Work-Related Musculoskeletal Disorders: Examining the Research Base Epidemiology: Physical Factors

Bradley Evanoff, MD, MPH Washington University School of Medicine, St. Louis, MO

This abstract will address four questions concerning the NIOSH document ''Musculoskeletal Disorders and Workplace Factors," as requested by the conference organizers. In my review of this document, I focused on the studies of elbow disorders, carpal tunnel syndrome (CTS), and hand/wrist tendinitis.

1. Has NIOSH missed or overlooked any important body of epidemiological evidence in its review?

The NIOSH review was comprehensive and inclusive; important epidemiological evidence was not overlooked.

NIOSH conducted a series of wide-ranging literature searches of appropriate databases, using suitable search terms. Articles were selected for detailed review if they pertained to the epidemiology of work-relatedness of musculoskeletal disorders, and contained explicit descriptions of the studied populations and assessments of health outcomes and exposures. Of the more than 2000 articles identified, NIOSH chose over 600 for detailed review. NIOSH did not provide a listing of which articles had been identified but rejected for detailed review; such a list would be helpful in evaluating their search and review strategy. I am unaware of any important articles omitted by NIOSH; their selection was generally more extensive and inclusive than that of previous published reviews.

2. Describe the study methods of the studies that have been heavily weighted in the NIOSH assessment. What is the general quality of these studies?

The general quality of these studies is good; a major drawback is the small number of longitudinal studies.

The design of most reviewed studies was cross-sectional. Many cross-sectional studies included temporal data on exposures and health outcomes, and can thus address the issue of whether the exposures preceded the health effects. More longitudinal studies would be helpful; these studies face a host of logistical hurdles including high turn-over of workers in many high-exposure jobs. In general, the weighted studies had appropriate and well-characterized referent groups.

The weighted studies chose appropriate epidemiological case definitions, which required both symptoms of a MSD as well as physical examination (PE) findings. For epicondylitis and hand/arm tendinitis, these case definitions were analogous to commonly used clinical diagnostic



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 152
--> Work-Related Musculoskeletal Disorders: Examining the Research Base Epidemiology: Physical Factors Bradley Evanoff, MD, MPH Washington University School of Medicine, St. Louis, MO This abstract will address four questions concerning the NIOSH document ''Musculoskeletal Disorders and Workplace Factors," as requested by the conference organizers. In my review of this document, I focused on the studies of elbow disorders, carpal tunnel syndrome (CTS), and hand/wrist tendinitis. 1. Has NIOSH missed or overlooked any important body of epidemiological evidence in its review? The NIOSH review was comprehensive and inclusive; important epidemiological evidence was not overlooked. NIOSH conducted a series of wide-ranging literature searches of appropriate databases, using suitable search terms. Articles were selected for detailed review if they pertained to the epidemiology of work-relatedness of musculoskeletal disorders, and contained explicit descriptions of the studied populations and assessments of health outcomes and exposures. Of the more than 2000 articles identified, NIOSH chose over 600 for detailed review. NIOSH did not provide a listing of which articles had been identified but rejected for detailed review; such a list would be helpful in evaluating their search and review strategy. I am unaware of any important articles omitted by NIOSH; their selection was generally more extensive and inclusive than that of previous published reviews. 2. Describe the study methods of the studies that have been heavily weighted in the NIOSH assessment. What is the general quality of these studies? The general quality of these studies is good; a major drawback is the small number of longitudinal studies. The design of most reviewed studies was cross-sectional. Many cross-sectional studies included temporal data on exposures and health outcomes, and can thus address the issue of whether the exposures preceded the health effects. More longitudinal studies would be helpful; these studies face a host of logistical hurdles including high turn-over of workers in many high-exposure jobs. In general, the weighted studies had appropriate and well-characterized referent groups. The weighted studies chose appropriate epidemiological case definitions, which required both symptoms of a MSD as well as physical examination (PE) findings. For epicondylitis and hand/arm tendinitis, these case definitions were analogous to commonly used clinical diagnostic

OCR for page 152
--> criteria, and included PE maneuvers such as tenderness to palpation and pain on resisted movement, as well as examination for other causes of upper extremity symptoms. For carpal tunnel syndrome, the use of case definitions based on symptoms and physical examination alone is more controversial. The most accurate definition of CTS would include electrodiagnostic studies (EDS) as well as symptoms (and possibly PE). This definition was used in some of the studies reviewed by NIOSH. EDS alone should not be used as a case definition. Several studies have evaluated the performance of case definitions based on symptoms and physical examination alone; although there is some misclassification of disease using this case definition, it is appropriate for studies where the disadvantage of misclassification can be offset by other factors, such as a larger sample size. The weighted studies did not use self-report or job titles as a measure of exposure, but used directly observed and measured data on specific exposures. The weighted studies controlled for potential biases and confounders through a variety of mechanisms. Individual risk factors such as age, gender, pre-existing disease, non-occupational activities, and metabolic diseases were controlled to some extent in all the weighted studies. Measurements of health and exposure status were done in a blinded manner. Weighted studies all had participation rates of 370%, decreasing the chance of respondent bias. 3. Would either the inclusion of any omitted studies or the assessment of the quality of those reviewed substantially alter the interpretation of the epidemiological evidence that certain physical stressors in the workplace increase risk of acquiring certain MSDs? Addition of omitted studies would not alter the interpretation of the evidence. More stringent requirements for study quality could change the conclusions only by excluding most or all current studies. The literature review and study weighting process adopted by NIOSH included most or all important evidence. Other authors, using different criteria for literature selection, have also concluded that certain physical stressors in the workplace increase the risks of acquiring certain MSDs. Authors who conclude that there is insufficient evidence to permit this interpretation have excluded most or all of the current epidemiological evidence, citing the lack of longitudinal data and questions about the significance of health outcomes measured in these studies. The well-designed cross sectional studies weighted by NIOSH provide data useful for causal inference. These studies took steps to avoid bias in the measurement of exposures and health outcomes. Data on the temporality of exposure and symptom or disease onset were available to ensure that the putative exposures occurred before symptom onset. These studies thus offer more than a "snapshot in time." Other potential causes of the observed associations were sought. To discount these studies, one must assume that measurements were significantly biased, or that persons prone to developing MSDs were preferentially placed into jobs with high physical demands. In cross-sectional studies, survivor effects will typically decrease the observed associations between symptomatic disorders and physically demanding jobs.

