Occupational and Environmental Medicine 1994;51:262-266

Case-control study of risk factors for disease inthe neck and shoulder area

K Ekberg, B Bjorkqvist, P Malm, B Bjerre-Kiely, M Karlsson, 0 Axelson

AbstractA case-control study was performed toelucidate the strength of the relationbetween musculoskeletal disorders in theneck and shoulders and physical, organi-sational, and psychosocial aspects of thework environment. Cases were identifiedas those persons who consulted a physi-cian in a community in southern Swedenfor new musculoskeletal disorders inthe neck and shoulders during a studyperiod from August 1988 to the end ofOctober 1989. One hundred and ninecases were collected and clinically exam-ined. The cases also answered the Nordicquestionnaire on symptoms as well as aquestionnaire on work conditions andbackground factors. Controls weredrawn as a random sample of the work-ing population in the community wherethe cases appeared. A total of 637 con-trols answered the same questionnairesas the cases. Odds ratios (ORs) were cal-culated by logistic regression. The oddsratios were 11P4 for women, 4 9 forimmigrant background, and 3-7 for cur-rent smoking. To exercise rarely, com-pared with often, appeared as apreventive factor with an OR of 03. TheORs for various determinants of physicalwork load were 7'5 for repetitive move-ments demanding precision, 13-6 forlight lifting, 3-6 for uncomfortable sittingpositions, 4-8 for work with lifted arms,and 3 5 for a rushed work pace.Regarding work organisational determi-nants, the ORs were 16-5 for ambiguity ofwork role (uncertainty whether the per-son could manage the work) 2-6 for lowquality work, and 3-8 for high demandson attention. Several of the determinantsshowed a significant dose-response rela-tion with disease. It seems that workorganisation and psychosocial work con-ditions are as important determinantsfor disease in the neck and shoulders asare the physical work conditions.

(Occup Environ Med 1994;51:262-266)

The relation between the development ofmusculoskeletal disorders in the neck, shoul-ders, and upper limbs and strained workloadhas become a subject of growing interest dur-ing the past decade. As well as the sufferingof the individual person there are consider-able economic costs to the community due to

sick leaves. Some organisational andergonomic work conditions such as repetitivetasks, force, static load, extreme joint posi-tions, high work pace, poor psychosocialwork environment and low job satisfactionseem to be determinants for development ofsymptoms.'4 Brisson et al 6 found strongassociations between disability due to muscu-loskeletal disorders and duration of employ-ment in piecework. A pronounced type Abehaviour among blue collar workers has alsobeen associated with a higher incidence ofmusculoskeletal symptoms.7 Dimberg et alfound that the duration of absenteeism due tosymptoms in the neck and upper extremitiesincreased with increasing physical strain inthe job.8 The symptoms seem to persist atleast over a period of three years after retire-ment from heavy physical workload.9Most of the studies mentioned are cross

sectional in character and based on preva-lence of symptoms and signs in various pro-fessional groups. Therefore they do not showthe temporal relations between exposure andoutcome. The similar results in several stud-ies may nevertheless be considered tentativeevidence for causal associations.

Hagberg and Wegman have reviewed stud-ies based on clinical examinations of the casesin different occupational groups.'0 Theyfound that the associations obtained betweenjob characteristics and disability suggest thathighly repetitive shoulder muscle contrac-tions, static contractions, and work atshoulder level are hazardous exposure factors.Work organisation and psychosocial condi-tions at work were not investigated in theseclinical studies, and there is a need to con-sider multifactorial determinants of muscu-loskeletal disease. It should be noted also thatan increased occurrence of symptoms in cer-tain work tasks does not necessarily meanthat these work conditions are generatingdisease in a clinical sense.The aim of our case-control study was to

elucidate the strength of the relation betweendisease in the neck and shoulder area andphysical as well as organizational and psycho-social aspects of the work environment.

