Evidence for Work-Related Musculoskeletal Disorders

Nortin Hadler's polemic offers a welcome opportunity to comment on the scientific basis for our view that work-related musculoskeletal disorders are a leading cause of preventable morbidity and disability in the American workplace today. Dr Hadler directs most of his attention to the relationship between keyboard use and carpal tunnel syndrome. There are flaws with Dr Hadler's analysis. Although we address some of his errors, we note that carpal tunnel syndrome among keyboard operators is a small fraction of the vast and growing number of disabling work-related musculoskeletal disorders. We would consider Dr Hadler's ruminations little more than a diversion of attention from the larger problem were it not for his efforts to generalize his beliefs about keyboard usage to virtually all other forms of repetitive motion and cumulative trauma in the workplace. Before turning to Dr Hadler's arguments, we set the context with what we believe to be some of the essential facts. First, the incidence and prevalence of work-related musculoskeletal disorders is intolerably large and growing. The Bureau of Labor Statistics reported 635,800 overexertion cases, 455,800 other strains and sprains, and 94,300 repetitive motion cases during 1993 that were severe enough to cause lost work days-cases acknowledged by employers to be work related.1 Forty-one thousand of these were lost work day carpal tunnel syndrome cases, constituting only 3% of the total musculoskeletal cases but 43% of the repetitive motion cases. Only 8000 of the carpal tunnel syndrome cases were attributed to repetitive typing or keying, constituting just .6% of the total musculoskeletal cases and 8% of the repetitive motion cases. In some high-risk industries, the rates for total repetitive motion disorders are extraordinarily high, such as in meat packing, where more than 12% of workers are affected every year. From 1993 to 1994, although most workplace injuries and illnesses were declining, the total number of repetitive motion cases (lost work day cases plus others) climbed 10%, from 302,000 to 332,000.2 Second, there is an impressive body of credible scientific evidence showing convincing associations between a variety of musculoskeletal disorders and such work-related factors as repetitive motion, stressful posture, and forceful exertion. Several recent reviews have reached this conclusion after applying standard scientific criteria (strength, specificity, temporal relations, consistency across studies and population, predictive performance, biological plausibility, dose-response) and after considering potential biases (selection, misclassification, recall, confounding).3-8 For some work-related musculoskeletal disorders, notably carpal tunnel syndrome, there are also associations with nonwork factors such as diabetes, pregnancy, rheumatoid arthritis, gender, and age.9-13 However, in many occupational epidemiology studies, these factors have been controlled through restriction in design or by control of confounders in analysis. Also, in some multivariate studies, the work-related risk factors are stronger predictors of adverse outcome than the nonwork factors. For example, in a case-control study, Wieslander et al found higher risks for work variables (odds ratios: 4.3 for vibration exposure 1 to 20 years; 16.0 for vibration exposure >20 years; 2.3 for repetitive work 1 to 20 years; 9.6 for repetitive work >20 years) than for most nonwork variables (odds ratios: 1.4 for diabetes; 2.3 for rheumatoid arthritis; 4.6 for thyroid disease; 2.0 for obesity; 1.5 for smoking).14 In addition to the epidemiology, numerous laboratory studies demonstrate that combinations of forceful, repetitive, and stressful postures contribute to musculoskeletal dysfunction.6,15-20 Third, there are many sound studies that focus on specific elements of proof. There is, for example, considerable information from prospective studies. In one such study of meat packing, the incidence rate of epicondylitis was seven times greater for those workers in physically strenuous jobs than for those in less strenuous jobs.21 Another prospective study of female light-assembly workers found exacerbation of neck/shoulder disorders among those who remained in jobs that involved constrained postures and improvement among those who changed to jobs with more varied tasks.22 There have also been studies exploring dose-response relationships. The risk of carpal tunnel syndrome, for example, increases with the duration of exposure to wrist flexion/extension23 or vibration.14 The risk for distal upper-extremity disorders increases as a function of force and recovery time within a work cycle24 and with the frequency of repetitive pinching or flexion/extension.25 Fourth, carpal tunnel syndrome has received extraordinary attention that is disproportionate to its relatively low incidence compared with other work-related musculoskeletal disorders. This is likely the result of the extended disability (32 median lost work days, more than for most amputations and fractures26) and expensive surgery associated with this condition. We share Dr Hadler's concern that some in the medical and rehabilitation communities have seriously mishandled management of this disorder, needlessly prolonging disability through excessive surgery or from ineffective nonscientific remedies. We are also troubled by inadequate attention to the correction of job risks. However, Dr Hadler's appropriate zeal with regard to disability management has apparently made it difficult for him to recognize the evidence for