Carpal tunnel syndrome (CTS) is the most commonly reported entrapment neuropathy of the upper limb.¹² Its effects on workers can be chronically debilitating both in the workplace and in their personal lives.³ The economic consequences are felt by workers, employers and insurers. CTS is among the greatest drivers of workers’ compensation costs, lost time, lost productivity and disability.⁴⁵ Many of these disorders could be prevented by identifying hazardous jobs and redesigning job tasks, tools and workstations to reduce physical stressors. A conceptual model describing the complex interaction of work requirements (exposure) leading to the doses (eg, physical loads) and responses (eg, tissue thickening) that can result in upper limb musculoskeletal disorders when tolerance (capacity) is exceeded was an organising theme for this study.⁶ Research has demonstrated associations between repetitive and forceful hand activity, non-neutral wrist postures and hand-arm vibration at work and CTS.⁷ Individual risk factors for CTS include age, gender, body mass index (BMI), pregnancy and other medical conditions, such as diabetes and thyroid disorders.⁸ A comprehensive review of the literature on work-related musculoskeletal disorders by NIOSH noted that exposure-response relationships were difficult to determine across studies due to a wide range of exposure assessment methods, different health outcome definitions and a restricted range of exposures in the jobs studied and other methodological issues.⁷ A more recent review suggests that there is still little guidance on the level of exposure to physical work demands that is associated with CTS.⁹ Limitations of past studies, including a lack of precision of estimation of potentially hazardous exposures, a lack of individual exposure assessment, a limited range of exposure variation, or cross-sectional study design have hindered attempts at exposure-response analysis.

The primary aim of this study is to quantify the relationship between recognised workplace physical factors, particularly repetitive or prolonged hand activity and forceful exertion, and CTS, while taking into account individual factors as well as psychosocial/work organisational factors such as job strain.

Potential study plants were identified from those who had volunteered to participate following announcement of the study in the Federal Register and communication with occupational safety and health professionals on the National Occupational Research Agenda (NORA I) Musculoskeletal Team. Workplaces were selected to represent a range of the two primary exposures of interest: repetitive or prolonged hand activity, and forceful exertion. During initial walkthrough evaluations, ergonomists used checklists to determine exposure categories of jobs. Selected workplaces had jobs that represented at least three of six exposure categories that combined low, medium, or high hand activity level (HAL) with low or high force. Additional workplace selection criteria included at least 100 employees and no anticipated lay-offs or major changes in production or reorganisation. Three worksites were selected: a hospital, a school bus manufacturing plant and an engine assembly plant. All full-time workers with at least 3 months on the job in selected departments were invited to participate in the study. Because of the detailed exposure assessment required, we excluded jobs with more than four different tasks, jobs that were too varied to characterise (eg, most maintenance jobs), or that could not easily be videotaped for subsequent analysis (eg, fork lift drivers). Prevalent CTS cases and those with polyneuropathy that were identified during the baseline cross-sectional study were excluded from the longitudinal study.¹⁰ The longitudinal study presented here included two follow-up exposure and health data collection visits 1 and 2 years after the baseline assessment. Interim visits to evaluate job changes, if they occurred, were made 6 months following the baseline and first annual full data collection visits. The prospective study population included 347 participants (figure 1). At the hospital, 120 workers from central and sterile supply, laboratory, pharmacy, engineering, surgical, kitchen, laundry and administrative support participated. In the engine plant, 70 workers in assembly, machining, quality control and manufacturing support were included. In the bus plant, 157 workers from assembly, sub-assembly, conveyor, machine fabrication, electrical, upholstery, engineering and transport jobs were included. Follow-up participation rates were 83% for at least 1 year of follow-up data collection beyond the baseline and 77% for both years. Fifty-six workers did not complete the study because they left employment, were absent from work or unavailable during the weeks of the annual data collection visits, or declined follow-up participation. Following approval by the NIOSH Institutional Review Board and written informed consent, each study participant underwent detailed job exposure assessment including direct observation and videotaping of job tasks, physical examination of the upper limbs, and nerve conduction testing of the median and ulnar nerves across the wrist. A detailed questionnaire was administered to collect information on work history, individual factors, work environment, physical activities outside of work and medical history, as well as neck, shoulder, arm and hand symptoms, including a hand diagram documenting type and location of symptoms.^11–15 Height and weight were measured. Participation in the study took place during working hours and was not compensated. Study participants were notified of their own health assessment results, interpretation and recommendations by mail. Employers were given group results that were not individually identifiable. The exposure assessment team was blinded to study participants’ health information, and the health assessment team was blinded to exposure information.