OCR for page 152
--> Another question has been whether epidemiological case definitions represent "real" clinical disorders. Many MSDs are self-limited or cause minimal disability, while others are more severe. Case definitions can be designed to capture a broad or a narrow spectrum of illness; it is probably not true that only the most severe cases are clinically "real." Case definitions used in the weighted studies required both symptoms and physical examination findings by an experienced clinician, and were analogous to common clinical diagnostic practices. For CTS, the issue of diagnosis without EDS is controversial as noted above. Many clinicians make the working diagnoses of CTS, and start therapy in individual patients, based on the same symptom and physical examination criteria used in those weighted studies which did not use EDS. 4. What does the evidence tell us about incidence in the general population versus specific groups of workers? MSDs are common in the general population. Epidemiological studies of occupational risk factors must include appropriate referent groups for valid comparisons. Many studies show much higher than expected rates of MSDs among workers with high exposures to physical risk factors. The prevalence and incidence of MSDs is highly dependent on the reporting mechanisms and case definitions adopted. Summary: The current best evidence shows consistent, strong associations between certain physical factors and MSDs. The quality of evidence is adequate but not perfect. It seems unlikely that future longitudinal studies will invalidate causal inferences drawn from current studies.

OCR for page 152
--> The Epidemiology of Workplace Factors and Musculoskeletal Disorders: An Assessment of the NIOSH Review Alfred Franzblau, MD Associate Professor of Occupational Medicine, University of Michigan School of Public Health A relationship between workplace factors and musculoskeletal disorders has been noted for hundreds of years, however, scientific (including epidemiological) study of the potential contribution of workplace factors to musculoskeletal disorders was only initiated in the early and middle decades of the current century. At present, a considerable body of work has accrued, and a critical review of the accumulated data represents a worthwhile effort. In 1997 the National Institute for Occupational Safety and Health published a review of the epidemiological literature pertaining to workplace factors and musculoskeletal disorders of the neck, upper extremity, and low back (U.S. Department of Health and Human Services, 1997). My comments will focus on this document, primarily those portions pertaining to the upper extremity since I am most familiar with this literature, and will be structured to address the four issues suggested by the workshop organizers. 1. Has NIOSH missed or overlooked any important body of epidemiological evidence in its review? The sole focus of the NIOSH document is epidemiological investigations. This is clearly stated, along with criteria for evaluating the quality of individual studies. I don't believe that the NIOSH document has overlooked any important body of epidemiological evidence pertaining to workplace factors and musculoskeletal disorders. Overall, I am impressed with the thoroughness of what was a massive and unprecedented review of literature. However, there are other categories of scientific evidence that contribute to the key questions of concern, such as biomechanical studies, engineering studies, and laboratory and/or psychophysical studies. Scientific knowledge and understanding is not derived solely from one type of study technique. In the current instance inclusion of information from these other types of studies would have, in my opinion, corroborated the conclusions of the NIOSH review of epidemiological literature. 2. Describe the study methods of the studies that have been heavily weighted in the NIOSH assessment. What is the general quality of these studies? There are many issues related to study methodology that one could comment on, however, I wish to focus my comments on one particular area that has received considerable attention. One of the major criticisms of epidemiological studies of workplace factors and musculoskeletal disorders is that most of the studies are cross-sectional, rather than prospective or longitudinal Although there is some basis to this criticism, and good prospective studies are of great value, in my opinion it is usually overblown in the present instance.

OCR for page 152
--> While, in theory, prospective studies have certain advantages, there are a number of practical reasons why they have been utilized so rarely to investigate relationships between workplace factors and musculoskeletal disorders. Aside from being far more expensive, such studies require on-going interaction between the research team and study site(s) over a prolonged period of time. Such interaction is usually time-consuming to the employer, and study procedures (especially screening medical examinations of workers) can interfere with production processes. Few employers are willing to extend this level of cooperation over a prolonged period of time, particularly when it interferes with production and thus may impact the 'bottom line'. Furthermore, when companies are bought, sold, or taken over (as has often been the case in the last 15 years), the new managers may not be committed to the study, which can lead to disruption or termination of a study. Additional complicating factors are that workers change jobs, and jobs also change. Obviously, such changes are not scheduled or designed for the convenience of the researcher. Thus, the 'exposures' are likely to be unstable over time, and tracking the 'exposure' of individual workers becomes very complicated and expensive, and, even if done, such information may be difficult to interpret. What some researchers have done is to perform cross-sectional studies among workers (and jobs) that are known to have been stable for some minimum period of time (e.g., six months or one year). This type of cross-sectional study design overcomes some of the shortcomings of cross-sectional studies relative to prospective studies, and serves to greatly strengthen the confidence one can have in the conclusions. Many of the studies that were most heavily weighted in the NIOSH assessment fall into this category. Finally, one needs to remember that although one of the weaknesses of cross-sectional studies relative to prospective studies is the greater potential for misclassification of exposures and outcomes, the effect of such misclassification is usually to reduce the apparent strength of association. Random misclassification almost always leads to a reduction in the apparent strength of the relationship between exposures and outcomes. Furthermore, one of the most common non-random effects on a cross-sectional study of workers is a survivor bias: study of a stable worker population is frequently biased by the 'healthy worker effect' since those workers with the worst 'effect' will have left the jobs in question. Thus, whatever statistical measure of risk is employed in analyses, if a positive (and statistically significant) relationship is found in a cross-sectional study, then one can assume that the true risk is probably even greater than what was found. Thus much of the criticism of the reliance on cross-sectional versus prospective studies is without foundation. A great deal that is useful has been learned from cross-sectional studies. And, in terms of utilization of resources, they are almost always cheaper and faster to complete than prospective studies. I believe that the studies most heavily relied on by NIOSH in its assessment of workplace factors and musculoskeletal disorders are of good quality. 3. Would either the inclusion of any omitted studies or the assessment of the quality of those reviewed substantially alter the interpretation of the epidemiological evidence that certain physical stressors in the workplace increase risk of acquiring certain MSDs? No.