on-the-job etiology. Well-designed studies, for example, have found carpal tunnel syndrome to be strongly associated (ie, more than fourfold difference in rates) with high-force/high-repetition tasks, segmental vibration, or prolonged wrist flexion/extension.14,23,27-30 Except for those reported by Cannon et al and Silverstein et al, these studies used electrodiagnostic criteria. In all cases, these studies accounted for known confounders and most of them addressed prior conditions through restriction or multivariate analysis. Daubert v. Dow-Merrell Given this body of evidence, we are mystified by Dr Hadler's repeated reference to the Supreme Court's Daubert decision31 as if it somehow fortifies his position that "no systematic study supports the contention that usage, on the keyboard or in general... increases the likelihood of suffering carpal tunnel syndrome" (emphasis added). The Daubert case concerned allegations that maternal ingestion of Bendectin, an antinausea medication, caused birth defects in children exposed in utero. The Supreme Court ruled that the traditional "general acceptance" test for admissibility of expert scientific testimony is no longer appropriate and valid. Instead, the Supreme Court ruled that under the Federal Rules of Evidence trial, judges have the responsibility and authority to "ensure that any and all scientific testimony or evidence admitted is not only relevant, but reliable." Although "general acceptance" is a factor that can be considered, it is not to be "an absolute prerequisite to admissibility." In other words, even if expert evidence is not generally accepted by the scientific community, it is admissible if the judge is convinced it is both reliable and relevant. Although it provided trial judges with considerable discretion to be used in determining whether expert testimony is scientifically valid, the Supreme Court did offer four "pertinent considerations" that may be used. First, can or has the theory or technique been tested? In other words, have hypotheses been generated and tested to see if they can be falsified? Second, has the theory or technique been subjected to peer review and publication? This is "not a sine qua non of admissibility... In some instances well grounded but innovative theories will not have been published." Publication, or lack of publication, "will be a relevant, though not dispositive, consideration." Third, does the known or potential rate of error in the methodology strengthen or weaken the reliability of the evidence? Fourth, is there "widespread acceptance" of the theory or technique? Although such acceptance is not a necessary element of admissibility, "widespread acceptance can be an important factor." The evidence that carpal tunnel syndrome and many other musculoskeletal disorders are associated with cumulative trauma at work easily meets these tests. We believe not only that Dr Hadler's contrary conclusions are incorrect but that his analysis is defective in two ways. First, he relies on an overly selective choice of studies and misstates or omits some of their key facts. Second, he commits rudimentary errors of logic. Selection of Studies and Treatment of Facts Dr Hadler offers a four-step argument: first, physical examination findings, symptoms, and history consistent with carpal tunnel syndrome have no clinical or epidemiologic value; second, only nerve conduction studies are sufficiently precise and predictive to study carpal tunnel syndrome; third, there is a body of electrodiagnostic studies that shows "that no pattern of usage studied can be shown to represent a hazard in terms of the conductivity of the median nerve at the wrist;" and fourth, there are no other studies, either electrodiagnostic or otherwise, that might lead to contrary conclusions. Dr Hadler is wrong at each step. 1. Physical Exam and Symptoms Dr Hadler proclaims that epidemiologic studies of carpal tunnel syndrome that use symptoms, history, and physical examination are "doomed from the outset." His rhetoric is inconsistent with his own references, as well as with the principles of epidemiology. Dr Hadler cites a study that, he claims, demonstrates that with the exception of muscle atrophy, no classic signs and symptoms will identify carpal tunnel syndrome with sufficient validity to be useful for epidemiology.32 The authors of this study, however, point out that the combination of a positive Tinel's sign and a probable or classic hand-pain diagram have a specificity of 89%, positive predictive value of 71%, and a negative predictive value of 69% in a high-prevalence population-more than adequate for powerful epidemiologic analysis of high-risk working populations. The authors also point out that experienced neurologists identify carpal tunnel cases with "impressive accuracy" by using signs and symptoms (ie, 67% positive predictive value and 87% negative predictive value). Other studies attest to the value of signs and symptoms for epidemiologic analysis of work-related musculoskeletal disorders. For example, in the only study of carpal tunnel syndrome in the general population published to date, Stevens et al used a clinical definition for case inclusion. In approximately half of these cases, electromyography was done and, in this subgroup, 73% of hands clinically diagnosed with carpal tunnel syndrome had positive electromyographic findings.33 This is consistent with other studies not mentioned by Hadler, which show reasonably high validity of self-reported symptoms when compared with NCV and other laboratory measures. Stetson et al found lower median nerve amplitudes and longer latencies among symptomatic automobile workers compared with asymptomatic workers.34 Hagberg and Kvarnstrom found symptoms to be well correlated with objective measures of trapezius muscle activity.35 Even powerful clinical tests often yield considerable numbers of false-positive results, particularly in low-prevalence groups. This can be a substantial problem for clinical screening. The impact on epidemiology is to introduce misclassification, which reduces the power of a study to find true associations. This drives results to the null and makes it difficult to find exposure-effect associations that are truly there. When an epidemiologic study is positive, despite this inherent tendency to understatement, it is often because the associations are especially strong. 