However, regional pain complaints and associated disabilities are not a simple consequence of physical stresses to tissues. There is strong evidence that they are influenced also by psychological factors such as low mood and a general tendency to worry about common somatic symptoms (somatising tendency) [3,4]. In addition, culturally determined health beliefs could also have an important role, and may explain large variations in the incidence and prevalence of pain and disability that have been observed between countries [5,6], and within countries over time [5]. It is important to understand the contribution of these psychosocial influences if preventive measures are to be optimised.

To help advance knowledge in this area, a multi-centre international study, CUPID (Cultural and Psychosocial Influences on Disability), has been established. The study, which is being carried out in 19 countries (both developing and developed) from six continents, involves a baseline cross-sectional survey that will allow comparison of rates of regional pain and associated disability in samples of workers who carry out similar physical activities, but in widely different cultural environments. This is followed by a longitudinal component, which explores predictors of persistent and newly incident pain.

Within each participating organisation, a manager agreed to act as a coordinator for data collection. The coordinator distributed a self-administered questionnaire to all employees in relevant jobs, with a covering letter from the survey team. Completed questionnaires were then returned to the survey team via the coordinator. A total of 3187 questionnaires were distributed to 1074 nurses, 425 office workers, 380 sales/marketing personnel and 1308 transportation operatives. No reminders were sent to non-responders.

The questionnaire was a Japanese translation of the survey instrument that is being used throughout the CUPID study. The accuracy of the translation was checked by independent back-translation to English and comparison with the original. Amendments were then made as necessary. Among other things, the questionnaire asked about demographic characteristics, hours of work and duration of employment in current job, whether the job involved certain specified activities in an average working day, job satisfaction, mental health, indicators of tendency to somatise, experience of pain during the past month and past year at each of six anatomical sites (low back, neck, shoulder, elbow, wrist/hand and knee), disability for specified everyday tasks arising from such pain, and absence from work in the past year because of musculoskeletal pain or for other reasons. Mental health (mood) was assessed from the relevant subscale from the SF-36 questionnaire [7], and was graded to three levels defined by approximate thirds of the distribution of scores in all subjects combined. Somatising tendency was assessed using a subset of items from the Brief Symptom Inventory (BSI) [8], and was graded according to the number of symptoms (out of a total of seven) that were reported as causing at least moderate concern in the past week.

Data from the completed questionnaires were entered onto computer, and after checks for errors, were analysed using SPSS Version 15 and STATA Version 10 software. Because a major focus of the study was pain and disability during the past year, subjects were excluded from the main analysis if they had worked in their current job for less than a year.

In addition to the compilation of simple descriptive statistics, logistic regression was used to explore associations with regional pain (classified in various ways) and associated disability and sickness absence. Pain at an anatomical site was considered disabling if during the past month, it had made at least one of the everyday activities specified in the questionnaire difficult or impossible. These activities were: getting dressed (all sites of pain), doing normal household jobs (all sites of pain), cutting toe nails (low back), combing or brushing hair (shoulder), bathing/showering (shoulder), opening bottles, jars or taps (elbow and wrist/hand), writing (wrist/hand), locking and unlocking doors (wrist/hand), walking up and down stairs (knee) and walking on level ground (knee). When looking at associations with occupational activities, we defined for each site of pain, an activity in an average working day that could cause physical stress to local tissues. These activities were: lifting weights of ≥25 kg by hand (low back); work with the hands above shoulder height for ≥1 hour in total (neck and shoulders); repeated bending and straightening of the elbow for ≥1 hour in total (elbow); use of a keyboard or other repetitive movements of the wrist/fingers for ≥4 hours in total (wrist/hand); and kneeling or squatting for ≥1 hour in total (knees). Associations in the logistic regression analyses were summarised by odds ratios (ORs) with associated 95% confidence intervals (CIs).

