British Journal of Industrial Medicine 1992;49:154-162

Individual and work related factors associated withsymptoms of musculoskeletal complaints. II Differentrisk factors among sewing machine operators

R H Westgaard, T Jansen

AbstractIndividual and work related risk factors in thedevelopment of occupational musculoskeletalcomplaints were studied in a group of 210female production workers, mainly sewingmachine operators. Another group of 35female employees performing secretarial orlaboratory duties were also included. Theproduction workers had significantly highersymptom scores with respect to self reportedmusculoskeletal complaints -than the groupwith more varied work tasks for the head, neck,shoulders, and arms, but not for the low back,hips, and the lower extremities. No significantdifferences were found in symptom level bet-ween geographically separate groups ofproduction workers with similar work tasks.The main individual risk factor identified inthis study was the experience of previous,similar symptoms in the same body region, butthis factor only accounted for 2-3% of totalvariance in symptom score for the neck andshoulders. Other individual factors ofimportance for symptoms in the neck andshoulders were "signs of psychologicalproblems" and "tendency of muscle tension,"but these only account for about 1% of totalvariance in symptom score. Symptoms in thehead and low back showed complex relationswith individual parameters.

It has long been recognised that workers withpredominantly repetitive work tasks, or those main-taining fixed postures for long periods, have an

Division of Organisation and Work Science, theNorwegian Institute of Technology, Trondheim,NorwayR H WestgaardNational Institute of Occupational Health, Depart-ment of Physiology, PO Box 8149 Dep, Oslo 1,

NorwayT Jansen

increased risk of developing work related musculo-skeletal illnesses.'2 Most of these studies havefocused on the detrimental health effects of the worksituations, whereas little attention has been given tothe considerable heterogeneity in response to thework strain. Work tasks considered to be a majorfactor in early retirement of some workers does not,apparently, affect other workers at all. A betterunderstanding of factors underlying this diversity ofresponses may help in the design ofthe work stations,perhaps by including some of the wider work organ-isational and psychosocial issues in the planning.The presence of individual risk factors in the

development of occupational musculoskeletal illnes-ses is only sparsely documented. Kilbom' reportedthat low muscle strength was a risk factor fordevelopment of shoulder, neck, and arm disorders ina group which frequently exerted high level externalforces. This was not the case for another groupexposed to low level postural loads. Psychosocialfactors in the work environment or at home have beensuggested as risk factors.45 The psychological state ofthe worker is another possible risk factor. In Norwaya large fraction of workers receiving disability all-owances due to chronic tension myalgia suffer fromelements of psychiatric illnesses.6 Muscle pain alsoappears to correlate with specific psychological statesor traits.7The present study of risk factors contributing to

the development of occupational musculoskeletaldisorders is based on interviews ofproduction work-ers in a clothing company, mostly sewing machineoperators. As well as symptom state,8 informationrelating to the life situation of the subject wasincluded. The operators had a mean static load in thetrapezius muscles of about 5% ofmaximal voluntarycontraction, and recorded frequent sick leaves due tomusculoskeletal injuries.9 It is also generally knownthat these work tasks are associated with a high rate ofmusculoskeletal injuries, particularly in the shoulderand neck region.'"" The study therefore allowed usto look for factors modulating the health effects of awork load known to be detrimental to the health of alarge fraction of the workers.Most studies showing an association between

154

on August 26, 2019 by guest. P

rotected by copyright.http://oem

.bmj.com

/B

r J Ind Med: first published as 10.1136/oem

.49.3.154 on 1 March 1992. D

ownloaded from

Symptoms of musculoskeletal complaints. II Different riskfactors among sewing machine operators

individual risk factors and the development ofmusculoskeletal injuries use study populations withlow or unspecified work loads. A low risk of workrelated musculoskeletal complaint may allowindividual risk factors to show up more clearly, and itwas of interest to see if individual risk factorsremained important when the aetiology was probablydominated by postural muscle strain at the work-place.

