Robert M. Szabo
The report of the Panel on Musculoskeletal Disorders and the Workplace of the National Research Council (NRC) and the Institute of Medicine (IOM) has used significant interpretations of the scientific literature that I consider inaccurate and misrepresentations, particularly with issues in the upper extremity. It also does not reflect the scientific evidence regarding the usefulness or urgency of interventions. I am also troubled by some of the methodological issues employed. For instance, in the epidemiology chapter, the panel set selection criteria for accepting which articles to review. Despite voicing my concerns at our panel meetings, numerous epidemiological articles that did not meet the selection criteria were used to support the biomechanics chapter. The panel agreed and we wrote, “Few high-quality intervention studies related to the primary and secondary prevention of low back pain are available in the literature.” No high-quality intervention studies related to the primary and secondary prevention of upper extremity disorders in general and carpal tunnel syndrome in particular are available in the literature. To circumvent the issue of not having reasonable scientific intervention studies available, the panel took a “best practices” approach, which I think is not very scientific. Prospective studies are of the greatest value; however, none is available at the present time. What needs to be emphasized is that no study has demonstrated that any intervention affects the short- or long-term outcome of developing an impairment, a disease, or a disorder with a positive physical examination correlate. In other words, symptoms of pain may have been reduced. For example, one study (Tittranonda et al., 1999)67 cited in the report on more than one occasion is particularly misleading. The authors of this small study explored the use of an alternative keyboard design on hand pain in patients already diagnosed with carpal tunnel syn-
drome and measured symptoms of pain. The participants abandoned the use of these alternative keyboards despite a “positive” effect. One might be led to believe by the NRC-IOM report that this study prevented carpal tunnel syndrome or provided evidence of a worthwhile intervention. I don't think so. Another study was misrepresented as demonstrating an association between computer keyboard use and slowed median nerve velocity. Not mentioned in discussing that study was that there were no abnormal median nerve velocities and there were no clinical carpal tunnel syndromes. Carpal tunnel syndrome is wrongly presented as being associated with computer keyboard use. Despite the panel's recognition of the lack of scientific evidence to link carpal tunnel syndrome to keyboard use, the report does not acknowledge this fact.
In order to determine the association between risk factors (exposures) and a disease (outcome variable), both the risk factors and the disease should be well defined. When a fall on the outstretched hand results in a fracture at the wrist, the relationship between the trauma and the injury is clear. This is often not the case for work-related musculoskeletal disorders. Some conditions, like carpal tunnel syndrome, have a pathogenesis that can be defined and measured objectively with electrodiagnostic studies. The majority of work-related musculoskeletal disorders of the upper extremity fall into a more amorphous category, such as hand pain when there is no objective way to define the condition or measure its severity and there is no clear anatomical basis for the symptoms.
Carpal tunnel's clinical picture of pain and paresthesias on the palmar-radial aspect of the hand, often worse at night, is readily recognized. Carpal tunnel syndrome is a condition of middle-aged people and is more common in females. In the first population-based study, the mean age at diagnosis was 50 years for men and 51 years for women; women accounted for 78.5 percent of the cases.62 Most middle-aged people work, so more often than not, carpal tunnel syndrome occurs in working people. The role that work-related activities play in its pathogenesis is controversial. On the basis of six cases, no controls, and a definition of occupation that included housewives, Brain in 1947 was the first to implicate occupation as a causal factor in the disorder.11 A high prevalence of work-related musculoskeletal disorders, including carpal tunnel syndrome, has been reported in professions requiring high-force wrist motions, such as assembly line workers, meatpackers, and material handlers. Much of the recent focus, however, has been on keyboard operators, whose activities, while extremely repetitive, do not require high force. It is not universally accepted that job-related factors are important determinants for predicting the appearance of carpal tunnel syndrome.25,68 In the general population, its prevalence is the same whether people perform repetitive activities or not.6 A recent study reported that its prevalence for repetitive
hand or wrist motion was 2.4 percent compared with 2.7 percent for nonrepetitive motion (95 percent CI for the difference, −2 percent, −1.5 percent; P = .69).6 Reports from workers and survey data tend to overestimate the prevalence of a disorder because social, cultural, and medicolegal factors have a major influence. In 1988 the National Health Interview Survey showed that 1.4 percent (1.87 million) of working adults in the United States reported that they had a “condition affecting the wrist and hand called carpal tunnel syndrome.” Only 675,000 indicated that a health care provider had made this diagnosis.66 Carpal tunnel syndrome must be distinguished from the vast array of upper extremity musculoskeletal complaints collectively called repetitive motion disorders. Many of these conditions are the product of somatization, the reporting of somatic symptoms that have no pathophysiological explanation, amplified by medicalization, whereby uncomfortable bodily states and isolated symptoms are reclassified as diseases for which medical treatment is sought.8,9
The association of carpal tunnel syndrome with work-related risk factors is a recurring theme of causation among workers, ergonomists, lawyers, and physicians. The majority of the literature that tries to establish this as a causal association fails to meet the appropriate standards of epidemiological validity.63,68 To conclude that carpal tunnel syndrome is a repetitive motion disorder, one must ask the question, “How significant a risk factor is repetition for the development of carpal tunnel syndrome?” To answer this question, one must consider the interaction of job exposures (extrinsic risk factors) with various innate anatomic, physiological, or behavioral characteristics of the worker (intrinsic risk factors) that render him or her more likely to develop the disorder. Occupational risk factors alone do not explain its occurrence; rather, it is the culmination of many distinct converging causal links. The majority of cases are likely due to intrinsic risk factors. One investigation concluding that carpal tunnel syndrome is closely correlated with health habits and life-style49 is supported by an analysis showing that 81.52 percent of the explainable variation in electrophysiologically defined carpal tunnel syndrome was due to body mass index, age, and wrist depth/width ratio, whereas only 8.29 percent was due to job-related factors.24 There may be important interactions between extrinsic and intrinsic risk factors that are yet to be understood. While there is a biologically plausible mechanism to relate forceful grip to compression of the median nerve,64,65 there is no such correlate to postulate biological plausibility with regard to repetition.
