Job task and psychosocial risk factors for work-related musculoskeletal disorders among newspaper employees.

OBJECTIVES
A cross-sectional study was conducted to assess the association of upper extremity musculoskeletal disorders and work-related factors among employees using video display terminals at a large metropolitan newspaper.


METHODS
The study included 1050 randomly selected workers from four departments. The workers were asked to complete questionnaires on symptoms, job tasks, and psychosocial and work organization conditions. Musculoskeletal disorders of the upper extremities were defined by frequency, duration, and intensity of symptoms not attributable to acute injury. Data were analyzed with the use of logistic regression.


RESULTS
A total of 973 workers completed the survey. The one-year period prevalence rate for any musculoskeletal disorder of the upper extremities was 41%. Neck symptoms (26%) were the most frequently reported, followed by hand or wrist (22%), shoulder (17%), and elbow (10%) symptoms. Greater time working at the video display station was associated with increased hand or wrist symptoms in a dose-response relationship. In addition, variables corresponding to increased work-load demands (eg, increased time working under deadline and increased job pressure) were associated with increased neck, shoulder, and hand or wrist disorders. Women were more likely to report symptoms in several areas, but this finding may reflect the concentration of women in jobs involving more risk factors.


CONCLUSIONS
The results suggest a high prevalence of musculoskeletal disorders of the upper extremities among newspaper employees, and they provide additional evidence that increased work load, time pressure, and greater hours of computer use are related to the occurrence of work-related musculoskeletal disorders among these workers, particularly for disorders in the hand or wrist area.

Computerization has altered the way newspaper employees write, edit, design, and sell newspapers. The video display terminal (VDT) provides both greater access to information and ease of information retrieval and processing, but it has also been associated with a high prevalence of musculoskeletal symptoms among newspaper employees. Several studies have reported musculoskeletal symptoms in 20 to 40% of this occupational group (1-3). However, little has been written about associated risk factors. In particular, few studies have attempted to untangle the effects of suggestive putative factors such as hours of VDT work, worker demographics, work organiza-tion, and the psychosocial environment in the newspaper workplace.
This cross-sectional study formed part of a larger investigation of work-related musculoskeletal disorders among newspaper employees with varying degrees of VDT use. This paper reports the prevalences of symptoms of work-related musculoskeletal disorders among this group and investigates the relationship of these symptoms to a variety of job conditions. The editorial staff was of special focus since previous data indicated a high rate of musculoskeletal problems among these employees (I, 2). Because previously collected data linked self-reported VDT exposure to musculoskeletal disorders (I), we also assessed the accuracy of self-reported VDT use by more objective job-sampling techniques.

Subjects and methods
The study population of 1050 was randomly selected from among the 3000 current full-time employees from four departments of the newspaper: editorial, circulation, classified advertising, and accounting. These departments were selected because the employees performed jobs with a wide range of hours of computer keyboard use, job tasks, and psychosocial and work organizational conditions for comparison purposes. The classified advertising and circulation departments consisted mainly of VDT operators who performed sales or clerical tasks, whereas the accounting department, like the editorial department, consisted of professional staff using VDT stations.
Randomly selected employees within the four chosen departments were notified of the study by mail and were asked to participate. They were also invited to participate by a letter signed by both an employee and a management ergonomic committee representative in their respective departments. Informed consent was obtained from each person prior to participation. Persons under 18 years of age and women who were currently pregnant were excluded from participation in the study.
Altogether 973 (93%) employees were included in the resulting study sample. All four departments had well above 85% participation. These response rates suggest that a high level of interest with respect to work-related musculoskeletal disorders exists among newspaper employees. The study population characteristics are shown by department in table 1. Over all departments, 1% were self-reported native Americans or Alaskan natives, 10% were Asian, 14% were Hispanic, 17% were African-American, 56% were of non-Hispanic European ancestry, and 2% listed themselves as "other." Questionnaire A self-administered questionnaire was used to obtain information on demographics, musculoskeletal symptoms of the upper extremities, job tasks, work history, work organization, and the psychosocial work environment. These questionnaire items were derived primarily from those used in previous investigations of the National Institute for Occupational Safety and Health (NIOSH) on work-related musculoskeletal disorders of the upper extremities (4-6) with modifications made after discussions with interested parties within the newspaper and the completion of pi-lot studies. Questions addressing psychosocial aspects of the work environment were taken from a separate NIOSH general job-stress instrument consisting of items which had extensive use in occupational stress research (7)(8)(9). Although information concerning workstation equipment was obtained, analyses were not performed on these variables because of changes that occurred in several departments both before and during the study period. The employees who reported symptoms were often given new furniture or equipment by the newspaper management. However, this practice was not consistent by department. Relationships of symptoms to workplace equipment were not evaluated because these changes reduced the reliability of exposure assessment. The questionnaire was administered to employees in groups of 5 to 40 in prescheduled, 1-h periods during workhours.