MethodsThe study is based on 210 production workers and 35other employees with varied work tasks, all women,employed by a Norwegian clothing company. The 35comprised all available female employees not per-forming production woork tasks. About half theproduction workers ( 118) were sewing machineoperators producing thermal clothing at three towns,Moss (18 subjects), Halden (44 subjects), and Rjukan(39 subjects), and one village, Tinn (17 subjects) insouthern Norway. Six subjects at Halden and one atRjukan were combining this work with other duties,such as other production work or supervisory duties.The remaining workers were all employed at Moss.Two groups of 34 and 25 subjects performed sewingand welding tasks in the production of off shoresurvival suits. The final 26 production workers wereperforming various sewing tasks not strictly coincid-ing with the above group definitions. The non-production ("office") group consisted of 31 subjectsperforming general office work, mainly secretarialduties at the company headquarters, or a combina-tion oflaboratory and office work (four subjects). Thework situation for all members of the office groupallowed considerable flexibility in alternating bet-ween different work tasks, although prolongedperiods with typing might occur.The sewing machine operators adopted a seated

posture with the lower arms moderately extended infront of the body. The table surface was adjustablefor height and was usually placed just below elbowheight. Thus the external load on the neck andshoulders was quite low in static biomechanics terms,but the need for continuous use ofthe arms over largeworking areas made these work situations strenuous.Some differences existed in the type of thermalclothing produced at the different departments: mit-tens were produced in Halden, other small garmentsin Tinn, and large garments in Rjukan whereas theproduction in Moss consisted of large garmentsincluding prototypes of new designs. Sewing opera-tions in the production ofoffshore survival suits weresimilar to those of thermal clothing, except that thegarments were heavier. Welding operations requiredpositioning of material on the welding machine everyone to two minutes, but the arms were resting whilethe welding took place. The welding task therefore

Table 1 Subject related parameters in the questionnaire

Work task variablesWork taskWork timeGeographical location

Health related variablesMusculoskeletal pain before employmentMusculoskeletal pain in present work situationCause of musculoskeletal painSpecific work task(s) associated with painUse of pain relieving drugsMedical consultation due to pain (date)Short term sick leave due to pain (date)Long term sick leave due to pain (date)Physical therapy (No last year and in total)

Background variablesAgeMarital stateNo of childrenLiving conditionWork load at homeEffort invested in leisure time activitiesRelationship to close familySigns of psychological problemsTendency to muscle tensionEconomic situationPhysical activities/training programmes

allowed regular rest pauses, at least with regard tomovement of the arms.Symptoms of pain or discomfort were scored for

eight body regions (the dorsal aspects of the head,neck, shoulders and upper arms, lower arms, lowback, hip, thighs, and knees and ankles) according tothe system described in the preceding paper.8 Apartfrom data on symptoms, three categories ofquestionswere included in the questionnaire: (1) related towork task, (2) related to symptoms, and (3) related toindividual factors (table 1). The variables related towork task gave factual information on geographicallocation, work task, and hours at work. Variablesrelated to symptoms included symptoms of painbefore present work situation (as a general question),whether the respondent related painful experiencesto her working conditions, whether specific worktasks were particularly strenuous, use of medicationto relieve pain, long term or short term sick leave dueto pain, treatment by physical therapy, and date oflast medical consultation due to pain (verified bymedical records). Finally, the individual factorsincluded age, marital state, number of children,living conditions, work load at home, leisure timeactivities, relationship to close family, signs ofpsychological problems, self reported tendency ofmuscle tension, economic situation, and amount ofphysical exercise.Most parameters that were not factual information

were scored on a three point scale indicating low,intermediate, and high values. The scoring wasperformed by the interviewer, after a discussion withthe respondent when the content of the differentquestions were explained. All interviews were carriedout by a consultant of physical medicine (TJ). The

155

on August 26, 2019 by guest. P

rotected by copyright.http://oem

.bmj.com

/B

r J Ind Med: first published as 10.1136/oem

.49.3.154 on 1 March 1992. D

ownloaded from

Westgaard, Jansen

parameter 'signs of psychological problems" wasscored on the basis of the interviewer's impression ofthe subject after detailed questioning. High valueswould indicate recurring depression or anxiety, oftenrelated to specific events. Information on psycho-logical problems was recorded before the recordingof pain symptoms. The scoring of this parameter wasperformed about two months later on the basis of therecorded information, when there was no recollec-tion of the pain symptoms. If in doubt the parametervalue was also considered by RHW.The parameter "relationship to close family" was

used as an indicator of the general psychosocial lifesituation of the respondent. The study did notinclude an indicator of psychosocial situation atwork, but the impression when carrying out theinterviews was that work satisfaction in general washigh. The parameter "tendency to muscle tension"was the response to a question as to whether thesubject had noticed herselfgenerating muscle tensionnot part ofa purposeful movement. The question wasincluded half way through the series of interviews asseveral of the first respondents volunteered thisinformation. Consequently many "missing" valueswere recorded. The measurement of amount ofphysical exercise was related to the level of conditiontraining (jogging, swimming, handball). The para-meter "economic situation" was quantified on thebasis of need to work due to financial strain.