OCR for page 152
--> 4. What does the evidence tell us about incidence in the general population versus specific groups of workers? Overall, there have been very few studies of the incidence or prevalence of specific musculoskeletal disorders in the general population. This is certainly a weakness in our overall understanding of these conditions, and it would be very helpful if such studies were performed. Even among those studies of musculoskeletal disorders in the general population which do exist, there is reason to believe that the published estimates may substantially miss the mark. For example, the best study we have of the incidence of carpal tunnel syndrome in the general population was published by Stevens et al., in 1988. This study was based on data collected from 1961 through 1980. It is notable that the incidence rates for carpal tunnel syndrome in this study increased substantially over this 20 year period, probably reflecting better diagnosis, and greater attention paid to this condition by both clinicians and patients. However, the data on which this study is based were collected prior to the massive increase in work-related carpal tunnel syndrome that began in the early 1980's and peaked in about 1994 or 1995 (Brogmus, 1995; Hanrahan, 1991; Bureau of Labor Statistics, 1989). With increased attention focused on this condition during the 1980's and 1990's, there is strong reason to believe that the true incidence of carpal tunnel syndrome in the general population may be substantially greater that the estimates provided by Stevens et al. There is even empirical evidence to support this contention (DeKrom et al., 1992). Despite the absence of a substantial body of data concerning the incidence or prevalence of specific musculoskeletal disorders in the general population, this does not detract materially from our understanding of the potential contribution of workplace factors to the development of such disorders. Almost all of the best studies of musculoskeletal disorders among workers have employed internal control groups (usually defined on the basis of ergonomic exposures) from the same company or a similar company in the same geographic location. Such internal comparison populations usually provide for better control of potential confounders since they are more likely to be better matched sociodemographically to the exposed population than would be the general population. References: Brogmus GE. Reporting of Cumulative Trauma Disorders of the Upper Extremities May be Leveling Off in the U.S. Proceedings of the Human Factors and Ergonomics Society 39th Annual Meeting—1995. pp. 591-595. Bureau of Labor Statistics. Occupational Injuries and Illnesses in the United States by Industry, 1989: U.S. Department of Labor, Bulletin 2379. Washington, D.C.: U.S. Government Printing Office, 1991. DeKrom MCTFM, Knipschild PG, Kester ADM, Thijs CT, Boekkooi PF, Spaans F. Carpal tunnel syndrome: prevalence in the general population. J Clinical Epidemiology. 1992;45(4):373-376.

OCR for page 152
--> Hanrahan LP, Higgins D, Anderson H, Haskins L, Tai S. Project SENSOR: Wisconsin Surveillance of Occupational Carpal Tunnel Syndrome. Wis Med J. 1991;90(2):80, 82-83. Stevens JC, Sun MD, Beard CM, O'Fallon WM, Kurland LT. Carpal tunnel syndrome in Rochester, Minnesota, 1961 to 1980. Neurology. 1988;38:134-138. U.S. Department of Health and Human Services/National Institute for Occupational Safety and Health. Musculoskeletal Disorders and Workplace Factors: A critical review of epidemiologic evidence for work-related musculoskeletal disorders of the neck, upper extremity, and low back . DHHS (NIOSH) Publication No. 97-141. July, 1997.

OCR for page 152
--> Workshop on Work-Related Musculoskeletal Injuries: Examining the Research Base Panel on Epidemiology: Risk Factors Fredric Gerr, MD Associate Professor, Rollins School of Public Health, Emory University, Atlanta, GA 1. Has NIOSH missed or overlooked any important body of epidemiological evidence in its review? No. NIOSH has been thorough in its collection of epidemiological evidence for inclusion in its review (NIOSH [Bernard, B., Ed.]. Musculoskeletal disorders and workplace factors. US Department of Health and Human Services, Washington, DC., 1997). 2. Describe the study methods of the studies that have been heavily weighted in the NIOSH assessment. What is the general quality of these studies? Prior to addressing this question, it may be useful to review the NIOSH methods for identifying studies for inclusion and its scheme for weighting more heavily those of greater methodological rigor. The authors of the NIOSH assessment described the strategy used for selection and inclusion of studies. They required that included studies 1) have well defined exposed and referent populations, 2) use well defined explicit criteria for identification of health outcomes, 3) evaluated exposure so that inferences could be made about specific ergonomic exposures, and 4) use the following designs: population based study, case control study, cross-sectional study, longitudinal cohort study, and case series. These guidelines, in general, are reasonable and appropriate. The majority of studies included by NIOSH in its assessment were cross-sectional in design. The specific methods varied considerably across studies, in particular with regard to the rigor of exposure assessment, health outcome assessment, control of confounding, and minimization of sample distortion occurring as a result of poor participation. NIOSH made efforts to addressed the methodological heterogeneity of the studies it reviewed by establishing a set of criteria for weighting studies' relative methodological quality. Specifically, NIOSH considered studies to be more influential if 1) participation was >=70%; 2) the health outcome of interest was defined by symptoms and physical examination, 3) the investigators were blinded to health or exposure status when assessing health or exposure status, and 4) the joint under consideration "was subjected to an independent exposure assessment, with characterization of the independent variable [i.e., exposure] of interest". These criteria for identifying studies of relatively greater methodological rigor are reasonable and appropriate.

