about 6 pounds for the average-sized adult, risk of CTS increases by 8%. Werner et al. (1997), in a study at five different worksites (four industrial, one clerical), concluded that obesity (BMI > 29), industrial work, and age were independent risk factors for median mononeuropathies. Their study, which did not define specific work-related exposures, showed no significant interaction between work activity and obesity. However, the authors caution interpretation of the data and urge more investigation. It has been suggested that relationship of CTS with BMI involves increased fatty tissue within the carpal canal or increased hydrostatic pressure throughout the carpal canal in obese persons compared with slender persons (Werner, 1994). Two other anthropometric risk factors, carpal tunnel size and wrist size have been suggested as risk factors for CTS; however, some studies have linked both small and large canal areas to CTS (Bleecker et al., 1985; Winn and Habes, 1990). Schierhout et al. (1995) found that short stature was significantly associated with pain in the neck and shoulder but not in the forearm, hand and wrist, or back, among workers in 11 factories. Height was not a factor for neck, shoulder, or hand and wrist MSDs among newspaper employees (Bernard et al., 1994). Kvarnstrom (1983) found no relationship between neck/shoulder MSDs and body height in a Swedish engineering company with more than 11,000 workers. A.4.b Low-Back Disorders Hrubec and Nashold (1975) found that height and weight were predictive of herniated disc disease among World War II U.S. army recruits compared with age-matched controls. Some studies have reported that people with back pain are, on the average, taller than those without it (Rowe et al., 1965; Tauber, 1970; Merriam et al., 1980; Biering-Sorensen et al., 1983). Heliovaara et al. (1987), in a Finnish population study, found that height was a significant predictor of herniated lumber disc in both sexes, but a moderately increased BMI was predictive only in men. Severe obesity (exceeding 30 kg/m2) involved less risk than moderate obesity. Kelsey (1975) and Kelsey et al. (1984) failed to reveal any such relationships involving height or BMI among patients with herniated lumber discs and control subjects. Magora and Schwartz (1978) found an association between obesity and radiological disc degeneration, but Kellgren and Lawrence (1958) did not. A study of Finnish white collar and blue collar workers found no association between workers who were overweight (relative weight > 120%) and lumbosacral disorders, either cross-sectionally or in a 10-year follow-up (Aro and Leino, 1985). Using sports medicine criteria, Craig et al. (1998) classified body composition (percentage body fat) as low (< 12%), medium (> 12% and < 19%), and high (> 19%). In a study of highly trained male high-frequency manual materials handlers, they demonstrated that high occurrences of injury were significantly correlated with high estimated percentage of body fat (> 19%). They randomly chose 25% of the study participants (58 individuals) and averaged skinfold measurements and body circumference measures for use in data analysis. A.5 Physical Activity The relationship of physical activity and MSDs is more complicated than just dose/response. Excessive physical activity may cause injury. However, the lack of physical activity may increase susceptibility to injury, and after injury, the likelihood of another injury increases. In construction workers, more frequent leisure time was related to healthy lower backs (Holmstrom, Lindell, and Moritz, 1993) and severe low-back pain was related to less leisure time activity (Holmstrom, Lindell, and Moritz, 1992). On the other hand, some standard treatment regimes have found that musculoskeletal symptoms are often relieved by physical activity. Good physical condition may not protect workers from the risk of MSDs. Baron et al. (1991) stated that persons with high aerobic capacity may be fit for jobs that require high oxygen uptake, but will not necessarily be fit for jobs that require high static and dynamic strengths and vice versa. Craig et al. (1998) stated that high occurrences of injury among material handlers were significantly correlated with low estimated relative maximal aerobic capacity. When physical fitness is examined as a risk factor for MSDs, results are mixed. For example, some early case series reported an increased risk of MSDs associated with playing professional sports (Bennet, 1946; Nirschl, 1993), or with physical fitness and exercise (Kelsey, 1975; Dehlin et al., 1978, 1981), while other studies indicate a protective effect and reduced risk (Cady et al., 1979; Mayer et al., 1985; Astrand et al., 1987; Biering-Sorensen et al., 1984). Boyce, Jones, and Hiatt (1991) reported that only 7% of absenteeism could be explained by age, sex, and physical fitness among 514 police officers 35 years or older. Cady et al. (1979, 1985), on the other hand, found that physical capacity was related to musculoskeletal fitness. Cady defined fitness for most physical activities as combinations of strength, endurance, flexibility, musculoskeletal timing, and coordination. Cady et al. (1979) evaluated male firefighters and concluded that physical fitness and conditioning had significant preventive effects on back injuries (least fit 7.1% injured, moderately fit 3.2% injured, and most fit 0.8% injured). However, the most fit group had the most severe back injuries. Low cardiovascular fitness level was a risk factor for disabling back pain in a prospective longitudinal study among aerospace manufacturing workers by Battié et al. (1989). Good endurance of back muscles was found to be associated with low occurrence of low back pain (Biering-Sorensen et al., 1984). Craig et al. (1998) stated that more specific validating research needs to be conducted; however, decreased percentage of body fat and increased maximal aerobic capacity can reduce musculoskeletal injuries. Few occupational epidemiologic studies have looked at non-work-related physical activity in the upper extremities. Most NIOSH studies (Hales and Fine, 1989; Kiken, Stringer, and Fine, 1990; Burt, Hornung, and Fine, 1990; Baron et al., 1991; Hales et al., 1994; Bernard et al., 1994) have excluded MSDs due to sports injury or other non-work-related activity or injury and have not included these factors in analyses. However, many of the risk factors that are important in occupational studies -- forceful, repetitive movements with awkward postures -- occur in sports activities. A combination of high exposure to load lifting and high exposure to sports activities that engage the arm was a risk factor for shoulder tendinitis, as well as osteoarthritis of the acromioclavicular joint (Stenlund et al., 1993). Kennedy, Hawkins, and Kristof (1978) found that 15% of competitive swimmers with repetitive overhead arm movements had significant shoulder disability, primarily due to impingement from executing butterfly and freestyle strokes. Epicondylitis in professional athletes has been well documented, and many of the biomechanical and physiological studies of epicondylitis have been conducted in professional tennis players and baseball pitchers (King, Brelsford, and Tullos, 1969; Nirschl, 1993). One prospective study of healthy baseball players has found slowing of the suprascapular nerve function as the season progresses (Ringel et al., 1990). Scott and Gijsbers (1981) found an association between athletic performance and pain tolerance, and suggested that physically fit persons may have a higher threshold for injury. In summary, although physical fitness and activity is generally accepted as a way of reducing work-related MSDs, the present epidemiologic literature does not clearly support that assumption. The sports medicine literature, however, does give a better indication that sports involving activities of a forceful, repetitive nature (such as tennis and baseball pitching) are related to MSDs. It is important to note that professional sports activities usually provide players (i.e., workers) with more substantial breaks for recovery and shorter durations for intense tasks, compared to more traditional work settings in which workers are required to perform repetitive, forceful work for 8 hours per day, 5 days per week. A.6 Strength Some epidemiologic support exists for the relationship between back injury and a mismatch of physical strength and job tasks. Chaffin and Park (1973) found a sharp increase in back injury rates in subjects performing jobs requiring strength that was greater than or equal to their isometric strength-test values. The risk was 3 times greater in the weaker subjects. In a second longitudinal study, Chaffin et al. (1977) evaluated the risk of back injuries and strength and found the risk to be 3 times greater in the subjects without lower static strength. Keyserling, Herrin, and Chaffin (1980) strength-tested subjects, biomechanically analyzed jobs, and assigned subjects to either stressed or non-stressed jobs. Following medical records for a year, they found that job matching based on strength criteria appeared to be beneficial. In another prospective study, Troup, Martin, and Lloyd (1981) found that reduced strength of back flexor muscles was a consistent predictor of recurrent or persistent back pain, but this association was not found for first-time occurrence of back pain. Other studies have not found the same relationship with physical strength. Two prospective studies of low-back pain reports (or claims) of large populations of blue collar workers (Battié et al., 1989; Leino, Aro, and Hasan, 1987) failed to demonstrate that stronger (defined by isometric lifting strength) workers are at lower risk for low-back pain claims or episodes. One study followed workers for 10 years after strength testing and the other followed workers for a few years. Neither of these studies included precise measurement of exposure level for each worker, so the authors could not estimate the degree of mismatch between workers' strength and task demands. Battie compared workers with back pain with other workers on the same job (by isometric strength testing) and did not find that workers with back pain were weaker. In two studies of nurses (Videman et al., 1989; Mostardi et al., 1992), lifting strength was not a reliable predictor of back pain. When examined together, these studies reveal the following: the studies that found a significant relationship between strength and back pain used more thorough job assessment analysis and focused on manual lifting jobs. However, these studies only followed workers for periods of 1 year, and whether this same relationship would hold over a much longer working period remains unclear. The studies that did not find a relationship, although they followed workers for longer periods of time, did not include precise measurements of exposure level for each worker, so they could not assess the strength capabilities that were important in the jobs. A.7 Smoking A.7.a Upper-Extremity MSDs In the Viikari-Juntura et al. (1994) prospective study of machine operators, carpenters, and office workers, current smoking (OR: 1.9; 95% CI: 1.0-3.5), was among the predictors for change from "no neck trouble" to "severe neck trouble." In a study of Finnish adults aged 30 to 64 (Makela et al., 1991), neck pain was found to be significantly associated with current smoking (OR: 1.3; 95% CI: 1.0-1.61) when the logistic model was adjusted for age and gender. However, when the model included mental and physical stress at work, obesity, and parity, then smoking (OR: 1.25; 95% CI: 0.99-1.57) was no longer statistically significant (Makela et al., 1991). With univariate analysis, Holmstrom (1992) found a prevalence rate ratio (PRR) of 1.2 (95% CI: 1.1-1.3) for neck/shoulder trouble in "current" smokers vs. people who "never" smoked. But using multiple logistic regression, when age, individual, and employment factors were in the model, only "never smoked" contributed significantly to neck/shoulder trouble. While investigating reasons for higher compensation claims for CTS in certain employee groups, Nathan et al. (1996) evaluated the effects of tobacco, caffeine, and alcohol on the prevalence of median entrapment neuropathy at the wrist, CTS symptoms, and CTS confirmed by nerve conduction studies among industrial workers (nonclaimants and working patients referred for upper-extremity symptoms) who volunteered for the study. Nathan et al. (1996) stated that greater use of tobacco combined with greater consumption of caffeinated beverages and alcohol abuse was associated with more median nerve slowing, more specific hand/wrist symptoms, and more electrophysiologically confirmed CTS. However, the effects explained only a small portion of the total risk. Toomingas et al. (1991) found no associations between multiple health outcomes (including tension neck syndrome, rotator cuff tendinitis, CTS, or problems in the neck/scapula or shoulder/upper arm) and nicotine habits among platers, assemblers, and white collar workers. In a case/referent study, Wieslander et al. (1989) found that smoking or using snuff was not related to CTS among men operated on for CTS. A.7.b Low-Back Disorders Several papers have presented evidence that a positive smoking history is associated with low-back pain, sciatica, or intervertebral herniated disc (Finkelstein, 1995; Owen and Damron, 1984; Frymoyer, Pope, and Clements, 1983; Svensson et al., 1983; Kelsey et al., 1984); whereas other papers have found a negative relationship (Kelsey, Golden, and Mundt, 1990; Riihimaki et al., 1989; Hildebrandt, 1987). Boshuizen et al. (1993) found a relationship between smoking and back pain only in those occupations that required physical exertion. In their study, smoking was more clearly related to pain in the extremities than to pain in the neck or the back. Deyo and Bass (1989) observed that the prevalence of back pain increased with the number of pack-years of cigarette smoking and with the heaviest smoking level. Heliovaara et al. (1991) only observed a relationship in men and women older than 50 years. Two studies did not find a relationship between sciatica and smoking among concrete reinforcement workers and house painters (Heliovaara et al., 1991; Riihimaki et al., 1989). Several explanations for the relationship with smoking have been postulated. One hypothesis is that back pain is caused by coughing from smoking. Coughing increases the abdominal pressure and intradiscal pressure and puts strain on the spine. A few studies have observed this relationship (Deyo and Bass, 1989; Frymoyer et al., 1980; Troup et al., 1987). The other mechanisms proposed include nicotine-induced diminished blood flow to vulnerable tissues (Frymoyer, Pope, and Clements, 1983), and smoking-induced diminished mineral content of bone causing microfractures (Svensson et al., 1983). Similar associations with diminished blood flow to vulnerable tissues have been found between smoking and Raynaud's disease. A.8 Vitamins In recent years, studies have addressed vitamins and their relation to CTS. There are conflicting views of the association between vitamin B6 status and CTS (Franzblau et al., 1996; Keniston et al., 1997; Jacobson, Plancher, and Kleinman, 1996). Some studies have not been designed well: they have no standardized case definitions, no appropriate controls, and a lack of objective outcome measures. Work tasks were not addressed, and the study populations were small and non-randomly chosen. Keniston et al. (1997) used multivariate analysis to investigate the relationship of vitamin B6 and CTS. Workers (n = 441) were recruited from six industries, described in a previous Nathan et al. (1996) study investigating causes of CTS. The subjects were questioned orally about symptoms and nerve conduction studies of the median nerve were performed. Specific and nonspecific symptoms were ranked as absent or present; a variable was calculated for the purpose of ranking subjects by total number of symptoms. Vitamin supplementation was self-reported. Blood was measured for plasma PLP (pyridoxal 5'-phosphate, a measure of vitamin B6 status) and plasma ASC (total ascorbate, a measure of vitamin C status). Keniston et al. (1997) stated that univariate analyses found no significant differences in vitamin B6 status between groups due to several confounders, including age, BMI, and vitamin supplementation. In multivariate analyses, they used unsupplemented subsets of the study population and vitamin interaction variables. Higher plasma PLP, particularly in males without vitamin supplementation, appeared to be associated with less frequent symptoms. In some analyses, higher vitamin C concentration or a higher ASC/PLP ratio were associated with more prevalent/frequent symptoms or greater sensory latencies. In contrast, Franzblau et al. (1996) examined 125 randomly selected workers in two industrial plants. The subjects completed a self-administered questionnaire and underwent nerve conduction studies of their median and ulnar sensory nerves. Blood was analyzed for plasma PLP and erythrocyte glutamic pyruvic transaminase (EGPT) index (abnormal would indicate vitamin B6 deficiency). Franzblau et al. (1996) stated there was no suggestion of a linear relationship between indices of vitamin B6 status and electrophysiologic measures of median/ulnar sensory nerve function. Using log-transformed electrophysiologic results did not alter their conclusion. Therefore, Franzblau et al. conclude that indices of vitamin B6 status were unrelated to self-reported symptoms, nerve conduction studies, and CTS defined on the basis of self-reported symptoms and electrophysiologic measurements. They suggest that CTS among active workers is not related to vitamin B6 status. A.9 Miscellaneous Articles and editorials have questioned the relationship between work and MSDs. Some argue that work does not cause cumulative trauma disorder or