In reviewing the published literature on work-related repetitive hand injuries, Hagberg et al. estimated the attributable fraction by (OR − 1)/OR (where OR was the estimated odds ratio) and concluded that exposure to physical workload factors, such as repetitive and forceful gripping, is probably a major risk factor for at least 50 percent, and as much as 90
percent, of all of the carpal tunnel syndrome cases in several types of worker populations.27 Greenland has pointed out that attributable fractions are dependent on the prevalence of cofactors of exposure (factors that enhance exposure effects on risk and causal mechanisms that do not involve exposure).26 Because the cofactors are inadequately accounted for in the investigations considered, the estimations offered by Hagberg, as well as the ones used in the NRC-IOM report, must be interpreted with caution.
Many factors have been identified in the pathogenesis of carpal tunnel syndrome but are rarely considered as possible confounders or effect modifiers in epidemiological studies of occupationally related disorders. Nathan and colleagues studied the relationship of tobacco, caffeine, and alcohol to the prevalence of carpal tunnel syndrome in 1,464 workers confirmed by nerve conduction examinations.46 They found when comparing workers with carpal tunnel syndrome to those without it that there was a 19 percent greater lifetime use of tobacco, a 75 percent greater history of alcohol abuse, and a 5 percent greater use of caffeine. In female workers, current tobacco, caffeine, and alcohol consumption independently predicted 5 percent of the explainable risk for definite carpal tunnel syndrome.
In a case-control study of 600 patients presenting for an independent medical exam, Stallings and colleagues found that obesity was associated with positive findings on nerve conduction exams for median neuropathy with an odds ratio of 3.92 (95 percent confidence interval = 2.65 to 5.79).59 In a study analyzing the computer records of all personnel on active naval duty in all Navy medical facilities from 1980 to 1988, Garland and colleagues found first hospitalization rates (those patients undergoing surgery) for carpal tunnel syndrome were strongly related to age, sex, and race.22 The occupations with high risk were boatswain's mates and enginemen, an occupation similar in description to civilian industrial painters and maintenance workers. The standardized incidence ratio for male boatswain's mates was 1.7, whereas for women in the same occupation it was 9.8. Female ocean systems technicians had a standardized incidence ratio of 5.6, whereas men in the same occupation had a standardized incidence ratio below 1.0. Gender was therefore a far more predictive risk factor for carpal tunnel syndrome than job exposure. The higher incidence rates for women could not be accounted for by the differential selection of high-risk occupations by women. Low-risk jobs included occupations analogous to secretarial work, including clerical duties and data entry using typewriters and computers.22
Silverstein and colleagues, in what has been regarded as a classic study, determined the prevalence of carpal tunnel syndrome among 652 active workers (in 39 jobs from 7 different industrial sites). Specific hand
force and repetitiveness characteristics were then estimated for different jobs. An equal number of men and women was identified. The prevalence of carpal tunnel syndrome ranged from 0.6 percent among workers in low-force, low-repetition jobs to 5.6 percent among workers in high-force, high-repetition jobs. Silverstein's data shows work to be a risk factor only when both high force and high repetition are present, but the precision of her estimated odds ratio of 15.5 (95 percent confidence interval = 1.7 − 141.5) suffers from an extremely small sample size. High repetitiveness appeared to be a greater risk factor than high force, but neither was statistically significant alone.58 Of particular concern, however, is that there were only 14 cases of carpal tunnel syndrome identified, which were associated with 11 jobs. Stetson et al. carried out a study of 240 industrial workers from the automotive industry. Electrodiagnostic studies were done on all subjects to assess median nerve conduction. Hand-intensive tasks correlated with decreased nerve conduction, but there were no significant differences in the measurement means for repetitiveness and pinch between the nonexposed and exposed groups.61
DeKrom et al. analyzed a population of primarily hospital patients to estimate the etiological relationship between workload and carpal tunnel syndrome.16 The diagnosis was based on clinical history as well as neurophysiological testing. Activities with the flexed wrist, such as grasping or the packing of products, resulted in an increased risk positively correlated with the duration of these activities, such that the odds ratio was as high as 8.7 for maximum exposure. Similarly, activities with the wrist extended, such as scrubbing or ironing, were associated with an increased risk of carpal tunnel syndrome, which was correlated with the duration of exposure with the odds ratio as high as 5.4. Obesity was also found to be a risk factor. Exposures were estimated, but the methodology was not discussed regarding the classification criteria.
The scientific evidence fails to be definitive in connecting occupational hand use and proven carpal tunnel syndrome. In 1991, Stock identified 54 potentially relevant studies on upper extremity cumulative trauma disorders and was able to retrieve 49. Of these 49, only 3 met her inclusion criteria emphasizing adequate definitions of populations, exposures, and outcomes. All three studies were cross-sectional in design. She concluded that when the results of these studies were compared, they provided strong evidence of a causal relationship between repetitive, forceful work and the development of musculoskeletal disorders of the tendons and tendon sheaths in the hands and wrists and nerve entrapment of the median nerve at the carpal tunnel.63 Only two of the studies, however, specifically examined carpal tunnel syndrome; one was the Silverstein study mentioned above. The authors of the other study in her review had opposite findings based on objective electrodiagnostic crite-
ria.50 Several concerns limit the use of existing data to resolve the relationship between job-related risk factors and carpal tunnel syndrome.
