Mechanical injury and psychosocial factors in the work place predict the onset of widespread body painA two-year prospective study among cohorts of newly employed workers.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 50, No. 5, May 2004, pp 1655–1664 DOI 10.1002/art.20258 © 2004, American College of Rheumatology Mechanical Injury and Psychosocial Factors in the Work Place Predict the Onset of Widespread Body Pain A Two-Year Prospective Study Among Cohorts of Newly Employed Workers Elaine F. Harkness, Gary J. Macfarlane, Elizabeth Nahit, Alan J. Silman, and John McBeth lifting >24 lbs with 2 hands, pulling >56 lbs, prolonged squatting, and prolonged working with hands at or above shoulder level. Of the psychosocial exposures, those who reported low job satisfaction, low social support, and monotonous work had an increased risk of new-onset widespread pain. In multivariate analysis, monotonous work and low social support were found to be the strongest independent predictors of symptom onset. Conclusion. Our findings demonstrate that the prevalence of new-onset widespread pain was high, but among this young, newly employed work force, both physical and psychosocial factors played an important role. Objective. Mechanical injury has been postulated as a risk factor for widespread pain, although to date, the evidence is weak. The aim of this study was to determine whether repeated exposure to mechanical trauma in the work place predicts the onset of widespread pain and to determine the relative contribution of mechanical trauma compared with psychosocial factors. Methods. In this prospective cohort study of 1,081 newly employed subjects in 12 diverse occupational settings, we collected detailed information on mechanical exposure, posture, physical environment, and psychosocial risk factors in the work place. Study questionnaires were completed at baseline and at 12 and 24 months. Individuals free of widespread pain at baseline and 12 months were eligible for followup. Generalized estimating equations were used to determine which factors predicted the new onset of widespread pain. Results. Of the 1,081 baseline respondents, 896 were free of widespread pain and were eligible for further study. Of these 896 subjects, 708 and 520 responded at 12 months and 24 months, respectively. The rates of new-onset widespread pain were 15% at 12 months and 12% at 24 months. Several work place mechanical and posture exposures predicted the new onset of widespread pain: lifting >15 lbs with 1 hand, Chronic, unexplained widespread pain is the clinical hallmark of the fibromyalgia syndrome. Adverse psychosocial factors, aspects of health beliefs and behavior, and a history of reporting somatic symptoms have been shown to predict the new onset of chronic widespread pain in the general population (1). It has also been hypothesized that physical trauma may be an important risk factor for widespread pain (2). Such trauma may occur as a result of, for example, a road traffic accident or a work place accident. Previous studies have shown an association between major physical trauma and the onset of fibromyalgia (3,4). For example, Al Allaf et al (3) found that prior traumatic events were reported more commonly in patients with fibromyalgia than in age- and sex-matched controls (39% and 24%, respectively). In contrast, the role of low-level physical trauma, which may occur by repeated exposure, for example, through manual handling activities within the work place, is unknown. Only 1 study has examined any Supported by the British Occupational Health Research Foundation and the Arthritis Research Campaign, Chesterfield, UK. Elaine F. Harkness, MSc, Gary J. Macfarlane, MD, Elizabeth Nahit, PhD, Alan J. Silman, MD, John McBeth, PhD: University of Manchester, Manchester, UK. Address correspondence and reprint requests to Elaine F. Harkness, MSc, ARC Epidemiology Unit, Medical School, University of Manchester, Oxford Road, Manchester, M13 9PT, UK. E-mail: firstname.lastname@example.org. Submitted for publication May 27, 2003; accepted in revised form February 10, 2004. 1655 1656 HARKNESS ET AL aspects of work-related risk factors and chronic widespread pain (5). That study found the prevalence of fibromyalgia to be increased in subjects who reported high levels of “physical work stress.” No associations were found for “mental work stress.” However, that study was cross-sectional, and other aspects of the work place environment were not considered. Recently, we reported that the role of workrelated risk factors in predicting chronic widespread pain in an unselected population sample was limited (6). However, that conclusion may have been influenced by the “healthy worker effect.” The majority of subjects had been in the same employment for at least 3 years and were therefore likely to have been well-established within the work force. As a consequence, we may have underestimated the true effect of work-related risk factors and the new onset