Material and methodsOur study was performed in a semiruralcommunity in southern Sweden. The labourmarket is characterised by many small manu-facturing companies and one large rubberindustry. Piecework is common. TheIndustrial Health Care Unit and the District

Department ofOccupational andEnvironmentalMedicine, UniversityHospital, S-581 85Linkoping, SwedenK EkbergM Karlssono AxelsonIndustrial Health CareUnit, S-334 00Anderstorp, SwedenB BjorkqvistB Bjerre-KielyDistrict Health CareUnit, S-332 00Gislaved, SwedenP MalmRequest for reprints to:Dr K Ekberg, Departmentof Occupational andEnvironmental Medicine,University Hospital, 5-58185 Linkoping, Sweden.Accepted 9 August 1993

262

on January 7, 2021 by guest. Protected by copyright.

http://oem.bm

j.com/

Occup E

nviron Med: first published as 10.1136/oem

.51.4.262 on 1 April 1994. D

ownloaded from

Case control study of risk factors for disease in the neck and shoulder area

Health Care Unit cover the primary healthcare needs in the area except for rubber work-ers, who had access to the health care facili-ties of the industry. The study period wasfrom August 1988 to the end of October1989 (15 months).

SUBJECTSThe common inclusion criteria applied were:(1) an age range between 18 and 59 years; (2)at least two months occupational experienceof the last held job and occupationally activeat 100% of the scheduled work times duringat least four of the past six months. For thecontrols these criteria were formulated as ayearly income of at least SEK 45 000 (about8000 US dollars) and not having been on sickleave for more than two months during thepast six months.) (3) not employed at thelarge rubber industry in the area. The crite-rion (2) was established to avoid the chroni-cally ill and occupationally unstable people.The third criterion was established to avoidproblems from special health care conditionsat the single large company in the area com-pared with the many small industries.

CasesThe cases were taken as those who consulteda physician in the community for muscu-loskeletal disorders in the neck, shoulder,arm, or upper thorax during the study period.To finally qualify as a case, the person alsoshould have become ill immediately beforeseeing the physician and should at most havebeen on sick leave for four weeks. Those whoalready had a long history of musculoskeletalproblems were not accepted as cases as thepattern of determinants might have beenobscured in the course of time. Also, the dis-order should not be caused by traumatic orinfectious agents or by an accident. Patientswith malignancy, rheumatic diseases, abuse,or pregnancy were also excluded.

All cases were examined by the samephysician and physiotherapist and diagnoseswere set according to predetermined criteriaas suggested by Waris et al.1 Cases were thengrouped according to their main diagnosisinto three groups: (1) disorders in the neckand shoulders, (2) disorders in arms andelbows, and (3) disorders in the hands andwrists. In our paper we present only caseswith their main diagnosis in the neck andshoulders (group (1)), as the determinants forthe different disease groups are likely to differ.

ControlsOne thousand people living in the samearea as the cases and who would have seenthe same physicians if ill, were randomlyselected for the study. We used the Swedishinsurance register after approval of the localethics committee. A stratified sample of 10%of the population was drawn, based on geo-graphic area so that each stratum corres-ponded to 10% of the population thatbelonged to each health care unit, in all 900subjects. The remaining 100 were saved forsubstitution if necessary. A letter of introduc-

tion about the study was mailed to these 900people in the middle of the study period, fol-lowed by a mailed questionnaire a week later.In all, 26 people in the sample had movedfrom the community and one had died.These subjects were replaced with 27 newsubjects, randomly drawn from the remaining100 people of the primary sample. Due tolimited funds the controls were not examinedclinically. If they had become ill during thestudy period, however, they would haveappeared at the health care units.

QUESTIONNAIREAll subjects answered a modified version ofthe Nordic questionnaire on musculoskeletalsymptoms.12 In the questionnaire the subjectsstated whether they had had symptoms dur-ing the past six months. They also answered aquestionnaire on their work situation, physi-cal demands at work, work organisation, andpsychosocial work conditions. The back-ground factors, age, sex, ethnic background,family situation, smoking habits, and exercisehabits were also included.