Carpal tunnel syndrome is a combination of signs and symptoms, and no single test absolutely confirms its diagnosis. The validity and reliability of many of the diagnostic tests used are not fully established. Since electrodiagnosis is considered the “gold standard,” all predictors of carpal tunnel syndrome are judged against it.36 Carpal tunnel syndrome verified by strict electrodiagnostic criteria is not common among workers. At first it would seem that the critical issue is whether only studies that use electrodiagnosis should be considered in attempts at resolving the role of job-related risk factors. Comparison of several different median nerve conduction study techniques, however, demonstrates a variety of sensitivities.35 The problem is that no single technique can identify all carpal tunnel syndrome patients without misclassifying an unacceptable number of normals. One must consider the specific techniques that investigators employ in their study. Using three separate electrodiagnostic criteria examining a normal population, one study showed that 23 of 50 (46 percent) had at least one false-positive test for carpal tunnel syndrome.53 The disorder is progressive and can present with symptoms before physical findings are present. Some of these symptoms, however, are non-diagnostic of carpal tunnel syndrome and can be present in other disorders. Therefore, studies that measure only symptoms may overestimate the problem. Besides nerve conduction studies, the diagnosis of carpal tunnel syndrome is based on history and physical examination. The history and the physical examination, including a variety of sensory tests, rely on a person's verbal reports of sensation or pain. In a motivated and honest patient, a combination of these tests may suffice to demonstrate the nerve deficit. However, if issues of secondary gain cloud the case, the patient's reports may be unreliable. The patient may have a very low threshold for perceiving discomfort, and suggestible individuals that are warned about the disorder from coworkers or trade publications may subconsciously amplify minor discomforts until they are perceived as significant symptoms. Electrophysiological studies measuring median nerve function are the only objective way to demonstrate the nerve deficit. When properly used, these tests have sensitivity and specificity near 90 percent.35,41 Unfortunately, the majority of epidemiological studies of carpal tunnel syndrome have not used these studies in their diagnoses, and it is therefore uncertain whether all reported cases in these studies were indeed carpal tunnel syndrome. The accuracy one requires in the diagnosis depends on its purpose. In surveillance of incidence of the disorder in an industry for monitoring purposes, case definition should maximize sensitivity at the expense of specificity.37 For epidemiological studies concerned with causation, specificity of case diagnosis should also take
precedence over sensitivity for the sake of validity.12 Of the nonelectrodiagnostic tests for carpal tunnel syndrome, sensory testing with Semmes Weinstein monofilaments or the Durkan pressure test seems to be best for this role, but neither has been used in epidemiological studies,17,23,42 nor has a hand symptom diagram, which may be an acceptable compromise.39 Accepting less specific diagnostic criteria when trying to establish causal relationships has led to many of the controversies surrounding work-related carpal tunnel syndrome.
Definition of the disorder based on clinical tests is liable to misclassification, because no clinical test has been proven valid. The National Institute for Occupational Safety and Health (NIOSH) proposed a surveillance case definition for work-related carpal tunnel syndrome: the presence of median nerve symptoms; one or more occupational risk factors; and objective evidence by physical examination findings, including the Tinel or Phalen signs or decreased pinprick sensation, or positive diagnostic nerve conduction studies. Using the NIOSH surveillance case definition, if a worker develops classical symptoms and has abnormal electrodiagnostic studies of the median nerve in the face of a work task that is highly forceful and repetitive, then the presumption is that the symptom complex is work related.43 The work by Silverstein and her colleagues58 formed the basis for this case definition; however, the diagnosis in their study was based on reported symptoms not confirmed with electrodiagnosis. When applying this case definition without using electrodiagnostic studies to a sample of symptomatic workers, Katz et al. found that 38 percent of subjects were misclassified; 50 percent of workers satisfying the case definition did not have carpal tunnel syndrome, while 25 percent not satisfying the case definition did have it.37 Misclassification if nondifferential would dilute any association if present; however, misclassification is likely to be differential in these studies due to detection bias: workers in at-risk jobs are more likely to get medical attention, be identified as subjects in a study, and be overdiagnosed because of workers' compensation and the current climate of medicolegal issues. Using Katz's data, this misclassification error could overestimate any association by 62 percent if differential.37 In a population of workers with a 10 percent prevalence of carpal tunnel syndrome, when the NIOSH case definition is applied to the entire population, 85 percent of the workers who meet the criteria for the NIOSH definition of the disorder will not have it.37 As pointed out by Wolens, Silverstein's study suggests only that individuals exposed to high rates of force and repetition more frequently meet the NIOSH definition of work-related carpal tunnel syndrome.72 Of remarkable interest concerning this study is the finding that so few workers even meet this case definition. Considering the poor positive predictive value of this definition for diagnosis of the disorder, there are even
far fewer industrial workers with true carpal tunnel syndrome. Since true carpal tunnel syndrome is an uncommon work-related disease, it is universally unacceptable to attribute a case of carpal tunnel syndrome to a person's work-related activities without evaluating all other avocational stresses and life-style risk factors. From my own experience, carpal tunnel syndrome is a frequent disorder in patients who are not employed.