of widespread pain, since some individuals may have previously left the work force as a result of their pain. The aim of the present study was to examine the effect of work-related risk factors as predictors of new-onset widespread pain within a cohort of newly employed workers. SUBJECTS AND METHODS Study design. This was a prospective cohort study of newly employed workers. At baseline, subjects completed a questionnaire that ascertained current pain status and measured various aspects of work-related mechanical, psychosocial, and environmental risk factors. Subjects who were free of widespread pain at baseline were identified and were sent followup questionnaires at 12 months and 24 months to ascertain whether there was new onset of widespread pain. Study subjects. The study population has been described in detail elsewhere (7). Briefly, newly employed workers from a range of 12 occupational settings were recruited. Work places were identified through a number of means, including newspapers, television reports, job centers, and contacts through the principal investigators or other work places. Occupations were chosen to include a large proportion of subjects who were taking up full-time employment for the first time (46%). Furthermore, occupations for which high prevalence rates of musculoskeletal pain had previously been reported were selected. Subjects engaged in these occupations were identified from a number of sources (see Appendix A, available on the Arthritis & Rheumatism Web site at http:// www.interscience.wiley.com/pages/0004-3591/suppmat/ index.html). Baseline information. Baseline information was collected by means of a self-administered questionnaire. The majority of workers were contacted in person, either in groups or individually, and were asked to complete the questionnaire, which included information across the following 4 domains: manual handling activities, posture and repetitive movements, psychosocial factors, and environmental factors. Manual handling activities. Individuals were asked about several manual handling activities they had performed during the last working day. These included lifting weights with 1 hand, lifting weights with 2 hands, carrying weights on 1 shoulder, lifting weights at or above shoulder level, and pushing or pulling weights. Subjects were asked to estimate (with the use of a guide) the weights they had lifted and the duration of each task (in minutes). These questions had previously been validated by comparing the responses on self-administered questionnaires with the responses taken through direct observation techniques (8). Posture and repetitive movements. Questions on postures and repetitive movements were based on the same validated instrument (8). Questions were based on postures and repetitive movements adopted during the last working day and the amount of time that was spent in each position. Postures included sitting, standing, kneeling, squatting, working with hands at or above shoulder level, stretching below knee level, and driving. Repetitive movements included actions of the arms and wrists. Psychosocial factors. Work-related psychosocial factors were assessed by asking individuals about their job demands, job control, and social support, all of which were based on the Job Strain model described by Karasek (9), using questions previously studied in relation to musculoskeletal disorders (10,11). In that model, subjects with high job demands, little control over their work, and low social support from colleagues and supervisors are at potentially high risk of having poor health outcomes (12). Job demands included questions about the work pace, stress/worry, and whether individuals found their work to be monotonous or boring. Job control included questions on whether individuals thought they could make their own decisions within the work place and whether they thought they had the opportunity to learn new things at work. Subjects were also asked about social support from their colleagues as well as their overall job satisfaction. The General Health Questionnaire (GHQ), which has previously been validated in a number of settings, was included as a measure of individual psychological distress (13). Environmental factors. To assess work place environmental factors, individuals were asked whether they had worked in very hot, cold, or damp conditions during the last working day. Pain status. Pain status was assessed by asking individuals the following question, “Thinking back over the past month, have you had any ache or pain which has lasted for one day or longer?” If subjects responded affirmatively, they were asked to indicate on a line drawing of the body any site of such pain. The “pain drawings” used to determine pain status were coded in a blinded manner (with regard to all baseline information) according to the American College of Rheumatology (ACR) definition of widespread pain that was used in the criteria for fibromyalgia (14). That is, subjects