Health outcomes
Information was obtained about symptoms of pain, numbness, tingling, aching, stiffness, or burning within the preceding year in specific areas of the neck and upper extremities (shoulder, elbow, hand, or wrist regions). The frequency of these symptoms during the preceding year was determined by an "anchored" five-point scale from "almost never (every 6 months)" to "almost always (daily)." The usual duration of the symptoms was ascertained with a seven-point scale of "less than 1 hour" to "more than 3 months." The intensity of the symptoms was evaluated with a five-point scale from "no pain" to "worst pain ever in life." Case definition for work-related musculoskeletal disorders For any of the relevant body areas (neck, shoulder, hand, or wrist), a work-related musculoskeletal disorder of the upper extremity was considered present if symptoms (pain, numbness, tingling, aching, stiffness, or burning) in the affected part occurred within the preceding year and all of the following criteria applied: (i) no previous accident or sudden injury that was not work-related (such as dislocation, sports injury, fracture, or tendon tear), (ii) symptoms began after the current job was started, (iii) symptoms lasted for more than one week or occurred at least once a month within the past year ; (iv) symptoms were reported as "moderate" (the midpoint) or worse on a five-point intensity scale. We did not exclude accidents or sudden injuries related to work from the cases if they met the other requirements for fulfilling the case definition. The case definition was developed by an internal NIOSH ergonomics medical team and has been used in other NIOSH studies on work-related musculoskeletal disorders (1, 4-6, 10).

Work sampling
To corroborate self-reports of computer keyboard use, 80 randomly selected participants, 40 with workrelated musculoskeletal disorders of the hand or wrist and 40 referents (with no neck or upper extremity symptoms) participated in the work-sampling evaluation. Employees chosen for the job sampling completed short questionnaires supplying information about usual job tasks and whether the observed day was "typical." Four different job evaluators, blinded to the case status, observed three to five participants over the course of the workshift, The job evaluators recorded the workers' tasks in summaries for l5-min intervals, attempting to obtain 30 observations per participant. During each observation interval, the job evaluators noted the workers' tasks and recorded whether the task involved computer keyboard typing for the length of that period. At the end of the work shift, a summation was performed on the number of observed 15-min intervals involving computer keyboard typing for each of the participants.

Statistical analysis
For the psychosocial variables (scales), all multi-item scales from the NIOSH general job-stress instrument were factor-analyzed to confirm that they were unidimensional. Further analyses revealed that all of the scales had acceptable reliability (Cronbach alpha coefficients of interval consistency ranging from 0.6 to 0.9). Sum scores were derived for each psychosocial scale. An unconditional logistic regression analysis was used to develop model s relating the predictor variables to the outcome measures for work-related musculoskeletal disorders in each of the three body areas (ie, neck work-related musculoskeletal disorders, shoulder work-related musculoskeletal disorders, and hand or wrist work-related mu sculoskeletal disorders). Analyses were done with SAS version 6.03 The development of regression models proceeded as follows. First, the independent variables were grouped into three sets : (i) work practice and work organization, (ii) demographic and individual factors, and (iii) psychosocial factors . Then , within each set , independent variables were tested individually for a Scand J Work Environ Health 1994. vol 20. no 6 relationship with each of the three outcome measures for work-related musculoskeletal disorders using Student's t-test, an analysis of variance, or the likelihood ratio chi-square test , as appropriate.
Independent variables not showing an association (P>O.IO) with outcome measures for the work-related musculoskeletal disorders were excluded from further analyses. With a backward elimination procedure, multiple logistic regression was then used to delete additional variables within set 1 until only variables with P-values of~0.05 remained. Next, if any continuous variables remained in the model, their quadratic forms were added, tested simultaneously, and deleted from the model if P >0.05. In later models, linear terms were forced into the model until quadratic terms were removed (which eventually occurred in all cases). The process was repeated for cubics if quadratics remained in the model. Two-way interactions among the remaining variables were then added to the model. If the P-value was ::;0.05 for the simultaneous test , interactions were removed one at a time as before. Next, the remaining variables in set 2 were added to the set 1 model , and the backward elimination procedure was repeated. The remaining variables in set 3 were then added to the combined set 1 and set 2 model and tested simultaneously to see if P~0.05. If so, backwards elimination was performed with the remaining set I and set 2 variables forced into the model. The set I and set 2 variables were checked, and if the P-value was >0.05 for some variable, that variable was removed from the model. A final check on interactions was made as follows. All two-way interactions with gender were added to the final models and tested simultaneously for significance. (Gender remained in each final model. This test was not significant for any symptom.) Then all other two-way interactions were added to the final models. (This test was also not significant for any symptom.) For all of the variables with multi-item response, odds ratios were derived from the 75-25% interquartile range difference of the scale sum score (ie, the odds ratio reflects the relative risk of a 75th percentile score compared with a 25th percentile score).