“The definitive cause (or even strong association) of work-related musculoskeletal disease has yet to be established.”29 To infer that a statistical association between job-related exposure factors and carpal tunnel syndrome is evidence of etiology, it should be demonstrated that the job exposure occurred before the disorder emerged. The temporal relationship between physical load factors and the onset of carpal tunnel syndrome has not been demonstrated in either cross-sectional or case-control studies examining prevalence. Armstrong and Chaffin noted that carpal tunnel syndrome subjects used a wrist position that deviated from the straight position more frequently and exerted greater hand forces in all wrist positions than nondiseased subjects; however, they could not establish whether the differences in work methods was a cause or an effect of the disorder.5 The majority of studies to date are prevalence studies in which exposures were measured at the same time that disease status was established. Only Nathan's study followed up a cohort of original participants (five years later); this study found no association between job and electrophysiological evidence of carpal tunnel syndrome.48 Other strong evidence of a causal relationship could be made if workplace modifications were shown to reduce the incidence of carpal tunnel syndrome; no study has demonstrated this.
Many epidemiological studies apply a single general class descriptor to different worker groups and then measure prevalence of carpal tunnel syndrome across groups. More precise measurement of exposure variables, such as position (of wrists, fingers, elbows, and neck), static loading, temperature, and vibration are necessary. These can be obtained with careful biomechanical characterizations of specific job tasks. Precise ergonomic measures, however, are of little value if ultimately subjects are grouped into broad categories based on what can be observed without such measurements.47 Harber et al. attempted to determine individual rather than average group exposures in a cross-sectional study of 50 supermarket checkers. With this approach, they failed to detect any association between wrist flexion/extension and symptoms. Instead they demonstrated an association between wrist pronation and symptoms and lumbar flexion (posture) and symptoms.29 Some studies perform a somewhat detailed biomechanical evaluation of the job classifications and then generalize the findings to all workers in that job category.5,14 Classification of exposure by industry is imprecise and may lead to erroneous
associations between exposure and outcome. Random exposure misclassification based on the subject's self-reports of occupational hand use could be responsible for biasing any association between hand use and electrophysiological evidence of median neuropathy at the wrist toward the null in the studies of Nathan et al. and Schottland et al.48,49 and 50,57 However, prevalence of exposure may generate a spurious trend associated with a third variable, such as age,19 as found in all of the studies of Nathan et al. and in the study of Schottland et al. In these studies, age may be increasing as the prevalence of exposure is increasing. If the true relative risk associated with exposure is the same in each age group, and the electrophysiological test has the same misclassification probabilities in each age group, then the apparent relative risk of carpal tunnel syndrome associated with the exposure will change systematically as age increases, producing a spurious trend with age.19 Furthermore, variation in prevalence exposure could mask a real trend in relative risk associated with some unidentified third variable. Weislander et al. found an exposure response trend for self-reported years of exposure to vibrating hand tool use and repetitive wrist motion.71 Since referents were less likely to be sensitive to occupational stresses, they would be less likely to recall and report symptoms of carpal tunnel syndrome. The evaluation of job exposure status was also not blinded as to the subjects' health status. Both of these factors could lead to information bias that is differential and that would overestimate the difference between the case and control exposures.
Generally there has also been a failure to measure background exposure. A full-time worker spends approximately 20 percent of the hours in a year doing his or her job; exposures to risk factors during the remaining 80 percent of the time have not been measured. Hales and Bernard concluded that 16 of 22 studies reported a positive association between occupational factors (repetition being only one) and carpal tunnel syndrome, but not a single one of these studies had quantified both exposure and disease ascertainment.28
Confounding can result in incorrectly attributing the etiology of carpal tunnel syndrome to the wrong risk factor. Obesity is identified in several studies as a significant risk factor,15,16,49,70,71 yet it is not controlled for in the majority of the studies as a potential confounder. With the exception of gender and age, very few other associations were even considered in the design or analysis phase of published studies. Cannon et al. performed a case-control study of the personal and environmental factors associated with carpal tunnel syndrome in workers at an aircraft engine manufacturing company.13 A total of 30 patients with carpal tunnel syndrome (3 men and 27 women) and 90 matched controls (9 men and 81 women) were compared. Information on age, sex, race, weight, occupa-
tion, number of years employed, worker's compensation status, history of diabetes mellitus, presence of hypertension, history of arthritis or hyperthyroidism, use of oral contraceptives or postmenopausal estrogens, and history of gynecological surgery was collected. The use of vibratory hand tools and a history of gynecological surgery, specifically hysterectomy and oophorectomy, were most strongly associated with the onset of carpal tunnel syndrome. Obesity, nonwhite race, and employment in jobs requiring repetitive motion tasks involving the wrist were more prevalent among subjects than controls but did not achieve statistical significance. The gender difference in this study highlights the problem of differences in the types of jobs assigned to women; they may have been assigned a disproportionate amount of the most highly repetitive work, and therefore the association between repetition and carpal tunnel syndrome may even be weaker than reported.
To test the hypothesis that carpal tunnel syndrome is associated with occupational risk factors, a study incorporating electrophysiological tests, physical examinations, and questionnaires was performed at a ski assembly plant where jobs were classified as repetitive and nonrepetitive.7 Repetitive jobs were defined as activities that required repetitive or sustained flexion, extension, or ulnar deviation of the wrist or use of a pinch-type grip. The conclusion drawn, based on a crude prevalence ratio of 4.92 (95 percent confidence interval = 1.17 − 20.7), was that carpal tunnel syndrome was associated with jobs requiring frequent and sustained hand work. This study, however, had several limitations. Diabetes mellitus, a known intrinsic risk factor, was present more commonly among those diagnosed with carpal tunnel syndrome and was not controlled for in the analysis. As people with diabetes mellitus are known to be more susceptible to nerve compressive lesions, that condition should be treated as a potential confounder or effect modifier and its association with exposure analyzed.