Statistical analysis was performed as a generalisedtest of differences in level of symptoms at differentbody locations with respect to the different indepen-dent variables. The Kruska-Wallis test was used inthe univariate analysis. When comparing frequenciesthe Fisher's exact test was applied. In the multi-variate analysis linear regression models with back-ward variable selection were used. All statisticalanalyses were performed with the SAS statisticalpackage and p < 0 05 was taken to indicate statisticalsignificance.

ResultsSYMPTOMS OF PAIN OR DISCOMFORT DURING WORKIn response to the general question on experience ofpain or discomfort at work, 200 of 210 (95 %)production workers answered in the affirmative and156 (74%) were convinced that their problem wasrelated to their working conditions. These percent-ages were clearly lower among the 35 subjects in theoffice group in which 25 (71%) reported pain ordiscomfort at work and 16 of these (46%) wereconvinced that the problem was due to workingconditions. Subjects convinced of a causal relationbetween their subjective symptoms and workingconditions were able to identify specific work tasksthat, in their opinion, were associated with thedevelopment of such symptoms.

Both the production workers and the office grouprecorded a high rate of medical consultations due topain or discomfort in the musculoskeletal system(70% and 51% respectively). The main differencebetween the two groups was the outcome of theconsultation; 76% of the production workers recor-ded a subsequent sick leave compared with 39% forthe office group. The remaining patients of the officegroup usually received physical therapy while carry-ing on working (61% v 16% for the productionworkers). This difference may be due to the moreflexible work situation of the office group, and alsoto less symptoms at the time of the medical con-sultation,8 which would reduce the need of sick leaveto cope with their complaint.

SYMPTOMS OF PAIN OR DISCOMFORT RELATED TO THEWORK SITUATIONFigure 1 shows the mean symptom score of differentbody regions for production and office workers. Thesymptom scale is based on two parameters, intensityand frequency of symptoms, which are added toproduce a combined symptom score ranging from nosymptoms (a score of 0) to daily occurrence of severesymptoms (a score of 6; see8). The individual differen-ces and the high frequency of symptoms in the studypopulation made it advantageous to treat the symp-tom scores as continuous parameters rather than to

4,

3-

0

E0

2-

(n

1-

01

| Production workers0 Office workers

I I I

04\

II

0II

0A.

0e

Figure 1 Mean symptom score at different body locationsfor210 production workers mainly operating sewing machinesand 35 office workers mainly performing secretarial duties.Symptom score is the sum of an intensity score (0-4) and afrequency score (0-2). See text8 for details on the scoringsystem.

156

4

on August 26, 2019 by guest. P

rotected by copyright.http://oem

.bmj.com

/B

r J Ind Med: first published as 10.1136/oem

.49.3.154 on 1 March 1992. D

ownloaded from

Symptoms ofmusculoskeletal complaints. II Different riskfactors among sewing machine operators

4-

3-

o

E0

(I)

w4'1

I I I

Figure 2 Mean symptom score at different body locationsforfour groups of sewing machine operators producing thermalclothing atfour geographically separate locations in southernNorway.

consider the symptom frequencies. The symptomvalues ofthe different body regions are joined by linesto produce a continuous curve, a "body profile ofpain," emphasising the most affected body regions.Seventy one per cent of the production workersreported a symptom score of 3 or higher for theshoulder region compared with 41% for the officegroup.The reported symptoms were mostly affecting the

neck and shoulders, less so the dorsal aspects of thehead, the lower arms, and the low back. Few workershad problems with the hip and lower extremities.The distribution of symptoms corresponded to thelocation of work strain: shoulders and arms were incontinuous use when performing the work tasks,generating static muscle loads. The lower arms andhands were at risk due to the need of continuouscontrol of the garment when performing the twistingand turning movements. The forward leaning seatedposture probably strained the low back. Symptomsin the dorsal aspects of the head often relate toshoulder and neck tension as several neck andshoulder muscles arise from this area.