Table 1 summarises various characteristics of the participants. Most of the nurses were women, whereas almost all of the sales/marketing personnel and transportation operatives were men. The majority of subjects were employed full-time, including 30% of the sample (mostly sales/marketing personnel and transportation operatives) who indicated that they worked for more than 60 hours per week. Reported occupational activities were much as would be expected, with a high frequency of keyboard use by office workers (89%). Transportation operatives and nurses had the highest prevalence of heavy lifting (83% and 66% respectively) and of repeated bending and straightening of the elbow (78% and 72%). Rates of job satisfaction were relatively low in office workers (28%) and sales/marketing personnel (31%). Poor mental health and tendency to somatise were most common among the nurses. In the study sample overall, the somatic symptoms most frequently reported as distressing were nausea or upset stomach (14%), weakness (12%) and faintness or dizziness (8%).

Table 2 shows the prevalence of pain at different anatomical sites in the study sample as a whole. The lower back was the site most commonly affected by pain, with a prevalence of 28% in the past month. Next most common were pain in the neck (21% in the past month) and shoulder (17%). In comparison, pain in the elbow and wrist/hand was much less frequent. The sites most commonly affected by disabling pain in the past month were the lower back (11%) and knee (8%). Only 4% of subjects had been absent from work during the past year because of low back pain, and absence because of pain in the elbow or wrist/hand was extremely rare.

The prevalence of regional pain by occupational group is summarised in Table 3 (data for men and women separately are given in Supplementary Tables 1 and 2). At almost all anatomical sites, pain in the past month was most common in nurses or transportation operatives, and least frequent in sales/marketing personnel. However, office workers had the highest prevalence of sickness absence in the past year attributed to regional pain (11%). A total of 251 subjects (11%) reported pain in the past month at ≥3 anatomical sites, 744 (32%) reported disabling pain at one or more sites during the past month, and 125 (5%) indicated that they had taken sickness absence during the past year because of regional pain.

Table 4 gives results from logistic regression analyses exploring risk factors for pain at different anatomical sites. For each site, two outcomes were examined – any pain in the past month and disabling pain in the past month – the comparator in both cases being no pain at the site in the past month. All analyses were adjusted for sex, age, mental health and occupational group. Significant associations with locally stressful physical activities were observed for pain in the low back (lifting ≥25 Kg), wrist/hand (use of use of keyboard or repeated movements of a hands/fingers for ≥4 hours) and knee (kneeling or squatting for ≥1 hour). However, the strongest and most consistent associations were with somatising tendency. For disabling pain in the low back, neck and shoulder, the ORs for report of ≥2 v 0 distressing somatic symptoms were all 4.5 or higher. Associations with poor mental health (not shown) were much weaker than those with somatising tendency, and not statistically significant.

Table 5 presents findings from two regression analyses, one for the risk of pain in the past month at ≥3 anatomical sites, and the other for disabling pain at one or more anatomical sites in the past month. In each case, the comparator was no pain at any site in the past month. Both variables were strongly associated with somatising tendency and showed a clear, progressive increase in risk in relation to the number of stressful physical activities reported. In addition, both were more frequent at older ages. Associations with poor mental health and job dissatisfaction were much weaker.

The occupational groups that were studied cannot necessarily be regarded as representative of the general population of working age in Japan. Nevertheless, they encompass a range of occupational tasks, both manual and non-manual, and provide useful insights into patterns of musculoskeletal symptoms and disability in a cultural environment that is notably different from that in, say, Western Europe. Furthermore, the high response rate that was achieved makes it likely that the samples of workers who participated were fairly typical of the occupational groups from which they were drawn.

A concern always in international studies of this type is that the meaning of questions may be distorted in translation between languages. Thus, care was taken to check the accuracy of the Japanese questionnaire by back-translation to English. It remains possible that a term such as “pain” is understood somewhat differently in Japan. However, this should not affect the relative frequency of the symptom at different anatomical sites, and is less likely to have been a problem in relation to more objective outcomes such as sickness absence.

Another possible source of error was incomplete recall of symptoms, particularly if they last occurred many months before the questionnaire was completed. For this reason, we based most of our analysis on pain and disability that was reported in the past month. An exception was sickness absence, for which a longer time period was required to give meaningful numbers of cases. However, we would expect spells of sickness absence to be more memorable than more minor episodes of pain.