MSDs, especially upper-extremity disorders: they assert that these disorders are a result of many influences, including individual attributes or risk factors, society, the media, the legal profession, and the government (Higgs and Young, 1996; Louis, Calkins, and Harris, 1992; Hadler, 1992, 1996, 1997; Nathan, 1992, 1996; Vender, Kasdan, and Truppa, 1995; Kasdan et al., 1996). Other editorials have refuted or challenged these claims (Egilman et al., 1996; Franklin et al., 1996; Mackinnon et al., 1997; Armstrong et al., 1996). Louis (1996) states that there are three positions on workplace musculoskeletal symptoms and causality: (1) work is not responsible for musculoskeletal symptoms; (2) musculoskeletal symptoms are multifactoral; or (3) work causes the symptom complex, no matter what it is. He suggests that much of the problem is due to imprecise use of language and self-interest, and recommends the use of the term "upper extremity pain" when a more specific diagnosis cannot be made. Louis states that a changing health care delivery system and a lack of epidemiologic education by orthopedists confound the issue. He encourages physicians to examine each individual discretely and specifically, taking a detailed medical and occupational history that includes non-work activities. Controversy over risk factors for low-back pain exists as well. Battie and Bigos (1991) and Bigos and Battie (1992) contend that both physical and nonphysical factors influence back problems, but few such problems have clearly and consistently been found to be associated with risk. In their Boeing study, work perceptions and psychosocial factors played a greater role in back pain complaints than physical measures. In this study the investigators looked at those employees who reported a back injury; the study did not investigate actual presence of symptoms. No physical job features were reported. In a review of psychosocial factors at work and low-back pain, Bongers et al. (1993) state that four of six longitudinal studies demonstrated a positive relationship between stress symptoms and low-back pain. Some psychosocial factors probably influence reporting of back pain. Leamon et al. (1994) wrote a critique of the association between manual handling tasks involving weights greater than 25 to 30 pounds and low-back pain. This review of the validity of various criteria for prevention of work-induced low-back pain acknowledged that repetitive lifting of heavy objects, particularly if associated with another risk factor such as twisting, is causally related to low-back pain. Hypotheses from the critique are: (1) the exposure-response relationship is inadequately defined; (2) strength is protective; (3) most low-back pain is not related to manual handling activities; and (4) physical occupational stressors related to lifting tasks are not the primary determinants of disability, but psychosocial or other factors may be. Despite points of disagreement or concern, there is general acknowledgment that multiple risk factors are associated with MSDs. A.10 References
B.1 Introduction The objective of this section of Appendix I is to review the evidence for work-relatedness of MSDs and symptoms in the neck and upper back. According to the World Health Organization, work-related diseases are multifactoral diseases that may be caused in part or aggravated by work or work conditions (Hagberg et al., 1995). This paper is a review of analytical and observational epidemiological research and experimental biomechanical studies of physical work factors, work organization, and psychosocial work factors, and some individual factors that have been found to be associated with work-related neck disorders and symptoms. However, the focus of the review is physical work factors, since those are the ones most likely to be under the direct influence of employers. B.1.a Neck and Upper-Back Pain in Workers The extent to which neck or back pain occurs in or affects workers depends to a great extent on the terms used to define the pain, in terms of intensity and duration, and on the methods used in determining the presence or occurrence (self-report, interview, or physical examination). Point prevalence of neck pain in a general U.S. population has been reported at 10%, matching point prevalence reports of workers in an aeroengineering factory and exceeding a 4% prevalence reported in a group of textile workers (Palmer et al., 1998). In the general Swedish population, chronic neck pain prevalence (pain lasting more than 3 months) was reported to be 14.5% in men and 19.1% in women, a significant difference (Andersson et al., 1993). Neck pain prevalence peaked in the 45- to 54-year age group. At 68% and 47%, respectively, the 3-month prevalence of MSDs in the neck and in the thoracic back in industrial workers performing unskilled tasks were each found to be twice that reported by the general population (Bjorksten et al., 1996). Specifically excluding neck pain associated with traumatic injury, the 1-year prevalence of MSD symptoms in a group of electricians was found to be 16%; with a less restrictive definition, the rate was 38% (Hunting, et al., 1994). In that group of workers meeting the less restrictive definition, 24% had missed work, had been on light duty, or had seen a medical provider as a result of neck pain. Based on self-report, 1-year prevalence of neck pain and upper-back pain in the Danish wood and furniture industry was reported to be 26% and 18%, respectively. This compared with 42% and 28% for low-back pain and shoulder pain (Christensen, Pedersen, and Sjogaard, 1995). Prevalence of regular discomfort in the posterior neck region was 6.3%, and 9.1% in the upper-back region, in a group of chicken-processing workers. However, the lifetime prevalence was 36%, the point prevalence was 18%, and 9% had sought medical treatment for discomfort (Buckle, 1987). Buckle (1987) also reported regular discomfort prevalence of 8.3% in department store staff, and 4.2% in supermarket staff. The prevalence of regular upper-back discomfort in the two groups was 4% and 5.6%, respectively. Many studies of neck pain have focused on employees working in health care. Milerad and Ekenvall (1990) reported cervical symptom prevalence of 45% of male dentists and 63% of female dentists, rates that were 2.6 and 2 times those of male and female pharmacists, respectively. The 1-year prevalence of neck discomfort was 62% in a group of Swedish dental hygienists who worked an average of 31.5 hours per week (Oberg and Oberg, 1993). Only right the prevalence of shoulder discomfort was greater (at 65%). Low-back discomfort prevalence was 39% in this group. Twelve-month prevalence of self-reported neck pain was 63.1% in a group of medical secretaries and hospital office personnel (Linton and Kamwendo, 1989). Sweden has set standards for working positions of the neck, for disability determination: > 15 degrees flexion for > 80% of work time, > 45 degrees rotation for > 50% of work time, and > 15 degrees rotation for > 80% of work time for dynamic work postures (Eklund et al., 1994). B.1.b Definitions of Neck Disorders Most of the studies reviewed in this paper did not investigate neck discomfort associated with a specific diagnosis, but rather self-reported pain or discomfort that occurred in a specific region, often using a graphical image to help subjects locate the region. Many papers reported using the Nordic Questionnaire, or some variant of it. There are, however, several specific entities of non-traumatic origin that occur in the cervical region. Thoracic outlet syndrome (TOS) is defined by Hagberg et al. (1995) as a "neurovascular impingement syndrome at different anatomical levels where the brachial plexus and subclavian vessels may be entrapped as they pass through, en route from the cervical spine to the arm." The syndrome involves compression of the subclavian artery and the lower trunk of the brachial plexus, at one or more locations between the neck and the axilla. Symptoms are experienced in the upper extremity. Cervical syndrome, or cervical radiculopathy, is defined by Hagberg et al. (1995) as "compressions of the nerve root by a herniated disc or a narrowed intervertebral foramen." Tension neck syndrome is defined by Hagberg et al. (1995) as myofascial pain localized in the shoulder and neck region. Cervical spondylosis refers to degenerative changes in the cervical spine that are apparent on radiological examination (Hagberg and Wegman, 1987). B.2 Epidemiological Research of Physical Factors Associated With Neck Disorders and Pain in Workers B.2.a Conclusions From Previous Reviews of the Literature Within the past decade and a half, several reviews have been conducted of the epidemiological literature of musculoskeletal disorders of the neck, neck/shoulder, or neck and upper-extremity regions. Five recent reviews focused, at least in part, on the neck: Bernard (1997), Grieco, et al. (1998), Hagberg et al. (1995), Hales and Bernard (1996), and Viikari-Juntura (1997). Hidalgo et al. (1992) reviewed the biomechanical literature of the neck. Conclusions from those reviews are presented below, as are conclusions from some other earlier reviews. Hagberg et al. (1995) reviewed epidemiological studies for evidence of work-relatedness of selected musculoskeletal disorders of the neck: TOS (neurogenic form), cervical syndrome, and tension neck syndrome. In compiling a list of valid papers for their review, the researchers considered the strength of each study based on minimization of bias (selection bias, information or misclassification bias, confounding or effect modification bias) and study power. Studies that met their validity criteria were then reviewed for causality (strength of association, demonstration of temporal association, consistency of association among studies, predictive power of exposure factors, and plausibility). Six cross-sectional studies of TOS (published between 1979 and 1991) met their inclusion criteria. From those studies, Hagberg et al. (1995) found the strength of association between work and TOS to be generally weak, based on low odds ratios (ORs). Since all studies were cross-sectional in design, temporal associations could not be confirmed. There seemed to be a consistent association between repetitive work and TOS across the studies. One study demonstrated a dose-response relationship between vibration and TOS. The authors also noted an association between TOS and age. Hagberg et al. (1995) concluded that the studies demonstrated the existence of a consistent association between repetitive arm movements, manual work, and TOS. Three cross-sectional studies of cervical radiculopathy (published between 1979 and 1983) met their inclusion criteria. Hagberg et al. (1995) observed that all studies showed a low prevalence for cervical radiculopathy. Low numbers meant wide confidence intervals, which made results difficult to interpret. They concluded that more directed research needed to be conducted in this area. Twelve cross-sectional studies and one laboratory study of tension neck syndrome (published between 1976 and 1988) met their inclusion criteria. From those studies, Hagberg et al. (1995) found the strength of association between work and tension neck syndrome to be moderate, based on ORs from 3 to 7. Since all studies were cross-sectional in design, temporal associations could not be confirmed. There seemed to be a consistent association between work with VDTs) and tension neck syndrome across several studies; there also seemed to be consistent associations between tension neck syndrome and repetitive work and static head and arm postures. The authors also noted that tension neck syndrome was found more commonly in women, but that finding may have been confounded by differences in work. Hagberg et al. (1995) concluded that the studies demonstrated the existence of a consistent association between repetitive work and tension neck syndrome caused by constrained head and arm postures. They also noted that tension neck syndrome had a high prevalence in both work and reference groups. Bernard (1997) reviewed epidemiological studies for evidence of work-relatedness of neck and neck/shoulder musculoskeletal disorders. In the process of identifying papers for this review, Bernard (1997) first considered the strength of each study based on whether it provided clear definitions of exposed and reference populations and clear definitions of outcomes, as well whether it evaluated exposures in such a way as to classify them with regard to force, repetition, posture, or vibration. Papers that met these standards were then evaluated based on four criteria: a 70% or better response rate in order to limit response bias, health outcome defined by symptoms and physical examination (PE)(1), investigators blinded were appropriate (exposure or health status), and the neck as a focus of the evaluation. Only one of the studies that focused on the neck and two that focused on the neck/shoulder region met all four criteria. The likelihood of bias in each study was examined. Finally, studies were summarized with respect to strength of association, demonstration of temporal association, consistency of association among studies, and exposure-response relationship. Bernard (1997) reviewed 46 epidemiological studies (1976 to 1995) reporting on the neck and 23 reporting on the neck/shoulder region. The authors of the review concluded that there was "reasonable evidence" for an association between highly repetitive work and neck/shoulder MSDs, where repetitiveness was most often defined in terms of hand activity. They also determined that there was "reasonable evidence" for an association between forceful exertion and neck/shoulder MSDs, where forceful work was conducted by the arms. Bernard (1997) determined there was "strong evidence" for an association between static loads and neck/shoulder MSDs, where "static load" referred to a static load of long duration, high intensity, or extreme amplitude. They found insufficient evidence of an association between vibration and neck disorders. A recent review of epidemiological studies by Grieco et al. (1998) concluded that cervical radiculopathy had not been shown to be associated with data entry work, dockers' work, or food production assembly line work. In contrast, tension neck syndrome was linked to static postures and static loads in several studies on populations of VDT workers, typists, and sewing machine operators. Study selection criteria were not discussed in that review. Viikari-Juntura (1997) reviewed both epidemiological and experimental studies focused on the neck (among other regions). The author mentioned studies that showed associations between degenerative changes or neck pain and heavy work, repeated impact loading, or static work. The author did not interpret findings from experimental research, other than to mention that one study demonstrated the predictive value of the response variables (electromyographic [EMG] gap analysis -- Veiersted, Westgaard, and Andersen, 1993). Study selection criteria were not discussed in this review. Based on consistency of association across several studies, Hales and Bernard (1996) concluded that neck disorders were associated with work involving repetitive motions, forceful repetitive work, and constrained or static postures. They noted inconsistent findings regarding neck disorder and work pace, which, they suggested, may be due to the many ways work pace can be quantified. Hales and Bernard also mentioned a consistent association between wearing bifocals, awkward neck postures, and neck disorders. Stock (1991) performed a review of epidemiological studies of workplace factors and neck and upper-limb MSDs. She found only three papers that met her inclusion criteria (explicit definitions of exposed and control working populations; explicitly defined workplace exposures, in terms of repetition, force, static loading, or joint posture; diagnosed outcomes that met explicit, clinically relevant criteria; and comparative study design). No significant findings were apparent from the single paper that considered neck-specific disorders. By pooling results across epidemiological studies, Hagberg and Wegman (1987) determined the OR for tension neck syndrome to be 2.0 in keyboard operators, whereas the OR for cervical spondylosis in cotton workers was 0.66 (protective). Several individual studies of workers performing heavy work (including meat carriers and miners) found increased ORs (most adjusted for age) for cervical spondylosis, as did one study of dentists. Tension neck syndrome was found to be related to a number of jobs, including VDT work, office work, scissors-making, and lamp assembly. Increased ORs were found for cervical syndrome in one study of office workers (with iron foundry workers as referents). Though no work factors were explicitly tied to the studies in the review, the authors provide hypotheses regarding etiology of each of the neck MSDs. They hypothesize that cervical syndrome may be due to extreme forward flexion of the cervical spine, cervical spondylosis due to high loads on the cervical spine, and tension neck syndrome due to static contraction of the neck/shoulder muscles. Hidalgo et al. (1992) reviewed papers that focused on occupational biomechanics of the neck. Based on that review, the authors recommended that prolonged static loading of the muscles of the neck should not exceed 1% of maximum voluntary contraction (MVC). B.2.b Search Methods and Inclusion/Exclusion Criteria for This Review The search conducted for this review was structured in such as way as to identify analytical and descriptive epidemiological research addressing neck pain and symptoms in workers, as well as experimental laboratory or field studies of factors (work or work-related) hypothesized to be associated with neck symptoms in workers. A review of potential pathophysiological mechanisms is outside the scope of this review.