Demographics and pertinent prior medical history by case status are shown in table 1. Obesity and prior non-dominant hand CTS were significantly associated with incident-dominant hand CTS, while age and gender were not. Smoking, thyroid disease and rheumatoid arthritis were higher among cases, but the differences were not statistically significant. There were no diabetics among cases, probably because we excluded those with polyneuropathy and baseline CTS cases from the prospective analysis. We observed 29 incident cases of dominant hand CTS over the 2-year follow-up period of the study, an incidence rate of 5.11 per 100 person-years.

As shown in table 2, job exposures significantly associated with CTS included worker peak and average RPE, force matching, forceful exertions per second, percent time in forceful exertion, the ACGIH TLV for HAL as a ratio (but not the cutpoints at the TLV and AL) and job strain.

Proportional hazards multivariate modelling resulted in final models (table 3) that included the percent of time working with forceful exertion and the ACGIH TLV for HAL as a ratio. Obesity was a significant predictor of CTS in both models. Working in a high-strain job was not significantly associated with CTS in the first quantitative job exposure model, but it was in the TLV ratio model (table 3). Obesity was also a confounder of the association between the percent time in significant exertion and CTS. With obesity in the model, the estimate for forceful exertion increased by 15%. The test of interaction between obesity and forceful exertion was not statistically significant (p=0.07). A proportional hazards model with the ACGIH TLV for HAL (table 3) shows an increased HR for CTS of 1.4 for each unit increase in the TLV for HAL as a ratio (model 2). Each unit increase represents a multiple of the TLV (eg, TLR=2 is twice the TLV).

Figure 2 shows the exposure-response relationship between percent time in forceful exertion and CTS, adjusted for obesity. The association between the percent time spent working with forceful exertion and CTS appeared to be linear (figure 2). BMI was not significantly associated with CTS as a continuous variable, but obesity was when using the standard cutoff of 30 kg/m² (table 2).

Physical job demands represented by the percent of time spent in forceful exertion or the ACGIH TLV for HAL as a ratio, adjusted for obesity and job strain, predicted dominant hand CTS. Recent studies have shown exposure-response relationships between ratings-based metrics, such as the ACGIH TLV for HAL among industrial workers²⁶²⁷; while an earlier study did not observe these relationships.²⁸²⁹ Garg et al²⁶ suggested the use of the TLV as a ratio if other researchers confirmed their findings that this is a useful metric; our study supports their suggestion. We are unaware of other prospective studies of CTS that reported exposure-response relationships with quantitative exposure measures such as percent time in forceful exertion based on detailed time studies of job tasks. Obesity also predicted CTS in our multivariate models. Proposed mechanisms for the impact of obesity on the carpal tunnel have not been fully developed, but some researchers suggest fatty tissue and/or swelling in the carpal tunnel impinges on the median nerve.³⁰ Obesity resulted in positive confounding of the association between physical job demands and CTS in this study, while in our cross-sectional study, obesity had a significant interaction with physical job demands.¹⁰ In both studies, the end result is that if someone is both obese and has high physical job demands, the risk is increased beyond what it would be if only one of these risk factors were present. The association between high job strain and CTS was significant in the TLV ratio model but not in the quantitative exposure model based on percent time in forceful exertion. High psychological demand/low control jobs, as defined by Karasek to represent job strain, also were more likely to have high physical demands as represented by the percent of time spent working with high force in our study, so our data cannot adequately address these issues.

This prospective study of dominant hand CTS has identified an exposure-response relationship with physical job demands, controlling for individual and work organisational factors. The risk for CTS increases with an increased amount of time spent in forceful exertion. These results and prior results from our cross-sectional analysis suggest that force may be the primary job exposure risk factor for CTS. A reduction in the amount of time spent in forceful exertion and the intensity of the required force of job tasks may reduce the occurrence of CTS. Obesity also increased the risk for CTS in this study as in our prior cross-sectional analysis. A reduction in obesity may also reduce the risk for CTS. Implications for job strain are not as clear. It is feasible to suggest that some of the jobs in this study that had high physical demands might also have offered workers little control, for example, over when they could take breaks and the pace of their work, and that these organisational factors may contribute to the risk for CTS. Although job strain has frequently been reported to be associated with upper limb musculoskeletal disorders in general, a comprehensive review of CTS studies found no associations between these or any psychosocial risk factors and CTS.⁹ Two cohort studies that included both physical job stressors and psychosocial factors reported no association between psychological job demands or control and CTS.³¹³² Silverstein et al reported associations between CTS and psychological job demands in descriptive data, but the analysis did not control for physical job stressors.³³ Further research may clarify these relationships.

