Associations between work-related factors and the carpal tunnel syndrome—a systematic review

A. Associations between work-related factors and the carpal tunnel syndrome—a systematic review. Scand J Work Environ Health 2009;35(1):19–36 . Objectives The aim of this study was to make a quantitative assessment of the exposure–response relationships between work-related physical and psychosocial factors and the occurrence of carpal tunnel syndrome (CTS) in occupational populations. Methods A systematic review of the literature was conducted on the associations of type of work, physical load factors, and psychosocial aspects at work to the occurrence of CTS. The associations between work factors and CTS were expressed in quantitative measures, namely, odds ratios (OR) or relative risks. Results Jobs with the highest risk of CTS included work in the meat- and fish-processing industry, forestry work with chain saws, and electronic assembly work (OR 76.5, 21.3, and 11.4, respectively). The occurrence of CTS was associated with high levels of hand–arm vibration, prolonged work with a flexed or extended wrist, high requirements for hand force, high repetitiveness, and their combination. No association was found between any psychosocial risk factor and CTS. Contradictory findings were reported for associations between computer work and CTS. Conclusions This review provides consistent indications that CTS is associated with an average hand force requirement of >4 kg, repetitiveness at work (cycle time <10 seconds, or >50% of cycle time performing the same movements), and a daily 8-hour energy-equivalent frequency-weighted acceleration of 3.9 m/s 2 .

Hand-arm symptoms are a common problem in society, especially among the working population. Of these, carpal tunnel syndrome (CTS) is the most frequently reported neuropathy of the upper extremity (1-3). CTS results from the compromise of the median nerve function at the wrist as a result of increased pressure in the carpal tunnel (1). The clinical diagnosis of CTS is based on a history of nocturnal pins and needles, numbness, or pain in the median nerve in the innervated area of the fingers and hand, which often causes the patient to awaken at night, supported by abnormalities appearing in an electrodiagnostic examination (4). Provocation tests do not necessarily contribute to the clinical diagnosis of CTS (5).
CTS can have serious economic consequences. Feuerstein et al (6) stated that 57% of all costs associated with occupational upper-extremity disorders were due to CTS. Furthermore, Silverstein et al (7) reported an average yearly claim rate for CTS of 27.3 per 10 000 full-time workers.
A review of occupational populations showed a wide range in the prevalence of CTS (0.6-61%), the lowest prevalence occurring for industrial workers and the highest for grinders, butchers, grocery store workers, and frozen food factory workers with workers using high-force, high-repetitive gripping (8). Hagberg et al concluded that exposure to physical load (such as repetitive and forceful gripping) is a major risk factor for CTS (8). Palmer et al (9) judged that there is reasonable evidence that regular and prolonged use of handheld vibrating tools increases the risk of CTS at least twofold and that there is substantial evidence for similar or even higher risks from prolonged and highly repetitious flexion and extension of the wrist, especially when combined with a forceful grip. At the same time, the balance of evidence concerning keyboard and computer work did not indicate an important association with CTS (9). Furthermore, the associations between physical risk factors (repetition, force, vibration) have been confirmed in several other reviews (10)(11)(12). The risk of developing upper-extremity symptoms is also influenced by psychosocial work characteristics. The reviews of Bongers et al (13) and Van den Heuvel et al (14) have presented evidence of an association between high job demands and low social support with upper-extremity symptoms. In addition, a review of the National Research Council concluded that high job demands and high job stress were associated with the occurrence of upper-extremity disorders (11). However, reviews on the association between psychosocial work characteristics and CTS are lacking.
The available reviews have presented overviews of the occupations in which workers are at risk of CTS and occupational risk factors for the occurrence of CTS, but they provide little guidance as to the duration and magnitude of exposure to risk factors that are associated with the development of CTS. Hence it remains a matter of debate whether it will be possible to derive exposure levels that will not increase the occurrence of CTS in occupational populations.
Therefore, this systematic review of the available evidence was conducted with the aim of providing a quantitative assessment of the exposure-response relationships between work-related physical and psychosocial factors and the occurrence of CTS in occupational populations.

