A Cross-Sectional Study of the Relationship Between Repetitive Work and Upper Extremity Musculoskeletal Disorders

Alfred Franzblau, Wendi A. Latko, Thomas J. Armstrong,
Sheryl S. Ulin, Robert A. Werner,James W. Albers,
James A. Foulke, Gary D. Herrin, Randy A Rabourn

The University of Michigan
Ann Arbor, Michigan 48109

ABSTRACT

A cross sectional study was conducted in which workers' exposure to repetition and other physical stressors was quantified using a new method, and compared to the prevalence of various upper limb disorders. Jobs were selected based on levels of hand repetition. Standardized medical evaluations were performed on all participating workers, and included a self-administered questionnaire, physical exam, and limited electrodiagnostic studies (EDS). Analyses focused on evaluating exposure-response relationships with adjustment for pertinent covariates.

Repetitiveness of work was found to be significantly associated with prevalence of discomfort in the wrist, hand, or fingers, tendinitis, and symptoms consistent with carpal tunnel syndrome as indicated on a hand diagram. There was no statistically significant relationship between repetitiveness of work and median mononeuropathy, but there was a borderline significant positive trend between hand repetition and carpal tunnel syndrome defined by EDS and hand diagram scores.

INTRODUCTION

Numerous studies have shown a relationship between various physical stressors and upper extremity musculoskeletal disorders (UEMSDs). It is generally agreed that a dose-response relationship exists between exposure to these stressors and the prevalence or incidence rates of these disorders (1). Most of these epidemiological studies, however, have only examined exposures in a binary classification, either present/absent or low/high (Figure 5.1). Thus, there is evidence to support the extremes of a dose-response curve for the various stressors, but there is relatively little information concerning the increased risk associated with intermediate exposure levels. Part of the reason for this lack of information is the difficulty in quantifying exposure to these stressors.

Repetition has been among the most widely studied of these stressors, yet no universal definition or quantification technique exists for it (2). The work presented in this paper was designed with the primary goal of providing more information on the shape of the dose-response curve for hand repetition with respect to specific medical outcomes.

METHODS AND RESULTS

A double-blind, cross-sectional epidemiological study was conducted to determine the relationship between exposure to physical stressors and prevalence of UEMSDs among industrial workers. Repetition was the stressor of primary interest, although other stressors were treated as covariates.

The selection of jobs for inclusion in this study consisted of two subtasks: 1) preliminary job selection/classification and 2) formal analysis and final job classification. The goal of job selection was to obtain examples of jobs in potential study plants encompassing three distinct levels of repetition: low, medium, and high. In order for a company to be eligible for the study, it was necessary that jobs representing three levels of repetition were present at the company, with at least thirty eligible workers per repetition level. Workers were eligible if they had been in their current job for at least 6 months prior to the study date. In addition, plant management had to agree to allow the medical evaluations to occur on company time, and during normal work hours to enhance worker participation (these required about 75 minutes per worker).

Quantification of the exposure levels for repetition and the other stressors was performed using an observational rating method developed for this study (2). The rating method utilizes a series of 10 centimeter visual-analog scales which range from 0, corresponding to no stress, to 10, corresponding to the most possible stress. In this method, repetition is defined in terms of hand activity, or how busy the hands are during the work cycle. Ratings of repetition take into account two factors: 1) amount of recovery time within the cycle, and 2) how fast the hands are moving.

Three manufacturing facilities were ultimately included in the study; they represented office furniture (OF), spark plug (SP), and industrial container (IC) manufacturing. Thirty-nine jobs were selected from these 3 facilities for inclusion in the study, and a total of 52 ergonomic variables were quantified for each of the jobs. The formal ratings were performed using a modified nominal group technique (3). The rating team consisted of 4 university faculty and research staff members who were experienced in ergonomic analysis in general, and this technique in particular. The job rating method used in this study has undergone both reliability and validity testing (4,5), with satisfactory results for both. Raters were unaware of medical survey results of workers in jobs rated for this study.

For the purposes of stratification in this study, three ranges of repetition were defined: low, medium, and high. These categories were defined by a strict division of the scale into thirds: 0 - 3.3 = low (n = 15 jobs), 3.3 - 6.6 = medium (n = 12 jobs), and above 6.6 = high (n = 12 jobs). The average of the four raters was used for all analyses.

After specific jobs were selected based on the above criteria, subjects were recruited for the medical survey from among the active workers performing those jobs. The only constraint on worker participation was that they had to have been performing the job of interest for at least the six consecutive months prior to the study date. All participants provided written informed consent, and all medical evaluations were performed on company time, during normal work hours.

The medical survey included: a questionnaire with a hand diagram, limited physical examination of the upper extremities, limited EDS at both wrists, and general anthropometry measurements. All clinical procedures were performed by appropriately trained health professionals. Details of the questionnaire, physical examination procedures, and EDS protocol are described elsewhere (6,7). Anthropometric data included height, weight, finger circumference and length (digit 2), wrist width and depth, and triceps skinfold thickness. Each member of the medical survey team was blinded to the medical and job-related data collected by other members of the study team.