Symptoms
Three hundred and ninety-five (41%) of the 973 participating employees reported significant work-related symptoms of the neck, shoulder, elbow, or hand or wrist in the last year. In addition, in the previous year, 419 employees reported some symptoms that did not meet our definition of work-related musculoskeletal disorders (figure 1). Table 2 lists the resulting prevalences of the work-related musculoskeletal disorders (participants could report symptoms in more than one body area) and the percentage of cases which reported daily pain in the affected areas.  a Due to the nonproport ional sampling, the percentages in this column are not necessarily unbiased estimates of the percentag es in th is newspaper establ ishm ent.
159 973 All subjecfs Table 3. Risk factors in the final log istic regress ion mod els for musculoskelet al symptoms in the neck, the shoulders, and the hand or wrist. (95% CI =95% confidence interval) Neck symptoms (26%) were the most frequently reported, followed by hand or wrist symptoms (22%), shoulder symptoms (17 %), and elb ow symptoms (10%) . Of those report ing significant symptoms, at least one-third had seen a health care provider, from 10 to 20% had missed one workday because of the disord er, and 6 to 10% had been assigned to a different j ob for one workday due to sympto ms. Table 3 describes the final logistic regression models relating personal, job characteristic, and psychosocial variables to the three muscul oskeletal disorder s (neck, shoulder, and hand or wrist areas). (The odds ratios in table 3 for shoulder symptoms were adju sted for race, which remained in the fina l model, but was felt to be a surrogate for some other variable. ) 81  were found only in the Editorial Department. The results of this analysis (table 4) revealed that the odds ratios for the gender variable from the models in the three body areas were not much different from those of the origi nal models. However, the results did not remain statistically significant. In fact, most of the variables in the models had odds ratio confide nce intervals which included one (the variables "hours spent under a dead line" and "perce ived lack of importance for ergonomic issue s by management" for work-related muscul oskeletal disord ers of the neck and the "n umber of hours spent typing" for workrelated musculoskeletal disorders of the hand or wrist did not). Neither age nor height were fou nd to be assoc iated with any of the work-related musculoske letal disorders. Other individual factors (incl udi ng existing medical conditions or hours spent typing away from work) were not found to be significant predictors.

Predictors of work-related musculoskeletal disorders
Job characteristics. The only significant jo b history risk factor was the numbe r of years employed at the newspaper, which was associated with shou lder symptoms (OR 104 ) (table 3). Of the 17 jo b-task and work-organization variables, the number of hours spent under a deadline was fou nd to be an important predictor for neck (OR 1.7) and hand or wrist work-related musculoskeletal disorders (OR 1.6) (table 3). Seventy-one percent of the employees reported working under a deadline , 47% of those with dead lines reporti ng 0-10 deadli ne hours per week and 2 1% of those with deadlines reporting more than 30 h per week under a dead line.
The risk of a work-rela ted musculoskeletal disorder of the hand or wrist increased in a dose-respo nse fashion as the number of hours typing increased (table 5). For the 23% of the participants who reported typing 6-8 h a day, the OR was 2.1 when they were compared with a reference group (who typed 0 to 2 h a day). For those with more than 8 h of typing (2% of the particip ants) the OR was 3.3 in a compariso n with the reference group.
Time spe nt on the telephone (the only significant work practice variable) had a modes t, but significant, association with work-related musc uloskele tal disorders of the neck (OR 104). Other work prac tice and work orga nization factors, incl uding number of breaks (brief or long), number of times arising from a chair , typing skill (slow , medium, or fast) , typing techni que ("hunt and peck" or touch ), use of eyeglasses, or training in posture technique, were not significa nt predictors.
Psychosocial work environment characteristics. A significantly higher prevalence of work-rela ted musculoskeletal disorders of the neck (tab le 3) was seen for those emp loyees who reported an increased workload variability (OR 1.7) and perceived a lack of importance given to ergonom ic issues by management (OR 1.9). The employees with a perceived lack of support from their immediate supervisors tended to have a higher prevale nce of work-related musculoskeletal disorders of the hand or wrist (OR 104 ) (ta-Scand J Work Environ Healt h 1994, vol 20. no 6 Table 4. Finallogislic regressio n models for work-related musculo skeleta l sym pto ms in th e neck , th e sho uld ers, and the hand or wrist in the analysis of jo bs having a comparab le number of men and women (N = 150 Perceived lack of su ppo rt fro m an imme dia te superv iso r (very much versus a little) 1.5 0.6-3.5 ble 3). A higher pre valence of shoulder work-re lated musculoskeletal disorders (table 3) occ urred for employee s who reported a lack of participation in job decision-making (OR 1.6) and increased job pressure (OR 1.5).
Other psyc hosoc ial variables -such as j ob control, worker partici pation , j ob security, and intera ction with co-workers or customers, group conflict, and lack of socia l suppo rt from spouses and friends -were not fou nd to be significant predictors of work-r elated musc uloske letal disorders.