Nathan et al. conducted a longitudinal cross-sectional study of the cause of carpal tunnel syndrome in industry, by evaluating sensory conduction of the median nerve in relation to age, gender, hand dominance, occupational hand use, and clinical diagnosis.48 In this study, the only one that addresses the issue of temporality, investigators reexamined 630 hands of 316 (67 percent) of the same workers from a five-year previous study group, which consisted of 942 hands of 471 industrial workers. The palmar segmental stimulation technique was employed, and slowing was defined as a maximum latency difference (MLD) of 0.4 msec or more after adjustment for temperature variation. No significant change in the prevalence of slowing between 1984 and 1989 (23 percent in 1984, 22 percent in 1989) was found, and slowing strongly correlated with increased age even in apparently healthy, symptom-free subjects, but not with gender. Al-
though slowing continued to be more prevalent in the dominant hand, there was no increase in the difference in prevalence of slowing between the dominant and nondominant hands between 1984 and 1989, suggesting that the role of industrial hand use in median nerve slowing is minimal. Slowing was no longer correlated in any fashion with occupational hand use. In particular, no correlation existed between rate of repetition and prevalence of slowing. The prevalence of probable carpal tunnel syndrome (symptoms and physical exam consistent with the disorder) was still strongly correlated with the degree of slowing. Nathan concluded that age and hand dominance were more important than any job-related factor in the prediction of slowing after five years. In a case-control study, Schottland et al. replicated Nathan's results and found that exposure to the repetitive tasks in a poultry plant was not associated with electrophysiological changes in median nerve function. A power analysis of their data suggests that if an association was present, it would have been observed with their sample population.57
Social determinants of carpal tunnel syndrome may confound the interpretation of many studies. Few studies address social determinants involved in predicting the disorder; only one specifically substantiated cultural differences as a possible predictive risk factor for the reporting of carpal tunnel symptoms.51 Since so much of the diagnosis of carpal tunnel syndrome relies on subjective data, it is a critical omission of most studies not to have considered psychosocial, legal, and cultural factors as confounders of exposure effects. “Teaching 25 years ago regarded carpal tunnel syndrome as a problem that would resolve following surgery when it was indicated.”43 When patients with carpal tunnel syndrome appear to have some relationship between their symptoms and their work, the problem is different. In one study of workers' compensation patients, individuals with less abnormal nerve conduction velocities were more likely to have persistent symptoms and more often changed jobs based on those symptoms than those with more prominent nerve conduction abnormalities.31 Higgs found that those workers undergoing carpal tunnel release in his study who had legal representation were twice as likely to have poor outcomes. He concluded that “for the average employee, the enticement of substantial financial gain, coupled with the legal premise that a more-disabled person may receive even greater rewards, provides ample incentive to prolong recovery.”30 There is no doubt that psychosocial issues prolong disability.10 The population of workers' compensation recipients experience worse outcomes following medical, surgical, and rehabilitative interventions,30,31 yet when Katz and colleagues studied the perceived improvement in quality of life and the perceived improvement in symptoms in a group of workers' compensation recipients and non-recipients after surgery for carpal tunnel syndrome, they found no differ-
ence.38 Subjects in this study, however, understood they were participating in a research project that would have no impact on their compensation benefits. Adams et al., in a retrospective cohort study on outcome of carpal tunnel surgery in workers' compensation patients, found no association between the biological severity of the disorder and the postoperative duration of disability.2 And 67 percent of the cases returned to the same job. The authors concluded that disability following carpal tunnel syndrome surgery may be related to other medical, psychosocial, administrative, legal, or work-related factors. Franklin similarly concluded that a number of factors need to be considered when interpreting outcome studies in workers' compensation patients: factors that predict good or poor outcome may not be the same as in the general population; comparable procedures have worse outcomes in the workers' compensation patients; outcomes after surgery are strongly correlated with the duration of preoperative disability but not the biological severity of the initial injury.21
Both people involved with repetitive activities and those who are not develop carpal tunnel syndrome. In order to deliver rational preventive measures for workers with the disorder, valid and scientifically sound information about the true association between repetitive exposures and median neuropathy must be established. Without incidence data, we cannot estimate the excess fraction that is the relevant attributable risk parameter to measure in planning and policy questions.26 Knowledge of the true associations has direct implications for the primary, secondary, and tertiary prevention. Primary prevention is aimed at reducing or controlling the workplace risk factors. Does job modification work? Would alternating between high-risk and low-risk jobs decrease the incidence of carpal tunnel syndrome in the workplace? Is there a critical exposure duration (threshold) that, once reached, should lead to retirement or change to a low-risk job? Although keyboarding appears to have a protective effect from developing symptoms of median nerve compression,16,51 a tremendous amount of money is being spent on the design, promotion, and use of new keyboards. Many of these ergonomic designs, which claim to reduce symptoms, are based on changing the forearm rotation to less pronation. One study suggests that 45 degrees of pronation is the ideal position to reduce carpal tunnel pressures.54 The usual shoulder position of 20-30 degrees abduction when seated in front of a computer terminal rotates the forearm to that optimal position with the common keyboard design. Some physical conditions tend to precipitate symptoms of the forearm, wrist, or hand when using a computer. Anatomical abnormalities, such as limitation of pronation of the forearm, may result in significant abduction of the shoulder and elbow in order to place the hand in a neutral position. For these few people, alternative keyboard designs may
prove useful. Wrist splints have long been used in the treatment of carpal tunnel syndrome but are now marketed as protection for people in high-risk occupations. Since wrist flexion/extension increases carpal tunnel pressure, which in turn inhibits median nerve function, it seems reasonable to use a splint that maintains the wrist in neutral position; however, one study demonstrated that carpal tunnel pressures were higher with splint use at baseline and during repetitive hand activity, perhaps suggesting some external compression.55
Epidemiological studies of occupational carpal tunnel syndrome have usually not isolated the issue of wrist or finger position from grip force and repetition. Motion analysis studies of sign language interpreters for the deaf and grocery checkers18,29 have shown that workers with symptomatic hands had more frequent and more extensive flexion and extension than nonsymptomatic workers, suggesting that extreme flexion or extension may be an extrinsic risk factor. However, the chain of causality can also be read in the opposite direction; that is, median nerve impairment may interfere with proprioception, resulting in exaggerated motions. One case-control study demonstrated increased risk factors related to activities associated with either wrist flexion or wrist extension but not to activities associated with both wrist flexion and extension.16 This demonstrates lack of consistency that needs to be explained by future studies.