Work-Relatedness of Neck MSDs
A search of Medline (using PubMed) and selected journals not appearing in Medline (Work and Stress, Clinical Biomechanics, and International Journal of Industrial Ergonomics) was undertaken, using sets of keywords. Two- and three-word sets were created by seeking all combinations among the following lists: (1) neck, epidemiology, cervical, upperback, upper back, thoracic outlet, or osteoarthritis, (2) work related, occupation*, job*, posture*, or work, (3) neck pain, neck strain, MSD, arthritis, muscle strain, facet joint irritation, disc disease, muscle strain, cervical pain, or cervical strain. Additional keyword sets were created by combining a word from each of the following lists: (1) neck, cervical, or upper back; (2) disc disease, joint disease, pain, or strain; (3) repetition, force, vibration, psychosocial, physiological, repetition strain injury, etiology, or cumulative trauma disorder. This created a large pool of citations from which papers were eliminated immediately if they appeared more than once, were not in English, or were unrelated to the neck/shoulder region (such as papers on cancer, animals, the bladder, the femur, etc.). The remaining titles and abstracts were then reviewed individually for relevance. This left a pool of approximately 450 papers. Using the titles and abstracts of the articles, two reviewers rated the priority for obtaining each remaining paper as high, medium, low, or undetermined based on lack information in the abstract. Eight citations were initially evaluated by both reviewers, providing good agreement on article priority. Except for those designated as low priority, all papers were obtained through local libraries or interlibrary loan, if they were not already in personal or workplace libraries. Those libraries were also reviewed for relevant papers, as were reference lists from papers as they were obtained. Once papers were obtained, they were reviewed for scope, content, and quality. Evaluations of epidemiological studies were based primarily on the criteria described in Bernard (1997). Papers that referred to "shoulder or neck" and that never referred to the neck region alone were excluded from this review, to minimize the inclusion of disorders of the shoulder joint, such as rotator cuff or bicipital tendinitis. The upper trapezius or descending trapezius was the focal region of several studies included in this review. Papers that referred to occupational cervicobrachial disorder (OCD) were excluded unless they provided a definition indicating that their primary focus was the neck region. This minimized the inclusion of disorders affecting the upper extremity (such as CTS, which is included in some definitions of OCD). It should be noted, though, that more than one paper referred either to correlation identified between neck problems and shoulder problems, or to little difference in the results of analyses whether authors assessed only the neck/shoulder region or the more inclusive neck/shoulder/upper-back/upper-extremity region (Jonsson, Persson, and Kilbom, 1988). The upper back was not the sole focus of any study, but that region was included in several of the studies. B.2.c Studies Specifically Assessing the Neck and/or Upper-Back Regions 1) Analytical epidemiological studies, 1996 to 1999 (cross-sectional, cohort, case-control) In analytical epidemiological studies, information on both exposure and health outcome collected from subjects. Given that two substantial reviews of the analytical epidemiological literature of work-related MSDs were recently conducted (Bernard, 1997; Hagberg et al., 1995), the current review includes only more recent analytical epidemiological studies, those published from 1996 to 1999. It should be noted that many of the studies examined several regions of the body and a variety of exposures; however, only those elements of the studies that relate to the neck or upper back are discussed here. a) Risk associated with job characteristic (posture, generic tasks, job design, classification, etc.) (1) Driving Scutter, Turker, and Hall (1997) conducted a cross-sectional survey of 179 farmers. Based on self-reports of symptoms and exposures, the authors concluded that driving a tractor was a risk factor for neckache and headache in farmers. There is also an extensive body of evidence documenting the association between whole-body vibration and driving in general, i.e., tractor trailers, other over the road vehicles. They hypothesized that this was due to exposure to whole-body vibration, as well as the twisted postures tractor drivers adopt when driving in order to be able to watch the implement behind the tractor. The researchers found no association between age and neck pain in this group. Magnusson, et al. (1996) using a cross-sectional design, studied the prevalence of neck pain in bus and truck drivers in the United States and Sweden, comparing them to sedentary controls. They were not able to find any association between the physical work factors they studied, including measured vibration dose or load handling, and the prevalence of neck pain. They did find an elevated prevalence of neck pain and lost work time across the work groups in the United States, with bus drivers having elevated prevalence compared to the truck driver or sedentary worker groups. Personal factors (overweight, smoking, regular exercise, or marital status) were also not associated with neck pain prevalence in any group from either country. Employing a cross-sectional design, Skov, Borg, and Orhede (1996) studied physical and psychosocial risk factors for neck pain in a group of 1,307 Danish salespeople. They identified a dose-response relationship between driving distance that exceeded 15,000 km per year and 1-year prevalence for neck pain. Some psychosocial work factors were also significant (low control over time and high perceived competition from other salespeople). They found a higher prevalence in women, but no association of neck pain with age. Jensen, Tuchsen, and Orhede (1996) studied a cohort of 89,146 male Danish drivers, through first admission for prolapsed cervical intervertebral disc (PCID), over a period of 10 years. The standardized hospitalization ratio (SHR) for drivers, compared with other male workers, was 142 (95% CI: 126.8-159.6), indicating an elevated occurrence in this worker group. Among the drivers, the SHR for drivers who performed heavy work (such as beverage delivery and garbage collection) was lower than the SHR for drivers who performed light work (such as driving taxis or buses). Based on a cross-sectional survey of workers, the PRR for self-reported whole-day vibration exposure was 7.1 for drivers (95% CI: 4.1-11.7), though these results were not linked to any neck MSDs. The authors suggested that vibration, rather than heavy work, was a risk factor in driving. (2) Time on task Finsen, Christensen, and Bakke (1997) conducted a cross-sectional study of 99 dentists to identify risk factors for MSDs in that type of work. One-year prevalence for "troubles in the neck" was associated with working with patients more than 25 hours per week. Age, years working as a dentist, total weekly work hours, and sex were not found to be associated with neck troubles. This study was supplemented with an observational study, described later in this review, of eight female dentists, in which awkward static postures and elevated muscle activity were documented during common procedures. In a group of auto assembly workers, Zetterberg et al. (1997) found self-reported neck pain increased with longer working time at the facility, though there was no similar finding for physical findings through blind PE. (3) Posture and workstation design Yu and Wong (1996) conducted a cross-sectional survey of a group of Hong Kong bank employees who used VDTs. They examined several work factors and found that -- after controlling for sex and age -- inclining the neck at work, non-adjustable keyboard height, and frequent VDT use were all associated with neck pain prevalence (time period undefined, but associated with current job). Sagittal and off-plane locations of the computer monitor were studied by Lu and Aghazadeh (1994). Subjects who placed the monitor at 45 degrees to the left of the sagittal plane with the keyboard away from the edge of the desk reported more discomfort than workers with other arrangements, for both the neck and upper back. (4) Physically demanding work de Zwart et al. (1997a) stratified a sample of workers based on sex, age, and type of work and found a higher prevalence of neck complaints in women for all types of work. Complaints increased with age for men and women. For men between 25 and 44 years old, jobs categorized as "mentally/physically demanding" or as "heavy physical work" had higher prevalence of neck complaints than the referent "mentally demanding" group. For women in the youngest age category, "mentally/physically demanding" work was protective, while heavy physical work had higher prevalence for neck complaints. In the 45 to 54 age group, "mentally/physically demanding work" had a higher prevalence of neck complaints than the referent category of work. de Zwart et al. (1997b) also reported results of a cohort study of 12,010 working men, stratified by age, in which only sedentary and heavy categories of work were compared. Changes (all increases) in prevalence of "regular stiffness or pain" between two self-report surveys, taken 4 years apart (on average), were greater for the heavy physical group for every age category except the fourth decade. (5) Managerial/professional work Lau, Sham, and Wong (1996) studied the prevalence of neck pain in the general Hong Kong population, through a household survey, using interviewers. Forty-five percent of the respondents did not work outside the home. Risk factors they found to be significant for lifetime prevalence of neck pain were: living in private housing (interpreted as greater affluence, OR: 1.64; 95% CI: 1.2-2.4), current occupation categorized as managerial or professional (OR: 2.1; 95% CI: 1.1-4.0), and hospitalized for accidents involving the neck (OR: 5.6; 95% CI: 3.3-9.4; pain during the year of the accident was excluded). Interestingly, secretarial work, light factory work, and heavy industry or construction work were not associated with lifetime prevalence of neck pain. It should be noted that there were only 15 people in the heavy work group, compared with 2 to 6 times that many in the other work categories. The authors also found those reporting pain reported spending significantly more hours reading each week, and significantly fewer hours watching television, than those who did not report pain. Given the cross-sectional design, no temporal conclusions can be drawn from the findings. In summary, these studies demonstrate that neck discomfort and pain are prevalent in workers performing diverse tasks. Based on consistency of association, driving appears to be associated with neck discomfort, although the one study that accounted for vibration dose was negative. The findings from the studies that focused on other factors were not inconsistent with findings from previous reviews, regarding concern for VDT work and physically demanding work. However, taking these studies separately, it is difficult to draw conclusions about any specific risk factors from them. b) Risk associated with job title (1) Physical therapy Bork et al. (1996) conducted a cross-sectional study among 928 physical therapists. Based on self-report, 12-month prevalence was 24.7% for "job-related ache, pain, discomfort, and so on" in the neck and 28.7% for the upper back. Based on self-report, "providing pediatric care," as opposed to adult care, was significant for neck and upper-back problems, and "routinely perform neurologic rehabilitation" was also significant for upper-back symptoms. Significantly greater symptom prevalence were found in women, for both neck and upper back. Symptom prevalence was found to decrease after age 45, in spite of no decrease in patient exposure. The authors hypothesized survivor effect. (2) Nursing Josephson et al. (1997), using both a fixed cohort and repeated cross-sectional surveys, presented results of a study of MSDs in nursing personnel over 3 years, based on surveys repeatedly collected over a 3-year period. Twelve-month prevalence of neck symptoms was 53% at the first survey (n = 565). The prevalence for the upper back was 30%. All analyses were performed based on having MSD symptoms in any body region, so they are not included in this review. (Several studies examined for this review provide neck symptom or disorder prevalence data on subject groups, but performed analysis only on a combination of body parts.) (3) Auto mechanic Torp, Riise, and Moen (1996) reported on the 12-month prevalence of neck and upper- back pain in Norwegian auto mechanics, based on a cross-sectional study design. At 62%, neck pain was second in prevalence only to low-back pain, while upper-back pain prevalence was 29% (shoulder pain prevalence was 52%). Although mechanics were questioned about the physical work environment, the study's authors hypothesized that lack of a reference group and low variance in work conditions among the mechanics precluded identifying associations between work factors and neck pain in the mechanics. (4) Agricultural work: tomato trainers Palmer (1996) reported finding a greater 12-month prevalence for upper-back symptoms in tomato trainers than in pickers. This study found no differences in the prevalence of neck symptoms between the two groups of workers. The author suggested the differences might be due to the greater amount of shoulder height work done by the tomato trainers. However, there were other differences between the two groups. The average years of employment for tomato trainers was 3.2 years vs. 0.5 years for pickers, and 89% of tomato trainers worked full time vs. 29% of pickers. The authors cited other studies that had shown no dose-response relationship for pain and overhead work, so they did not feel comparison between the two groups was invalid. (5) Industrial work Alund, Larsson, and Lewin (1994) performed clinical assessments on a group of former steelworks grinders and a group of white collar workers, studying the "possibility of a work-acquired acceleration of the physiological degenerative changes taking place in the cervical spine with age." Grinders had performed that work for an average of 8 years, at a relatively young age (approximately 18 years earlier). There was a greater prevalence of "neck pain/stiffness, pain/stiffness in areas other than neck or shoulder, recurrent neck/shoulder and/or low back pain/stiffness, and numbness or tingling sensations peripherally" in the group of former grinders than in the control group. Pain in the former grinders did not appear to be muscular in origin. Grinders demonstrated reduced active head/neck range of motion (ROM) in every direction, and required more time to perform maneuvers than did controls. Grinders also received higher scores for foraminal encroachment in the cervical region due to posterior osteophytes and spondylarthritis than another group of controls (judged to be similar to the white collar controls). In a cross-sectional study of MSDs in workers at an aluminum smelter, Hughes, Silverstein, and Evanoff (1997) found a low prevalence of neck MSDs in three different groups of workers. The highest prevalences were 6.5% based on interview, and 1.6% based on interview and PE. Based on such low levels, no exposure/health outcome analysis was conducted. In summary, it is difficult to draw any general conclusions from this group of studies. Findings in the two studies of industrial workers are quite different, illustrating the need for task description and other exposure information in