The symptom scores of the head, neck, shoulders,and lower arms were clearly lower for the office grouprelative to the production workers (p values from0-006 to < 0 001). By contrast, no difference was

found in symptom score for the low back and thelower extremities indicating no effect (or a similar

effect) of working conditions with regard to reportedsymptoms in these body regions.The four geographically separate groups of work-

ers producing thermal clothing had nearly identicalmean symptom scores for all body regions (fig 2). Theworkers at Tinn, and to some extent Rjukan, repor-ted lower symptom scores in the head, neck, andshoulders, but the difference was not statisticallysignificant. The two groups of workers performingsewing and welding operations in the production ofoffshore survival suits had symptom profiles similarto those producing thermal clothing, except that thewelding group had significantly less problems withtheir lower arms consistent with the reduced strainon the arms for this work task (see Methods).

Figure 3 shows the location ofsymptoms in the leftand right shoulders and lower arms for the produc-tion workers. Distribution of symptoms for theshoulders was symmetrical (fig 3A) whereas problemsin the lower arms were biased towards the right arm(fig 3B). The symmetrical distribution of shouldersymptoms is consistent with the electromyographload recordings, which showed static load patterns inboth trapezius muscles.

It was reported in a separate publication of thisstudy that production workers with a four to five hourworking day on average recorded their first sick leavedue to musculoskeletal illnesses about six monthslater than those on eight hour work schedules but thisdifference disappeared after the first few years forcomplaints in the shoulder and neck region. Bycontrast, differences for sick leaves with a complaintlocated to the low back appeared to last longer.15Figure 4 shows symptom profiles for full time andpart time production workers. The mean symptomscore for the neck was somewhat lower for the part

60 AO Bso®

50E L0

U>300

Figure 3 Distribution of symptoms of discomfort or pain onleft and right shoulders (A) and lower arms (B).

157

1 -

OJ r--b

on August 26, 2019 by guest. P

rotected by copyright.http://oem

.bmj.com

/B

r J Ind Med: first published as 10.1136/oem

.49.3.154 on 1 March 1992. D

ownloaded from

Westgaard, Jansen

4-

3-

O Full time workers* Part time workers

2_ ~~~~C

11

v',

00-11

Figure 4 Mean symptom score at different body locationsforproduction workers workingfull time (115 subjects) and parttime (45 subjects).

time workers (p = 0 06) and significantly lower forsymptoms in the low back, consistent with the sickleave analysis.

SYMPTOMS OF PAIN AND DISCOMFORT ANDINDIVIDUAL FACTORSWorkers who have suffered symptoms beforeemployment may have established a health conditionat the time of employment that makes themparticularly susceptible to similar injuries at theworkplace. Many workers shared this opinion withregard to their own situation, as only 49% of thosewith musculoskeletal symptoms at work and withsimilar experiences before employment believed thata clear link existed between symptoms and workingconditions compared with 88% of those who had notexperienced such symptoms previously.

Figure 5A shows the fraction of production work-ers reporting symptoms at different body locationsbefore employment in the current work situation.Symptoms in the head and low back were mostcommon, at about 30% each. About 15% of allsubjects had experienced symptoms in the neck andshoulders, symptoms in knees and ankles were repor-ted by 11% of all subjects, and symptoms in lowerarms, hips, and thighs were infrequent (6 to 8%).Figure 5B shows the mean symptom score for thedifferent body regions of subjects with and withoutsimilar symptoms before employment. The same

subject may contribute in both categories depending

on the location of previous pain symptoms. In allbody locations the mean current level of symptomswas higher when similar symptoms had occurredpreviously. Many production workers without suchproblems previously, however, had experienced con-siderable discomfort at work, particularly in theshoulders and neck.Another a priori risk factor is age, as it is known

that the incidence of musculoskeletal complaintsincreases with age in the general population. Figure 6shows the mean symptom score for productionworkers less than 30 years, 30-40 years, and morethan 40 years old. The three upper body regions,which appear to be most affected by the workingconditions, have the same symptom level at all agegroups. A statistically significant negative correlationwith age was found for the low back, possibly due to a"healthy worker" effect. Hips, thighs, and knee andankles showed a positive correlation with agealthough this was not significant for the thighs (p =007).

Personal psychological or psychosocial problemsare frequently postulated to be a confounding factorin the development of muscle pain. Figure 7 showsthe symptom profile of production workers con-sidered to have psychological problems, and ofproduction workers with no apparent sign of suchproblems. Those with psychological problems had ahigher mean symptom score for the head and neckwith no significant differences in other body regions.