A major criticism of the term “repetitive stress” is that it implies that the etiology is due to repetition. The word “injury” implies damage to tissues. “Repetition and stress” imply that repetitive mechanical forces applied to tissues cause the injury, yet no information exists regarding the frequency, magnitude, duration, or rate that renders these forces harmful. We know that forces exceeding the mechanical limits of tissue lead to irreversible damage, yet physiological forces allow for normal maintenance and enhanced wound healing.3 The relationship between physical loads and musculoskeletal disease has still not been quantified, and contradictory evidence persists due to poor measurement of exposures and lack of specific diagnoses. Arguably, symptoms of aches and pains that precede the status of disease or injury are important to recognize, as many patients present themselves with such symptoms. Progression to disease, however, is the exception rather than the rule, or one would expect a greater prevalence in reports of disease states. The relative contribution of occupational and nonoccupational physical loads to causing symptoms has not been addressed sufficiently in epidemiological studies. Nevertheless, ergonomists promote the concept that from the perspective of prevention and treatment, all work-related factors that can be modified should be.69
There has been an increased advocacy of ergonomics as a solution to the incidence of repetitive motion injury in the workplace. This has re-
sulted in a proliferation of ergonomic literature in which conclusions are based on associations and suggestions.56,67 Ergonomics is defined as the study of how human beings use machines;4 however, it often refers to the craft of designing workplace equipment, including computers, to minimize health problems or injuries.1 Despite the widespread use of ergonomic methods in industry, there is controversy over their effect. Some have asserted that there is little substantive evidence that these methods are either valid or reliable.60 Others have found that ergonomics have resulted in substantial improvements in the workplace.52 There is little doubt that most ergonomic interventions increase comfort in the work environment, which is of great benefit to the worker. While ergonomists may create a more comfortable environment, they have not lowered the incidence of well-documented medical conditions, such as carpal tunnel syndrome. In fact, one study from Australia reported an increase in the incidence of “repetitive stress injuries” even after ergonomic redesign of the workstation and instituting rest periods from keyboarding every hour.32
On one final note, I think it important to look at the historical perspective with regard to what we've learned about work-related musculoskeletal injuries over time. Both the scientific community and nonmedical journalists are reporting an epidemic of disability from work-related musculoskeletal injuries. In order to understand why more people are becoming work-disabled, one must examine overall societal trends in disability. Since the middle of the 20th century, consecutive generations in the United States have become more disabled as assessed from self-reports, and it is unclear whether this is due to improved survival of the chronically ill, to lowered cultural thresholds for defining disability, or to real increases in disability incidence.20
The United States is not the first country to experience an epidemic of so-called work-related musculoskeletal problems. Between 1960 and 1980, Japan experienced an epidemic of cervicobrachial disorders.45 The epidemic became so widespread that in 1964 (even before the widespread introduction of personal computers) the Japanese Ministry of Labor set ergonomic standards for keyboard operators. These standards ultimately failed, however, to decrease the number of new reported cases.
In Australia at a telecommunications company, Telecom Australia, which had 90,000 workers, the rate of “repetitive stress injuries” began to rise in late 1983, peaked in late 1984 (30 times higher than the 1982 rate), and declined in 1985, reaching 1983 levels in 1987. The patients were most often keyboard operators who complained of pain that was neither consistent from patient to patient nor did it conform to any known neurological pathway, anatomical structure, or physiological pattern. There were no objective clinical findings other than random tenderness; clinical in-
vestigations were negative, and symptoms failed to respond to any form of physical treatment.34
In Australia, there was little evidence of a dose-response relationship of repetitive stress injury to keystroke rate, age, or job duration.32 Neither ergonomics, new technology, nor psychosocial theory explained the Australian epidemic. Miller and Topliss studied 229 consecutive patients referred with a label of repetitive stress injury.44 Of these patients, 29 fulfilled the usual criteria for a specific disorder, such as deQuervain's tenosynovitis or rheumatoid arthritis. Evaluation of the remaining 200 workers showed that 100 percent had anxiety, irritability, and/or lowering of mood; 91 percent had sleep disturbances; 84 percent had chronic fatigue; 61 percent had frequent tension headaches; and 78 percent had decreased sensation involving both hands in a nondermatomal distribution. Ergonomic measures that the investigators instituted for all office workers, including cessation of keyboard use, writing, or other activities that aggravated arm pain, failed to relieve symptoms in 78 percent of patients. Medications were used in all cases, and physiotherapy was used in 94 percent. All the patients reported that the treatment had little effect on their long-term progress.