Several specific health outcome measures were modeled in this study: symptoms in the wrists, hands or fingers; tendinitis; and, carpal tunnel syndrome (CTS). Tendinitis was defined as symptoms (e.g., pain, stiffness, burning, tightness, etc.,) plus physical examination findings consistent with tendinitis in the elbow, forearm, wrist, hand, or fingers (e.g., local tenderness, or pain with resisted motion). Three different case definitions of CTS were analyzed: symptoms consistent with CTS indicated by a "Classic" or "Probable" hand diagram score; median mononeuropathy ("MM5"-defined by a difference in peak latency of at least 0.5 milliseconds between ipsilateral ulnar and median nerves); and CTS defined by symptoms and median mononeuropathy ("Classic" or "Probable" hand diagram score and MM5 in the same limb).

Data were analyzed following the approach outlined by Hales et al. (8). There were a total of 109 covariates: 10 anthropometry parameters, 25 medical history parameters, 5 demographic parameters, 13 psychosocial parameters, 4 tobacco use parameters, and 52 ergonomic parameters. Only models for the dominant hands are presented here. Subjects who reported a history of physician­diagnosed diabetes on the questionnaire were excluded from analyses which included electrophysiologic parameters (n=16). Ergonomic parameters were modeled as continuous effects.

Four hundred thirty-eight workers were employed in the selected jobs at the time of the study and met the inclusion criteria (i.e. 6 month job tenure), and 352 (80%) participated in the study. Table 1 summarizes the demographic characteristics by plant and repetition category.

Table 2 shows the general linear trends for the five health outcome measures in relation to hand repetition. The linear trends were highly significant for wrist/hand/finger discomfort, hand diagrams suggestive of CTS, and tendinitis. There was no linear trend for MM5, but there was a borderline trend for CTS defined by hand diagrams and MM5.

The final logistic regression model using wrist/hand/finger discomfort as the outcome measure was significant (n = 351; -log likelihood = -208). Significant covariates included: history of CTS (odds ratio (OR) = 2.11, 95% confidence interval (CI) = 1.00-4.47); history of tendinitis (OR = 1.98, CI = 1.07-3.63); gender (females = 1, males = 0; OR = 2.07, CI = 1.27-3.35); and, hand repetition (OR = 1.17 per unit of repetition, CI = 1.06-1.29). Age was not significant in this model (OR = 1.00, CI = 0.97-1.02).

The final logistic regression model for tendinitis was also significant (n = 348; -log likelihood = -101). History of soft tissue disease (i.e. tendinitis, epicondylitis, or rotator cuff syndrome; OR = 2.66, CI = 1.27-5.55), triceps skinfold thickness (OR = 1.04 per millimeter, CI = 1.01-1.07), and hand repetition (OR = 1.21 per unit of repetition, CI = 1.03-1.44) were significant in the final regression model for tendinitis. Age (OR = 1.03, CI = 0.99-1.07) was borderline significant in the model, and gender (OR = 0.70, CI = 0.27-1.85) was not a significant covariate in this model.

Wrist ratio (OR = 2.59, CI = 1.35-4.96) and hand repetition (OR = 1.16 per unit of hand repetition, CI = 1.00-1.34) were significant in the final model for CTS based on hand diagram scores (n = 351; -log likelihood = -127). Wrist ratio was defined as the ratio of wrist depth to width and was modeled as a binary variable, with ratios at or below the 75th percentile (0.73) of the study population modeled as 0 and ratios above the 75th percentile (>0.73) modeled as 1. Age (OR = 1.00, CI = 0.97-1.03) and gender (OR = 1.88, CI = 0.96-3.70) made borderline significant contributions to the model.

The final model for CTS based solely on MM5 was significant (n = 336, -log likelihood = -164). Significant covariates in the logistic model for MM5 included age (OR = 1.03, CI = 1.00-1.06), gender (OR = 0.54, CI = 0.31-0.97), body mass index (OR = 1.11 per kilogram/meter squared, CI = 1.06-1.16) and wrist ratio (OR = 2.77, CI = 1.56-4.92). None of the ergonomic covariates made a significant contribution to this model.

The final model for CTS based on hand diagram scores and MM5 was also significant (n = 336; -log likelihood = -69). Gender (OR = 0.86, CI = 0.32-2.31) and age (OR = 1.02, CI = 0.97-1.06) did not contribute significantly to this model. However, wrist ratio (OR = 2.53, CI = 0.97-6.57) and hand repetition (OR = 1.22 per unit of repetition, CI = 0.98-1.53) were borderline significant in the model.

CONCLUSIONS AND RECOMMENDATIONS

In this study repetition was found to be associated strongly with discomfort in the upper limbs, tendinitis, and symptoms consistent with CTS as reported on a hand diagram. The model which examined MM5 as an outcome did not show a statistically significant relationship to ergonomic exposures. Hand repetition was borderline significant for logistic and linear models combining MM5 and hand diagram scores for defining CTS. The ORs for hand repetition ranged from 1.16 to 1.22 per unit of repetition. If the overall OR per unit of repetition is assumed to be 1.20, then a change from 'low' to 'medium' (i.e., 3.0 units of repetition) yields an OR = 1.73, and from 'low' to 'high' (i.e., 5.6 units of repetition) results in an OR = 2.78.