OCR for page 152
--> Before examining the literature as a whole, NIOSH examined statistical significance and control of confounding and other biases present in each study. Finally, NIOSH applied well established criteria to the literature reviewed to determine whether "strong evidence", "evidence", "insufficient evidence" or ''evidence of no effect" suggestive of a causal association was present. As described, the system NIOSH used to identify studies for inclusion in its review, its methods of assessing their methodological rigor, and its process for evaluating the literature as a whole were reasonable and appropriate. In terms of application, however, NIOSH was, at times, less rigorous than its multi-level system of identifying and considering studies of high epidemiological quality might otherwise suggest. Two examples are provided for illustration: NIOSH included numerous studies of neck and neck/shoulder MSDs in which the exposure variable was a measure of ergonomic exposure of the hand and/or wrist in which no measures of force, repetition, or posture directly applicable to the neck or neck/shoulders were obtained. NIOSH acknowledged that such studies were more prone to exposure "misclassification" and indicated that they were given lesser weighting because of it (p. 2-2). The utility of including such studies at all, however, is unclear as they add little or nothing to the clarification of the relationship between ergonomic exposure to the neck or neck and shoulders and disorders of these regions. NIOSH indicated that only 2 of the 27 studies of the relationship between repetition and neck or neck/shoulder MSDs met all four "evaluation criteria" (indicators of methodological rigor) and that only one of those reported odds ratios for the exposure (p. 2-4). Despite this relatively low number of studies meeting criteria for epidemiologic rigor, NIOSH reported in its Conclusions Regarding Repetition that 27 studies found ORs greater than one and that statistically significant increases in risk were found in 19 studies (p.2-12). Basing such summary statements on a body of literature that includes at least some studies of questionable methodological quality may overstate the actual strength of inference that can be made from the limited number of high quality studies available. Other examples similar to those provided here can be found in other sections of the NIOSH report. The body of literature describing the association between work and musculoskeletal disorders is of heterogeneous quality. This has been acknowledged by many investigators and authors, and has been the subjects of numerous published reviews. It appears that NIOSH has included many studies of questionable quality in its comprehensive review although it did make efforts to identify studies of higher and lower quality and to weight those of higher quality in its assessments. NIOSH may have been insufficiently rigorous, however, in its efforts to consider only those studies of sufficiently high methodological quality to render their results useful and to eliminate from consideration those so flawed that their results add little to the resolution of the issue of work-relatedness. In doing so, NIOSH may have weaken the case that it hoped to make from its literature review.

OCR for page 152
--> 3. Would either the inclusion of any omitted studies or the assessment of the quality of those reviewed substantially alter the interpretation of the epidemiological evidence that certain physical stressors in the workplace increase risk of acquiring certain MSDs? More rigorous elimination of those studies that failed to meet criteria for methodological rigor would substantially reduce the number of studies available for review for each of the body locations (i.e., neck, shoulder, elbow, etc.) considered. Review of this smaller, but more rigorous body of literature, would not, in my opinion, substantively change the conclusions drawn from the literature regarding work and musculoskeletal disorders. Because of a relative paucity of high quality investigations, some specific associations, such as that between repetition and neck disorders or force and epicondylitis might be considerably less firm than currently reported if weak studies were excluded from consideration. The conclusion that the literature supports an association between work characterized by forceful and repeated use of the hands and arms and a range of painful musculoskeletal disorders of the upper extremities would still be justifiable, however. 4. What does the evidence tell us about incidence in the general population versus specific groups of workers? No studies of which I am aware have applied identical criteria for identification of MSDs to specific working groups and to the "general population". Some studies have attempted to examine the relationship between work and musculoskeletal disorders in the general population. Virtually were cross-sectional and few, if any, such population based studies used objective assessment of both health and ergonomic exposure, however. As a result, such studies are subject to reporting bias in which those with a painful MSD may be more likely to report exposure than those without such a condition.

OCR for page 152
--> Epidemiologic Studies of Physical Ergonomic Stressors and Musculoskeletal Disorders Laura Punnett, Sc.D. Department of Work Environment, University of Massachusetts Lowell The extensive literature review edited by Bernard and colleagues at NIOSH (1997) was rigorous in its methodology. The four review criteria applied to the literature (page 1-10) reflected well-accepted, important principles of epidemiology. They gave greatest emphasis to studies in which selection bias and information bias were unlikely to have exerted a large influence on the results. Almost all of the studies judged to be of highest quality (Armstrong 1987a; Bovenzi 1995; Chatterjee 1982; Chiang 1990, 1993: Jonsson 1988; Kilbom 1986; Juopajärvi 1979; Ohlsson 1994, 1995; Osorio 1994; Punnett 1991; Silverstein 1987), and many of the other papers as well, employed standard epidemiologic and statistical techniques to control for multiple covariates, making confounding also an unlikely explanation for the reported associations. Thus, the most heavily weighted investigations were rigorously conducted according to standard scientific principles and are highly appropriate for NIOSH and OSHA to rely upon. A number of them also followed quasi-experimental procedures in selecting study subjects, in order to obtain good contrast between exposure levels, and thereby were particularly informative with regard to good contrast between exposure levels, and thereby were particularly informative with regard to exposure-response relationships. In addition, some of the papers that NIOSH judged not to meet all 4 criteria could have been evaluated in a more positive light if they had been read as part of a body of work, rather than as stand-alone studies (e.g., Punnett 1985; Riihimäki 1989b). In sum, this group of 13 (or more) high quality papers represents a larger body of epidemiologic evidence than OSHA has been able to rely upon for almost any prior rule-making in the agency's history. Although it is always possible to overlook some papers, despite one's best efforts, there were few omissions proportionate to the volume of literature reviewed. I have not found any body of evidence relevant to the stated scope of the review was systematically overlooked. I have noticed the apparent omission of a few studies, although some of them at least may have been excluded intentionally because they examined health outcomes that did not fit the specific body part categories designated for the review. Most of the omitted studies that I have noted were investigations of changes in health status following reductions in exposure, either from intentional ergonomic interventions in the workplace or from evolving production processes, equipment, and work environments. The documentation of such health benefits speaks directly to the question of temporal relationship and is often held to provide very strong evidence of a causal relationship. Thus, if these studies had been included, they would likely have strengthened NIOSH's conclusions. Some of them were in fact reviewed, but the fact that they provided evidence of health benefits following reduction in occupational exposure was not highlighted as much as it might have been. Since the intervention literature will be reviewed in a later session, I will only mention here a sample, such as a large cohort of visual display unit (VDU) operators, in which the use of new keyboards led to the recovery of some neck, shoulder and upper arm problems over a 7-year follow-up period (Bergqvist 1995). Among other groups of VDU operators, multifaceted ergonomics programs produced benefits such as decreased

