Occupational causes of low-back pain

of low-back Scand J Work Environ Health 1989;15:54-59. Associations between occupational activities and low-back pain (LBP) were examined in a retrospective postal survey of 545 randomly selected adults. Each subject gave a lifetime occupational history on six specified physical activities and also a lifetime history of LBP. Among the 436 subjects answering the questionnaire, the lifetime incidence of LBP was 63 070. The occur rence of LBP was related by Cox's proportional hazards regression model to occupational activities in the year prior to the onset of symptoms. For the men the strongest associations were with heavy lifting and prolonged car driving. There was also an association with heavy lifting among the women. These risks were exaggerated in the subset of subjects whose LBP followed a chronic unremitting course. The findings are consistent with previous reports linking LBP with manual materials handling and driving at work. However, in this sample, less than 20 % of the cases could be attributed to such activities.

Low-back pain is a common and economically important problem in industrialized societies. It generates substantial demands on medical services and is a major cause of time lost from work (1-3). Most cases of lowback pain in adults of working age are ascribed to degenerative changes in the lumbar spine, but it is unclear whether these changes represent a single disease process or whether a variety of pathogenetic mechanisms independently affects different spinal structures. With the exception of neurological abnormalities of the lower limb due to prolapsed lumbar intervertebral disc, it has so far proved impossible to delineate patterns of symptoms and signs associated with specific pathology of the lumbar spine (4).
Mechanical stresses on the spine, in particular as a result of occupational activities, are widely believed to cause low-back pain, but their impact is hard to assess epidemiologically. Cross-sectional surveys of industrial groups may underestimate the contribution of mechanical factors because subjects with severe lowback pain have been selected out of the most physically demanding jobs. On the other hand, studies which ascertain cases through sickness absence or presentation to a doctor will tend to exaggerate the effects of occupation since subjects with physically demanding jobs are more handicapped by a given level of symptoms and are therefore more likely to take time off work and seek medical advice. a general population sample and relating the incidence of low-back pain to occupational activities prior to the onset of symptoms. We have also explored the possibility that the syndrome of low-back pain encompasses several etiologically distinct diseases by examining subcategories of low-back pain distinguished by the time course and radiation of pain.

Materials and methods
The survey was carried out in Whitchurch, a small market town in the south of England where the main sources of employment are agriculture, a paper mill, a silk mill, and service industries. A postal questionnaire was sent to a sample of 267 men and 268 women who were in the age range of 20-70 years and who were selected at random from the list of the local general practice. The subjects were asked to give a lifetime occupational history and to indicate for each job their age at its start and finish and whether an average workday entailed standing or walking for more than 2 h; sitting for more than 2 h; driving a car or van for more than 4 h; driving a truck, tractor or digger; lifting or moving weights of 25 kg or more by hand; or using handheld vibrating machinery such as chain saws or pneumatic drills.
A history of back pain was elicited with the aid of a diagram outlining the area between the lower costal margins and gluteal folds. The subjects were asked whether they had ever had pain that lasted for more than a day in this area.
Those who had were asked at what age their symptoms began; whether they first came on suddenly or gradually; whether the pain had ever spread down the leg and, if so, to mark on a diagram how far; whether they had suffered from pain for more than a total of 30 d ; what was the longe st period th at they had been completely free of pain since the on set of th eir symptom s; a nd whether the y had suffered from low-back pain in the past 12 months. Subjects who reported lowba ck pain in the past 12 months were asked whether th e pa in had made it difficult or impossible to walk, get up from an armchair, get out of a bath, get in and out of a car, put on shoes, or go up and down stairs.
Nonresponders were sent two reminders at intervals of six weeks and two months.
The an alysis was ba sed upon Cox's proporti on al hazards regression model (5) with allowance, where appropriate, fo r time-dependent cov a riates. As an illustration , when looking at occupational activities at the time of or shortly before the onset of sympto ms, we first considered sub jects who attained 15 years of age while employed and without a previous history of back pain . Thi s su bset of the sa mple provided information about th e risk of fir st develop ing back pa in at the age of 15 yea rs in relation to occup ational activities at th e time of the 15th birthday. Information was deri ved simila rly for each year of age up to and including 70 years. The method th en combined dat a from each age group to produce a summa ry estimate of risk. In on e mod el each acti vity was exam ined independently, a nd in a second all activities were examined simultaneou sly with the assumption that the risk from a mult iple expo sure (eg, to both liftin g and vibration) was the product o f the risk associated with each exposur e in isolatio n.
As well as examining associations with recent occupational activity, we also loo ked for longer term effects of wor k. Instead of classifying subjec ts by occup ational activities at th e time of their last bir thday, we grouped them according to whether or not they had been exposed to activities for more than half the ir pre viou s worklife. Otherwise the method of an alysis was similar, risk estimates bein g derived for su bjects with such exposur e relati ve to those without.