Ultimately the incidence of repetition strain injury in Australia fell just as precipitously as it increased. What caused the Australian epidemic? In a time of relative prosperity, with technological changes and computerization of clerical tasks that threatened those less adaptable to change, and in a country with as many physicians and pharmacists per capita as any industrialized nation, the inability to work because of a physical ailment became more socially acceptable.33,34 Kiesler and Finholt concluded that the repetitive stress injury epidemic in Australia was more indicative of social problems than of workplace factors, and dissatisfaction was a major contributor, as was social legitimization of complaints related to repetitive stress injury.40 Political and social factors can act in both directions. The single factor that had the greatest influence on the decline of repetitive stress injury in Australia was a judicial decision in the case Cooper v. the Commonwealth (1987). The Supreme Court found that the employer was not guilty of negligence and the plaintiff had not suffered an injury. All costs were awarded against the plaintiff, and soon thereafter the Australian repetitive stress injury epidemic disappeared.34
1. Review & outlook: ergo, lawsuits. In The Wall Street Journal. Edited, New York, 1999.
2. Am J Ind Med, 25(4): 527-36, 1994.; ; and : Outcome of carpal tunnel surgery in Washington State workers' compensation.
3. Repetitive motion disorders of the upper extremity, pp. 217-230 . Edited by Gordon, S. L.; Blair, S. J.; and Fine, L. J., 217-230, Rosemont, Illinois, American Academy of Orthopaedic Surgeons, 1994.; ; and : Effect of loading on metabolism and repair of tendons and ligaments. In
4. Hand Clinics, 2(3): 553-565, 1986.: Ergonomics and cumulative trauma disorders.
5. J Occup Med, 21(7): 481-6, 1979., and : Carpal tunnel syndrome and selected personal attributes.
6. JAMA, 282(2): 153-158, 1999.; ; ; ; ; and : Prevalence of carpal tunnel syndrome in a general population.
7. Scand J Work Environ Health, 17(1): 46-52, 1991.; ; ; ; and : Carpal tunnel syndrome among ski manufacturing workers.
8. Ann Intern Med, 130: 910-921, 1999., and : Functional somatic syndromes.
9. JAMA, 274: 1931-1934, 1995., and : Somatization and medicalization in the era of managed care.
10. J Hand Surgery, 22A(1): 30-34, 1997.; ; ; and : Factors prolonging disability in work-related cumulative trama disorders.
11. Lancet, 1: 277-282, 1947.; ; and : Spontaneous compression of both median nerves in the carpal tunnel.
12. Am J Epidemiol, 132: 181-192, 1990., and : The effects of sensitivity and specificity of case selection on validity, sample size, precision, and power in hospital-based case-control studies.
13. J Occup Med, 23(4): 255-8, 1981.; ; and : Personal and occupational factors associated with carpal tunnel syndrome.
14. Scand J Work Environ Health, 19(2): 126-31, 1993.; ; ; ; ; and : Prevalence of shoulder and upper-limb disorders among workers in the fish-processing industry.
15. Occup Health Saf, 62(9): 116-20, 1993.: Solving ‘white collar' pain problems.
16. Am J Epidemiol, 132(6): 1102-1110, 1990.; ; ; and : Risk factors for carpal tunnel syndrome.
17. An instrumented device for diagnosing carpal tunnel syndrome. Orthop Rev, 23(6): 522-5, 1994.: The carpal-compression test.
18. J. Occup Med, 34: 257-264, 1992., and : Biomechanical factors affecting upper extremity cumulative trauma disorders in sign language interpreters.
19. Am J Epidemiol, 123: 736-751, 1986.; ; and : The effects of exposure misclassification on estimates of relative risks.
20. Scan J Work Environ Health, 21(3): 3-14, 1996.; ; ; ; and : Occupational back pain - an unhelpful polemic.
21. Am J Ind Med, 29: 642-648, 1996.: Outcomes research in Washington state workers' compensation.
22. Arch Environ Health, 51(5): 395-407, 1996.; ; ; ; ; ; and : Carpal tunnel syndrome and occupation in U.S. Navy enlisted personnel.
23. J Bone Joint Surg [Am], 68(5): 735-7, 1986.; ; ; and : Carpal tunnel syndrome. An evaluation of the provocative diagnostic tests.
24. J Occup Med, 34(11): 1117-9, 1992., and : Risk factors for carpal tunnel syndrome in industry: blaming the victim?
25. Annu Rev Public Health, 12: 543-66, 1991.; ; and : Upper-extremity musculoskeletal disorders of occupational origin.
26. Am J Epidemiol, 128(6): 1185-1197, 1988., and : Conceptual problems in the definition and interpretation of attributable fractions.
27. Scand J Work Environ Health, 18(6): 337-45, 1992.; ; and : Impact of occupations and job tasks on the prevalence of carpal tunnel syndrome.
28. Orthop Clin North Am, 27(4): 679-709, 1996., and : Epidemiology of work-related musculoskeletal disorders.
29. J Occup Med, 35(8): 805-11, 1993.; ; ; ; ; and : Supermarket checker motions and cumulative trauma risk.
30. J Hand Surg, 20A: 354-360, 1995.; ; ; and : Carpal tunnel surgery outcomes in workers: effect of workers' compensation status.
31. J Hand Surg, 22A(2): 216-221, 1997.; ; ; and : Relation of preoperative nerve-conduction values to outcome in workers with surgically treated carpal tunnel syndrome.
32. Med J Aust, 147: 218-222, 1987.: Epidemiological aspects of repetition strain injury in Telecom, Australia.
33. Occupational Disorders of the Upper Extremity, pp. 79-66 . Edited by Millender, L. H.; Louis, D. S.; and Simmons, B. P., 79-66, New York , Churchill Livingstone , 1992.: The Australian experience with cumulative trauma disorders. In
34. J Hand Surg, 20A(3): 553-556, 1995.: Repetition strain injury: the Australian Experience - 1992 update.
35. AAEM Quality Assurance Committee. Muscle Nerve, 16(12): 1392-1414, 1993.; ; ; ; and : Literature review of the usefulness of nerve conduction studies and electromyography for the evaluation of patients with carpal tunnel syndrome.