with contralateral pain and pain in the axial skeleton were classified as having widespread pain. Followup. Subjects who were free of widespread pain at baseline were eligible for followup at 12 and 24 months. At followup, questionnaires were mailed to the study participants. Up to 2 reminders were sent to those who did not respond. Pain status at followup was assessed in the same way as at PREDICTORS OF NEW-ONSET PAIN IN NEWLY EMPLOYED WORKERS baseline. In addition, individuals were asked whether they had changed their job and, if so, whether this job change was because of aches and pains. Analysis. For the purpose of analysis, exposure variables were categorized as follows. For manual handling activities, the referent group consisted of subjects who did not perform these activities. The remaining subjects were dichotomized according to the midpoint of the average amount of weight that was lifted during the last working day. For postures and repetitive movements, the referent group consisted of subjects who did not adopt these postures or repetitive movements. The remaining subjects were dichotomized (or were split into 3 groups) on the basis of categories of time spent in each position. For psychosocial factors, subjects were dichotomized by collapsing the categorical scales into those with low exposure (referent group) and those with high exposure (comparison group). For environmental factors, subjects were dichotomized into those who did not work in these conditions (referent group) and those who did (comparison group). For the GHQ, the referent group consisted of subjects who scored 1657 zero. The remaining subjects were dichotomized according to the midpoint of the distribution of the GHQ scores. The study included repeated measures of exposures and outcomes and, hence, was subject to within-subject correlation. Generalized estimating equations were used to assess predictors of new-onset widespread pain at followup in order to take into account the repeated measures (15). As a result, one summary measure is obtained for the relationship between baseline exposures and outcome at 12 months, and exposures at 12 months and outcome at 24 months. The results are expressed as odds ratios (ORs) with 95% confidence intervals (95% CIs), adjusted for age, sex, and occupational group. To identify factors that best predicted the new onset of symptoms, the analysis was conducted in the following way. First, univariate associations were assessed for each potential risk factor, adjusting for age, sex, and occupational group. Second, to identify factors that predicted the new onset of widespread pain within each domain, multivariate models were constructed for each separate domain. Where 2 variables were strongly correlated, those with the stronger point estimate in the univariate analysis were included in the domain-specific models. Third, a final multivariate model was constructed by including factors from the individual domain-specific models that were considered biologically plausible, were statistically significant, and the OR was ⱖ1.5 or was ⬍0.67. Each predictor variable in the final multivariate model was examined for interactions with the followup period and within specific domains. Interactions were included in the final model if they were considered to be biologically plausible, they were statistically significant, the OR was ⱖ1.5 or was ⱕ0.67, and they contributed significantly to the final model. All analyses were conducted using the Stata statistical package (version 7.0; Stata, College Station, TX). RESULTS Figure 1. Distribution of subjects in the study of new-onset widespread body pain in newly employed workers. Newly employed workers were identified from 12 sources and invited to participate in the study. Subjects were asked to complete a questionnaire at baseline, 12 months (followup 1), and 24 months (followup 2). Prevalence of widespread pain. Figure 1 shows the distribution of subjects and the rates of new-onset widespread pain from baseline through to the second followup at 24 months. Approximately one-third of the study population were men. The median age of subjects who were eligible for followup at 12 months was 23 years (interquartile range 20–28 years). At baseline, 167 individuals reported widespread pain (15% prevalence; 18% in men, 11% in women) (P ⫽ 0.007). In a further 18 subjects, pain status was undetermined. Thus, 896 subjects (83%) were eligible for followup. Over all occupational groups, the followup rates at 12 and 24 months were 79% and 87%, respectively (Figure 1). Complete information on pain status was available for 703 (78%) and 476 (79%) subjects at 12 and 24 months, respectively (Table 1). There was wide variation in the rates of new-onset widespread pain by occupational group. At 12 months, symptom onset was lowest among postal workers (6%) increasing to ⬎30% in podiatrists and the army infantry; however, the prev- 1658 HARKNESS ET AL Table 1. Participation and rates of