Results
Four hundred and thirty-six que stionnaires were returned , giving an overall response rat e of 81 0,70 (75 % of the men and 88 % of the women). Th e response tended to be higher in the older age groups, ran ging from 78 % for subjects age d 20-29 years to 89 IT/o for those aged 60 years and over.
Th e lifetime incidence of symptoms amon g the subjects who completed the questionnaire is given in table I . Sixty-four percent of the men and 61 % of the women reported that they had suf fered from low-back pain at so me tim e. Subjects who respo nded to th e first mailing reported symptoms more frequently (65 %) than tho se who only replied to reminders (56 %). Sciatica, defined as pain radiating below the knee, occurred more oft en in the women (20 % ) than in the men (10 % ), but the time course of symptoms was similar for both sexes. We excluded from furth er analysis eight subjects whose back pain first sta rted before 15 years of age and II who could not remember when their sympto ms began . Table 2 shows the risk of low-back pain in successive birth cohorts. At an y given age, recent generations tend ed to report a higher inciden ce of symptoms than tho se born earlier. Th is patt ern was ob served both fo r pain of sudden onset and also for pa in that cam e on gradua lly. To co ntr o l any confounding effects arising from differences between birth coho rts, subseq uent analyses were stratified by year of birth (in lO-year groupings), as well as by age at the onset of symptoms. Table 3 lists the mo st common occupation s within the study sample and shows the freq uency with which different act ivities were reported in association with these jobs. The distribution of activities confor ms with what would be expect ed from a knowledge of work practices. The apparent anomaly tha t not all truck and van drivers repor ted driving trucks ari ses becaus e light vans are usuall y cla ssed as cars .
Th e risks of back pain in relation to occupational ac tivities on the birthday precedin g th e onset of symptoms ar e given in table 4. The strongest associations were for lifting and mo ving weights over 25 kg with relat ive risks of 2.0 for bo th the men and th e women in th e simultaneo us a nalysis of all activities. Car dri ving also carried an incr eased risk for the men, but not Low-back pain with the lon gest pain-tree period lasting 30 d or less' ("u nremitti ng" pain)    a For each activity, ri sks have been ca lculated for the subjects who reported exposu re relative t o those wh o were in work but une xposed .   for the women. The re was a trend towards more back Table 8. Levels of disab il it y among th e subj ect s wi th low-back pain in tho se whose jobs requ ired them to sit for more pain during the past 12 months accordin g t o th ei r employment than 2 h per day, but not to the point of stati stical sig-on th e birthday prior to the onset of sy mptoms." nificance at a 5 070 level. based on on ly one exposed case.

Persons in jobs
Possible wilh diffic ulty 23 37 29 38 We looked at subcategories of back pain to see if there were any differences in their association s with Getting up from an armchair occupational activity. The distinctions between pain No diff iculty 18 29 29 38 of sudden and gra dual onset and between pain lasting In this subcategory the associations described for back among the men and 2.9 among the women. jects who ha d suffer ed from low-back pain dur ing the a Subjects without employment havebeen excluded.
past 12 months, according to their employment on the birthday prior to the first onset of symptoms. Disability was slightly greater among the subjects who se jobs had entailed hea vy lifting th an among those with other types of work.