Work sampling results
Work sampling was performed for 36 randomly selected persons with hand or wrist symptoms and 40 randomly selected referents. Four selected cases refused to participate. Based on questio nnaire data, the mean number of hours typing was 4.5 h for the symptomati c perso ns and 3.9 h for the referents, a significant difference (P < 0.05). For the observed number of hours spent typing , the means were 2.5 h for the symptomatic persons and 1.9 h for the refere nts (table 6). The Studen t' s t-test analysis of the observed times revealed that the symptomatic persons spent significa ntly more time typing than the referents (P < 0.02). The Pearson corre lation coefficients for the observed and reported times were 0.39 for the sympt omatic person s and 0.40 for the referents (P<0.05 ).

Prevalence
Approximately 40 % of the newspap er employees reported moderate to severe symptoms which fulfilled our case definition for a work-related musculoskeletal disorder in the neck or upper extremities. Neck symptoms were the most prevalent, followed by hand or wrist, shoulder, and elbow symptoms. The predominance of work-related musculoskel etal disorders in the different upper-extremity areas were similar to the results of previous news paper VDT surveys (3, written communication, Sauter et al 1981), symptoms in the neck and shoulder (usually combined in most studies) occurring more frequently than in the hand or wrist region. This rankin g by upper extremity area was also similar to the results of most previous VDT studies. In a comparison of the hand or wrist results with those newspaper studies which focused only on work-related musculoskeletal disorders of the hand or wrist, the findings were similar, with prevalences in the 17 to 26% range (table 7). It is important to note that only one-third of those with work-related musculoskeletal disorders sought health care; this finding may be related to issues of severity or to an unwillingness to seek care for a myriad of reasons.

Typing hours
The odds for having a work-related musculoskeletal disorder of the hand or wrist increase d in a doseresponse fashion as the number of hours typing increase d, with a greater than twofold risk for those typing more than 6 h. Other studies have examined typing hours as a risk for work-related musculoskeletal disorders and found conflicting results (15)(16)(17)(18)(19).
Factors such as small sample sizes (18,20) , which may have lacked statistical power to detect rea l differe nces among the groups, or inadequate varia nce of computer time ( 10) to account for differences may pain and stiffness and current VOT usage 17% " dullness in arms"; 15% "dullness In fingers" compared w ith 12% and 5%, respectively, of nonus ers 63% response rate ; 26% lower arm tendin it is or carpal tunnel syn drome 40% symptoms of work -related musculoskeletal disorders; percentage of time typ ing and being a reporter , significant facto rs 26% symptomatic, majori ty hand or wr is t , deadlines important have been responsible for the conflic ting results in the cross-sec tional studies. Comp arisons of typing hours in studies before the mid-I 980s may also have been problematic becau se of differen ces in VDT equipment at that time (slower equipment, only nonrout ine tasks performed on computers) and less reliance on a VDT to complete work tasks by the newspaper work force than in recent years. Our findings concur with those of another recent NIOSH study (I), which also found an increased risk of work-related musculo skelet al disorders with increasing time spent typing at computer keyboards in VDT work on a newspaper. A follow-up in 1991 at the same newspaper (2 1) found a significant relationship in the preval ence of work-related musculoskeletal disorders of the hand or wrist on the basis of the percentage of time keying. Despite these recent findings, controversy still remain s as to whether a causal relationship exists between the time spent typing on computer keyboards and a greater likelihood of develop ing musculoskeletal symptoms. Our results support a dose-response relationship; that is, the more hours one spends typing per day, the higher the odds are of developing work-related musculoskeletal disorders of the hand or wrist.
Through job sampling, we obse rved that the randomly selected persons with hand or wrist symptoms spent significa ntly more time typing than the randoml y selected referents (P<0.05). It has been suggested that VDT users with musculoskeletal symptoms might overestimate their time spent typing as compared with those without symptoms. Although both the symptomatic per sons and the referent s reported that their average daily typing time was about twice the amount they were observed to type during a single workshift, the degree of overreporting was similar in both groups (table 6). This finding verifies that the symptomatic perso ns were not overestimating the time spent typing when compared with the referents. These work sampling results concur with results from the recent NIOSH study (I ) which found that news reporters with symptoms of workrelated musculoskeletal disorders did not overreport in a comparison with those without such symptoms.