new-onset widespread pain, by followup period 12-month followup 24-month followup Occupation No. of eligible subjects Response rate, no. (%)* Prevalence rate, no. (%) Response rate, no. (%)* Prevalence rate, no. (%) Firefighter Police officer Army officer Army infantry Army clerk Dentist Podiatrist Nurse Forestry worker Retail worker Postal worker Shipbuilder Total 139 41 75 59 62 92 68 77 28 98 61 96 896 120 (86) 36 (88) 50 (67) 27 (46) 54 (87) 69 (75) 53 (78) 61 (79) 20 (71) 83 (85) 48 (79) 82 (85) 703 (78) 12 (10) 4 (11) 7 (14) 9 (33) 9 (17) 10 (14) 17 (32) 8 (13) 3 (15) 10 (12) 3 (6) 10 (12) 102 (15) 90 (83) 28 (88) 27 (63) 7 (39) 38 (84) 53 (90) 33 (92) 44 (83) 13 (76) 58 (79) 25 (56) 60 (83) 476 (79) 9 (10) 3 (11) 1 (4) 1 (14) 5 (13) 8 (15) 7 (21) 4 (9) 2 (15) 8 (14) 3 (12) 7 (12) 58 (12) * Excluding 49 subjects with missing information on pain status. alence did not differ by occupational group (P ⫽ 0.751). At 24 months, the new-onset rates ranged from 4% in army officers to 21% in podiatrists, and the prevalence did not differ according to occupational group (P ⫽ 0.492). Furthermore, there was no difference in the proportion of subjects reporting the new onset of wide- spread pain by followup period for any occupational group. The new onset rate of widespread pain was 15% and 12% at 12 and 24 months, respectively (Figure 1). At 12 months, the rate of new-onset widespread pain was significantly higher in women (19%) than in men (12%) Table 2. Work-related mechanical risk factors and new onset of widespread pain, univariate associations for manual handling activities* No. of subjects with new-onset widespread pain 12 months Manual handling activity Lifting with 1 hand Never ⱕ15 lbs ⬎15 lbs Lifting with 2 hands Never ⱕ24 lbs ⬎24 lbs Carrying on 1 shoulder Never ⱕ30 lbs ⬎30 lbs Lifting weights at or above shoulder level Never ⱕ23 lbs ⬎23 lbs Pushing Never ⱕ65 lbs ⬎65 lbs Pulling Never ⱕ56 lbs ⬎56 lbs 24 months Not exposed Exposed Not exposed Exposed OR 95% CI 296 137 161 42 31 25 192 116 104 24 18 16 1 1.7 1.9 Referent 1.1–2.7 1.1–3.3 270 165 153 43 30 27 202 105 104 30 12 16 1 1.3 1.7 Referent 0.8–2.1 1.0–2.8 482 53 58 76 11 14 330 44 37 47 5 6 1 1.1 1.6 Referent 0.6–2.0 0.9–3.0 448 63 80 71 16 14 324 47 39 43 9 6 1 2.0 1.7 Referent 1.2–3.3 0.9–3.2 393 92 106 69 18 14 284 70 57 36 11 11 1 1.5 1.7 Referent 0.9–2.5 0.96–3.0 448 65 80 72 13 16 321 51 36 42 7 9 1 1.6 2.3 Referent 0.9–2.9 1.3–3.9 * Adjusted for sex, age group, and occupation. OR ⫽ odds ratio; 95% CI ⫽ 95% confidence interval. PREDICTORS OF NEW-ONSET PAIN IN NEWLY EMPLOYED WORKERS 1659 Table 3. Work-related mechanical risk factors and new onset of widespread pain, univariate associations for posture and repetitive movements* No. of subjects with new-onset widespread pain 12 months Posture and repetitive movements Sitting Do not sit as part of job ⬍2 hours ⱖ2 hours to ⬍4 hours ⱖ4 hours Standing Do not stand as part of job ⬍15 minutes ⱖ15 minutes to ⬍2 hours ⱖ2 hours Driving as part of job No Yes Kneeling Never ⬍15 minutes ⱖ15 minutes Squatting Never ⬍15 minutes ⱖ15 minutes Bending Never ⬍15 minutes ⱖ15 minutes Stretching below knee level Never ⬍15 minutes ⱖ15 minutes Working with hands at or above shoulder level Never ⬍15 minutes ⱖ15 minutes Repetitive wrist movements Never ⬍2 hours ⱖ2 hours Repetitive arm movements Never ⬍2 hours ⱖ2 hours 24 months Not exposed Exposed Not exposed Exposed OR 95% CI 194 113 143 145 29 17 29 25 120 116 65 112 17 16 5 20 1 1.0 1.0 0.9 Referent 0.6–1.6 0.6–1.7 0.5–1.6 54 166 194 184 8 29 33 31 48 140 117 106 7 21 17 13 1 1.6 1.5 1.6 Referent 0.8–3.1 0.8–2.9 0.8–3.2 521 79 93 9 336 75 46 12 1 0.9 Referent 0.5–1.6 371 153 74 51 34 17 243 107 62 32 16 10 1 1.7 1.6 Referent 1.1–2.6 0.9–2.6 353 167 62 53 29 20 258 107 45 29 17 12 1 1.6 2.9 Referent 1.0–2.5 1.8–4.9 280 173 135 39 38 24 213 106 91 26 12 20 1.0 1.3 1.4 Referent 0.9–2.0 0.9–2.1 330 198 67 50 42 10 221 149 42 33 17 8 1 1.2 1.3 Referent 0.8–1.8 0.7–2.4 328 145 119 55 27 20 231 128 53 27 16 15 1 1.3 1.8 Referent 0.8–2.0 1.1–2.8 188 220 183 24 41 37 162 124 127 21 16 21 1 1.1 1.1 Referent 0.7–1.7 0.7–1.8 248 192 152 37 41 24 225 103 84 22 16 18 1 1.7 1.5 Referent 1.1–2.5 0.9–2.3 * Adjusted for sex, age group, and occupation. OR ⫽ odds ratio; 95% CI ⫽ 95% confidence interval. (P ⫽ 0.012), although this difference did not persist at 24 months (11% prevalence in men compared with 14% in women). Symptom onset was consistently higher among women for all age groups, but this difference was not statistically significant. Univariate associations. Table 2 shows the univariate associations between mechanical exposures and the new onset of widespread pain, adjusted for age, sex, and occupation. Many of the manual handling activities conferred an increased risk of symptom onset, including lifting ⬎15 lbs with 1 hand (OR 1.9, 95% CI 1.1–3.3), lifting ⬎24 lbs with 2 hands (OR 1.7, 95% CI 1.0–2.8), and pulling heavy weights of ⬎56 lbs (OR 2.3, 95% CI 1.3–3.9). Of the postures and