In the limited material from production workersreporting on their general tendency to generatemuscle tension, 29 definite negatives and 40 definite

| Previous complaintsNo previous complaints

t~ 4- "44j

e

Figure S (A) Fraction of production workers reportingsymptoms of discomfort or pain before present employment atdifferent body locations. (B) Mean symptom score at differentbody locations for production workers with and withoutsymptoms of discomfort or pain at the same body locationbefore present employment.

40-

CA

o 30-

E

CA:*-- 20-

-10

4,

'u 3

0

E0

2

cn

0u

E0

Ecn

158

1

.a3 II I I I I I I

b -* C* to 'k- Q C* -C,*.b c.i 4, $' ci " 105

le qi 5b b zz ;-.. 'b

ll'- 11',. '4 le lb",

.? op-,O`. -4

on August 26, 2019 by guest. P

rotected by copyright.http://oem

.bmj.com

/B

r J Ind Med: first published as 10.1136/oem

.49.3.154 on 1 March 1992. D

ownloaded from

Symptoms of musculoskeletal complaints. II Different riskfactors among sewing machine operators

4-

3-

a,f0E0

EC,

2-

1 -

0-... . . * ~~. |

Figure 6 Mean symptom score at different body locationsforthree different age groups ofproduction workers ( < 30years,84 subjects; > 30-40 years, 72 subjects; > 40years,54 subjects).

positives were identified. The symptom score wassignificantly higher in the head, neck, and shouldersfor those with a positive response. A possible inter-action between this parameter and psychologicalproblems was examined by cross tabulating necksymptoms as a function of positive or negativeresponses to the two parameters. Positive inter-actions were not indicated-that is, a raised level ofneck symptoms was correlated with a positive res-ponse to either parameter with no further increase forthose with positive responses to both parameters.Table 2 presents a summary of the univariate

analyses for the 210 production workers. Thus thelarge differences in level of symptoms in the upperbody regions between production and office workersare not included. Apart from previous problems,which appeared to affect all body regions, onlypsychological problems and tendency of muscletension showed a significant contribution to symptomlevel in the head and neck (muscle tension also in theshoulders). The only differential effect of the para-meters of work task was the reduced symptom scoreof the low back for part time workers and ofthe lowerarms for welders of survival suits. The correlationbetween symptoms in the lower arms and physicalactivity may be a type I I error-namely, a falsepositive-as most of the activities described in theinterview (jogging, swimming) would not strain thearms. Also, -the initial hypothesis was that physical

activity would reduce the level of musculoskeletalcomplaints at the work place. Low back symptomsshowed a relatively complex relation with othervariables.The univariate analyses were used as a basis for a

multilinear regression analysis. Individual factors(except physical activity) shown to correlate withsymptom level at any body location on a univariatebasis were included in the initial run for all bodylocations. Table 3 gives the results ofthe final analysisafter elimination of those parameters not contribut-ing to the variance in level of symptoms in the firstanalyses.Low back and head symptoms retained a relatively

complex relation with other factors, whereas symp-toms at the other body regions only correlated withone or two parameters. Age showed a negativecorrelation with symptoms in the head and low back,and a positive correlation with symptoms in knee andankles. The parameter "previous pain symptoms"correlated with present symptoms at all body regionsexcept the neck, but the correlation was low for theshoulders and the lower arms. Working part timeappeared helpful in reducing problems in the lowback, and possibly also in the neck. Psychosocial(family) problems remained a risk factor for low backpain. The variance explained by the independentparameters (adjusted r2) varied between body regionsand was highest for the lower body regions with lowsymptom score. By contrast, only 2-4% of totalvariance was explained by the individual parametersfor neck, shoulders, and lower arms.

4.

3.

0

E0a 2-E

1-

0.

* Psychological problems

O No psychological problems

o eto-

½ o'5~~~~~~~~1

-o0 0

Figure 7 Mean symptom score at different body locations ofproduction workers indicating (28 subjects) or not indicating( 93 subjects) personal psychological problenms. -

159

on August 26, 2019 by guest. P

rotected by copyright.http://oem

.bmj.com

/B

r J Ind Med: first published as 10.1136/oem

.49.3.154 on 1 March 1992. D

ownloaded from

160 Westgaard, Jansen

Table 2 Parameters correlating with pain/discomfort

Body region Variable p Value

Head Previous pain symptoms < 0 001Muscle tension 0-014Psychological problems 0-031

Neck Previous pain symptoms 0-008Psychological problems 0 010Muscle tension 0 013

Shoulders, upper arms Previous pain symptoms 0 017Muscle tension 0 035

Lower arms Physical exercise 0 002Previous pain symptoms 0 005Welding work task 0-042

Low back Previous pain symptoms < 0 001Working hours 0-002Family relationship 0-029Children (negative) 0 041Age (negative) 0-047

Hip Previous pain symptoms < 0 001Age (positive) < 0 001

Thighs Previous pain symptoms < 0 001Knee and ankles Previous pain symptoms < 0 001

Age (positive) 0-015

Univariate analysis of variance for 210 production workers. All parameters resulting in p 0-05 are included in the table.