Deadlin es
The result s also showed that the number of hours spent working under a deadl ine was associa ted with an increased risk of developing work-related musculoskeletal disorders of the hand or wrist. Fifty-three percent of the participants reported 10 h or more per week under a deadlin e. More hours under a deadline may contribute to the developm ent of symptoms through mult iple mechani sms. For exa mple, increased deadline work may mean greater psychological stress and increa sed musculoskeletal tension, or simply more exposure to VDT use, such as increased typing and repetition, time in co nstrained postures, and fewer rest breaks. In his study of jo urnalists SCQndJ Work Environ Health 1994, vol 20, no 6 using a VDT, Buckl e (22) also found that deadline hours, particularly among those journalists facing daily deadl ines, predicted the extent and severity of reported pain and discomfort.

Gender
The analyses revealed that the women were at a higher risk of experiencing symptoms of work-related musculoskeletal disorder s. Becau se the proportions of women and men in the different newspaper j obs varied, we examined the possibil ity that gender may have been a surrogate for job title. This analysis (tables 5-7) found results similar to those of our original models; however, the results were not significant. The smaller sample size (approximately ISO comp ared with about 1000 in the full data set) may have reduced the statistical power substantially. Although this result may indicate that gender was not a surrog ate for job title in the models based on the complete data, it does not give a clear answer.
The importance of gender as a risk factor for workrelated musculoskeletal disorders has also been controversial. Earlier studies (23,24) have reported that female gender was an important risk factor for carpal tunnel syndrome, but these studies did not look at occupational risk factors, and they did not eva luate men and women performing similar jo bs. In a few studies that did concern men and women performing comparable jo bs, no gender difference was found ( I, 25). On the other hand, a recent NIOSH study ( 10) of VDT users found that in jobs with similar proporti ons of men and women, femal e gender was also a risk factor fo r musculo skeletal symptoms.
It has been suggested that gender may be a surrogate for other important factors, such as a differing set of social stressors for women and different (more stressful) nonoccup ational upper extremity usage than men, especially that related to child care and housekeeping (26). Because nonoccupational factors were not examined in detail in this study, it is possible that the gender differences noted may result from confounding from nonoccupational stres sors or job factors such as job title, as previ ously mentioned .

Psychosocial and work organi zational fa ctors
Variables correspondin g to increa sed work-load demands (eg, increase d time working under a deadline and increased jo b pressure) were associa ted with increased neck, shoulder, and hand or wrist disorders. Lack of support from immediat e supervisors and upper management were important factors for work-related musculoskeletal disorders of the neck and the hand or wrist. Overall, our results provide additional evidence that aspects of the psychosocial environment , especially variables corresponding to work or time pressures, lack of social support, and lack of participation in decision making, are importa nt contributors to work-rela ted musculo skeletal disorders among office workers. Several studies of office personnel have related relatively high levels of worker stress to work-related musculoskeletal disorders of the upper extremities (9,(27)(28)(29)(30)(31)(32). In this study, the logistic regression models showed psychosocial factors to be more influential in predicting musculoskeletal disorders in the neck and shoulder regions than in the hand or wrist area.
The causal linkages between a demanding psychosocial environment at work and symptoms of workrelated musculoskeletal disorders are not well understood, but several plausible mechanisms can be postulated. Psychosocial demands resulting in job stress may increase the awareness (and possibly overreporting) of musculoskeletal symptoms or may affect perceptions of their cause (ie, increased attribution to the workplace). In addition, stressful work conditions may be associated with more physically (ie, biomechanically) demanding situations or may produce increased muscle tension and consequent biomechanical strain as caused by stress-induced static muscle loads in the neck and shoulders (33,34).
Such factors as lack of control over many aspects of the job, increased isolation, reduction of task diversity, and increased work load have been attributed to the introduction of VDT workstations into the workplace and to the increase in reports of work-related musculoskeletal disorders (35,36).