MEASURES OF EXPOSUREMeasures of exposure are based on the infor-mation given in the questionnaire. Statisticalanalysis of the exposure questions has beendescribed in detail elsewhere.5 Briefly, 20items on physical work conditions werereduced to seven factors by factor analysis.The seven finally considered determinants(factors) were: (1). uncomfortable sitting posi-tion, (2) uncomfortable standing position, (3)physically demanding work, (4) light lifting(< 6 kg), (5) repetitive movements demand-ing precision, (6) work with lifted arms, and(7) monotonous work position. Each bodyposition was described by a graph in thequestionnaire. The rating scales, rangingfrom one to four, were based on averageduration in hours per day of each item ofexposure. Physically demanding work (heavylifting) was defined by weights of objects(6-12, 12-30, > 30 kg) and how often theywere lifted. Fifty two items on organizationaland psychosocial work conditions werereduced by factor analysis to eight factorsdescribing (1) psychological work climate, (2)quality of work content, (3) work pace, (4)demands on attention, (5) work planning, (6)job security, (7) job constraints, and (8) workrole ambiguity. Each item was rated on ascale ranging from 1 (not at all) to 7 (to alarge extent).

Background determinants that could havebeen of importance for development ofmusculoskeletal disorders and hence affectthe associations between disease and workenvironment were also included. The deter-minants were age, sex, ethnic background(born or not born in Sweden), smokinghabits, exercise habits, and having preschoolchildren.

STATISTICAL ANALYSISThe distribution on each exposure factor was

263

on January 7, 2021 by guest. Protected by copyright.

http://oem.bm

j.com/

Occup E

nviron Med: first published as 10.1136/oem

.51.4.262 on 1 April 1994. D

ownloaded from

Ekberg, Bjorkqvist, Malm, Bjerre-Kiely, Karlsson, Axelson

coded into tertiles based on the distributionsof the cases. Because the cases were likely tobe more highly exposed in at least some ofthe exposure factors, the distributions of thecases were used to make sure that the highexposure categories would be covered. Thelowest 30% were coded as 1, the middle 40%were coded as 2, and the highest 30% of theexposure distributions were coded as 3.Odds ratios (ORs) based on multiple logis-

tic regression and confidence intervals at the90% level (90% CIs) were calculated. Twocontrol groups were established: (1) controlgroup A consisting of those who, according tothe questionnaire, were free of musculoskele-tal symptoms in all parts of the body and (2)control group B consisting of control group Aplus those subjects who had indicated somemusculoskeletal symptoms in other parts thanthe neck and shoulder area.The lowest levels of exposure were used as

reference for the risk factors. For the variousbackground determinants, age was coded into

Table 1 Demographic data of the study groups

Cases Control group A Control group BDeterninant (%, n = 109) (%, n = 136) (%, n = 327)

Sex (female) 80 39 39Immigrants 28 4 8Current smokers 62 23 25Preschool children 23 23 19Type of employment:

Blue collar 77 28 29Service, health care 18 19 29White collar 5 53 49

Table 2 Odds ratios (ORs) and 90% CIs for determinants for diseases in the neck andshoulders (n = 109).

OR (90% CI) v

Determinant Control group A Control group BLevel of exposure (n = 136) (n = 327)Sex (female) 15-5 (3-4-71) 11-4 (4 7-28)Immigrants 28-3 (3-1-257) 4 9 (1-8-14)Current smoker 8-2 (2 3-29) 3-7 (1-8-7-5)Exercise:

> 5 hours/week 1 0 1.0< 5 hours/week 1 (0 02-0-62) 0-4 (0 11-1 1)seldom 0 1 (0 01-0 38) 0 3 (0 11-1 0)

Repetitive precision movements:Low 1 0 1 0Medium 3-8 (0-74-20) 2-7 (1 1-6-4)High 15-6 (2-2-113) 7-5 (2 4-23)

Light lifting:Low 1.0 1.0Medium + high 49-7 (9 0-273) 13-6 (4 8-39)

Uncomfortable sitting:Low 1.0 1.0Medium 0 5 (0-13-2-3) 0-8 (0 34-2 0)High 0 7 (0-15-3-8) 3-6 (1-4-9-3)

Lifted arms:Low 1 0 1.0Medium 5-9 (0 94-37) 2-4 (0-80-7-1)High 3-7 (0-44-30) 4-8 (1-3-18)

Work pace:Low 1.0 1.0Medium 7-6 (1-6-36) 2-5 (1 0-6-2)Rushed 10-7 (2-2-52) 3-5 (1-3-9-4)

Ambiguity of work role:Low 1.0 1 0Medium 0-8 (0-13-4-7) 2-2 (0-62-7-9)High 22-9 (4-9-107) 16-5 (6-0-46)

Demands on attention:Low 1 0 1.0Medium 0 7 (0-17-3-2) 1-7 (0-64-47)High 2-6 (0-59-11) 3-8 (14-11)

Work content:Good 1 0 1.0Medium 1 0 (0-16-5-9) 1 1 (0-33-3-9)Low 10-5 (1-4-79) 2-6 (0-71-9-4)

decades and the youngest age group consti-tuted the reference for age, males were usedas the reference for sex, Swedes as the refer-ence for immigrants, non-smokers as the ref-erence for smokers, and the highest level ofleisure time exercise (at least five hours perweek) was the reference for exercise. Not hav-ing preschool children was the reference forhaving preschool children.