Further analyses were performed using slight variations in exposure and health outcome criteria in order to test the robustness of the associations found. For example, expanding the hand diagram criterion for CTS to also include scores of "possible", or restricting the criterion to include only scores of "classic" yields similar results. When the definition of CTS is changed to reflect a hand diagram score of "classic" or "probable" and median mononeuropathy based on 0.8 millisecond threshold difference (MM8), repetition becomes more significant, with little impact on the other parameters in the model. When CTS is modeled as reported numbness, tingling, burning, or pain and MM5, repetition is again significant, while wrist ratio is not. The similarity between models when employing alternative dependent variables suggests that the findings are robust. Similarly, modeling repetition as two or three categories (low/high or low/medium/high) yields similar odds ratios to those reported in the above models, again suggesting that the findings are robust.

No other ergonomic stresses (e.g. force, posture) were found to be associated with health outcome measures. This is not surprising. Job selection was stratified only on repetition to insure a wide range of this exposure. No effort was made to insure a wide spectrum for other stressors.

There are several limitations to this study. This was a cross-sectional study. There is also the possibility of a survivor effect or healthy worker effect. Exposure to ergonomic stressors was evaluated for one representative worker in each job classification. Although the investigators attempted to ascertain that the worker evaluated was representative, it is possible that exposures varied considerably between workers, although it is unlikely that repetition rates would vary significantly because most of the jobs included in this study were machine paced, performed in a workcell (i.e. group paced), or had strict production standards. Stresses such as posture, however, may have had considerable variability due to individual work style and anthropometry. Subjects in this study were limited to employees at companies which were receptive to participation in the study; there is a possibility that the management attitude and culture at these companies may be different than at other companies, thus biasing the results. Finally, for some outcomes (e.g., CTS defined by hand diagrams and MM5) this study lacked power (only 19 subjects met this case definition), which probably explains the borderline relationship between repetition and this outcome.

The results of this study indicate there is a definite exposure-response relationship for certain measures of UEMSDs and hand repetition. There is a definite need for future studies examining exposure-response relationships between a wide range of ergonomic exposures and UEMSDs among workers.

ACKNOWLEDGMENTS

This study was supported by grant number 1 R01 OH02941-01 from the National Institute for Occupational Safety and Health. We would like to thank the workers and managers at the plants who made this study possible, and also the many members of the study teams who worked on this project.

REFERENCES

1. Armstrong, T.J., P. Buckle, L.J. Fine, M. Hagberg, B. Jonsson, A. Kilbom, I.A. Kuorinka, B.A. Silverstein, G. Sjogaard, and E.R. Viikari-Juntura: A conceptual model for work-related neck and upper-limb musculoskeletal disorders. Scandinavian Journal of Work & Environmental Health. 19:73-84 (1993).

2. Latko, W. A., T.J. Armstrong, J.A. Foulke, G.D. Herrin, R.A. Rabourn, and S.S. Ulin: Development and evaluation of an observational method for assessing repetition in hand tasks. American Industrial Hygiene Association Journal. 58(4):278-285. (1997).

3. Gustafson, D. H., R.K. Shulka, A. Delbecq, and G.W. Walster: A comparative study of differences in subjective likelihood estimates made by individuals, interacting groups, delphi groups, and nominal groups. Organizational Behavior and Human Performance. 9:280-291. (1973)

4. Latko, W. A.: "Comparison of observational and instrumental based measurements of repetition, force, and wrist posture in manual work (Chapter 4)". University of Michigan [unpublished doctoral dissertation]. (1997).

5. Latko, W. A.: "Evaluation of interrater reliability and test-retest reliability of observational ratings of physical stressors in manual work (Chapter 3)". University of Michigan [unpublished doctoral dissertation]. (1997).

6. Franzblau, A., R.A. Werner, J.W. Albers, C.L. Grant, D. Olinski, and E. Johnston: Workplace Surveillance for Carpal Tunnel Syndrome Using Hand Diagrams. Journal of Occupational Rehabilitation. 4(4):185-198 (1994).

7. Franzblau, A., R. Werner, J. Valle, and E. Johnston: Workplace surveillance for Carpal Tunnel Syndrome: A comparison of methods. Journal of Occupational Rehabilitation. 3(1):1-13 (1993).

8. Hales, T. R., S.L. Sauter, M.R. Peterson, L.J. Fine, V. Putz-Anderson, L.R. Schleifer, T.T. Ochs, and B.P. Bernard: Musculoskeletal disorders among visual display terminal users in a telecommunications company. Ergonomics. 37(10):1603-1621 (1994).


Table 1: Age and Gender Distributions By Plant and Repetition Category

* statistically significant difference between levels at p<0.05.

Table 2: Linear Trend of Outcomes by Repetition Category