OCR for page 152
--> prevalence of symptoms ("probable CTS") and the severity of slowed median nerve conduction velocity (p<0.001 for both 1984 and 1989 data)(1992a). In addition, it has been argued that clinical medicine does not yet provide the necessary diagnostic techniques for all of the soft tissue disorders, especially in their early stages, and that self-reported symptoms or functional impairments may often be more useful indicators than the available physical examination maneuvers and objective tests (e.g., Mackinnon 1994). Both the NIOSH document itself and several panelists here today have addressed the fact that a large majority of the reviewed studies were cross-sectional in design. Rather than repeat those points, I would simply like to note that—once a study has found a positive association between exposure and outcome—the final question is whether there are reasonable alternative explanations for that finding, or whether the most plausible explanation is a causal relationship. Although the ambiguous temporal direction is often cited as a weakness of cross-sectional studies, it is highly unlikely that workers who have already developed musculoskeletal pain would preferentially transfer into jobs with higher physical exposures. On the contrary, it has been shown that workers tend to leave or transfer to lower exposure jobs after MSD onset (Ostlin 1988; Punnett 1996; Silverstein 1987), which would bias toward, rather than away from the null hypothesis. Thus, positive bias by this mechanism is not a plausible alternative explanation for the findings of these studies. There is very little direct evidence as to incidence rates in the general population versus specific groups of workers, both because so few longitudinal studies have been published to date and because in the United States and many other countries there are no (reliable) registries that can provide data on population incidence of MSDs. However, the literature reviewed by NIOSH clearly demonstrates that people occupationally exposed to physical ergonomic stressors, especially at the extremes of magnitude, duration and frequency, have higher prevalences of many MSDs than people without or with lower exposures. In fact, it is impressive that a group of over one dozen studies, utilizing different study designs and data collection protocols and carried out on several continents, have all produced findings consistent with the hypothesis that occupational exposure to physical stressors increases the risk of MSDs. An issue sometimes raised is that these disorders are so common in the general population, because of the many nonoccupational risk factors, that occupational factors cannot account for a large proportion of the musculoskeletal disease burden in general. This is irrelevant to the question at hand. Numerous studies have shown that, after accounting for age, gender, body mass index, smoking, recreational activities, systemic disease, and other individual characteristics, there are still impressive associations between MDSs and occupational exposure to physical stressors. The question of whether or not these factors account for few or many MSDs in the general population has no bearing on the question of whether or not people at work can be protected from preventable risks in their workplaces. If physical ergonomic stressors are shown to increase the risks in their workplaces. If physical ergonomic stressors are shown to increase the risk for otherwise generally healthy working people—as I believe the literature shows—then OSHA has a responsibility to consider formally how it can most effectively pursue primary prevention of these often-disabling disorders.

OCR for page 152
--> In closing, I would like to refer to Adrian Renton's article, "Epidemiology and Causation," in the J Epi Comm Health. She pointed out that—because we define disease in terms of abnormal variation in structure or function—a cause must be understood as a factor that influences a biological process or mechanism to evolve toward that abnormal state. In other words, the search for causation is not merely at the level of biological induction from repeated observations, but for factors that are directly involved in biological mechanisms. We should not overemphasize problems of inferential logic at the expense of understanding the material causes of disease. Epidemiology must be integrated with other basic medical sciences in order to inform rational public health decision-making as it relates to disease prevention. To quote, "The consistent association between a factor and a disease occurring in correct time order in observational studies, where bias has been minimized, suggests a causal or confounded relationship. A strong relationship which persists in the face of strenuous attempts to control confounding in observational studies and through intervention studies shifts the balance towards causation. A knowledge of the mechanisms of pathogenesis of the disease, and the demonstration that a factor will materially influence these mechanisms through the material laws which govern them, adds further to our confidence in causation. Hill's criteria of biological gradient, plausibility, and coherence shift the epidemiologist's attention towards the real material basis of disease causation. Where there is evidence, either from basic medical science or epidemiology of causation, policy makers will consider whether the use of public health technology to modify the distribution of the factor or to identify those exposed might be possible and appropriate. Where there is both of these, properly designed public health programs might certainly be expected to yield some success and controlled trials, where ethical, are likely to be the best way to assess their effectiveness." I congratulate the Academy for its wisdom in bringing together these multiple disciplines, all of which contribute insights into the biological mechanisms involved in the etiology of work-related musculoskeletal disorders. It is precisely through the convergence of these approaches and knowledge bases that we will be able to arrive at an appropriate public health solution to this pressing problem. References Note: Citations are listed below only if they did not appear in the NIOSH review itself. Bergqvist U. (1995). Visual display terminal work—a perspective on long-term changes and discomforts. Inter J Industr Ergonomics 16: 201-209. Bernard B. (1997). Musculoskeletal disorders and workplace factors: A critical review of epidemiologic evidence for work-related musculoskeletal disorders of the neck, upper extremity, and low back. 97-141, Department of Health and Human Services, National Institute of Occupational Safety and Health, Cincinnati, OH.