Literature search
Comprehensive literature searches were conducted by the first author (RMvR) in MEDLINE (from 1966 to September 2007), EMBASE (from 1984(from to September 2007, and the Cochrane Central Register of Controlled Trials (September 2007). The following keywords were used: (carpal tunnel syndrome OR median nerve) AND (work related OR physical load OR psychosocial load OR exposure) AND (association OR risk factors OR odds ratio OR relative risk). The result of the complete search strategy is available on request.
Two reviewers (RMvR and BMAH) independently selected the articles, initially based on title and abstract (figure 1). For final inclusion, the articles had to fulfill all of the following criteria: (i) the occurrence of CTS was reported in occupational populations, (ii) a quantitative description of measures of exposure or a description of a distinct exposure pattern at job level was presented, (iii) the association between work-related risk factors and CTS was expressed in a quantitative measure, such as odds ratio (OR) or relative risk (RR), or sufficient raw data were provided with which to calculate these associations, and (iv) the article was published in a peerreviewed scientific journal written in English, German, French, or Dutch. A consensus method was used to resolve disagreements.   (17) and the Dutch Cochrane Centre (18), which were adapted to the specific aim of this review (appendix I). The list covers five topics with 16 items concerning the study population, the assessment of exposure, the assessment of outcome, the study design and analysis, and the data presentation (table 1). Two reviewers (AB & BWK) independently assessed the quality of each study by scoring each criterion as positive, negative, or unclear. Disagreements were resolved by consensus. The quality score for every study was calculated by summing the number of positive criteria.

Data extraction
Relevant information on study population, study design, outcome ascertainment, exposure characteristics, measure of association, and confounding factors was extracted from the articles by the first author (RMvR) using a standardized form. The core findings in each article were expressed by measures of association (OR or RR value) with corresponding 95% confidence intervals (95% CI). Where possible, these associations were directly extracted from the original article. For articles in which this information was not presented, associations were calculated if sufficient raw data were provided.

Data analysis
In this review three types of statistical associations were distinguished. The association was described as positive when a higher value of the risk factor was statistically associated with the occurrence of CTS. In a negative association, a higher value of the risk factor was statistically associated with a lower occurrence of CTS. In null associations, the risk estimate did not differ statistically from unity. The null associations were further evaluated as to whether the results actually suggest the absence of an effect or whether the studies were inconclusive due to a lack of information or a lack of statistical power. First, we focused on the association between type of work (based on job descriptions with a distinct exposure pattern) and the occurrence of CTS. Second, we evaluated the associations of five types of exposure, namely, force, repetitiveness, hand-arm vibration, combined exposure measure, and awkward postures, with the occurrence of CTS. Finally, we addressed the associations between psychosocial risk factors and the occurrence of CTS.
Pooling the results of individual studies was considered only when health outcomes were clinically homogeneous, the measures of exposure were sufficiently similar according to the reviewers, and comparable study designs were used. Furthermore, the outcome of the quality assessment was used in a sensitivity analysis to evaluate whether design characteristics and the methodological quality of the studies had an impact on the reported associations between work-related risk factors and CTS.

Characteristics of included studies
Our search of the literature resulted in 985 potentially relevant articles (figure 1). Of these, 44 publications met our inclusion criteria (30 cross-sectional studies, 9 case-control studies, and 5 cohort studies (appendix II). A total of 22 articles compared the occurrence of CTS across occupations (table 2), and 23 articles reported on the association between physical risk factors and CTS (table 3), of which two articles also compared the occurrence of CTS across occupations. Four articles reported on the association between psychosocial factors and CTS (table 4), of which three articles also reported on the association between physical risk factors and CTS. Table 5 presents the methodological quality assessment of the included studies. The initial agreement of the two reviewers was 68% (480 of 704 items). The initial disagreements were all solved in a consensus meeting.       Figure 2 shows that the methodological quality score of the included articles did not improve over time. Furthermore, the methodological quality score is not related to the level of significance of the association between occupation or risk factors and the occurrence of CTS.

Outcome assessment
In tables 2-4, the included studies have been classified by the method of outcome assessment. In 19 studies (43%), the diagnosis of CTS included both the presence of typical symptoms (numbness, tingling, burning, or pain in median innervate fingers) and a positive electrodiagnostic finding (impairment of median nerve function at the wrist). In the rest of the articles, the diagnosis of CTS included symptoms (N=9, 20%), a physical examination (N=2, 5%), an electrodiagnostic examination (N=2, 5%), or a combination of symptoms and a physical examination (N=12, 27%). Of the studies with an accurate diagnostic method (ie, typical symptoms and electrodiagnostic examination), 58% (N=11) reported a significant association between work-related factors and CTS, against 64% (N=16) of the studies with a less accurate diagnostic method.