repetitive movements, squatting for ⱖ15 minutes was the strongest predictor of new-onset widespread pain (OR 2.9, 95% CI 1.8–4.9), and those working with hands at or above shoulder level for ⱖ15 minutes had an 80% increased odds of developing widespread pain at followup (Table 3). Several work-related psychosocial factors showed an increased risk of new-onset widespread pain (Table 4). A statistically significant association was found for 1660 HARKNESS ET AL Table 4. Work-related psychosocial risk factors and new onset of widespread pain, univariate associations for job demand, job satisfaction, social support, control over work, and individual distress* No. of subjects with new-onset widespread pain 12 months Psychosocial factor Job demand Stressful work Never/occasionally At least half of the time Monotonous work Never/occasionally At least half of the time Hectic work Never/occasionally At least half of the time Job satisfaction Satisfaction with job Not dissatisfied (Very)/dissatisfied Social support Support from colleagues Not dissatisfied (Very)/dissatisfied Control over work Control over own work At least sometimes (Very)/seldom Learn new things At least sometimes (Very)/seldom Individual distress, by GHQ GHQ score 0 1–2 ⱖ3 24 months Not exposed Exposed Not exposed Exposed OR 95% CI 480 113 67 34 346 66 50 8 1 1.5 Referent 0.99–2.3 531 61 80 21 359 51 47 11 1 2.4 Referent 1.5–3.9 417 173 62 39 309 103 42 16 1 1.4 Referent 0.97–2.1 580 12 96 5 392 18 54 4 1 2.1 Referent 0.9–4.6 584 7 98 3 399 12 54 4 1 2.4 Referent 0.96–6.0 531 60 90 9 394 17 52 6 1 1.4 Referent 0.8–2.5 570 24 95 6 390 22 55 3 1 1.7 Referent 0.7–4.1 348 163 90 45 28 27 271 77 69 37 12 8 1 1.2 1.5 Referent 0.8–1.9 0.97–2.4 * Adjusted for sex, age group, and occupation. OR ⫽ odds ratio; 95% CI ⫽ 95% confidence interval; GHQ ⫽ General Health Questionnaire. monotonous work (OR 2.4, 95% CI 1.5–3.9), and increased, although not statistically significant, associations were apparent for stressful work, hectic work, low job satisfaction, and lack of support from colleagues. In addition those with high levels of individual psychological distress, that is, those who scored ⱖ3 on the GHQ, also had an increased risk of symptom onset (OR 1.5, 95% CI 0.97–2.4). Of the work place environmental factors examined, those working in cold conditions had a lower risk of symptom onset (OR 0.5, 95% CI 0.3–0.98) (Table 5). Domain-specific models. Several exposures were excluded from the domain-specific models because of strong correlations with other variables. Lifting with 1 hand was correlated with lifting with 2 hands (r ⫽ 0.56), and the latter was excluded from the manual handling model, pushing was correlated with pulling (r ⫽ 0.50) and was also excluded from the manual handling model. Lifting weights at or above shoulder level was correlated with working with hands at or above shoulder level (r ⫽ 0.47), and the former was excluded from the manual handling model. Kneeling was correlated with squatting (r ⫽ 0.60) and was excluded from the posture model. Therefore variables from the domain-specific models that met our criteria for inclusion in the final multivariate model were: lifting with 1 hand, pulling, working with hands at or above shoulder level, squatting, monotonous work, support from colleagues, hot working conditions, and cold working conditions. In the environmental factors model, there was evidence of a protective effect of cold working conditions. This is in contrast to the findings of other investigators (16,17), and there is no plausible hypothesized mechanism of action for such a protective effect (18). Therefore, cold working conditions was excluded from the final multivariate model. PREDICTORS OF NEW-ONSET PAIN IN NEWLY EMPLOYED WORKERS 1661 Table 5. Work place environment as risk factors for new onset of widespread pain, univariate associations* No. of subjects with new-onset widespread pain 12 months Environmental factor 24 months Not exposed Exposed Not exposed Exposed OR 95% CI 452 147 70 31 309 101 44 12 1 1.5 Referent 0.9–2.2 473 107 83 15 340 69 53 3 1 0.5 Referent 0.3–0.98 435 116 80 13 314 94 47 10 1 0.6 Referent 0.4–1.1 Work in hot conditions No Yes Work in cold conditions No Yes Work in damp conditions No Yes * Adjusted for sex, age group, and occupation. OR ⫽ odds ratio; 95% CI ⫽ 95% confidence interval. Final multivariate model. In the final multivariate model (Table 6), those who pulled heavy weights had an 80% increased, but not statistically significant, risk of symptom onset compared with those who did not perform these activities. Those who squatted for ⱖ15 minutes and those who thought their work was monotonous or boring had a significantly increased (approximately double) odds of developing new-onset widespread pain. Those who reported low social support from colleagues also had an increased risk of symptom onset, but this association was not statistically significant. None of the interaction terms we assessed contributed