36. The gold standard [editorial]. Am J Phys Med Rehabil, 72(1): 1, 1993.: Diagnosis of carpal tunnel syndrome.
37. Am J Public Health, 81(2): 189-93, 1991.; ; ; and : Validation of a surveillance case definition of carpal tunnel syndrome.
38. Am J Public Health, 86(1): 52-56, 1996.; ; ; ; ; and : Workers' compensation recipients with carpal tunnel syndrome: the validity of self-reported health measures.
39. J Hand Surg [Am], 15(2): 360-3, 1990., and : A self-administered hand diagram for the diagnosis of carpal tunnel syndrome.
40. Am Psychol, 43: 1004-1015, 1988., and : The mystery of RSI.
41. Brain, 102(3): 619-35, 1979.: The carpal tunnel syndrome: localization of conduction abnormalities within the distal segment of the median nerve.
42. Clin Orthop, (251): 157-61, 1990.; ; ; ; and : Carpal tunnel syndrome. Evaluation of a quantitative provocational diagnostic test.
43. Hand Clinics, 12(2): 305-308, 1996.; ; and : Carpal tunnel syndrome in the work place.
44. J Rheumatol, 15: 1705-1712, 1988., and : Chronic upper limb pain syndrome (repetitive strain injury) in the Australian workforce: a systematic cross sectional rheumatological study of 229 patients.
45. J Hum Ergol, 11: 7-16, 1982.; ; and : History of occupational cervicobrachial disorders in Japan and remaining problems.
46. 35(11): 1084-6, 1993.: Intrinsic causes of keyboard injuries [letter]. J Occup Med,
47. 35(10): 1034-42, 1993.; ; ; and : Validation of occupational hand use categories. J Occup Med,
48. J Hand Surg [Am], 17(5): 850-7, 1992.; ; ; and : Longitudinal study of median nerve sensory conduction in industry: relationship to age, gender, hand dominance, occupational hand use, and clinical diagnosis.
49. J Occup Med, 34(4): 379-83, 1992.; ; ; and : Obesity as a risk factor for slowing of sensory conduction of the median nerve in industry. A cross-sectional and longitudinal study involving 429 workers.
50. The Journal of Hand Surgery, 13B: 167-170, 1988.; ; and : Occupation as a risk factor for impaired sensory conduction of the median nerve at the carpal tunnel.
51. J Hand Surg, 19B(1): 30-4, 1994.; ; ; ; and : Slowing of sensory conduction of the median nerve and carpal tunnel syndrome in Japanese and American industrial workers.
52. Human Factors, 40(2): 337-350, 1998., and : Workplace changes associated with a reduction in musculoskeletal symptoms in office workers.
53. Muscle Nerve, 11(5): 511-8, 1988., and : False positive electrodiagnostic tests in carpal tunnel syndrome.
54. J Hand Surg, 23A: 38-42, 1998.; ; ; and : Effects of forearm pronation/ supination on carpal tunnel pressure.
55. J Hand Surg, 19A: 106-110, 1994.; ; ; ; and : The effects of wearing a flexible wrist splint on carpal tunnel pressure during repetitive hand activity.
56. JOEM, 41: 11-119, 1999.; ; ; ; and : Effect of keyboard design on hand pain.
57. J Occup Med, 33(5): 627-31, 1991.; ; ; ; and : Median nerve latencies in poultry processing workers: an approach to resolving the role of industrial “cumulative trauma” in the development of carpal tunnel syndrome.
58. Am J Ind Med, 11(3): 343-58, 1987.; ; and : Occupational factors and carpal tunnel syndrome.
59. J Hand Surgery, 22A(2): 211-215, 1997.; ; ; and : A case-control study of obesity as a risk factor for carpal tunnel syndrome in a population of 600 patients presenting for independent medical examination.
60. 399(May 20): 197-198, 1999., and : What price ergonomics? Nature,
61. American Journal of Industrial Medicine, 24: 175-189: 175-189, 1993.; ; ; ; and : Median sensory distal amplitude and latency: comparisons between nonexposed managerial/professional employees and industrial workers.
62. Neurology, 38: 134-138, 1988.; ; ; ; and : Carpal tunnel syndrome in Rochester, Minnesota, 1961 to 1980.
63. Am J Ind Med, 19(1): 87-107, 1991.: Workplace ergonomic factors and the development of musculoskeletal disorders of the neck and upper limbs: a meta-analysis.
64. J Hand Surg, 14A(4): 624-627, 1989., and : Stress carpal tunnel pressures in carpal tunnel patients and normal patients.
65. J Orthop Res, 1: 172, 1983.; ; ; and : Effects of increased systemic blood pressure on the tissue fluid pressure threshold of peripheral nerve.
66. Am J Public Health, 84: 1846-1848, 1994.; ; ; ; ; and : The US prevalence of self-reported carpal tunnel syndrome: 1988 national health survey data.
67. Am. J. Ind. Med., 35: 647-661, 1999.; ; ; and : Effect of four computer keyboards on computer users with upper extremity musculoskeletal disorders.
68. J Hand Surg, 20A: 543-541, 1995.; ; and : Upper extremity disorders: a literature review to determine work-relatedness.
69. Clin Orthop, 351: 39-43, 1998.: Risk factors for upper limb disorders. Implications for prevention and treatment.
70. Muscle Nerve, 17(6): 632-6, 1994.; ; ; and : The relationship between body mass index and the diagnosis of carpal tunnel syndrome.
71. Br J Ind Med, 46(1): 43-7, 1989.; ; ; and : Carpal tunnel syndrome (carpal tunnel syndrome) and exposure to vibration, repetitive wrist movements, and heavy manual work: a case-referent study.
72. Hand Clinics, 12(2): 308-311, 1996.: Invited epidemiologic comment.