Limitations
A high response rate (93%), minimizing the potential for selection bias, was obtained in the study. Inherent in this type of study is the potential for "survivor bias" (ie, not including people who left their jobs because of the health problems of interest). This possibility may result in an underestimation of an association between risk factors and health problems. However, survivor bias was not considered to be a major problem in this study, since both records and accounts of newspaper employees indicate that the number of individuals who left the work force during the year preceding the study was small.
Because risk factors and health problems are measured at the same time in cross-sectional studies, it is not always possible to determine which occurred first. For example, an important association was found between hand or wrist symptoms and lack of support by the immediate supervisor, but we were unable to ascertain whether symptoms or the perceived lack of support came first.
Because the prevalence rates for work-related musculoskeletal disorders were determined solely by selfreported symptoms from questionnaires, the prevalence of work-related disorders could have been over-or underestimated. However, our rates agree fairly well with those in prior research (l, 2). No reference group outside the newspaper industry was studied, and therefore it is unclear whether there are certain characteristics (such as frequent deadlines) 424 which put newspaper employees at increased risk for symptoms when compared with other computer users. In addition, musculoskeletal symptoms are fairly common in the general population and may arise from nonoccupational activities. Therefore, it is not possible to determine precisely the proportion of all the disorders which are work-related.
This study did not fully address the impact of variables that are not work-related and their possible associations with work-related musculoskeletal disorders. For example, we did address social support issues of spouses, friends, and relatives, but we did not cover issues such as child care or home responsibilities, which may have some impact on the occurrence of neck and upper extremity symptoms (25).
Because the analysis in this study included a variable reduction scheme, it is possible that some of the variables from the questionnaire were not included in any final model because of their elimination in an intermediate stage, when perhaps they could conceivably have attained a P-value of :50.05 when all three sets were combined. Therefore, a possibility of "false negatives" exists, and variables not included in the final models should not be confused by the reader with a definite statement of "no-effect." In a study that examines numerous potential associations between possible exposure factors (like work pressure or hours of typing) and multiple different health outcomes, it is likely that a few of the associations are "statistically significant" by chance alone and are thus spurious. Differentiating spurious and nonspurious associations is a difficult task involving an examination of observed effects in light of past findings, determining biological plausibility, and considering the strength of associations. However, recurring associations confirmed by others (such as the relationship between typing under a deadline and work-related musculoskeletal disorders of the hand or wrist) are not as likely to be spurious.

Concluding remarks
Despite the inherent limitations, this study provides additional evidence that increased work load or job pressures and hours of computer use are related to the occurrence of work -related musculoskeletal disorders, particularly for symptoms in the hand or wrist area. This study also supports the conclusion that work-related musculoskeletal disorders among newspaper employees is dependent on a variety of factors, among them job task, work organization, and psychosocial factors. An approach to the prevention of work-related musculoskeletal disorders among these employees, which systematically includes all these elements, has the greatest likelihood of being successful in reducing the magnitude and severity of work-related musculoskeletal disorders. The jobsampling conducted by us documents that there was no selective overreporting of time spent typing by th o se e m p loyees wi th sy m pto ms in a c om parison w ith those w ith out sy m p to ms (respo nse bias ). Finally , psychosocial facto rs app ear to be more important risk factors for neck and sho ulde r di sorders th an for hand or wrist diso rders in relation to w ork .
Overall , the ph ysical and psych o so cial as so ciatio ns identified in th is st udy are cons is te nt wi th find in g s of si m ila r stud ies . H o we ver, the m ode st streng th of the associati o ns between each indi vidual predict or and health outcome mea sure sugges ts that a substa ntial reduction in sy m pto m rates may be di ffi cult to achie ve . Nevertheless, prevention efforts cannot wait until th e etiology of w ork-related musculosk el etal disorders is completel y understood. Effec tiv e pr evention intervention m ay she d more light on the complex web of causal factors th an any othe r typ e of action.