OCR for page 152
--> Faucett J, Rempel D. (1996). Musculoskeletal symptoms related to video display terminal use: An analysis of objective and subjective exposure estimates . AAOHN Journal 44(1): 33-39. Fine LJ, Silverstein BA, Armstrong TJ, Anderson CA, Sugano DS. (1986). Detection of cumulative trauma disorders of upper extremities in the workplace. J Occ Med 28: 674-678. Gamberale F. (1985). The perception of exertion. Ergonomics 28: 299-308. Kamwendo K, Linton SJ. (1991). A controlled study of the effect of neck school in medical secretaries. Scand J Rehab Med 23: 143-152. Mackinnon SE, Novak CB. (1994). Clinical commentary: Pathogenesis of cumulative trauma disorders. J Hand Surg 19A: 873-883. Ong CN. (1984). VDT work place design and physical fatigue: a case study in Singapore. Ergonomics and Health in Modern Offices, E Grandjean, ed., Taylor & Francis, London. Oxenburgh MS, Rowe SA, Douglas DB. (1985). Repetition strain injury in keyboard operators: Successful management over a two year period. J Occ Health Safety-Aust NZ 1:106-112. Punnett L. (1998). Ergonomic stressors and upper extremity disorders in vehicle manufacturing: Cross-sectional exposure—response trends. Occup Environ Med 55(6): 414-420. Silverstein BA, Armstrong TJ, Flaschner D, Woodland D, Burt S, Fine LJ. (1990). Upper limb ergonomic stressors in selected newspaper jobs: A pilot study. National Institute for Occupational Safety and Health and The University of Michigan, Cincinnati OH. Stellman J, Klitzman S, Gordon GC, Snow BR. (1987). Work environment and the well-being of clerical and VDT workers. J Occ Behav 8: 95-114. Torgén M, Alfredsson L, Köster M, Wiktorin C, Kilbom A. Reproducibility of a questionnaire for assessment of present and past physical work loads. 25th International Congress on Occupational Health, Stockholm, Sweden, 140. Westgaard RH, Jansen T. (1992). Individual and work related factors associated with symptoms of musculoskeletal complaints: I. A quantitative registration system. Brit J Industr Med 49: 147-153.

OCR for page 152
--> Analysis of the Scientific Approach in Assessing Epidemiological Evidence for the Relationship Between Work and Musculoskeletal Disorders Howard M. Sandler, M.D. Richard S. Blume, M.D., M.P.H. Occupational and Environmental Medicine Introduction Musculoskeletal injuries, disorders and/or symptoms affect virtually everyone at various points in their lives. There are various known and suspected causes including trauma, genetics, metabolic disorders, psychogenic factors, lifestyle factors, body habits and occupational as well as non-occupational activities. Workers with MSDs and MSD symptoms suffer significant morbidity and generate costs in billions of dollars annually in workers compensation and lost productivity. Increasing interest about the relationship between work and musculoskeletal disorders or injuries (MSDs) over the past few decades has generated significant research in biomechanics, ergonomics and the epidemiology of MSDs. Numerous studies have been published examining the potential work—MSDs relationship as well as non-work factors. Several reviews of the epidemiologic data, including those employing meta-analytic techniques have yielded conflicting conclusions about the work—MSD association for specific disorders, e.g., carpal tunnel syndrome and low back pain. The NAS workshop session on physical factor epidemiology will endeavor to assess the overall contribution of published epidemiological studies associating physical stressors with musculoskeletal disorders. Specifically, the "quality of the science" will be examined. A focus of the session will involve assessment of the National Institute for Occupational Safety and Health (NIOSH) review of work-related MSD epidemiologic evidence, released by the Institute in July, 1997 including methods for evidence weighting, completeness of the literature base, MSD incidence in working and general populations and quality of the literature base. Methodology for Causal Determination of Work and MSDs Causal association from an epidemiological standpoint as described by Hill, Rothman and others, requires careful critical analysis of individual studies comprising the body of literature, determination of "causal criteria", and development and utilization of a weighting scheme to objectively assess the evidence for associating exposure and effect. The NIOSH document has identified, assembled, and critiqued the epidemiological literature base regarding work, specifically physical factors and MSDs. And given the constraints of the literature as acknowledged by NIOSH, the document did not assess specific dose-relationships, relative contributions of work versus non-work factors, and thresholds (e.g., "triggers") among other items of considerable import. These important issues may or may not be able to be evaluated

OCR for page 152
--> given the state of the present epidemiological database. The NIOSH document also examined a limited number of specific MSDs in certain anatomical locations, e.g., the back, hands/wrists, elbows and neck; carpal tunnel syndrome, epicondylitis, shoulder tendonitis, low back pain, etc. Study Critique NIOSH chose four criteria to qualitatively critique each study: participation rate, health outcome definition, blinding, and exposure status. Each criterion was further specified but limited in their characterization and requirements, e.g., health outcome was limited to symptoms and physical examination. MSDs such as carpal tunnel syndrome may require additional evidence such as electrodiagnostic studies for diagnosis. The exclusion of electrodiagnostic results from assessment of health outcome in screening studies may therefore significantly limit the reliability of associations. This has been demonstrated in recent studies and the potential impact of overstated health outcome presence may be substantial. Other important factors in study methodology such as confounding were not systematically addressed in the NIOSH document. The selection of criteria for determining study inclusion and weighting in a formal assessment is a key element and one which requires validation. The NIOSH document does not provide a basis for the limited critique approach or describe epidemiologic sources validating the approach. The NIOSH document states that it applied the greatest qualitative weight to those studies which met the four criteria considered in the NIOSH document. The methodology used to perform this weighting is not specified in the NIOSH document. In addition, application of the four critique criteria demonstrated significant deficiencies in the body of literature cited in the NIOSH document. This significant observation deserves discussion beyond that provided in the document. Despite the known limitations of certain epidemiologic study designs in causal analysis, e.g., cross-sectional, such studies were included. The importance of using a tested and validated critique methodology cannot be underscored. For example, would the criteria approach in the NIOSH document yield expected results if applied to a more well-recognized exposure-effect relationship, e.g., smoking and chronic obstructive lung disease? Without such confirmation, the validity of the methodology remains to be established. Furthermore, the actual weighting process used in applying the criteria to derive causal associations needs to be fully defined. It is important to note that other scientific investigators utilizing comprehensive critique criteria, e.g., Stock, could as a result employ but a few studies in the epidemiologic assessment of work and carpal tunnel syndrome while the NIOSH document employed a much greater number of studies. Weight of Evidence Approach Based upon available information provided in the document, it appears that the NIOSH document's overall approach in weighting the causal evidence is as follows: Selected studies are critiqued using a limited quality criteria scheme; Using undefined methods, the selected studies are weighted for quality based upon the critique;