The CTS incidence rate of this study (5.11 per 100 person-years) is within the range reported by other prospective work-place studies. A similar prospective study of CTS by Silverstein et al reported a 1-year dominant hand CTS incidence rate of 7.5%,³³ and another prospective study of CTS by Garg et al²⁶ reported a 1-year CTS incidence rate of 2.55 per 100 person-years. Violante et al³⁴ reported a 1-year incidence of CTS of 7.3%, although the case definition differed from ours in that it did not require nerve conduction testing. Gell et al²⁸ reported a CTS incidence rate of 1.2% per year for a cohort of industrial and clerical workers over a 5-year period, but approximately 50% of their study group was lost to follow-up. Werner et al²⁹ reported a CTS incidence of 4.5% if the case definition required nerve conduction testing on the industrial cohort from the same study.

One of the challenges of a study like ours in which there is an abundance of detailed quantitative exposure data is how to reduce the raw data to summary exposure variables, and then how to select exposure variables for the final multivariate analyses. Our primary analysis used subtask level force and frequency of exertion data to determine peak and TWA exposures at the task level. It may be fruitful in future analyses to use the actual force and frequency required for particular exertions within each task. Another challenge is that the summary exposure variables are still numerous, and many of them are correlated with each other; with a relatively small number of study subjects, there is a need to reduce the number of variables analysed. We took the approach of limiting the variables for multivariate analysis to a maximum of one variable per ‘domain’ of exposure, with each domain representing either force (intensity of exertion), HAL (repetition, duration of exertion, etc), or posture (degree of postural deviation). Some combined variables were also analysed, such as the % time at different levels of exertion or postural deviation. We did not find significant associations between wrist posture and CTS, although it has been cited as a risk factor for CTS in other studies.⁷⁹³⁵ Our posture observations from video still frames may not be the best method to evaluate wrist posture; the inter-rater reliability of our posture observations was good for shoulder and arm postures, but lower for wrist postures.¹⁰²² It has also been demonstrated in past studies that wrist flexion and extension lead to increased carpal tunnel pressure, a proposed mechanism leading to CTS.³⁶ Neither did we find an association between vibration and CTS, although there is substantial evidence of this from prior studies.⁷ Our study was designed to focus on HAL and forceful exertion, and the only vibration measure used here was the observation of whether a vibrating tool was being used or not. We selected jobs to provide a range of HAL and force since these were the primary exposures of interest. The lack of associations observed in our data for vibration and wrist posture should not be construed as evidence of no association between these risk factors and CTS.

The limitations of our study include a reduced sample size from the original study plan to include 1000 study subjects based on power and sample size calculations. Due to resource constraints, we needed to cut our sample size in half. We still were able to see significant results probably due to reduced mis-classification of exposure because of detailed individual quantitative exposure measures, and a relatively wide range of exposures among the jobs selected for the study. We selected jobs to represent a range of exposures to two primary risk factors—force and HAL. This study is part of a collaborative research effort, and analysis of pooled data from multiple research organisations is underway and may provide further answers to remaining questions on work-related CTS.

Multiple risk factors contribute to the occurrence of CTS. In the general population, the emphasis is usually on individual factors, such as age, gender, obesity and prior medical conditions, such as diabetes; these factors are easy to measure. However, general population studies do not take workplace exposures into account. High-quality workplace research studies require extensive effort and expense, particularly to collect detailed quantitative exposure measures at the task level and even at the level of specific events within tasks for each individual study participant, and to analyse the multitude of data that result. This study shows that when such efforts are taken, specific quantitative measures of job physical exposures, such as the time spent in forceful exertion, reveal significant risk for CTS that can be greater than individual risk factors if the job exposure is high. Summary measures of the physical demands of jobs, such as the ACGIH TLV for HAL, offer a more practical means of identifying jobs that represent higher risk for CTS and, possibly, other musculoskeletal disorders that share similar risk factors of a high level of hand activity and high force. For those who have both personal risk factors such as obesity, and high physical job demands, such as a high percent of time working with forceful exertion, the risk for CTS is greater than it would be if only one risk factor were present. In workplaces, preventive efforts should target job tasks that require a high level of force or a high percent of time working with force. Our study also supports the use of the ACGIH TLV for HAL (especially when expressed as a ratio) as a guide to targeting jobs for preventive interventions. Workplaces might also consider organisational factors that facilitate opportunities for healthier lifestyles, such as exercise and healthier food choices, that might reduce obesity.