significantly to the final model. Using the factors with an odds ratio of ⱖ1.5 in Table 6. Final model of the predictors of new-onset widespread pain, multivariate associations* Variable Mechanical load Pulling Never ⱕ56 kg ⬎56 kg Squatting Never ⬍15 minutes ⱖ15 minutes Psychosocial factors Job demand Monotonous work Never/occasionally At least half of the time Social support Support from colleagues Not dissatisfied (Very)/dissatisfied Odds ratio 95% confidence interval 1 1.2 1.8 Referent 0.7–2.3 0.98–3.2 1 1.3 2.0 Referent 0.8–2.1 1.1–3.6 1 1.9 Referent 1.1–3.2 1 2.2 Referent 0.8–5.8 * Adjusted for sex, age group, occupation, and all other factors in the model. the final multivariate model, we were able to determine the rate of new-onset widespread pain in subjects exposed to a combination of these factors. The new-onset rate in any 12-month period increased from 11% (95% CI 9.3–13.7) in those exposed to none of these factors to 60% (95% CI 26–88) in those exposed to ⱖ3 factors (pulling ⬎56 lbs, squatting ⱖ15 minutes, monotonous work, and low support from colleagues). Respondents versus nonrespondents. Those who responded to the questionnaire and those who did not were found to differ on a number of potential risk factors at 12 and 24 months (see Appendix B, available on the Arthritis & Rheumatism Web site at http:// www.interscience.wiley.com/pages/0004-3591/suppmat/ index.html). However, due to the nature of the data collection, we were able to further examine the relationship between baseline predictors and outcome at 12 months stratified by response status at 24 months. We found no significant differences for the factors that were included in the final multivariate model. DISCUSSION This is the first study conducted within an occupational setting to examine the relative effects of workrelated mechanical, psychosocial, and environmental factors in relation to the new onset of widespread pain. Widespread pain is common in this young, newly employed population, and the risk factors are multidimensional, with work-related mechanical factors and workrelated psychosocial factors playing an important role. In interpreting these findings, there are a number of methodologic considerations. The current study was conducted among newly employed workers to minimize the “healthy worker effect.” Individuals working in 1662 well-established work forces may be exposed to different risk factors than those experienced by young, newly employed workers. For example, in the current study, relatively few subjects reported being exposed to adverse work-related psychosocial factors. We did, however, assess whether the healthy worker effect influenced the present findings by investigating the proportion of subjects who changed their job at followup and determining whether such job changes were due to aches or pains. A total of 79 subjects left their baseline occupation during followup, but only 1 of them reported any musculoskeletal symptoms. It therefore seems that the influence of the healthy worker effect in the present study was minimal. However, due to the nature of the jobs in the current study, this cohort may have been healthier than the general population, and some jobs required the completion of rigorous fitness tests prior to employment. Consequently, we may have underestimated the prevalence of new-onset widespread pain in relation to the general population. Due to the prospective design of the current study, with exposure being measured a year prior to the new onset of widespread pain, we were able to determine the temporal relationship between exposures and outcomes. We made no attempt to ascertain widespread pain status in the intervening months and are therefore unable to say anything about the new-onset rate during this time. However, this does not affect the internal comparisons between predictors and outcomes. Misclassification of pain status at followup may have made it more difficult to detect an effect. Changes in exposure may also have occurred in the intervening months. The suitable length of time over which to measure exposures, particularly work-related psychosocial exposures, is something that has recently been debated (19). We assessed interactions between the followup period and predictor variables to determine whether the relationship between exposures and outcomes had changed during the followup period. We found that the relationship had not changed significantly for any of those variables included in the final predictive model. The outcome ascertained in this study was widespread pain, which was defined according to the ACR guidelines used for the classification of fibromyalgia. Estimates of chronic widespread pain in the general population have been reported to be in the region of 10–13% (20–22), but we did not identify any studies conducted within the work place. Previously, we found that the majority of individuals who report widespread body pain have experienced their pain for 3 months or longer (20). HARKNESS ET AL The “new prevalence” rates observed in this study of young, presumably healthy, workers was high. Furthermore, a substantial proportion reported that their pain had limited their normal activities, either at work or at home (27% at 12 months and 31% at 24 months). Nevertheless, the proportion of these subjects with chronic pain is likely to be smaller, and it is unlikely (although not investigated) that more than a very small proportion would be positive for the other features, such as tender points, necessary to classify their condition as fibromyalgia. In addition, the prevalence of widespread pain at baseline was more common in men, whereas chronic widespread pain and fibromyalgia tend to be much more commonly reported in women. Thus, we caution that the high prevalence rates we obtained should not be taken as an indication of severe morbidity. Despite this, it is thus of substantial interest that work place psychological factors should show the positive associations observed. It would be necessary to undertake longer-term followup of these subjects to determine whether such factors have a similar, or even enhanced, role in predicting the future development of disabling, chronic widespread pain of the nature of fibromyalgia. The work place exposures we analyzed may not have been typical of a normal working day. However, in an attempt to assess this and in response to a direct question, we found that the majority of individuals reported that the demands of the last working day were much the same “as usual” (78% and 92% at 12 and 24 months, respectively). Importantly, the proportion of subjects reporting demands as being less physically demanding or more physically demanding during the last working day did not differ according to pain status at followup. We have found within this population, as with other study populations, that monotonous work is one factor that consistently predicts new-onset musculoskeletal pain (10,23). Although monotonous work could be a marker of repetitive tasks, such as movements of the arms or wrists, it was found to be independently predictive. Monotonous work may lead to increased psychological job stress, which might explain adverse health outcomes, including the onset of musculoskeletal pain (18). Another possible explanation is that subjects who perceive their work as monotonous or boring have a lower pain threshold than those who do not perceive their work in the same way. Other investigators observed a trend toward an increasing prevalence of fibromyalgia in association with increasing levels of physical work stress, but not mental stress (5). However, that study was cross-sectional, which makes it difficult to establish the PREDICTORS OF NEW-ONSET PAIN IN NEWLY EMPLOYED WORKERS temporal relationship and introduces the possibility of recall bias. It also was not clear how levels of physical and mental stress were measured. More recently, we examined the relationship between work place factors and the new onset of chronic widespread pain among an unselected population sample (6). As in the current study, the risk of the new onset of chronic widespread pain was found to be multifactorial, with the strongest associations for repetitive movements of the wrists, other regional pain symptoms, and individual psychosocial factors, in particular, illness behavior. However, that study was conducted among a stable work force, with the majority of subjects (95%) being in the same employment for the previous 36 months. In addition, job histories were collected retrospectively and may have been influenced by recall bias. Furthermore, that study did not include as extensive measurements of work-related risk factors as those collected in the current study. In that study, however, as in the current study, we did not measure major physical trauma (e.g., work place accidents), which have been implicated in the onset of widespread pain syndromes (3,4). Our conclusions are restricted to low-grade workrelated trauma. The current study has potential implications for interventions designed to prevent the new onset of widespread musculoskeletal pain in the work place. Such studies are currently limited. One small study of fibromyalgia patients (n ⫽ 86) found no difference in the job difficulty subscale of the Fibromyalgia Impact Questionnaire in 3 groups of subjects (control, education, and education plus physical therapy) (24). Further research is required to investigate the effectiveness of workrelated psychosocial interventions, for example, reducing the perception of monotonous work may be achieved through more-frequent job rotation or more-varied work tasks, whereas increased social support may be achieved through more contact with line managers or availability and accessibility to occupational health professionals. In summary, this study is the first to examine the relationship between work-related mechanical, psychosocial, and environmental factors and the new onset of widespread pain in a cohort of newly employed workers. We demonstrated that the new onset of widespread pain is common and the risk is multifactorial. The strongest independent predictors of symptom onset were, however, work-related psychosocial factors, and these associations have implications for the development of possible interventions. 