Between 27 and 53% of the subjects reported a during the past 12 months. Thus the symptom scoresdevelopment towards worse symptoms in different reported in this paper also represent an approxima-body regions after they first appeared, with the tion of a 12 month period prevalence.highest percentages for complaints in the neck,shoulders, and arms. Between 3 and 16% reportedimproved symptom levels after the first episode of Discussionpain. The questions on symptoms at the interview This study confirms the results of other studies,were phrased in terms of symptoms throughout the which show a high rate of musculoskeletal com-working period, which would be variable. It was plaints among sewing machine operators (seetherefore not possible at the outset to calculate period Introduction). A particular interest of the presentprevalence. For the 84 to 97% of the subjects with study was the intersubject variation in symptomaggravated or unchanged problems, however, the levels and the dependence on factors not a priorireported symptom score would also be representative related to the work tasks. The increased symptomof symptoms experienced during the last 12 months. level of the upper body regions of the productionAlso, some subjects indicating an improved symp- workers relative to the office group, and the neartom level would have experienced this improvement identical symptom profiles of different groups of

Table 3 Multilinear symptom analysis

Body region Variable Adjusted r2 p Value

Head Previous pain symptoms 0-10 < 0 001Muscle tension 0 035Age (negative) 0-038Psychological problems 0-052

Neck Psychological problems 0 04 0-038Working hours 0 055

Shoulders, upper arms Previous pain symptoms 0 02 0-041Lower arms Previous pain symptoms 0 03 0 011Low back Previous pain symptoms 0.19 < 0 001

Working hours 0 001Age (negative) 0-002Family relationship 0 012

Hip Previous pain symptoms 0 33 < 0 001Thighs Previous pain symptoms 0 18 < 0 001Knee and ankles Previous pain symptoms 0-17 < 0 001

Age (positive) 0 025

Multivariate linear regression analysis of pain symptoms at different body locations (dependent variable) v previous pain symptoms, age,psychological problems, tendency of muscle tension, children, relationship to close family, and daily working hours (independentvariables), based on 210 production workers. All parameters with p < 0 1 are included in the table.

on August 26, 2019 by guest. P

rotected by copyright.http://oem

.bmj.com

/B

r J Ind Med: first published as 10.1136/oem

.49.3.154 on 1 March 1992. D

ownloaded from

Symptoms of musculoskeletal complaints. II Different riskfactors among sewing machine operators

production workers with similar work tasks,however, provide strong evidence for a dominant,work related aetiology in the development of thesecomplaints. This study thereby serves as anotherindication of the inherent detrimental qualities ofwork situations with continuous arm movements,even when the static postural parameters are nearoptimal (see Methods).The most important predictors of individual

susceptibility to musculoskeletal complaints wereprevious episodes of similar pain syndromes, effectspresent in all body regions. The multivariate analysisindicated that previous pain symptoms contributedto a considerable fraction of total variance for symp-toms in the lower extremities, but explained verylittle of the total variance of pain symptoms in theneck and shoulders. The reason for this result ispresumably that the risk of developing work relatedmusculoskeletal disorders in the neck and shouldersis high, regardless of latent problems present at thetime of employment. This interpretation is sup-ported by the results ofa preliminary investigation inwhich a medical examination to identify signs ofmusculoskeletal complaints was carried out on 22production workers in employment. Two years laterlittle difference in regard to symptom level or need ofmedical attention was found between workers classi-fied as risk or non-risk subjects when employed.The only parameters apart from previous pain

symptoms correlating with problems in the neck andshoulders in the univariate analysis were psycho-logical problems (neck), and the self reported tend-ency ofmuscle tension (neck, shoulders). The effect ofpsychological state only accounted for 2% of totalvariance in symptom level for the neck and had noeffect in the case of the shoulders according to themultivariate analysis. The low explanatory powercan be due to our choice of parameter representationnot pinpointing stronger underlying effects. Moststudies of psychogenic muscle tension indicate,however, that such tension is of low intensity6 17 andmay therefore only result in a small additionaltension component if static tension due to posturalfactors is already present in the muscle activitypattern. Accordingly, this parameter may appear as amuch more important risk factor in populationswhere postural loading of the shoulder muscles is nota problem.The development of muscle tension is associated