addition to job title. In combination with results from the previous section, though, at least the grinder study is consistent with other studies indicating that risk of neck-related problems can be associated with heavy work, vibration, or both. Several of the studies illustrate problems with drawing conclusions when there is no reference group, or when exposure within a group of subjects is invariant. c) Intervention impact (1) Workstation design Williams and Dickinson (1997) reported on the prevalence of neck symptoms and upper-back symptoms in female lock assemblers. From an initial survey they compared the prevalence of symptoms across several plants. The weekly prevalence of symptoms in both the neck and upper back differed across the plants, possibly indicating effects of differences in work design, work organization, or plant policies. One facility was surveyed a second time, 1 year after "ergonomic changes to the workstations," which included adjusting VDT screens to reduce neck extension and providing adjustable height chairs. Based on the limited information in the paper, it is difficult to draw conclusions about the effect of those changes. The changes did not seem to affect the occurrence of annual or weekly prevalence of neck or upper-back symptoms in workers who had been employed prior to the changes. This may mean that the changes were ineffective, or it may mean that employees that are exposed to problem work conditions can continue to experience symptoms even after work conditions have been improved, at least over a period of several months. Prevalence of neck or upper-back symptoms in workers hired after the changes was lower than that in the other employees, but whether this was due to the workstation changes or a shorter exposure time cannot be discerned. Harms-Ringdahl and Arborelius (1987) reported on the success of arm supports (suspenders or balancers) introduced in an electronics plant. The authors characterized the success of the intervention in terms of acceptance and continued usage (94%) over a year's follow-up and reduced pain perception ratings in the neck (and shoulders), to one-third to one-half of pre-intervention ratings. (2) Production methods Ólafsdóttir and Rafnsson (1998) studied the impact of introducing of a flow line production method to fish processing in the Icelandic fish-fillet industry. In general, changes for female workers included a reduction in work scope from performing several tasks to performing a single task, change from individual to group incentive pay, and provision of adjustable-height workstations and chairs. Production rates increased, and most female workers (71%) thought the modifications brought improvement. Only 34% of the men thought so. Women and men typically did different work in the plants, and the changes mainly affected the women's work. There were no changes in prevalence of neck or upper-back pain as a result of the workplace modifications. (3) Ergonomics training Rundcrantz et al., (1991) studied the effect of two types of intervention on dentists with pain in the neck, head, or shoulder. Dentists either received individual ergonomic instruction alone or ergonomic instruction and physiotherapy specific to their condition. Only those who received both interventions showed a decrease in neck pain 5 weeks after treatment or instruction. The authors did not discuss the specifics of the ergonomic intervention instruction, which makes it difficult to interpret their results. In summary, it is also difficult to draw any general conclusions from these studies. Intervention study outcomes are difficult to interpret when, as is often the case, several factors are changing at the same time (as in the fish processing or lock assembly studies), or when reference groups are not included (these can be difficult to arrange). However, arm supports seemed to be a positive intervention, based on decreased discomfort and (importantly) acceptance/usage by employees. 2) Supplemental research and reports (descriptive studies, observational studies, case series, others) Reports in this section generally collected information on either health outcome or exposure in groups of workers. a) Heavy physical work In a series of papers, Waskiewicz (1996, 1997) reported on the health effects observed over several years in various categories of dock workers, in particular stevedores, equipment operators, and hoistmen. Stevedores' work was reported to require energy expenditure of 7200 to 11200 kJ/shift, compared with 4650 kJ for operators and 2570 kJ for hoistmen. Stevedores loaded "from several to several dozen tonnes" of cargo per shift. The author noted that only 27% of the stevedores remained on the job over a 7-year period, while 91% of the operators and all of the hoistmen remained. In dock workers he examined over the years, he noted "deviations and changes within the locomotory functions" in the neck region at about twice the rate in stevedores as in the other groups of workers, and in the thoracic region at about 2 to 3 times the rate in stevedores as in the other two groups. b) Seated industrial work van Riel et al. (1995) sought to characterize and contrast the impact of the combined effects of work task and workstation design on head posture in three groups of workers within a container corporation: crane operators, straddle carrier drivers, and sedentary office workers. The most commonly adopted head flexion postures of the crane operators, office employees, and straddle carrier drivers were 60, 30, and 18 degrees, respectively. Crane operators were found to spend more than 80% of their days with heads flexed more than 30 degrees. The most common rotation position was 45 degrees for the straddle carrier drivers, and essentially neutral for the other workers. The authors did not attempt to associate these posture data with specific health outcomes in the groups, but did present their data with respect to parameters identified previously as representing "strenuous postures" (more than 30 degrees in the sagittal or frontal plane; more than 45 degrees in the transverse plane, per Karhu, Kansi, and Kourinka, 1977). Alund, Larsson, and Lewin (1992) sought to characterize the functional capabilities of a group of crane operators on long-term sick leave, and to compare those capabilities to a group of working operators and a control group of clerks. All subjects were female. Neck motion assessment demonstrated reduced active ROM in each direction for the workers on sick leave, compared to the other two groups. Women in the sick leave group also required more time to perform the maneuvers. Physical examination of all the subjects revealed a higher prevalence of muscle tenderness in the sick leave group, including tenderness in the right trapezius. Prevalence of subjective pain assessment in the neck and in the shoulder was greater for the sick leave group as well. The authors pointed out the fact that symptoms and functional limitations persisted in these women even after being away from the work (the presumed source of exposure) for at least 1 year (an average of 3 years). They suggested that primary prevention should be the preferred mode of prevention for work-related MSDs. c) Professional work Finsen et al. (1997) supplemented their cross-sectional study of dentists with an observational study of eight female dentists. One-year prevalence of neck pain was 50% in this group of eight, and shoulder pain prevalence was 88%, both higher than in the cross-sectional group. The dentists were observed to work with cervical spines flexed more than 15 degrees for 97% of the time during procedures, and more than 30 degrees over 80% of the time. About one-quarter of the time the neck was flexed greater than 30 degrees, it was also positioned in either lateral bend or rotation. Shoulder abduction exceeding 30 degrees was observed about 30% of the time. Postures were relatively static. Static EMG levels in the splenius exceeded 5% of MVC in each of the tasks studied. Static trapezius activity ranged from 7% to 10% of MVC, on average. These exceed recommendations for 2% to 5% MVC static loads. Finsen (1997) assessed dentists' neck extensor activity and flexion moment loads while working in the patient's mouth (neck flexion about 23 degrees at C7-T1), and while handling instruments above the patient (neck flexion about 46 degrees at C7-T1). They found no differences in splenius muscle activity between the two work locations, but did find a significantly higher moment about C7-T1. They cautioned that muscle activity, alone, was not sufficient for characterizing tissue loading in the task they studied. Problems with heavy work, recounted by Waskiewicz (1996, 1997), are consistent with results from many analytical epidemiological studies. With regard to industrial or professional seated work, this group of reports shows that neck posture is clearly driven by the interaction of the task and workstation design. It is also clear that neck posture can be measured, as can muscle activity required to sustain neck postures, for different tasks within a job. From this type of data collection effort, more taxing tasks can be identified and targeted for redesign. However, specific guidelines for workstation or task redesign are not apparent from these studies or any of the studies in this review that focused specifically on the neck or upper back. B.2.d Neck/Shoulder (Not Intended To Include the Glenohumeral Joint, Does Include Region Covered by Descending Part of Trapezius) 1) Analytical epidemiological studies from 1996 (cross-sectional, cohort, case-control) a) Risk associated with job characteristic (posture, task, job design, classification, etc.) (1) Awkward postures Brulin et al. (1998) performed a cross-sectional study among home care personnel in Sweden. Based on self-reporting of symptoms and exposures, the authors determined that 7-day prevalence was 44% for neck symptoms, 47% for the shoulders, and 30% for the upper back (low back was 40%). After accounting for age, workplace, and work hours per week, only one of four indices of physical work load on the body was found to be significant for neck/shoulder complaints. The adjusted OR for "standing in forward-bent and twisted postures" was 2.0 (90% CI: 1.2-3.1). The authors also found an interaction between this factor (index) and one of the psychosocial indices, "possible to influence planning of work." If both indices were "poor" the adjusted OR was 3.5. Ylipaa et al. (1997) performed a cross-sectional study of Swedish dental hygienists, seeking to determine if there were associations between MSD prevalence and either physical or psychosocial work factors. Based on analysis employing stepwise regression, neck/shoulder/upper-back pain was predicted by "strainful ergonomics" (incremental RR of 0.19), solitary work, patient treatment hours, and working hours per week, which together explained 29% of the variance among respondents. Factors not found to be significant included age (dichotomized at 43 years), marital status, co-worker and supervisor relations, and delivery system (public vs. private). In contrast, Vasseljen, Jr., and Westgaard (1997) found no differences in posture (assessed as arm elevation), or in muscle activation, between manual workers (postal workers performing sorting activities, and paper mill workers with similar repetitive demands) designated as either cases (shoulder and neck pain) or controls, when monitoring workers performing representative tasks during a 30-minute session. The authors concluded that median arm elevation below 45 degrees was not a sensitive indicator for shoulder and neck pain. The first two studies in this group tend to support the notion of a relationship between awkward or strainful postures and neck/shoulder discomfort, consistent with other papers in this review. It is unclear whether the third paper contradicts the first two, since it is the only one that specifically described the risk factor of interest. It is consistent with several studies that have found no differences in muscle activation between cases and controls. b) Risk associated with job title (1) Forestry work In a national cross-sectional study, Hagen, Magnus, and Vetlesen (1998) assessed the 12-month prevalence of neck/shoulder disorders (pain occurring for more than 30 days within the last 12 months) in Norwegian forestry workers. Workers were categorized as manual workers (chain saw work), machine operators (heavy equipment operators), and administrative workers. With administrative workers as the reference group, and controlling for age, adjusted ORs for the manual group and the machine group were 2.34 and 3.37, respectively. Prevalence of neck/shoulders disorders increased with age. Some psychosocial factors were found to be significant as well. The authors mentioned that the chain saw workers were typically exposed to localized vibration, while the machine operators were subject to whole-body vibration. However, these factors were not explicitly explored, nor were differences in other work factors or non-work factor differences, including level of formal education and pay structure. c) Intervention impact A paper by Kuorinka, Alaranta, and Erich (1995) illustrated some of the difficulties in conducting intervention research. They attempted to compare the preventive effects of a dynamic training/relaxing program to effects of workstation improvements in groups of manufacturing employees. In the training program, only half the group completed the training; in the workstation change group, only one-third of the workstations were changed. Additionally, production changes occurred during the course of the study that were outside the control of the researchers. (1) Workstation design Orgel, Milliron, and Frederick (1992) reported reductions in self-reported "neck/upper- back/shoulder" symptom intensity, pain medication requirements, and number of days to recover from symptoms, in a small group of grocery store cashiers, following workstation design modifications. Modifications consisted of strategic placement of a barrier to keep items from moving too far away from the cashier, a height-adjustable base for the keyboard, and a training videotape to improve work practices. These modifications were expected to reduce awkward shoulder postures (flexion and abduction) and trunk flexion. The authors recounted limitations of their work, including lack of a control group, small sample size, and reliance on self-reports of health status. They also suggested that ergonomic intervention is an iterative process, and that employees can be an important resource in the process. Based on measurement of trapezius muscle activity pre- and post-intervention, Aarås (1994) demonstrated an association between a reduction in trapezius activity post-intervention and reduced sick leave in a group of female assembly workers. The intervention included provision of adjustable-height workstations supporting sit-stand options, suspension of hand tools, and provision of arm supports. A significant reduction in trapezius activity was also observed in a group of VDT operators, who reported reduced shoulder pain intensity following table and chair adjustment for relaxed shoulder positioning, and additional tabletop work surface area. Aar&229;s et al., (1998) monitored the effects of a series of interventions -- new lighting, followed by new workstation design, followed by optical correction in groups of VDT operators -- while following a control group that did not receive any intervention. For the intervention group, workstations were redesigned with the monitor in the corner. The keyboard was located in front of the monitor away from the edge of the desk (to provide a support surface for the forearms) and height was individually adjusted for each worker. Chairs with height and back inclination adjustability were provided. While there did not appear to be changes in neck or back posture after the intervention, shoulder abduction and flexion angles increased; and descending trapezius muscle activity was significantly lower, decreasing from 1.5% of MVC to 0.3% of MVC. Additionally, neck pain intensity decreased in the groups receiving the intervention, while there appeared to be an increasing trend in neck discomfort in the control group. The authors of this study found no association between median head flexion (1 to 26 degrees) and neck or shoulder pain; found no association between neck and shoulder pain and psychosocial indices; found a correlation between visual discomfort and neck and shoulder pain; and found a correlation between muscle activity and neck and shoulder pain. Taken together, these intervention studies seem to indicate that changes in workstation design that reduce repeated reaching (awkward postures) or that reduce the load on the trapezius muscle to levels below those which are commonly considered acceptable (2% to 5% of MVC), can reduce neck or neck/shoulder discomfort in workers. 