The prevalence of pain at most of the anatomical sites considered was somewhat lower than has been recorded in UK workers who were surveyed using similar questions [6]. For example, low-back pain in the past month was reported by 28% of the Japanese workers as compared with 28% in a sample of white UK office workers and 37% in a group of white UK manual workers; while the corresponding figures were 21% v 26% and 23% for neck pain, 17% v 20% and 24% for shoulder pain, and 5% v 10% and 9% for elbow pain. More remarkable, however, is the much lower prevalence of wrist/hand pain in Japanese workers (7% v 30% and 23%). This lower prevalence extended to Japanese office workers (6% with wrist/hand pain), most of whom were regular users of computer keyboards. The difference in the prevalence of wrist/hand pain between Japanese and UK office workers was much larger than that between manual and non-manual workers in the UK, or between white workers in the UK and those of South Asian origin [6].

Also notable is the low rate of sickness absence that was attributed to regional pain complaints. Overall, only 4% of study participants had been absent from work in the past year because of low back pain, 2% for neck pain, 1% for shoulder pain, 0.3% for elbow pain and 0.4% for wrist/hand pain. In comparison, reported rates in UK workers were more than three times higher [6]. Workers from Japan tend to claim compensation and take time off work for illness attributed to occupation less often than their counterparts in the United States [9]. However, the differences we found are not explained simply by low overall rates of sickness absence in Japan – 16% of participants reported absence in the past year because of non-musculoskeletal illness. Rather the proportion of absence attributed to musculoskeletal disorders was much lower than in the UK.

Earlier studies of musculoskeletal symptoms in Japan have focused mainly on low back pain [10-22], with prevalence rates varying from 13% (in female nursing students [18]) to 83% (in nurses [19]), according to the population studied and case definition. Where assessed, rates of neck pain have been lower than those for low back pain in the same study [16-19], and the prevalence of pain in the wrist or hand has been even lower [19,21].

Although there are many published surveys of regional pain in other countries, few studies to date have compared rates of musculoskeletal illness between countries, using standardised methods for data collection. In an analysis of data from surveys of the general adult population in 10 developed and seven developing countries, the age-standardised prevalence of chronic back pain was somewhat higher in developing countries (24.3%) than in developed countries (18.5%) [23]. A comparative survey of nursing personnel found a higher 12-month prevalence of back complaints among Greek hospital nurses (75%) than in Dutch nurses and caregivers employed in nursing homes (62%) [24]. And in another study, rates of pain among manual workers were substantially lower in Mumbai, India, than in the UK, at each of five anatomical sites (low back, neck, shoulder, elbow and wrist/hand) [6]. For office workers, the differences were much smaller.

Within our Japanese sample of workers, analysis of risk factors for regional pain revealed expected associations with stressful physical activities. However, associations with somatising tendency were stronger, especially when pain was disabling. Given that the data analysed were cross-sectional, it is possible that the observed associations between physical activities and regional pain arose in part because of greater awareness, and therefore more frequent reporting, of such activities among workers who found them painful. It seems less likely, however, that the presence of back, neck or arm pain would cause a person to over-report worry about somatic symptoms such as nausea, weakness, or faintness and dizziness. Furthermore, in other countries, longitudinal studies have found that somatising tendency predicted the future incidence and persistence of musculoskeletal pain [3,4,25,26], and was associated with subsequent poor outcome in patients presenting to primary care or treated by physiotherapy for musculoskeletal disorders [27-30]. Tendency to somatise has also been associated with other complaints, including irritable bowel syndrome [31] and report of symptoms following exposure to pesticides [32]. In comparison with somatising tendency, low mood was a much weaker risk factor for regional pain in the Japanese workers.

In contrast, neither physical activity nor somatising tendency were clearly related to sickness absence because of regional pain, which was associated much more strongly with absence attributed to non-musculoskeletal disorders. It may be that in Japan, the major determinant of variation in rates of absence ascribed to musculoskeletal symptoms is not differences in the occurrence of such symptoms, but differences in workers’ general inclination to take sickness absence when they perceive a health problem.

In summary, this study provides further evidence that the prevalence of musculoskeletal symptoms varies importantly between countries, and suggests that, as in the UK, a major risk factor for musculoskeletal complaint in Japan is tendency to somatise.