2. The Reliability of Electrodiagnostic Tests Just as vigorously as Dr Hadler criticizes sign/symptoms based epidemiology as "doomed," he defends the reliability of electrodiagnostic test results for carpal tunnel syndrome. However, even with very high specificity and sensitivity, the positive predictive value of nerve conduction tests can be very low, especially in low-prevalence populations. In addition, there are many reasons why nerve conduction velocity studies can be falsely positive or negative, making them even less reliable. One study, for example, found more than 45% of healthy, asymptomatic subjects to have at least one false-positive electrodiagnostic test for carpal tunnel syndrome.36 3. A Series of "Compelling" Studies Dr Hadler describes a series of electrodiagnostic studies that he finds "compelling," but which we find questionable or inaccurately described. The study of meat-packing workers by Masear et al, for example, does not show that applicants have abnormal electrodiagnostic tests with the same frequency as active workers, as Dr Hadler suggests. Instead it shows that 15% of applicants had at least mildly abnormal tests, whereas 15% of employees had carpal tunnel-release surgery.37 The true frequency of abnormal tests among active workers, which may be much higher than 15%, was unreported. Dr Hadler describes the study by Schottland et al as showing no discernible difference in median nerve conduction between workers and applicants for work in a poultry-processing plant. In fact, female workers had significantly greater mean palm-wrist distal sensory latencies in the dominant wrist, compared with applicants. In addition, twice as many female workers had an absolute abnormality in distal sensory latency compared with female applicants.38 Radecki's study, which found associations between nerve latency and various personal attributes such as aging, cannot be used to conclude an absence of associations with job attributes such as repetitive trauma.39 By defining job risk according to type of industry rather than specific job duties, Radecki introduced profound misclassification because many workers have high-risk jobs and occupations within low-risk industries. Because these subjects from high-risk subgroups would have been assigned to low-risk categories for analysis, this study, by its design, would have little likelihood of documenting even strong job associations. Dr Hadler's characterization of the case-control study of carpal tunnel syndrome in the Netherlands is extraordinarily misleading.23 as Dr Hadler there were no specific associations between carpal tunnel syndrome and there were substantial cumulative trauma that Dr Hadler to than of cases in with extended and for both be the risk at least to for in these for more than 20 A for and was but not Dr Hadler that the for and at the no electrodiagnostic test between symptomatic and employees that be by keyboard In fact, the authors reported that number of typing at the was significantly associated with work musculoskeletal among the cases and for and dominant the case definition for carpal tunnel syndrome was found to be significantly associated with abnormal distal median sensory in distal median latency and median conduction velocity were also but the of these was This not only keyboard usage but it the of a case definition based on signs and In of the that was not by Dr Hadler, employees were using this case definition but no electrodiagnostic There was a strong dose-response between or wrist disorders and of keyboard The fourfold in risk among those with keyboard use to was greater than that for any of the The prospective study by et al of workers that, according to Hadler, no clinically or is, in fact, positive when the are For example, there was a in the of subjects with abnormal latency and abnormal sensory velocity and The of those with abnormal results from one to two for virtually all tests, although most did not The studies by et al, which show inconsistent between electrodiagnostic test results and occupational have flaws that may very well true First, for occupational which has been in other questionable despite recent efforts to For example, more than half of the workers in the of et al are from but the of workers to the high-force/high-repetition is at with our with in this This type of exposure misclassification can it to find true relationships. these et al did find a strong between occupational categories and nerve conduction results in workers and Second, et al use an overly definition of abnormal sensory latency which prevalence rates for carpal tunnel syndrome that are higher than those of virtually any other published studies. The prevalence of in the et al study for example, with the prevalence rate of of Stevens et on clinical or the incidence rate of of et on This based on overly and