Exposure and carpal tunnel syndrome
Force. Three articles found a significant association between exposure to force and CTS, with OR values of 2.11 to 9.0 (19,41,42). In the case-control study of Roquelaure et al (42), handling loads of >1 kg at least 10 times/hour was a risk factor for CTS. Moderate and heavy manual work, as reported by Lam et al (41), and precision grip, as reported by Abbas et al (19), were associated with CTS, but these factors were not described in more detail. Null associations were reported in four articles, in which the duration of work with high loads on the wrist, the duration of using a pinch grasp, heavy lifting, and the relative intensity of force were described, the OR values ranging from 0.39 to 3.50 (43)(44)(45)(46).
Repetitiveness. Five articles found a significant association between exposure to repetition and CTS, the OR values ranging from 0.50 to 9.39 (42,(44)(45)(46)(47). The strongest associations were found for jobs with a work cycle of <10 seconds (42,47). Contradictory results were observed for jobs with work cycles of <30 seconds, one study reporting an elevated OR value of 2.18 and other studies reporting OR values close to unity, with 0.7 and 0.96, respectively (47)(48)(49). Five articles reported null associations between exposure to repetitiveness and CTS (46)(47)(48)(49)(50).
Hand�arm vibration. Three articles reported a significant association between exposure to vibration and CTS, with OR values of 2.52 to 4.8 (46,51,52). An increased risk of CTS was found for workers using vibrating tools (1-20 years and >20 years) and power tools or machinery (6-11 hours/day) (46,52 (55) significant associations were also found between exposure to low force-high repetition (OR 4.72) and high force-low repetition (OR 3.21) and CTS, but these associations were not confirmed in other studies. The case-control study of Cosgrove et al (56) did not corroborate any of these findings.
In two studies, self-reported exposure to computer work for >8 hours/day and mouse use for >20 hours/week were associated with CTS, with OR values of 3.6 and 2.6, respectively (20,57). In contrast, in five studies, no significant associations were found between CTS and the duration of keyboard use, the duration of mouse use, or the frequency of mouse use (OR 0.9-3.4) (41,45,57,59,60).  1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 Methodological quality score 16 Table 4 presents the characteristics of four studies that reported associations between psychosocial risk factors and CTS. None of the reported associations were statistically significant.