2) Supplemental research and reports (descriptive studies, observational studies, case series, others) a) Light repetitive work Muscle activity requirements from letter sorting tasks were studied by Jørgensen, Fallentin, and Sidenius (1989). Based on assessments of six male letter carriers, the researchers determined that static EMG levels were at the upper limit (5% of MVC) of typical recommendations, while median levels were at the lower end of recommendations for dynamic work (10% of MVC). Based on this assessment and an observation of workers working above shoulder height 25% of the time, recommendations for elimination of higher rows of mail slots were made. However, the authors acknowledged the potential adverse impact that additional lower rows might have on neck postures. Bao et al. (1997) characterized two assembly line workplaces, one Swedish and the other Chinese, with respect to exposure to neck/shoulder risk factors, in an effort to explore potential trade-offs between ergonomic improvements and traditional industrial engineering improvements (geared toward production efficiency). They found that shoulder flexion and abduction were less extreme for the workers on the Swedish line (SL), but that the better balance of the SL provided the workers with an almost uninterrupted work pace. This was evidenced in part by EMG data from the trapezius showing that workers on the Chinese line (CL) exhibited low EMG activity 10.8% of the time, compared to only 0.6% of the time for the workers on the SL. The trade-off may have manifested itself in the identical rankings of subjective fatigue in the neck/shoulder region at the end of the day from each group (3.8, which is considered moderate to somewhat strong). Both reports illustrate conflicts that can occur when trying to incorporate ergonomics into the design of work. In the first report, the authors foresaw conflict between designing for good shoulder posture and good neck posture, while the second report presented a potential conflict between highly efficient production methods and sufficient work/rest time ratio for workers. b) VDT work Similar to their findings in two industrial samples, Vasseljen, Jr., and Westgaard (1997) found no differences in muscle activation or arm elevation between VDT operators designated as either cases (shoulder and neck pain) or controls, when monitoring workers performing representative tasks during a 30-minute session. c) Professional work Preceding Finsen's work on dentists (which was specific to the neck), Milerad et al. (1991) characterized activity in the left and right trapezius muscles of 11 dentists performing six different dental procedures. Static EMG levels were about 2% of MVC, and median values were 5.5 to 6.3 % of MVC. The authors also observed that the dentists typically worked with upper arms unsupported in abducted and flexed positions. They also noted individual differences in postures among dentists. B.2.e Summary In the last 4 years, the vast majority of epidemiological research in the area of work-related MSDs in the neck and neck/shoulder regions, as in previous years, has been of a cross-sectional nature, which makes causal inference virtually impossible. Many studies lacked a reference group. Many studies relied on self-report of symptoms and exposure. Most did not focus on specific diagnosed entities. Most of the studies did not examine generic factors (such as postures and muscle activity) but, instead, asked questions specific to the tasks that the sample of workers performed. This information can be helpful in developing appropriate work practices for specific occupations, but does not improve the understanding of injury mechanisms or symptom development. In spite of this, a few conclusions can be drawn from this group of studies:
B.3 Experimental Research of Physical Factors Associated With Neck and Neck/Shoulder Disorders and Pain Risk factors can be identified through epidemiological work, addressing what may be causing(2) a problem, but experimental work can help explain how factors may cause adverse health effects. There has been much work in the area of the cervical spine, including anatomical and clinical research, and basic and applied biomechanical research. However, many fundamental questions about cervical spine pain remain, including questions regarding diagnoses, diagnostic methods, and the etiology of cervical pain (Vasseljen and Westgaard, 1995). According to Garfin (1998), this is due to a "limited understanding of the effects of compression, ischemia, chemical alterations, and age on the spinal cord, spinal column, and surrounding structures." The scope of this portion of the review is limited to those biomechanical research studies that specifically address occupational issues. Papers included in this section describe the effects of researchers' manipulation of conditions or other factors that were expected to affect the neck, upper back, or descending/upper trapezius regions of the body. Postural measurements and EMG often used to objectively assess subjects' responses to work changes, while preferences and perceptions of discomfort or task difficulty are used to capture subjective responses. Many researchers have demonstrated direct relationships between joint posture and muscle activity in controlling muscles. Most acknowledge significant individual differences in muscle activation between subjects. Vasseljen, Jr., and Westgaard (1997) concluded that arm elevation (upper arm raised in any combination of flexion and abduction) and trapezius activity were "largely independent representations of biomechanical exposure in the shoulder and neck region" for the tasks they studied. B.3.a Static Seated Work: Keyboarding and Pointing Device Use Numerous studies related to VDT work have been conducted, in both laboratory and field settings. VDT workstation design has been shown to directly affect operator posture and muscle activity, though interpretation of the results of these studies -- particularly studies of various monitor locations within the sagittal plane -- remains a matter of debate, in terms of the impact on workers. 1) Adjustable furniture Field studies have found adjustable furniture to have a positive effect on neck and upper- back discomfort in field studies, with days or weeks of exposure to conditions (Shute and Starr, 1984). Based on heart rate, systolic blood pressure, and discomfort data from 25 young men seated in four seat inclination conditions (0, 5, 10, and 15 degrees) for 15 minutes, Naqvi (1994) recommended a seat with a forward inclination of 5 degrees to obtain a comfortable neck position, 10 degrees to obtain a comfortable low back position, and 15 degrees to obtain a comfortable position for the whole body. 2) Computer monitor location VDT location in the sagittal plane has been studied by many, mostly in laboratory settings. If the work task requires operators to focus on the monitor, research generally shows, changes in head tilt and (in some studies) neck angle, augmented by changes in eye rotation, compensate for changes in VDT location. Disagreement, based often on models, theory, or conjecture, exists regarding the significance or appropriateness of these induced postures. For example, some biomechanical models suggest that a 30-degree-downward gaze is best because the moment arm of the head and moment arm of the neck extensor muscles are equal in that position (Vital and Senegas, 1986). Others argue for balancing the head over the spine by inducing an upward gaze, thereby "eliminating" the need for muscle activity. (Snidjers, Hoeck van Dijke, and Roosch, 1991). Burgess-Limerick, Plooy, and Ankrum (1998) demonstrated that small changes in head angle (3 to 4 degrees) accompanied changes of 13 to 17 degrees in gaze angle, from eye level to 15 degrees below and from 15 degrees to 29 degrees below eye level (5 minutes on task). Neck angle was not significantly affected. Burgess-Limerick et al. (1999) reported similar findings: an average 5-degree change in head flexion with an 18-degree declination of the monitor, and non-significant change in neck angle. In this study, the role of the document holder (primary or secondary point of focus) was not clarified. Kietrys, McClure, and Fitzgerald (1998) determined that there were no practical biomechanical differences between working with a monitor on a 29-in. (74-cm) desktop or working with one located on the CPU, because, although they found small but significant differences in various head and neck postures between the two conditions, the flexion moment of the head about T1 was similar in the two conditions (lab study, 10 minutes each condition). They suggested that subjects seemed to adjust between the two conditions with a combination of changes in screen tilt and eye rotation. Based on user preference, biomechanical modeling, inducement of more erect postures, and no added restriction of movement, de Wall et al. (1992) suggested the center of the monitor be located 0 to 15 degrees above eye level, at least for operators who look at the screen most of the day (field study, exposure of 2 days). Comparing discomfort and performance during typing for the monitor placed at eye level, and 250 mm above or below eye level, Lu and Aghazadeh (1993) found that subjects performed better at either the higher or lower location, but felt less discomfort in the neck at the eye-level location, followed by the lower location. Subjects felt the least discomfort in the upper back when the monitor was placed at eye level, and similarly felt more in the other two locations (lab study, exposure of 35 minutes each condition). Neck and shoulder muscle activity has also been used to evaluate the effects of monitor placement. Aarås et al. (1997) compared locations of 15 and 30 degrees below eye level, during typing text from a document holder located at about 30 degrees below horizontal, and found no differences in head angle or trapezius muscle activity. However, this was most likely because subjects focused on the document holder during the task. Based primarily on lower splenius and trapezius muscle activity and no differences in subjective assessment of strain, Bauer and Wittig (1998) recommended that the monitor be centered between 0 and -17.5 degrees, rather than below -17.5 degrees, based on results from a laboratory study of a reading task and a visual search task (subjects tested 5 minutes in each location). When subjects in a laboratory performed a 2-hour-long reading task that required them to focus exclusively on the monitor, Turville et al. (1998) found, in comparison to a -15 degree location, a -40 degree monitor location was associated with greater activity in the right and left cervical and thoracic erector spinae muscles, as well as the right sternocleidomastoid and levator scapulae. Eye-ear line was angled at 12 degrees above the horizontal for the -15 degree location and at about 6 degrees below for the -40 degree location, and postural shifts were infrequent in each location. Subject discomfort in the neck and upper back was also similar for the two locations. Villanueva and colleagues (Villanueva et al., 1997; Villanueva et al., 1996) also observed significant increases in neck extensor and trapezius muscle activity with a reduction of monitor height, from 120 cm to 80 cm above the floor (corresponding to average viewing angles of eye level and -39 degrees, respectively) in a laboratory study. Postural changes also occurred, including increases in neck angle, trunk inclination, and thoracic bending. Based on muscle activity reductions, they recommended the more upright placement, although they acknowledged the trade-off of an increase in ocular surface area with higher placement. The interaction of sagittal monitor placement and bifocal use was studied in a laboratory setting by Kumar (1994). This author recommended a sunken monitor placement, finding that such a placement resulted in reduced subjective task difficulty and discomfort, lower muscle activity, and a reduction of the angular span of view that was within the span of the bifocal. For tasks in which the VDT is not the primary focus, Hamilton (1996) recommended that the source document be placed directly in front of the operator, and the monitor off to the side. This recommendation was based on findings of higher neck extensor activity when the document holder was placed to the side, in both reading or typing tasks. For the mid-sagittal document holder location, muscle activity in the neck extensors was lower for a higher document location (taking the place of a monitor set with the top at eye level) than a lower location (described as "between monitor and keyboard"). Taken together, these studies indicate that if lower muscle activity is better than higher activity, then an eye-level or slightly lower placement of the monitor is preferable, except for bifocal wearers. Although one could argue that muscle activity has been shown to be quite low for VDT use, the intervention study by Aarås et al. (1998) demonstrated that even reductions from 1.5% to 0.3% of MVC were beneficial (associated with reduced discomfort) in workers performing assembly work. 3) Sit vs. stand VDT workstation design Aarås et al. (1997) compared locations of 15 and 30 degrees below eye level, in subjects typing text from a document holder, in both standing and sitting, and found a tendency (not significant) for lower trapezius activity when standing. However, neck flexion was significantly greater when standing (31 degrees vs. 18 to 22 degrees). They also found a significant decrease in trapezius muscle activity when subjects' forearms were supported on the work surface, compared to unsupported conditions. 