not very specific electrodiagnostic was in study by et workers, when using the of hands had an whereas only of hands were to have carpal tunnel syndrome when a combination of electrodiagnostic and clinical criteria was used. In a review of studies of the most useful nerve conduction tests for the of carpal tunnel syndrome, the American of that use of the of only the a specificity to the other tests in this This that the of et al those with and making the of Although it is not an electrodiagnostic study of carpal tunnel syndrome, the of employees is by Dr Hadler for its important negative with regard to use and Dr Hadler out that but no physical of keyboard use were associated with the disorders However, Dr Hadler does not point out that the physical factors were not In the case of did no analysis. In the case of duration of keying, found little in the study to The study was also as the operators very of Studies There are important electrodiagnostic studies that not Dr Hadler's and not in his In addition to those et al have reported that duration of vibration exposure among and workers is a of prolonged median distal on sensory latency et al reported a and prevalence of 2.0 for carpal tunnel syndrome for workers in repetitive jobs compared with those in Stetson et al has reported an between median nerve and in the latency among symptomatic workers compared with asymptomatic workers, for and There is also a substantial scientific in addition to electrodiagnostic studies, that easily meets the Daubert tests but is not consistent with Dr Hadler's conclusions and is not by For example, the only study of work-related carpal tunnel syndrome in the that shows in incidence rates across industry Although the industry rate was the rates were for and packing, for meat and and for Other studies, electrodiagnostic and otherwise, are by Hagberg et and Dr Hadler that the context of and repetitive motion is as a to be with and he that a keyboard is is not In the is This is that some repetitive motion is repetitive motion is is the of that some are and can some all can be In making his Dr Hadler of dose-response relationships. of A are necessary for whereas larger can it that some strong a of and a does This is not just and and who physical are at high risk of repetitive trauma from physical many times their the that the risk of of and increases after only about In fact, there is a scientific documenting repetitive motion disorders among prospective studies of which have found of nerve conduction as the such as or also risk the between physical and repetitive be that workplace who are to or type in that the of the body to are not just in but are in Dr Hadler that is a and of and that with musculoskeletal is prerequisite to criticizes workers with for their to of most of the adequate that we see no to the of There are two errors of First, it is incorrect to believe that or is to well or This one to the of or other high-prevalence disorders. such Dr Hadler would be to recognize that in many of the is and Second, despite Dr Hadler's "not to these he has the that workers are the and associated with and of by the of a that Although we with Dr Hadler that many workers are an of and we vigorously his to problem There is and about the physical of work in meat work in repeated by workers, and and by is that working with these extraordinary physical very to with their and hands on the very well that their jobs are in they the of medical and who to be with and Dr Hadler that with symptoms and signs are more likely to have median than to have carpal tunnel epidemiologic studies on such symptoms and signs from the Dr Hadler apparently that with high false-positive rates are for scientific By using such he would all epidemiologic studies of high that on or studies of that the American One of the of epidemiology is the to in the of variables in large in in which the test is and not useful for clinical screening. we out misclassification caused by false-positive tests reduces the likelihood that an epidemiologic study than a prevalence will be This the that a positive epidemiologic is falsely studies more but also more reliable. Dr Hadler absolute beliefs when for no hazard for carpal tunnel syndrome has been is because is either there is no risk or it is small to be the that is Even if Hadler were not about an absence of scientific evidence, this would be who have and in the or other to the have been by the that of evidence is not the same as evidence of This is true not just for the of from the to a of from in in what is study evidence of of and excessive strains caused by the of Other evidence to the of in a the of physical that was also although all of Dr Hadler's with carpal tunnel syndrome among keyboard his conclusions are to all cumulative trauma and all workplace The is is is Dr Hadler might as well be that because only of are by only of are by the same Nortin Hadler has that who is the of the is one can well who the they are at risk for is We share Dr Hadler's concern and with the workers who and disability on the job are often We that our of and often the for and However, although we may on some of and and we with Dr Hadler the of and of we use the criteria by the Daubert or traditional scientific criteria for there is convincing evidence that a vast number of musculoskeletal are at least in substantial by exposure to repetitive motion and cumulative trauma at difference with Dr Hadler has because beliefs about cause for beliefs to the where the of are to physical risk factors at their Dr Hadler would His is not the of There is a of Labor and and of of of and