with psychiatric disorders,"8 with various psycho-logical factors,'9 with anxiety,202' and with mentallydemanding work tasks.'6222' Muscle pain has beenassociated with parameters relating to the physio-logical state or trait.7 Primary fibromyalgia has beenassociated with specific psychological profiles in theMinnesota multiphasic personality inventory.2425 Ahypothesis to be tested in future is that the para-meters "psychological problems" and "tendency of

muscle tension" serve as indicators of the sameunderlying risk factor-namely, a subject dependenttendency of excess tension in the upper trunkmuscles'626 with a subsequent development of painsymptoms. The psychosocial work environment isanother factor which may contribute to neck andshoulder pain through such a mechanism.Both psychological problems and tendency to

muscle tension correlated with symptoms in thehead. This is consistent with the above hypothesis ashead symptoms were delineated as tension headachein the dorsal aspects of the head. The negativecorrelation of head and low back symptoms with agemay be explained as a "survival effect"-that is, aselection towards those who tolerate the work strainrelatively well.The low back was the only body region where

symptoms correlated with our indicator of psycho-social life situation, relationship to close family.Psychosocial factors have also been previouslyimplemented in the development of low backpain." 2 The reduction in low back pain of part timecompared with full time workers is consistent withthe results of the sick leave analysis.'5Many parameters often postulated to influence the

susceptibility to musculoskeletal problems were notcorrelated with the level of discomfort at any bodyregion. The economic situation, amount of physicalexercise, marital state, children, living conditions,work load at home, and other activities outside workwere not related to the level of musculoskeletalcomplaints (disregarding the correlation betweenphysical activity and symptoms in the lower arms).The results may be influenced by the choice of

subjects or the system for measurement of theseparameters. For instance, physical exercise wasdefined in terms of a certain level of conditiontraining. Thus the negative result for this parameterdoes not rule out the possibility that a dedicatedand well formulated training programme will bebeneficial in preventing the development ofmusculoskeletal symptoms, but suggests that condi-tion training does not have the general effect ofprotecting against occupational musculoskeletalcomplaints. Other individual parameters may showup as significant in larger populations, in populationswith different life situations, or with different delinea-tion of the parameters in the questionnaire.Even so, parameters similar to those included in

this study would probably not explain much of thetotal variance in the level of symptoms from the neckand shoulders. It is therefore not understood whythere are "survivors," or symptom free workers inthe face of severe work strain. Physiological factors atthe workplace, like the number or pattern of micro-pauses in the muscle activity patterns may beimportant, but such hypotheses remain speculative atpresent.

161

on August 26, 2019 by guest. P

rotected by copyright.http://oem

.bmj.com

/B

r J Ind Med: first published as 10.1136/oem

.49.3.154 on 1 March 1992. D

ownloaded from

Westgaard, Jansen

We acknowledge the support of employees andmanagement of Helly-Hansen A/S in carrying outthis project. We are grateful to Mr Leiv Sandvik foradvice on the use of statistical methods.

Requests for reprints to: RH Westgaard, Division ofOrganisation and Work Science, the NorwegianInstitute of Technology, Trondheim, Norway.

1 Wallace M, Buckle P. Ergonomic aspects ofneck and upper limbdisorders. International Review Ergonomics 1987;1:173-200.

2 Hagberg M, Wegman DH. Prevalence rates and odds ratios ofshoulder-neck diseases in different occupational groups. Br JInd Med 1987;44:602-10.

3 Kilbom A. Isometric strength and occupational muscle dis-orders. Eur J Appl Physiol 1988;57:322-6.

4 Linton SJ, Kamwendo K. Risk factors in the psychosocial workenvironment for neck and shoulder pain in secretaries. JOccupMed 1989;31:609-13.

5 Karasek R, Gardell B, Lindell J. Work and non-work correlatesof illness and behaviour in male and female Swedish whitecollar workers. Journal of Occupational Behaviour 1987;8:187-207.

6 Kogstad 0, Hartvig P. Psychiatric and social factors in rheu-matological and physical medical disability. Tidsskr NorLageforen 1978;98:943-5 (In Norwegian).

