Panel Response to the Dissent
Dr. Szabo's dissent focuses on whether the panel was consistent in evaluating the literature relevant to this report. His dissent deals almost exclusively with only one of the musculoskeletal disorders considered in the report; specifically he ascribes to the panel overstatements about the research findings relating carpal tunnel syndrome to work exposures of a variety of types.
Dr. Szabo states correctly that criteria for the inclusion of studies in the report differed for the analysis of biomechanical exposures and for the analysis of epidemiologic associations. The four bodies of literature reviewed—tissue mechanobiology, biomechanics, epidemiology, and workplace interventions—have differing study designs, measurement techniques, and outcome variables. The selection criteria used in determining the quality of particular studies necessarily varied among these literatures (see Chapter 1, pp. 22). These criteria were set early in the panel's deliberations. Specifically, the biomechanics papers required detailed measures of biomechanical exposure, while the epidemiologic studies did not require that same kind of detail. Similarly, the epidemiologic papers had to meet criteria for epidemiologic inference that were not required of the biomechanics papers. The panel discussed the distinction carefully before agreeing to adopt it. The distinction would be problematic only if the panel made epidemiologic inferences from studies included in the biomechanics section that failed to meet criteria for epidemiologic studies. We did not do that.
Dr. Szabo contends that the panel concluded that interventions examined in this study prevented carpal tunnel syndrome; this misstates our report, which clearly states otherwise (see Chapter 8, pp. 313). The report states that interventions influenced pain reports and not the occurrence of
specifically defined disorders of the upper extremities. The studies are summarized in Table 8.3. The report does not state that interventions prevent carpal tunnel syndrome or, indeed, any other upper-extremity disorder. The emphasis, rather, is on amelioration of symptoms, which is the end point in the relevant literature. Furthermore, the comments on upper extremity interventions carefully state that interventions influence symptoms, not the incidence of specific disorders ( Chapter 8, p. 313):
Studies of engineering interventions for computer-related work that reduce static postural loads, sustained posture extremes, and rapid motions have demonstrated decreases in upper extremity pain reports. Further study of these interventions is needed to determine the amount of pain reduction possible, the duration of salutary effect, and which upper extremity clinical conditions could benefit from these interventions.
Dr. Szabo uses the case of carpal tunnel syndrome with regard to lowforce, high-repetition exposures (primarily the use of computer keyboards) as the causal factor to suggest that the relationship of musculoskeletal disorders to work exposure may not be sound. The panel has recognized that the evidence for low-force, high-repetition exposures is weaker than for other relationships among risk factors and musculoskeletal outcomes; however, strong evidence for causal relationships between physical work and musculoskeletal disorders is provided throughout the report.
The epidemiology section as it relates to the upper extremity was carefully written. We discuss the cross-sectional designs of most studies and possible implications for causal inference, including the potential for the “healthy worker” effect. In 9 studies, carpal tunnel syndrome was defined by a combination of a history of symptoms and physical examination or nerve conduction testing. In these studies there were 18 estimates of risk based on various specificities of carpal tunnel syndrome diagnosis and varying degrees of work exposure. Of these, 12 showed significant odds ratios greater than 2.0 (range 2.3 to 39.8), 4 showed nonsignificant odds ratios of greater than 2.0, and 2 showed nonsignificant odds ratios of between 1.7 and 2.0. The epidemiology section, however, does not draw specific conclusions regarding carpal tunnel syndrome. The report points out that just three articles dealt with keyboard work; indeed, keyboard work is not a major consideration or focus in the report.
Dr. Szabo's dissent provides an incomplete view of a study published in the Journal of the American Medical Association (Atroshi, 1999). He states: “In the general population the prevalence of Carpal Tunnel Syndrome is the same whether people perform repetitive activities or not.” In the panel's view, the nature of the design in that study and its survey instruments were such that the power to demonstrate this association was not
high. The study, however, did show a significant risk for carpal tunnel syndrome for blue-collar work, use of excessive force of the hands, working with excessively flexed or extended wrist, or the use of hand-held vibratory tools; these findings are not mentioned by Dr. Szabo.
Dr. Szabo cites the paper of Greenland and Robins (1988) to suggest that without knowledge of cofactors that contribute to carpal tunnel syndrome, “estimates offered by Hagberg as well as the ones used in the NAS report must be interpreted with caution.” In fact, the thrust of the Greenland and Robins argument is that such attributable risk calculations may severely underestimate (not overestimate, as implied by Dr. Szabo) the proportion of cases in which the etiologic factor is important because of possible interactions between that factor and the cofactors. Greenland and Robins cite numerous examples in which a small excess risk masks a much larger effect of a primary study factor.
Several articles cited by Dr. Szabo in his discussion of the epidemiology literature on carpal tunnel syndrome did not meet the quality criteria (insufficient participation and inadequate exposure measures were common problems) used by the panel in selecting articles for the epidemiology review and so are not included in the report.
In his dissent, Dr. Szabo states, “More importantly, reliance on ergonomics to the exclusion of medical and health risk factors can have adverse consequences for the patient.” Nowhere in its report does the panel suggest the exclusive use of ergonomic interventions.
It is important to reemphasize the fact that we made a major effort to base our conclusions on literature that met accepted scientific criteria and that the report represents consensus of all of the panel members except for Dr. Szabo. At the same time, the report makes plain the panel's view that the literature about musculoskeletal disorders is incomplete, as all clinical and scientific literatures are, and also emphasizes the importance of continuing research on a variety of fronts. There is, however, sufficient basis in the research to date to support our conclusions and recommendations.
Jeremiah A. Barondess, Chair
Mark R. Cullen
Barbara de Lateur
Richard A. Deyo
Sue K. Donaldson
Colin G. Drury
John W. Frymoyer
Jeffrey N. Katz
Jeffrey C. Lotz
Susan E. Mackinnon
William S. Marras
Robert G. Radwin
David H. Wegman