OCR for page 152
--> Selected causal criteria are then applied to the cited literature and studies meeting the criteria are identified; again, the weighting system used to apply the results of the causal criteria is undefined; Finally, exposure-effect, evidence or "strength" is classified according to a rating scheme apparently unique to the NIOSH document, with scores including "-," "0/+," "++, " and ''+++." The NIOSH document employed two components in assessing the work physical factors—MSD relationship once the studies were critiqued and selected: causal criteria and "evidence classification categories". Six causal criteria were employed by NIOSH. The criteria chosen are consistent with published criteria and employ the areas principally discussed throughout the scientific literature on causal association. Limitations are noted in the NIOSH document's utilization of these criteria. For example under temporality, the one criterion which absolutely has to be met to establish causation, i.e., the exposure must precede the effect, the NIOSH document recognizes that cross-sectional studies cannot satisfy this criteria. However, the MSD epidemiological literature employed by the NIOSH document primarily consists of cross-sectional studies. Nonetheless, the NIOSH document states that from such studies "reasonable assumptions" regarding temporality can be made and specifically cites two authors who have published studies which are said to support the document's assumption. For example, Rothman (1996) was reported to state that some assumptions about timing for exposure and disease must be made. However, there is no indication that worker interviews or work histories are reliable in terms of predicting presence or absence of pre-existing conditions or symptoms. Disorders such as tendonitis do come and go. Recall of prior medical history is variable as has been noted in a variety of other health issues, such as recall of spontaneous abortion, which is undoubtedly a more significant health event. Citing Kleinbaum (1991), the NIOSH document states that with additional information such as biomechanical findings, temporality can be established. Unfortunately there is no scientific evidence offered to evaluate the proposal that individuals with underlying hand/wrist tendonitis will not seek work which may require significant use of the hand and wrist. This assumption may or may not be operative. Workers with certain skill sets can also be assumed to be qualified in, and primarily gain entry to the same type of job/industry. A critical consideration which affects the basis for any conclusions derived in the NIOSH document is its use of a four-point category scheme to classify "evidence". To our knowledge this approach has not been previously published, defined or validated in other scientific settings. Unfortunately, the NIOSH document does not define the methodology it employed to classify each physical factor or physical factor combination and associated health outcomes. Without a clearly defined epidemiologic basis for this approach and a defined methodology for applying the critique and causal criteria to derive the ratings, it is not possible to determine whether the ratings are appropriate and scientifically supported. As in all scientific endeavors, replication of the results using the same methodology is crucial to assure that the conclusions are indeed well-founded. It is not possible to replicate the findings of the NIOSH document as the methods used in the document are not fully defined. It is critical to the discussion at hand to be able to

OCR for page 152
--> replicate the NIOSH document's evidence determination for work—MSD causation. However, without such stated methodology, the process becomes merely a "black box" from which certain data goes in and a classification rating emerges from the other end. The use of a four-point scale is to some extent quantitative. Interestingly, the NIOSH document had referred earlier in the analysis to qualitative weighting, but does not offer any additional explanation. Further, in describing the ratings, the NIOSH document employs the terms "very likely," "convincing," "insufficient" and "adequate." Again, an epidemiologic basis for applying these terms to the ratings is not provided. No definitions or triggers are supplied to describe when those terms apply, nor how those terms satisfy the rating structure. To draw an analogy, similar rating schemes are used in medicine, for example, to grade reflexes or muscle strength. There is significant variability between examiners when evaluating reflexes and muscle strength, as a result of the lack of well-established definitions and validation that the system indeed is accurate. Various validated tools are now increasingly employed in such evaluations. Other well-defined rating systems such as the ILO system for describing and rating pneumoconiosis, the latter also using a proficiency testing program have nevertheless been shown to be subject to significant inter-reader variability. These observations beg the question, what would be the reproducibility of the approach employed in the NIOSH document, even if it was fully described and therefore could be independently replicated? Work-Related and Non-work-Related MSD Incidence One question posed to this workshop session concerns the incidence of MSDs in workers versus that in the general population. Incidence generally cannot be derived from cross-sectional studies. The overwhelming number of studies use in the NIOSH document's assessment are cross-sectional in design. Additionally, there is a dearth of data in the scientific literature from which to determine the incidence of MSDs in the general population. Various studies cited in the NIOSH document employed workers with different exposures such as clerical workers to serve as controls. Lastly, confounding variables were not consistently considered in the cited studies to assure that other factors would not account for any differences found. Thus, determination of incidence and incidence differences is hampered by such factors. SUMMARY It is imperative that specific relationships between work physical factors and potentially other factors, and the development and aggravation of MSDs and related symptoms be determined. Such information is vital to the effective protection of worker health and safety. The question is not just, are MSDs and/or their symptoms related to work? But, more importantly, how are work physical factors, psychosocial factors and other nonwork factors related to MSDs development and aggravation? This includes dose-response, specific factor or factor combination relationships and safe levels, as well as other areas of inquiry. An assessment of the NIOSH document's approach to the weight of the epidemiologic evidence to determine the causal associations between various work factors and MSDs is a key element in this scientific undertaking. We have been provided a valuable exercise in assessing the NIOSH

OCR for page 152
--> document's findings regarding specific MSDs including those of the back, neck and upper extremity. Aspects of the approach described in the NIOSH document and concerns regarding the resulting weight of evidence findings inhibit use of the document in causal determination. Fully defined and validated scientific methodology for epidemiologic literature critique, classification and weighting should provide a valid assessment of the present state of the art for various physical work factors and MSD development. Additionally, NIOSH, NIOSH-funded, and various other studies currently underway by a vast array of academic, government, labor and business researchers should start to provide answers to specific causal and prevention questions which are so vital in establishing scientifically-based regulatory and enforcement approaches. We are pleased to have had the opportunity to participate in this scientific process to attempt to analyze the relationships between work and MSDs. Establishing and understanding the potential causal associations in this area is essential in the process to prevent and manage work-related musculoskeletal disorders.