4) Upper-extremity support In contrast to findings by Aarås et al. (1997) regarding benefits of forearm support, Bendix and Jessen (1986) found trapezius activity increased with use of a "wrist support" (pictured as supporting the distal half of the forearm) with a typewriter, in spite of subjects being given a week to become acclimated to the wrist support. Wells, Lee, and Bao (1997) did not see a similar increase with use of a wrist support for mousing, but found support at the elbow was the most effective support location for reducing static trapezius activity. Erdelyi et al. (1988) found that female word processors who were experiencing shoulder pain (local tenderness and reduced mobility upon clinical exam) at the time of their study, had higher trapezius muscle activity (assessed in µv) than did their pain-free counterparts, when performing a standardized typing task. Both arm rests and arm suspension devices reduced muscle activity in all cases, but increased activity was observed in some of the controls. The greatest reduction occurred when elbows were open to 105 degrees. However, neither cases nor controls preferred the supports. Preference of the cases was for the 105-degree elbow angle without support; controls preferred either a 105- or 90-degree unsupported condition. In a review of the literature concerning upper-extremity supports, Sommerich (1999) concluded that the presence of arm supports in the workplace did not guarantee benefits or usage, due to a few points of conflict. One conflict occurs between the need for movement and the need for support, and another between the preferences of the individual workers and the workplace designer. Height and point of application are both key design features of arm supports, yet because they are so variable, general conclusions about arm supports cannot be made. 5) Type of computer: notebook vs. desktop Inherent differences between notebook and desktop PCs (NPCs and DPCs) include screen size and either screen location or keyboard location or both, relative to the user. These design characteristics have been shown, in other studies and the ones reviewed herein, to affect body segment postures and muscle activity. Straker, Jones, and Miller (1997) compared assumed postures and discomfort after subjects worked for 20 minutes at a DPC and an NPC, with subjects making workstation adjustments prior to working on each type of computer. Discomfort and performance did not differ between computers, although neck angle and head tilt both increased during NPC use. Comments from some non-touch typists were favorable toward the NPC computer due to the proximity of screen and keyboard. Studies by Saito and colleagues (Saito et al., 1997; Villanueva et al., 1998) contrasted smaller and larger NPCs and a DPC. Neck extensor activity was greater with the NPCs, though trapezius activity was not affected. Neck angle was also significantly increased, but only with use of the smaller NPCs (which had smaller screen sizes and lower monitor heights). In looking at configurations that would improve comfort while working with NPCs, Price and Dowell (1998), found that it was better (causing less subjective discomfort in the neck, upper back, and other locations) to optimize keyboard location (table top location) rather than monitor location (raised computer). 6) Keyboard design Fernstrom, Ericson, and Malker (1994) evaluated cervical erector spinae and trapezius muscle activity during short-term use (10 minutes per keyboard) by typists (experienced on mechanical typewriters) of several types of keyboards, including mechanical and electric typewriter keyboards and standard and alternative computer keyboards. The standard computer keyboard did not provide any reduction in trapezius muscle activity, compared to the typewriter keyboards. The alternative computer keyboard did show reduced activity in comparison to some of the typewriter keyboards. Problems precluded analysis of cervical erector spinae activity. 7) Work duration and work/rest cycle design Zipp et al. (1983) reported an increase in trapezius muscle activity over time, during a 1-hour period of uninterrupted typing. This finding shows that typing, a light physical activity, can induce localized muscle fatigue. One of the means for reducing or slowing the fatigue process is to introduce breaks into the work cycle. Work organization was studied by Hagberg and Sundelin (1986), in a field study carried out over 3 days. The researchers compared a 5-hour work period, a 3-hour work period, and a 3-hour work period with short pauses every few minutes. Muscle activity in the upper trapezius was the same in all test conditions. Discomfort increased in the neck over time under all test conditions, but subjects felt less discomfort at the end of the work period with pauses compared to the other two. Sundelin and Hagberg (1989) assessed effects of different types of rest -- passive pauses (sitting with hands in lap), active pauses (sitting while performing exercises), and diverting pauses (walking away from the workstation) -- on muscle activity and subjective discomfort. Pauses lasted 10 to 12 seconds, and occurred every 6 minutes for a 30-minute test period. The authors found nothing in their data, other than subjective preference, to lead them to recommend one type of pause over another. Subjects preferred the more active pauses, but found the frequent introduction of pauses disruptive. 8) Task type: computerized vs. paper-based Wærsted and Westgaard (1997) compared a task in two forms, paper-based and computerized. They found equal or reduced trapezius muscle activity and a more erect seated posture in the computerized task. The computerization also resulted in an increased pace with no significant change in error rates. However, long periods of static EMG activity in the computerized task were also found. Static loading was discussed as a musculoskeletal hazard for the spine and low-threshold motor units. Postural fixity was highlighted as a confounding factor to the increased productivity of such a paperless task. 9) Mousing The effect on the neck and shoulder of using alternative input devices, such as the mouse, has been the subject of several studies. Cooper and Straker (1998) found decreased upper trapezius activity, yet no significant differences in discomfort, in comparing subjects performing a mousing and a keyboarding task. The study used short task durations (10 minutes), ten subjects, and non-normalized EMG data in grouped data sets. Cook and Kothiyal (1998) examined the effects of three different mouse locations on muscle activity in the neck and shoulder. Ten subjects performed mousing tasks in each location for 20 minutes. While trapezius muscle activity was not affected by mouse location, the authors recommend a mouse location adjacent to a keyboard without a numeric keypad, based on reduced deltoid muscle activity and subjective reports of discomfort. Karlqvist et al. (1998) studied the influence of six mouse positions (near and distant versions of each of the following: central to the body, in front of the right shoulder, and to the right of the right shoulder) on posture, muscle activity, and perceived exertion. Twenty subjects performed the tasks in each position for 2 minutes. The far locations tended to induce subjects to support the forearm on the work surface, which generally meant lower muscle activity in those locations. It was noted that subjectively uncomfortable locations (those inducing external rotation) could produce low trapezius muscle activation. Most subjects preferred the location directly in front of the right shoulder, with taller subjects preferring the far location and shorter subjects the closer position. It was also noted that subjects who restricted the height of the work surface to less than 0.03 m above elbow height displayed lower trapezius muscle activity. The majority of the VDT studies reviewed herein employed EMG data. Where conditions were shown to affect muscle activity levels, those conditions producing lower muscle activity were typically recommended (eye-level rather than much lower monitor locations; use of arm support; incorporation of rest pauses during work; neutral mouse location). Several studies also assessed posture, and where conditions were shown to affect posture, those conditions producing more neutral postures were recommended (higher rather than much lower monitor location; laptop configured with peripheral, which provided for neutral head and arm posture). B.3.b Static Work/Sustained Postures 1) Seated work -- other than keyboard a) Precision work Kumar and Scaife (1979) studied posture and muscular effort during microscope work. A precision wire-threading task was performed at a combination of three work surface heights and five work surface angles. Although the changes in the workstation correlated poorly with cervical muscle activity, changes in the head tilt and neck angle were predictable. Increasing the tabletop height decreased the neck angle and head tilt; tilting the table top forward resulted in a decrease in neck angle and an increased head tilt. The authors suggested adjusting chair and table heights for each individual and using a microscope with a prism designed to deflect the optical path. A series of studies (Schüldt, 1988; Schüldt et al., 1986) was performed to determine muscle activity requirements in various seated postures during a simulated circuit board assembly task. In general they found that neck flexion was accompanied by increased activation of the cervical erector spinae muscles, regardless of trunk posture. Slumped-forward trunk posture increased activity in neck, shoulder (trapezius sampled in several locations), and upper-back muscles, compared to a vertical trunk orientation. Reclining the trunk slightly reduced activity in those muscles (Schüldt, 1988). Elbow supports and arm suspension devices were both shown to reduce neck and shoulder muscle activity, though the extent of reduction depended on the subject's trunk posture (Schüldt et at., 1987). Abduction of the shoulder increased activity in the trapezius and upper back muscles; work targets further away and higher elicited more muscle activity than those that were lower and nearer. Takala and Viikari-Juntura (1991) used a case-control design for studying differences in muscle activation between female bank cashiers with shoulder-neck pain of muscular origin and those with no pain. The researchers found no difference in trapezius or upper-back muscle activity between the cases and their matched controls when performing a simulated circuit board assembly task. They found that both responded similarly to changes in location of the circuit board and the location of targets on the board. These findings are consistent with those of Vasseljen, Jr., and Westgaard (1997). Working at arm's length increased activity in the trapezius muscle. Consistent with Schüldt's findings, work targets further away and higher elicited more muscle activity than those that were lower and nearer. Orr and Wells (1994) studied the interaction of work speed and task (stapling) precision with a variety of biomechanical indicators. The trapezius muscle activity was not influenced by changes in task speed or precision. For this task, the wrist-related measures were affected by both speed and precision. Milerad and Ericson (1994) evaluated the influence of precision, along with force, grip size, and arm support on shoulder and upper-extremity muscle activity. Twelve subjects without professional manual work experience performed 14 simulated work tasks for 30 seconds each. Precision, force, hand support, and grip diameters did not significantly affect trapezius activity. (Precision did influence other extensor muscles studied.) Arm support significantly reduced all shoulder muscle activity, including that of the trapezius muscle. Results from these experiments are consistent with those of the VDT experiments described previously and with the descriptive studies of dentists and crane operators that showed that workstation design and location of visual targets influence neck and/or shoulder postures. These studies also found that neck muscle activity increased with neck flexion, and trapezius muscle activity increased with shoulder flexion. Degree of precision did not seem to impact trapezius muscle activity in either study in which it was a factor. b) Sustained postures Harms-Ringdahl et al. (1986) studied the impact of a common work posture on the development of pain in the lower cervical and upper thoracic spine. Ten female subjects maintained a posture of relaxed, forward neck flexion for up to 60 minutes. Pain developed in all subjects within 15 minutes, beginning in the dorsal cervical region, increased with time in position, and disappeared 15 minutes after returning to neutral posture. However, all but one subject experienced post-provocation pain that evening or the next morning. Muscle activity was generally low (mean of 2% of MVC in trapezius, mean less than 4% of MVC in splenius), but tended to increase over time (more pronounced increase in splenius). The effects of sustained concentration were investigated by Bansevicius, Westgaard, and Jensen (1997), through use of a 1-hour-long, two-choice reaction time task. The researchers noted that although muscle activity in the neck extensors was low throughout the task (median value of 1.8% to 2.1% of MVC) and did not increase over time, ratings of discomfort did increase, fairly steadily, from 1 to 10 mm on a visual analog scale (VAS). Shoulder muscle activity was also low (task median was 0.6% of MVC), but increased with time, as did discomfort ratings (0 to 10 mm). This study demonstrated that even with low levels of muscle activation, discomfort can develop, but that muscle activity is not necessarily indicative of discomfort, at least at low levels (the authors suggested below 10% of MVC). While one of these studies showed neck muscle activity to increase over time and the other did not, this might be explained by the severe posture subjects adopted in the study that found an increase in activity over time. In both studies muscle activity was very low, but both studies subjects felt discomfort -- those in the extreme posture felt more (their VAS at quitting was 58 to 99 mm). Taken together, these two studies could lead one to conclude that maintaining any neck posture for an hour can result in development of discomfort, with the degree of discomfort a function of time and the severity of the posture. 2) Standing work Whistance et al. (1995) studied postural adaptations to restrictive standing workstations in a group of fairly young subjects. Restrictions included location of work object (near or far), and foot position (restricted, unrestricted, or foot rail provided). Subjects performed a simple assembly task (jigsaw puzzle), with the work surface at elbow level, for 10 minutes in each condition. Neck flexion was greater in the near work location by about 10 degrees, and the neck was the most frequently identified body region of discomfort for two of the conditions involving near work (in the near work with footrail location, the supporting leg was identified most often). B.3.c Repetitive Work 1) Light work: effects of workstation and tool design In a study of different types of cash register systems, Lannersten and Harms-Ringdahl (1990) found that the static load for the cervical erector spinae was below the commonly recommended limit of 2% to 5% of MVC regardless of the system in use. The static load for the trapezius fell in the 2% to 5% range and both muscle groups fell within the recommendations for the median load and peak load. However, the repetitive merchandise handling influenced static loading more than the type of cash register system. Standing reduced activity except for the cervical erector spinae. Using the vertical scanner minimized cervical erector spinae activity and maximized all but one measure associated with trapezius activity. Hagberg (1981) evaluated work load and shoulder muscle exertion in six subjects performing repetitive arm flexions. Perceived exertion increased faster than heart rate, indicating an increased importance of local factors such as muscle and tendon strain. All subjects reported discomfort in the descending trapezius after the experiment, 77% after 24 hours, and 33% after 48 hours. Hagberg concluded that exertions of the descending part of the trapezius could be an explanation for worker reports of neck discomfort during tasks with repetitive arm elevations. Wiker, Chaffin, and Langolf (1989) also looked at a task that required arms to be raised during work. They examined the effects of vertical work location (defined as hand elevation from just below shoulder height to overhead) in combination with tool weight, tolerance requirements, and work cycle design on upper trapezius muscle activity, postural tremor, and global and regional discomfort. Postural tremor and global discomfort were significantly influenced by elevation, stylus weight and duty cycle. They also found that upper trapezius muscle activity was reduced with the shorter duty cycle (working 20 seconds of each minute vs. working 40 seconds of each minute) and when work was performed with hands below shoulder level. Residual discomfort in the trapezius was also reduced with the shorter duty cycle. Lee et al. (1997) studied a task that was less stressful, in terms of vertical work height, but one that was highly repetitive: applying spray-on finishes (as in furniture manufacturing). In particular, they examined the effects of spray gun modifications, comparing a traditional spray gun to a new one that was lighter and afforded both inline and pistol gripping. They found significant reductions in median frequency shift for the new spray gun and reduced discomfort in the upper back. Neck discomfort scores were associated with time on task and years of work experience, with novice sprayers experiencing greater discomfort than experienced sprayers. These studies make it clear that there are many opportunities to improve work design in areas that affect neck/shoulder comfort, areas such as tool design and location of the work piece. It is also clear that interactions among workplace factors, and between workplace and personal factors (such as experience and skill), should be considered. 