Discussion
This review evaluated the associations between exposure to physical and psychosocial risk factors and CTS. Frequent handling of loads, highly repetitive work with and without force requirements, hand-arm vibration, and activities with a flexed or extended wrist were associated with CTS. Occupations with the highest prevalence of CTS were jobs in the meat-and fish-processing industry, forestry work with chain saws, and electronic assembly work. Contradictory findings were observed for the duration of computer work.
The aforementioned conclusions are primarily based on evidence presented in cross-sectional and case-control studies, since only a small number of cohort studies was identified. Therefore, the causality of the reported associations between exposure and the occurrence of CTS cannot be established, especially since most of the studies relied on self-reported measures of exposure. The lack of cohort studies can partly be explained by the relatively modest incidence of CTS in occupational populations at risk. Nathan and his colleagues found 34 incident CTS cases among 256 industrial workers during an 11-year follow-up, resulting in an estimated incidence of 12.0 per 1000 person-years (45). In the prospective study of Gell et al (35) among 432 industrial and clerical workers with an average follow-up of 5.4 years, a similar incidence of 12.4 per 1000 person-years was observed. Given this low incidence, cohort studies with a low exposure prevalence will be strongly underpowered with respect to demonstrating significant associations, even with OR values of 2.0 or higher, unless very large populations are studied. The lack of sufficient power in studies is also demonstrated by the large confidence intervals in many cross-sectional studies, whereby even a study population with 652 workers was not sufficiently large enough to establish the significance of a meaningful association (OR 3.3) (54).
A larger heterogeneity among the studies was observed in the assessment of exposure to physical and psychosocial risk factors. With respect to exposure assessment, only 4 of 14 articles on the influence of force and repetitiveness used the same definition, based on the classical study of Silverstein and her colleagues from 1986 (61). Another important drawback is that 29 studies (66%) used questionnaires or interviews to determine the magnitude, frequency, or duration of exposure, and there is ample evidence that self-reports will introduce substantial misclassification in exposure and, thus, considerable attenuation of true associations (62). The heavy reliance on questionnaires for exposure assessment is also reflected in the fact that 20 of the 32 (67%) exposure measures with a significant association with CTS (as reported in table 3) were defined by duration in terms of hours per week or years, which is the exposure dimension with the least random misclassification when estimated by a questionnaire (62).
In about half of the studies (N=25, 57%) in this review, the diagnosis of CTS was determined with the use of a questionnaire, a physical examination, or an electrophysiological examination. In the remaining studies (N=19, 43%), the diagnosis was based on both symptoms and the results of an electrophysiological examination. The combination of a positive electrodiagnostic study and characteristic symptoms appears to have the best predictive value for a case definition of CTS (4). However, this strict method of diagnosing CTS leads to a low prevalence of the outcome. For example, in a large prospective study by Gerr et al (63), only three prevalent cases, and three incident cases were reported among 630 workers. It is of interest to note that only six studies (14%) relied on questionnaires only to estimate the presence of CTS, whereas 29 studies (66%) used questionnaires for exposure assessment. Thus it is reasonable to assume that the heterogeneity among studies is much larger for exposure assessment than for the ascertainment of CTS.
The scores on the methodological quality assessment ranged from 4 to 15 (on a scale from 0 to 16). As a result of a lack of cohort studies, items 12 (follow-up period ≥1 year) and 13 (prospective or retrospective study design) scored positive in only eight articles. Other critical items were "blinding to exposure status" during the verification of CTS (N=10) and "blinding of case status" during the exposure assessment (N=14). Blinding to exposure often failed since the physicians were aware of the occupation of the worker under examination, and blinding of CTS status often failed since the diagnostic procedure started with a questionnaire with self-reported exposure measures. The quality assessment of the exposure assessment strategy was restricted to three items, and therefore did not take into account several other issues that are important in the evaluation of exposure assessment, such as measurement error and the contrast in exposure magnitude and duration (64). The methodological quality was not associated with the presence of a significant association between work-related factors and CTS. It is an interesting observation that the methodological quality score of the included articles did not improve over time (figure 2), predominantly due to the publication of various studies in recent years that only compare the CTS prevalence across different jobs without any further investigation into underlying exposure patterns and, thus, exposure-response relationships.
Factors that may influence whether a study reports a significant association or not between a physical risk factor and CTS are lack of power, low incidence of outcome, low exposure prevalence, measurement error, and presence of confounding factors. However, the results of our review support the conclusions presented by Palmer et al (9). An excess risk to CTS was reported for assembly workers, meat-processing workers, and food-processing workers. In addition, prolonged use of handheld vibratory tools and prolonged and highly repetitious flexion or extension of the wrist increased the occurrence of CTS. The current review extends existing knowledge with a quantitative assessment of exposure-response relationships between work-related factors and CTS.
A meta-analysis with a pooling of data was considered only for the duration of computer work for >20 hours/week (20,57,59) and combined exposure to high force-high repetitiveness (53)(54)(55). With regard to computer work, the procedures for case ascertainment differed too much across studies, as was reflected in the prevalence estimates, which varied from 1.4% (57) to 13.1% (20). The contradictory findings for computer use and the development of CTS are in agreement with the conclusion of a recent review (65). In two crosssectional studies (53,54), a similar definition of force and repetitiveness was used, but, although significant associations were reported in the separate studies, the large heterogeneity of the risk estimates across studies made the pooling of results not possible. In comparison with other studies on repetitive jobs, it is suggested that harmful cycle times with repetitive movements are <10 seconds (42) rather than <30 seconds (48,49).
Pooling study results was not possible for exposure to hand-arm vibration and the duration of flexion or extension of the wrist, although several studies have been carried out on these risk factors. With respect to hand-arm vibration, all of the studies consistently reported increased OR values, albeit not always significant associations. A comprehensive exposure assessment strategy was used by Bovenzi and his colleagues (51), who showed that an 8-hour energy-equivalent frequency-weighted acceleration of 3.9 m/s 2 was associated with CTS (51). This value is well above the action value of 2.5 m/s 2 described in the European directive on physical agents (68). The studies on flexion or extension of the wrist show large differences in exposure levels that are already associated with an increased occurrence of CTS, as was found for three case-control studies. Nordstrom et al (52) presented an increased risk of CTS with bending the wrist for >3 hours/day, Blanc et al (58) reported hand bending >2 hours/day as a risk factor, whereas, in the study of De Krom et al (43), the risk was already elevated when the wrist was extended or flexed for 1 hour/ week. These results, all based on self-reported exposure, are too contradictory for drawing any meaningful conclusion about harmful exposure levels. In addition, the four studies on psychosocial risk factors have presented no indications for an association with CTS.
In summary, this systematic review provides consistent indications that CTS is associated with the following physical risk factors: average requirements for a hand force of >4 kilograms, repetitiveness at work (cycle time <10 seconds, or >50% of a cycle time during which the same movements are performed), and a daily 8-hour energy-equivalent frequency-weighted acceleration of 3.9 m/s 2 . Prolonged or repeated flexion and extension of the wrist is a risk factor under scrutiny, but the available evidence does not permit us to provide some guidance on the levels of hazardous exposure levels.