Discussion
In this study, cases of low-back pain were ascertained solely on the basi s of reported symptoms. With thi s method th e wording of key questions can have a substantial effect on estimates of disease frequency (6). Nevertheless, our findin g of a lifetime incidence of lowback pain of 64.5 ltfo is comparable to the rates of 62 070 found for residents of Copenhagen aged 30-60 year s (7),69.9 0J0 for men aged 19-54 years in Vermont in the United States (8), and 49 070 for women in Goteborg, Sweden (9). Moreover, the precise definition of cases is likely to be less crucial in comparisons of disease rates between occupations than it is for overall estimates of incidence. Our finding that successive birth cohorts reported the development of low-back pain at any given age with increasing frequency ha s been described before (10). It is po ssible that the incidence of low-back pain is rising. Certainly in Great Britain over the pa st 30 years there ha s been an increase in sickness absence attributed to low-back pain (11,12) and in rates of consultation with general practitioners for low-back pain (13)(14)(15). Alternatively, the phenomenon may be due to deficiencies of recall, older subjects failing to remember remote epi sodes of low-back pain. To exclude any bias arising from this cohort effect, we allowed for period o f birth in our analysis of occupation.
Unfortunately, there was no reli abl e sta nda rd against which we cou ld assess the ac curacy of occupational act ivities reported by our subjects . However, the activities which they described were consi stent with the jobs in which they said they worked (table 3), and previous studies have indicated that ability to recall past occupations is generally good (16,17). The reliabilit y of self-repo rted occupational activities ha s been asses sed directly by Baty and his colleagues in a study of nurses (18). They found that estimates o f time spent sitting and sta nding or walking agreed clo sely with those of trained observer s, but reports of time spent in other acti vities, such as kneeling, were less reliable. Our questionnaire required subjects only to dichotomize their exposure to each activity and, as such, should have been more valid. However, it is possible that the crude classification which this response entailed tended to obscure some associations with lowback pain.
The subj ects may also have had difficult y in recalling accurately the timing of jobs and the on set of lowback pain. However, th e fact that associat ion s of lowback pain with activities in th e year preceding the on -58 set of symptoms were broadly similar to tho se with lifetime occupational activity suggests that no serious bias resulted from inaccuracies in the timing of symptoms or job changes, The strongest occupational associations that we found were with heavy lifting, particularly when the low-back pain was unremitting. One explanation for thi s finding might be that subjects whose jobs ent ailed heavy liftin g had a lower threshold for reporting symptom s, becau se the symptoms interfered more with the ir work. However , the reported levels of disability for everyday ta sks , such as getting out of a bath (table 8), do not support thi s hypothesis. If anything, the indi viduals who se jobs involved lift ing reported higher levels of disability. Nor is it likel y that the asso ciation with lifting resulted from a response bias, since it was present not only among first-time responders but also among subjects who only replied after a reminder. A similar association has been observed in other studies (19)(20)(21)(22)(23), though not consistently (24)(25)(26). Possibly th e failure to det ect an effect in some investigations stems from differences in the definition of lifting. It ma y only. be th e handling of heavier weight s that is important.
Driving a car for more than 4 h a da y was associated with low-back pain in men but not with low-back pain in women . However, the number of women reporting the activity was small, and the confidence intervals around their risk estimate were correspondingly wide. Car driving ha s previously been reported as a risk factor for lumbar disc prolapse (27) . Postulated mechanisms include mechanical stresses on the spine from the sitting posture and from vibration. Exposure to vibration is even greater among tractor and truck drivers, occupations which have frequently been identified as prone to low-back pain (28)(29)(30)(31) and prolapsed lumbar disc (27). Ho wever , no such risk was apparent in our study . We did find associat ion s with jobs that entailed sitting for more than 2 h a day, but they were weak and not statistically significant, except among women with prolonged exposure. The results of other studies which have looked at the effects of sitting have been inconsistent (19,20,28,32) . In compariso n with other sedentary occupations, car dri ving allows little opportunity for changes of po sture or intermittent standing. Perhap s it is the continuous nature of po stura l stresses which cau ses back pain in drivers.
Of the subcategories of low-back pain which we examined , onl y unremitting back pain showed clearl y distinctive features . This pattern of symptoms is not known to indic ate any specific spinal pathology , and it is possible that the higher risks which were found fo r unremitting pain occurred by chance. However, the observation is worth testing in future studies.
Our data give strong support for a role of regular heavy lifting in the etiology of low-back pain and add weight to the evidence implicating occupational driving as a risk factor. At the same time , ho wever, the y suggest that suc h acti vities acc ount for onl y a small proportion of the total burdenof low-back pain in the gener a l populati on . Our es t imates of the f ractio n o f d isease attributable to heavy lifting and car driving a re 14 a n d 4 %, re sp ectivel y . E ven if a llowance is m ade for uncerta in tie s in the se figu res , a su bs ta ntial p roport ion of cases rem ain unexplai n ed .