1663 ACKNOWLEDGMENTS We would like to thank the individuals who permitted us access to their work forces and to all the workers who participated in the study. We would also like to thank Professor Nicola Cherry, who was involved in aspects of study design and conduct, and Christina Pritchard and Stewart Taylor, who were involved in the data collection. REFERENCES 1. McBeth J, Macfarlane GJ, Benjamin S, Silman AJ. Features of somatization predict the onset of chronic widespread pain: results of a large population-based study. Arthritis Rheum 2001;44:940–6. 2. Greenfield S, Fitzcharles MA, Esdaile JM. Reactive fibromyalgia syndrome. Arthritis Rheum 1992;35:678–81. 3. Al Allaf AW, Dunbar KL, Hallum NS, Nosratzadeh B, Templeton KD, Pullar T. A case-control study examining the role of physical trauma in the onset of fibromyalgia syndrome. Rheumatology (Oxford) 2002;41:450–3. 4. Buskila D, Neumann L, Vaisberg G, Alkalay D, Wolfe F. Increased rates of fibromyalgia following cervical spine injury: a controlled study of 161 cases of traumatic injury. Arthritis Rheum 1997;40:446–52. 5. Makela M, Heliovaara M. Prevalence of primary fibromyalgia in the Finnish population. BMJ 1991;303:216–9. 6. McBeth J, Harkness EF, Silman AJ, Macfarlane GJ. The role of workplace low-level mechanical trauma, posture and environment in the onset of chronic widespread pain. Rheumatology (Oxford) 2003;42:1486–94. 7. Nahit ES, Macfarlane GJ, Pritchard CM, Cherry NM, Silman AJ. Short term influence of mechanical factors on regional musculoskeletal pain: a study of new workers from 12 occupational groups. Occup Environ Med 2001;58:374–81. 8. Pope DP, Silman AJ, Cherry NM, Pritchard C, Macfarlane GJ. Validity of a self-completed questionnaire measuring the physical demands of work. Scand J Work Environ Health 1998;24:376–85. 9. Karasek RA. Job demands, job decision latitude and mental strain: implications for job redesign. Adm Sci Q 1979;24:285–311. 10. Macfarlane GJ, Hunt IM, Silman AJ. Role of mechanical and psychosocial factors in the onset of forearm pain: prospective population based study. BMJ 2000;321:676–9. 11. Leino PI, Hanninen V. Psychosocial factors at work in relation to back and limb disorders. Scand J Work Environ Health 1995;21: 134–42. 12. Karasek RA, Theorell T. Healthy work: stress, productivity and reconstruction of working life. New York: Basic Books, 1990. 13. Goldberg DP, Williams P. Users’ guide to the General Health Questionnaire. Windsor (UK): NFER-Nelson; 1988. 14. Wolfe F, Smythe HA, Yunus MB, Bennett RM, Bombardier C, Goldenberg DL, et al. The American College of Rheumatology 1990 criteria for the classification of fibromyalgia: report of the multicenter criteria committee. Arthritis Rheum 1990;33:160–72. 15. Liang KY, Zeger SL. Longitudinal data analysis using generalised linear models. Biometrika 1986;73:13–22. 16. Jin K, Sorock GS, Courtney T, Liang Y, Yao Z, Matz S, et al. Risk factors for work-related low back pain in the People’s Republic of China. Int J Occup Environ Health 2000;6:26–33. 17. Pienimaki T. Cold exposure and musculoskeletal disorders and diseases: a review. Int J Circumpolar Health 2002;61:173–82. 18. Zakaria D, Robertson J, MacDermid J, Hartford K, Koval J. Work-related cumulative trauma disorders of the upper extremity: navigating the epidemiologic literature. Am J Ind Med 2002;42: 258–69. 19. Hoogendoorn WE, Bongers PM, de Vet HC, Twisk JW, van 1664 Mechelen W, Bouter LM. Comparison of two different approaches for the analysis of data from a prospective cohort study: an application to work related risk factors for low back pain. Occup Environ Med 2002;59:459–65. 20. Croft P, Rigby AS, Boswell R, Schollum J, Silman A. The prevalence of chronic widespread pain in the general population. J Rheumatol 1993;20:710–3. 21. Hunt IM, Silman AJ, Benjamin S, McBeth J, Macfarlane GJ. The prevalence and associated features of chronic widespread pain in the community using the ‘Manchester’ definition of chronic widespread pain. Rheumatology (Oxford) 1999;38:275–9. HARKNESS ET AL 22. Buskila D, Abramov G, Biton A, Neumann L. The prevalence of pain complaints in a general population in Israel and its implications for utilization of health services. J Rheumatol 2000;27:1521–5. 23. Harkness EF, Macfarlane GJ, Nahit ES, Silman AJ, McBeth J. Mechanical and psychosocial factors predict new onset shoulder pain: a prospective cohort study of newly employed workers. Occup Environ Med 2003;60:850–7. 24. Burckhardt CS, Mannerkorpi K, Hedenberg L, Bjelle A. A randomized, controlled clinical trial of education and physical training for women with fibromyalgia. J Rheumatol 1994;21: 714–20.