2) Work cycle design Many types of work use predetermined time systems (PDTS) to set the standard pace of work. Sundelin and Hagberg (1992) found that four of six subjects who performed a simulated work task, with task pace set by PDTS (2466 cycles per hour), showed signs of muscle fatigue after 1 hour. Median values for peak activity for the cervical and lateral portions of the trapezius muscle were 37% and 31% of the reference MVC, respectively; static loads were 17% and 4%, respectively. Although there is a strong concern regarding static work and sustained postures, as evidenced by many of the papers in this review, these authors suggested that dynamic work can also cause muscle fatigue, at least dynamic work constituted by continuous, repetitive short-cycle work. Workstation design, of course, is an important interacting factor. 3) Load bearing Rice et al. (1996) found that using four-person military stretcher-bearing teams instead of two-person teams reduced neck discomfort (but had no effect on upper-back or shoulder discomfort). Many other dependent variables were studied, and in consideration of all of those, the authors suggested that four-person teams were best for mass-casualty scenarios, and that harness systems should be made available to two-person and female teams. B.3.d Active Seated Work Eklund et al. (1994) were specifically interested in studying the effect of vehicle and equipment cab design on operator head postures. They studied small groups of forklift operators (forward-seated and sideways-seated), forestry machine operators (with or without rotatable cab), and crane operators (with or without rotatable cab). The cab designs were shown to impact head postures directly. Less head rotation was exhibited, and less time was spent with head rotated, with the forward-seated forklifts and the rotatable cabs. For example, the sideways-seated forklift operation resulted in head rotation exceeding 45 degrees as much as 50% of the time, while the forward-seated design reduced that proportion to less than 20%. B.3.e Summary Taken together, the experimental studies in this review clearly demonstrate that workstation design, tool design, and work design all affect neck and arm postures, and can affect subjective discomfort. However, questions always arise as to whether discomfort in experimental subjects is similar to what skilled, experienced workers would experience (given differences that have been demonstrated between skilled and unskilled workers), and whether discomfort is a health outcome about which there should be concern. Clearly, many researchers who conduct epidemiological studies feel that self-reported discomfort in workers is an important health outcome. It is more difficult to dismiss concerns regarding inexperienced subjects, or subjects working under novel conditions for brief periods of time. Some studies in this review did employ workers as subjects, though most required even experienced workers to work in novel situations for short (to brief) periods of time, making assessments of discomfort suspect. Subjects are more likely to be able to assess discomfort over a longer exposure period (more than 5 to 10 minutes). Additionally, muscle activity patterns are more likely to be lower in people who are familiar with a task. Given these concerns, it can still be said that the studies were generally supportive of the findings in the epidemiological work. For example, neck and shoulder discomfort is common in VDT workers, and many of these studies demonstrated that certain workstation designs were more comfortable than others and that some required less muscle activation than others. The observations of Bao et al. (1997) and the experimental work of Sundelin and Hagberg (1992) highlight the potential for conflict between designing jobs for worker comfort/health and designing for efficiency improvements. B.4 Additional Factors Associated With Neck or Upper-Back Disorders and Pain B.4.a Psychosocial or Psychological Factors A number of studies have focused on psychosocial work factors or psychological factors in an effort to understand or predict the occurrence of work-related neck or neck/shoulder discomfort. Only a few are mentioned here. Psychological distress was found to be predictive of neck-shoulder disorder development in a group of Finnish farmers followed prospectively for 10 years (Manninen et al., 1997). Through univariate analysis, Zetterberg et al. (1997) found that "neck-myalgia" identified through PE increased with stress at work, as did 12-month prevalence of self-reported neck pain and upper-back pain. They also found pain in both locations to decrease with "good relation to workmates/foremen" and with work satisfaction. Vasseljen, Westgaard, and Larsen (1995) found few psychosocial factors that were important (explanatory) for both types of work they studied (manual vs. office). In their cross-sectional study of medical secretaries and hospital office personnel, Linton and Kamwendo (1989) found perceptions of workload, work demands, influence over work, and social support at work to be related to neck pain reports. Based on both PE and symptom reporting, Toomingas et al. (1997) found perception of social support (high vs. low) to be related to tension neck syndrome, in a mixed group of workers. B.4.b Personal Factors 1) Previous injury Previous traumatic injury to the neck has been shown to impede circulation in the upper portion of the trapezius muscle in a group of subjects with persistent neck/shoulder pain that interfered with work, but for whom there were no clinical findings (Larsson et al., 1994). 2) Muscle activity Studies have shown differences in muscle activation patterns between novices or unskilled workers and skilled workers, with novices and unskilled workers typically exhibiting more activity than skilled workers. Development of localized muscle fatigue appears to differ as well. Kadefors, Petersén, and Herberts (1976) set out to document differences in fatigue development in inexperienced and expert welders working at shoulder level, 40 cm above shoulder level, and in overhead work locations; most of their data were collected in the overhead condition. They assessed fatigue, in terms of shift in spectral EMG and based on subjective response of subjects. Inexperienced welders exhibited objective signs of fatigue in trapezius, deltoid, and supraspinatus muscles, while experienced welders only tended to exhibit fatigue in the supraspinatus. Subjective fatigue developed earlier in the overhead task for inexperienced welders. Only a few of the experienced welders reported subjective fatigue, and that was toward the end of the task. Veiersted, Westgaard, and Andersen (1993) conducted a prospective study of a healthy cohort of new employees performing light manufacturing tasks, seeking to determine if muscle activity patterns would be predictive of development of trapezius myalgia. They found that initial static levels of activity differentiated between future cases and those who remained healthy. They also found that EMG gap patterns (gaps were periods of at least 0.2 seconds during which EMG was less than 0.5% MVC), which were fairly consistent over 50 weeks, were also different, with non-cases exhibiting more gaps per minute at each measurement interval (every 10 weeks) than those who became cases. In this group of new employees, there was a 25% conversion to case status in the group of employees exhibiting more than 10.8 gaps per minute, and an 82% conversion to case status in the group of employees exhibiting 10.8 gaps per minute or fewer. The "risk of becoming a patient decreased by 6% for each additional gap per minute." It is important to note that static muscle activity in all workers was below 2.5% of MVC, and most subjects were below 1.7% of MVC, which suggests that typical 2% to 5% MVC recommendations may not be protective. Hägg and Åström (1997) employed a cross-sectional design to examine differences in trapezius muscle activation patterns (gaps) in a group of medical secretaries (n = 23), dichotomized by prevalence of pain (7-day or 12-month) in the areas of the upper trapezius muscle. EMG data were recorded on the job, for 40 minutes for each worker, while workers performed typing and paper-handling activities. Gaps were defined as activity below 0.5% of MVC. Dichotomized by 12-month prevalence, the accumulated EMG gap time was significantly lower in the disorder group, compared to the non-disorder group, during typing. The authors suggested that the duration of individual gaps was also important, and that gaps in the range of 1 to 3 seconds in duration were the most relevant, in terms of differences between the subject groups. They also found lower pressure pain thresholds (PPT) in the 7-day prevalence group, compared with the nondisorder group. The authors also included information regarding individual differences, which indicated that one non-symptomatic subject exhibited no gaps, and at least one symptomatic subject interjected gaps whose cumulative duration exceeded the average total duration in the non-disorder group. From this the authors concluded that "an accumulated gap time during typing > 20% of total time is protective: however, lack of gaps does not unerringly imply disorders." The authors did not mention the potential for EMG pattern alteration simply due to measuring activities or the presence of the researchers. 3) Personal perception of discomfort Hansson et al. (1992) found that workers who were diagnosed with "tension neck" exhibited shorter endurance times for sustained arm abduction than did a group of workers who performed similar repetitive work, but were diagnosed as healthy. Changes in trapezius muscle amplitude and median frequency over the course of the test did not differ between the two groups of workers. Nakata, Hagner, and Jonsson (1993) discovered an inverse relationship between PPT and perceived musculoskeletal discomfort in the trapezius muscle when comparing two groups of subjects who performed light repetitive work for 2 hours. Subjects were divided into two groups based on the levels of perceived discomfort they reported during and after the task. Subjects in the group that reported the lowest discomfort also had higher pressure pain thresholds. 4) Sex Some studies have shown differences between men and women in prevalence of neck or neck/shoulder discomfort, while others have not. These contrasting findings occur even among studies where men and women have the same jobs. Bork et al. (1996) reported a higher prevalence in female physical therapists than in male physical therapists for both the neck (30% vs. 19%) and upper back (36% vs. 20%), with no difference for the shoulder. Skov, Bork, and Orhede (1996) studied physical and psychosocial risk factors for neck pain in a group of 1,307 Danish salespeople. They found a higher prevalence in women, but no association of neck pain with age. de Zwart et al. (1997a) stratified workers based on sex, age, and type of work and found a higher prevalence of neck complaints in women for each category of work (mental, mental physical, light physical, and heavy physical). From graphs they presented, prevalence of neck complaints in women appeared to be about twice that of men, within the five age categories and within job categories. Finsen, Christensen, and Bakke (1997), however, found no prevalence differences in a sample of Danish dentists, and Yu and Wong (1996) found no differences in bank employees who used VDTs, although men and women held different jobs. In a household survey, Lau, Sham, and Wong (1996) used interviewers to study the prevalence of neck pain in the general Hong Kong population. In the portion of the sample under age 60, 12-month neck pain prevalence in women was greater than men, while in the 60-or-older portion neck pain prevalence in women was less than that in men. Prevalence in men was similar across age groups. In women, prevalence was lower in the group aged 60 or older. 5) Age In their study of neck pain in farmers, Scutter, Turker, and Hall (1997) found that age had no effect. Ylipaa et al. (1997) did not find a difference in discomfort prevalence in the neck/shoulder/upper back in a sample of female dental hygienists, dichotomized by age (under or over 43 years). Finsen et al. (1997) found no prevalence differences in a sample of Danish dentists. In contrast to these negative findings, de Zwart et al. (1997a) stratified workers based on sex, age, and type of work, and found that neck complaints increased with age for men and women. On the other hand, Bork et al. (1996) reported that 1-year prevalence of neck and upper-back symptoms decreased beyond age 45 in a cross-sectional study of physical therapists. These authors, however, suggested their finding was likely to represent a survivor effect. 6) Personal habits Magnusson et al. (1996) studied the prevalence of neck pain in bus and truck drivers in the United States and Sweden, comparing them to sedentary controls. Personal factors (overweight, smoking, regular exercise, or marital status) were not associated with neck pain prevalence. 7) Physical make-up Based on findings of excessive postural sway following vibratory stimulation of the neck or lumbar proprioceptors in a group of forestry workers diagnosed with tension neck syndrome, compared with healthy referents, Koskimies et al. (1997) suggested that tension neck may be the result of inadequate proprioceptive and vestibular activation of the cervico-collic reflex (CCR). It should be noted, however, that postural sway assessment was determined after tension neck diagnosis. Norlander et al. (1997) found that consistent findings of inverse C7-T1 function was predictive of neck-shoulder pain in a group of laundry workers followed for 2 years. 8) Visual considerations Willford et al. (1996) found that multifocal lens wearers in the general population adopted a more forward head posture than non-multifocal lens wearers. They also noted a tendency (non-significant) for pain intensity to increase with increased forward head posture. Potential subjects were excluded if they had medical conditions that would have affected posture or caused pain. Shurts and Bommarito (1996) reported a reduction in neckache in a group of VDT users who changed from multifocal lenses to a prescription specifically designed for VDT work. B.5 Summary The objective of this paper was to review the evidence for work-relatedness of MSDs and symptoms in the neck and upper back. This paper surveyed previous comprehensive reviews of the epidemiological literature in the area, examined the epidemiological research published since those reviews were written, and examined the experimental research in the area, with particular emphasis on physical work factors. The review also included information from less rigorous work, including descriptive, observational, and case study reports. Conclusions that can be drawn based on this review include the following:
There is clearly a need for more intervention research to determine how effective laboratory solutions are in practice. There is also clearly a need for epidemiological research that relies on researchers' evaluations of exposures, and includes ways of assessing the temporal relationship between exposure and health outcomes for the neck and upper-back regions. B.6 References
C.1 Introduction Work-related lower-extremity MSDs are common. In a survey of nursing staff in nursing homes, 20% of all participants regularly suffered from leg complaints. Specifically, 9% of respondents reported hip and upper-leg complaints, 13% reported knee and lower-leg complaints, and 4% reported ankle and foot complaints (Engels et al., 1994). In a survey of garment workers, female machine operators reported prevalence of 22.2% for knee pain lasting at least 6 weeks, and 13.7% for knee swelling. When compared to age-, race-, and gender-matched controls these resulted in ORs of 1.84 and 9.98 for knee pain and knee swelling, respectively (95% CI: 1.34-2.52 and 6.01-16.5, respectively). A survey of electricians found that 33% reported knee pain, aching, stiffness, burning, numbness, or tingling that occurred more than three times or lasted longer than 1 week; 16% reported the same symptoms occurring once per month with no traumatic injury (Hunting et al., 1994). When construction workers were compared to white collar workers, ORs of 1.907 (95% CI: 1.413-2.573) were found for foot and leg complaints. In the same study, blue collar workers had ORs of 1.417 (95% CI: 1.207-1.662) for foot and leg complaints as compared to white collar workers (Petersen and Zwerling, 1998). C.2 Osteoarthritis Osteoarthritis is one of the most common joint disorders, affecting close to 16 million persons in the United States and producing significant morbidity, typically in the elderly (Felson, 1988; Imeokparia et al., 1994). In a study of men between the ages of 51 and 61 years, 24.1% of white collar workers reported arthritis in some part of their body. However, 36% of construction workers and 32% of blue collar workers reported complaints of arthritis, suggesting that physical strain on the body may be associated with arthritis (Petersen and Zwerling, 1998). Osteoarthritis is a disease affecting a joint's cartilage and underlying bone. It results from a number of different pathological processes. The first observable pathological change due to the disease is loss of cartilage. Changes in the skeletal structure include eburnation or increased sclerosis of underlying bones, osteophyte formation, and occasionally subchondral bone cysts (Felson, 1988; Silman and Newman, 1996). There can be few symptoms, but suffering, decreased quality of life, and a change in work habits and patterns are usually observed (Vingärd, 1996). Work by Vingärd et al. (1991) hypothesizes why dynamic and static physical loads may contribute to the osteoarthritis process. Dynamic forces due to short-lived stresses on the joints may result in microtraumas of the cartilage and of the subchondral bone when the capacity to absorb shock by these structures is exceeded. Static compression may hamper the movement of the joint, resulting in a decrease in the elimination of metabolites and the intake of nourishment, since cartilage has neither blood vessels nor lymphatic drainage. In this paper, the term "osteoarthritis" is used and includes both osteoarthritis and osteoarthrosis. Osteoarthritis of the Knee. Work-related and personal factors have both been associated with osteoarthritis (see Tables D-1 and D-2). Repetitive work activities have been linked with an increase in osteoarthritis of the knee (Cooper et al., 1994). In a population-based case-control study, men and women were studied who responded positively to a mailed questionnaire asking if they had pain in or around the knee. Of those with knee pain, 109 subjects were diagnosed with osteoarthritis of the knee; 218 age- and gender-matched controls with knee pain but without osteoarthritis were also studied. An OR of 6.9 (95% CI: 1.8-26.4) was found relating knee osteoarthritis with more than 30 minutes of squatting at the subjects' main job. Likewise, ORs of 3.4 and 2.7 (95% CI: 1.3-9.1 and 1.2-6.1) were found for kneeling more than 30 minutes per day and climbing more than 10 flights of stairs, respectively. When the data for male and female subjects were examined separately, the ORs were similar to the combined group. The lifting of heavy loads was not found to have an increased association by itself with knee osteoarthritis. However, when the lifting of loads greater than 25 kg was combined with kneeling, squatting, or climbing stairs, the OR rose to 5.4 (95% CI: 1.4-21.0) (Cooper et al., 1994). The First National Health and Nutrition Survey (HANES I) provided information on osteoarthritis of the knee and suspected risk factors. HANES I was a multi-stage, stratified- probability, cross-sectional study. To study osteoarthritis of the knee, 5,193 adults were examined, x-rayed, and questioned. Only African-American and Caucasian subjects were included in the sample. Personal and work-related factors were studied including race, income, educational level, marital status, uric acid level, diabetes, early menopause and parity for women, BMI, skinfold thickness, smoking, number of cigarettes per day, occupation, strength demand of job, and knee-bending demand of job (see Tables D-1 and D-2). Occupations were categorized according to the U.S. Bureau of the Census three-digit occupation codes. For both men and women, laborers and service workers had the highest percentage of diagnosed osteoarthritis of the knee. Further analysis using stratification by age revealed that strength demands of the job were not associated with knee osteoarthritis in men but were associated with women over age 55 (OR: 3.13; 95% CI: 1.04-9.39). The knee-bending demand of the job was associated with knee osteoarthritis for both men and women over 55 years of age (OR: 2.45; 95% CI: 1.21-4.97 for males; OR: 3.49; 95% CI: 1.22-10.52 for females). The authors note that the increased risk of osteoarthritis due to job demands appearing in only older workers suggests that osteoarthritis may develop only after a long period of repetitive occupational exposure. It is also noted that knee-bending may be a more accurate indicator of knee stress than strength demands of the job (Anderson and Felson, 1988). The Clearwater Osteoarthritis Study (COS) looked at an ongoing prospective cohort; information from COS was used in a case-control study of subjects 40 years and older (Imeokparia et al., 1994). This study looked at past physical activity in the workplace, at home, and during sports and exercise and determined an overall physical activity rating using energy requirements as calculated in metabolic equivalents (METS). No significant results were found with the male subjects. In the female subjects, the association of physical activity and knee osteoarthritis was found to have an adjusted OR of 1.66 overall (CI: 1.02-2.72) and higher ORs when the data for the female subjects controlled for age, education, BMI, diabetes, and marital status. The authors note that caution should be used when interpreting these results since duration of physical activity was not considered. In a registry-based cohort study (Vingärd et al., 1991), men and women were classified by high exposure and low exposure to dynamic and static forces on the knee. The subjects were tracked to determine if they sought hospitalization for osteoarthrosis of the knee during a 3-year period. Occupations with an RR over 1.0 for men were firefighters, farmers, and construction workers (RR: 2.93, with 95% CI: 1.32-5.46 for firefighters; RR: 1.46, with 95% CI: 1.23-1.98 for farmers; RR: 1.36, with 95% CI: 1.13-1.79 for construction workers). For women, only the occupation titled "cleaners" had an RR greater than 1.0 (RR: 2.18, with 95% CI: 1.26-3.00). Osteoarthritis of the Hip. Hip osteoarthritis is a common cause of pain and disability, especially in the elderly (Coggon et al., 1998). As in knee osteoarthritis, risk factors comprise both personal and work-related conditions (see Tables D-3 and D-4). In a case-control study (Coggon et al., 1998), 611 patients requiring hip replacement surgery for osteoarthritis were age- and gender-matched to controls. Only persons aged 55 or older were used in the study. The subjects were questioned as to the average weight they lifted more than 10 times in a normal working week. Following adjustments for confounders, men who lifted more than 25 kg for 10 years before age 30 and for more than 20 years over the course of their working life were found to have a higher association with hip osteoarthritis (OR: 2.7; 95% CI: 1.4-5.1 for the first group; OR: 2.3; 95% CI: 1.3-4.4 for the second group). Interestingly, this association was not found in the female subjects. There were indications of increased risk of osteoarthritis with frequent stair-climbing for both genders and with large amounts of walking on the job (more than 2 miles per day) for females. A study in Sweden also defined cases as those requiring total hip replacements (Vingärd, Alfredsson, and Malchau, 1997). In this study, only female subjects (n=230) between 50 and 70 years old were used. Controls were women without hip problems (n=273). Participants were questioned about their activities, both at work and at home, and grouped into high exposure (the highest 25% of the aggregate loads experienced before age 50), medium exposure (the middle 50% of the aggregate loads) and low exposure (the lowest 25% of the aggregate loads). High exposure to jumping or moving from one level to another was found to have an OR of 2.1 (95% CI: 1.1-4.2). High exposure to stair climbing by the subjects also resulted in an OR of 2.1 (95% CI: 1.2-3.6). When physical activities at work were combined with physical activities at home, an interesting pattern emerged. Medium physical work loads combined with high physical loads from sporting activities resulted in an OR of 2.7 (95% CI: 1.2-5.9). High physical work loads combined with medium and high physical loads from sporting activities resulted in ORs of 2.7 and 4.3, respectively (95% CI: 1.1-7.0 for medium sports loads and 95% CI: 1.7-11.0 for high sports loads). The authors suggest that the results imply that there is a dose-response relation between life-long physical load and osteoarthritis of the hip.
* Other studies did not find this effect or reported a protective effect (Silman and Newman, 1996).
*1 Croft et al. (1992b) found no relationship between osteoarthritis of the hip and BMI. Anderson and Felson (1998) suggested that there were "conflicting reports" concerning hip osteoarthritis and obesity. When hip osteoarthritis cases in men were diagnosed by minimal joint space less than or equal to 1.5 mm and compared to male controls with minimal joint spaces greater than or equal to 3.5 mm, standing and material handling were determined to have positive associations with hip osteoarthritis (Croft et al., 1992b). Standing for more than 4 hours per day for over 40 years had an OR of 2.7 (95% CI: 1.0-7.3). Material handling (lifting or moving weights) involving more than 25 kg with exposure greater or equal to 20 years had an OR of 2.5 (95% CI: 1-15.7). The registry-based cohort study reported for knee osteoarthritis also considered hip osteoarthritis (Vingärd et al., 1991). For male subjects, the occupations exhibiting significant associations with hip osteoarthrosis were farmers; firefighters; mill workers, butchers, and meat preparers; and construction workers (RR: 3.78 with 95% CI: 2.91-3.88 for farmers; RR: 2.52 with 95% CI: 1.38-4.64 for firefighters; RR: 2.17 with 95% CI: 1.29-3.65 for mill workers, meat preparers, and butchers; RR: 1.66 with 95% CI: 1.32-1.87 for construction workers). In the female subjects, only mail carriers had an association with hip osteoarthrosis (RR: 3.83; 95% CI: 1.19-12.05). C.3 Venous Pooling Discomfort and edema of the lower extremities has been attributed to jobs with constrained standing work (Ryan, 1989). To minimize or alleviate the problems of discomfort and edema of the lower legs and feet, anti-fatigue mats and shoes with shock absorbing qualities have been developed (Hansen, Winkel, and Jorgensen, 1998). In a study (Hansen et al., 1998), female students were exposed to 2 hours of standing work and 2 hours of standing/walking work. Standing work significantly increased the discomfort in the legs and the feet; standing work combined with walking work also significantly increased the discomfort in the legs and feet. Interestingly, no effect due to shoe or mat softness was found to significantly alleviate the discomfort. However, edema was significantly decreased in both work situations with the use of soft shoes. The soft mat actually increased the edema in standing work. This effect may be attributed to the lack of a sufficient venous pump due to the lack of natural swaying movements: workers on hard surfaces tend to sway and move more than workers on soft surfaces. The authors note that although soft shoes tend to alleviate the edema, the greatest factor in discomfort and swelling is the prolonged time standing in a constrained posture. With that in mind, they recommend organizing the workday to include time limits for work in an upright position and changes between standing, walking, and seated work. C.4 Vibration Syndrome Vibration syndrome has been reported in the foot due to direct vibration exposure to the foot (Hedlund, 1989). However, increasing evidence suggests that workers exposed to hand-arm vibration, particularly those with vibration-induced white finger (VWF), may have circulatory disturbances in the foot due to the sympathetic nervous reflex (Sakakibara et al., 1991). In a study examining 11 patients with VWF, 12 patients who used vibratory tools but showed no signs of VWF, and 20 referents without vibration exposure, the authors found that the VWF patients had more complaints of coldness in the lower extremities and had lower skin temperatures in their toes as well as their fingers (Sakakibara et al., 1991). C.5 Tarsal Tunnel Syndrome Tarsal tunnel syndrome results from compression of the posterior tibial nerve when it passes through the fibro-osseus tunnel beneath the flexor retinaculum located on the medial side of the ankle (Forst and Hryhorczuk, 1988). The risk factors for the syndrome are largely unknown, but there is evidence to suggest that repetitive motions of the foot, as seen in foot pedal operation, may be related (Forst and Hryhorczuk, 1988). C.6 References
Footnote (1) Prevalence based on symptoms alone tends to be about twice the prevalence from symptoms combined with PE findings (Hales and Bernard, 1996). (Back to Text)
Footnote (2) In epidemiology, conclusions are made on the basis of causal inference, which depends on prior knowledge, intuition, insight, and uncertainty. As such, epidemiologists often refer to "risk associated with," rather than "cause of" an adverse health outcome because it is not possible to establish a basic link between cause and effect via inductive inference. (Back to Text) |