7 Ursin H, Endresen IM, Ursin G. Psychological factors and self-reports ofmuscle pain. Eur J Appl Physiol 1988;57:282-90.

8 Westgaard RH, Jansen T. Individual and work related factorsassociated with symptoms of musculoskeletal complaints. I Aquantitative registration system. Br J Ind Med 1992;49:147-53.

9 Westgaard RH, Waersted M, JansenT, Aaris A. Muscle load andillness associated with constrained body postures. In: CorlettN, Wilson J, Manenica I, eds. The ergonomics of workingpostures. London: Taylor and Francis, 1986:5-18.

10 Miyashita K, Shiomi S, Kasamatsu T, Itoh N, Iwata H. A studyon occupational cervicobrachial disorder among female sew-ing machine operators in a small scale industry. WakayamaMedical Reports 1980;23:81-8.

11 Vihma T, Nurminen M, Mutanen P. Sewing-machine oper-ators' work and musculo-skeletal complaints. Ergonomics1982;25:295-8.

12 Punnett L, Robins JM, Wegman DH, Keyserling WM. Softtissue disorders in the upper limbs offemale garment workers.Scand J Work Environ Health 1985;11:417-25.

13 Blider S, Barch Holst U, Danielsson S, Ferhm E, Kalpamaa M,

Leijon M, Lindh M, Markhede G, Mikaelsson B. Neck andshoulder complaints among sewing-machine operators. In:Buckle P, ed. Musculoskeletal disorders at work. London:Taylor and Francis, 1987:110-1.

14 Sokas RK, Spiegelman MS, Wegman DH. Self-reportedmusculoskeletal complaints among garment workers. Am JInd Med 1989;15:197-206.

15 Wazrsted M, Westgaard RH. Working hours as a risk factor inthe development of musculoskeletal complaints. Ergonomics1991;34:265-76.

16 Westgaard RH, Bj0rklund RA. Generation of muscle tensionadditional to postural load. Ergonotnics 1987;30:911-23.

17 Warsted M, Bj0rklund RA, Westgaard RH. Shoulder muscletension in a VDU-based task under reward and non-rewardconditions. Proc Int Union Physiol Sci 1989;XVH:331.

18 Malmo RB, Shagass C. Physiologic study of symptom mechan-isms in psychiatric patients under stress. Psychosom Med1949;11:25-9.

19 Rimehaug T, Svebak S. Psychogenic muscle tension: thesignificance of motivation and negative affect in perceptual-cognitive task performance. Int J Psychophysiol 1987;5:97-106.

20 Sainsbury P, Gibson JG. Symptoms of anxiety and tension andthe accompanying physiological changes in the muscularsystem. J Neurol Neurosurg Psychiatry 1954;37:216-24.

21 Malmo RB. On emotions, needs, and our archaic brain. New York:Holt, Rinehart, and Winston, 1975.

22 ShawWA, Kline LH. A study ofmuscle action potentials duringthe attempted solution by children of problems of increasingdifficulty. J Exp Psychol 1947;37:146-58.

23 Eason RG, White CT. Muscular tension, effort, and trackingdifficulty: Studies of parameters which affect tension level andperformance efficiency. Perceptual Motor Skills 1961;12:331-72.

24 Ahles TA, Yunus MB, Riley SD, Bradley JM, Masi AT.Psychological factors associated with primary fibromyalgiasyndrome. Arthritis Rheum 1984;27:1101-6.

25 Wolfe F, Cathey MA, Kleinheksel SM, Amos SP, Hoffman RG,Young DY, Hawley DJ. Psychological status in primaryfibrositis and fibrositis associated with rheumatoid arthritis.J Rheumatol 1984;11:500-6.

26 Veiersted KB, Westgaard RH, Andersen P. Pattern of muscleactivity during stereotyped work and its relation to musclepain. Int Arch Occup Environ Health 1990;62:31-41.

27 Biering-Sorensen F, Thomsen C. Medical, social and occu-pational history as risk indicators for low-back trouble in ageneral population. Spine 1986;11:720-5.

28 Biering-S0rensen F, Thomsen CE, Hilden J. Risk indicators forlow back trouble. Scand J Rehabil Med 1989;21:151-7.

Accepted 17 June 1991

162

on August 26, 2019 by guest. P

rotected by copyright.http://oem

.bmj.com

/B

r J Ind Med: first published as 10.1136/oem

.49.3.154 on 1 March 1992. D

ownloaded from