OCR for page 152
--> Summary Comments David H. Wegman, MD, MSc Professor and Chair, Department of Work Environment,University of Massachusetts Lowell, Lowell, MA 01854 NIOSH staff accepted an important charge and a sizable task when they undertook a comprehensive review of musculoskeletal disorders and workplace factors. They appear to have spent substantial effort in determining an appropriate approach and carrying out that approach in a transparent, public manner with continuous internal and external review. As a member of today's panel I have been asked to evaluate several central aspects of the process and the final product. Regarding whether NIOSH has missed or overlooked any important body of epidemiological evidence in its review, this will inevitably be a matter of opinion. Their approach to gathering information was systematic and comprehensive using electronic literature searches supplemented by suggestions from a large number of subject-matter experts. I am not aware of any omissions. In an effort to evaluate comprehensiveness I compared the NIOSH report with a literature review I co-authored some years ago [1]. Articles not included by NIOSH that we considered had been appropriately excluded and the risk factor approach undertaken by NIOSH was superior to the exposure considerations used in our review. Since study methodology was not the sole criterion used to select studies to be given greatest weight in determining work-relatedness, there is no way to identify the study "methods" that have been heavily weighted in the NIOSH assessment. To select studies, NIOSH developed a comprehensive approach that combined several factors to serve as the basis for judgements about work-relatedness. The quality of the studies that were most heavily weighted was generally quite high because they met the multiple criteria set out by NIOSH for weighting: high participation rates, appropriate "blinding" of investigators, health outcomes defined by symptoms and physical exam and independent exposure measures. Those that met all criteria were weighted most heavily, but thankfully, NIOSH did not eliminate all other studies. Rather studies that met a minimum threshold were still considered, with appropriate lesser weighting. In its weighting, NIOSH paid attention to explicit criteria for causality applying a standard definition of causality for epidemiologic studies. Possibly unique in such an effort, and a substantial contribution to any review of work and musculoskeletal disorders are the extensive tables in the body and the appendix. These tables summarize each study's adherence to the four criteria as well as provide a comprehensive, readable summary of many detailed aspects of each study, thus allowing the reader to develop independent judgements. Furthermore, the authors have developed an excellent graphical presentation format that permits rapid review of the large number of studies. It is unlikely that any reasonable reassessment of the quality of the studies reviewed would substantially alter the interpretation of the epidemiological evidence that certain physical stressors in the workplace increase risk of acquiring certain musculoskeletal disorders. While it

OCR for page 152
--> is clear that judgement is involved, the demands made of each study are quite severe and the summary evaluations that are presented are quite conservative. Serious disagreements in how these data should be interpreted may come from those who question the legitimate use of epidemiology in any forum such as this, but epidemiologists can be expected to see these judgements as soundly based and determined on reasonable and well articulated criteria. The scheme developed to evaluate strength of evidence is clear and consistent with similar efforts by agencies such as the International Agency for Research on Cancer or organizations such as the American Conference of Governmental Industrial Hygienists. NIOSH appropriately does not directly address the question of incidence in the general population versus specific groups of workers. There are a variety of data sources that might be used to estimate population prevalence (for example, NHIS, NHANES, and other surveys from NCHS, SSA Disability Reports, data accumulated as byproducts from health insurance) but none provide incidence data and none serve as a useful comparison for NIOSH's evaluation. NIOSH is careful to provide explicit information on the comparison populations (for example no or low exposure groups in cohort studies, unaffected controls in case control studies). Since the assessment examined specific physical factors, several of which may have been included in a particular study, the reviewers take care to determine that a proper unexposed or low exposed group is defined in terms of the specific exposure factor studied in cohort or cross-sectional studies. In case control studies the assessments made of both cases and controls are with regard to the specific factor under study. I want to answer a question not asked: Is the approach taken by NIOSH to evaluate and summarize the epidemiologic literature on work-related musculoskeletal disorders of the neck, upper extremity, and low back appropriate and the most desirable? One might ask why NIOSH did not attempt several meta-analyses rather than use their more qualitative review? Meta-analysis is not appropriate when the question under study is as broad as the one NIOSH addressed. In my judgement Shapiro provides the answer which, in his words is: "I question whether quantitative methods can ever be as thoroughgoing, probing and informative as qualitative methods."[2]. In addition, in his final comments from a symposium on "Meta-analysis of Observational Studies" he summarizes apparent agreement with two other leading epidemiologists that synthetic meta-analyses (efforts to arrive at a single risk estimate by combining multiple studies) are of questionable validity and more likely to be misleading than helpful. One might also ask why did NIOSH not exclude case series from their review? The answer is provided by Checkoway, et al. in their classic textbook [3]. In discussing causal inference, he says: "Attempts to codify guidelines for assessing research quality are invariably detrimental to the practice and application of epidemiologic methods." He goes on to illustrate, through a hypothetical example, how a case series can easily provide information at least as important as a well-designed epidemiologic study. There is no "correct" way to carry out a literature review particularly with as large a scope as the one undertaken by NIOSH. The authors of the NIOSH report are to be commended for developing a methodology that is reasonable, understandable, clearly presented, open and conservative. It is hard to imagine a more effective way to summarize this literature.

OCR for page 152
--> Notes:     Hagberg M, Wegman DH. Prevalence Rates and Odds Ratios of Shoulder Neck Diseases in Different Occupational Groups. British Journal of Industrial Medicine, 44:602-610, 1987.     Shapiro S. Is there is or is there ain't no baby?: Dr. Shapiro replies to Drs. Pettiti and Greenland. Amer J Epid (1994) 14:788-791 (see also preceding three articles by Shapiro, Pettiti and Greenland in the same issue)     Checkoway H, Pearce NE, Crawford-Brown DJ. Research Methods in Occupational Epidemiology. Oxford University Press, New York, 1989 (pp 13-14).