ResultsEach case obtained a main diagnosis and insome cases also additional diagnoses.Altogether 109 subjects had a main diagnosisof neck or shoulder disorders. The most com-mon diagnoses were cervical syndrome or rhi-zopathi (18%), tension neck syndrome(47%), and humeral tendinitis (27%).The response rate was 73% for the controls

(655 subjects responded). Eighteen subjectswere excluded as they did not fulfil the occu-pational criteria (due to studies, militaryservice, etc). The final group therefore com-prised 637 persons. As the control group wasobtained as a random sample of the popula-tion it is not surprising that many reportedsymptoms from the neck (43%) and shoul-ders (32%), and also from other parts of thebody. One hundred and thirty six controls(21%) did not report any musculoskeletalsymptoms in any part of the body.

Table 1 shows the demographics of thegroups. The median age of the cases was 39(range 18-57) years. The age distribution ofthe control groups were similar (controlgroup A: 38 (18-58) years, control group B:39 (18-59) years). The average workinghours were 36 hours among the cases and 38hours in both control groups.

Table 2 shows ORs and 90% CIs for dis-ease in the neck and shoulders. To make theresults easier to comprehend, only determi-nants with significant associations in relationto one or both of the control groups are pre-sented.

Female sex and immigrant backgroundwere associated with diseases in the neck andshoulder area. To be a smoker was also a sig-nificant determinant. For exercise, less thanfive hours per week seemed preventivewhereas more extensive exercise, at least fivehours per week, was significantly associatedwith disease in the neck and shoulders. Ageand having preschool children were not asso-ciated with disease.

Repetitive movement demanding precisionwas a significant physical determinant with adose-response relation showing higher risksfor higher degrees of exposure. Light liftingwas also a strong determinant for disease.The medium and high exposure categorieswere combined for this factor as there weretoo few cases in one of the exposure cate-gories. Long durations of uncomfortable sit-ting and work with lifted arms weresignificant determinants compared with thelarger control group B. To work standing inuncomfortable positions, monotonous workpositions, and physically demanding work

264

on January 7, 2021 by guest. Protected by copyright.

http://oem.bm

j.com/

Occup E

nviron Med: first published as 10.1136/oem

.51.4.262 on 1 April 1994. D

ownloaded from

Case control study of risk factors for disease in the neck and shoulder area

(heavy lifting) were not significant determi-nants for disease in the neck and shoulders.

Quantitative demands expressed as selfrated work pace had a significant dose-response relation with diseases in the neckand shoulders. Work role ambiguity-that is,uncertainty about how to best perform thework and uncertainty about whether the per-

son could manage the work-was a strongdeterminant in the high exposure category.High demands on attention also seemed to bea determinant when compared with controlgroup B. Low quality work-namely, lack ofstimulation and variation in the job, lowopportunities for development, and a low jobdiscretion-were significant determinantswhen compared with the controls with no

symptoms (control group A). Deficits in thepsychological work climate, work planning,job security, and job constraints were not sig-nificant determinants for disease in the neckand shoulders.As shown in table 2, there are distinct

dose-response relations between disease inthe neck and shoulders and some determi-nants-namely, repetitive precision move-

ments, work pace, and low qualitywork-irrespective of control group. Whencompared with the larger control group B,work with lifted arms, work role ambiguity,and demands on attention also showed dis-tinct dose-response relations with disease.

DiscussionThere are few studies of musculoskeletal dis-eases in the neck and shoulders where theoutcome variable is based on clinical exami-nations. In this study attempts were made to

follow distinct diagnostic criteria as presentedby Waris et al."I A diagnosis in the neck andshoulder area was, however, often accompa-

nied by other, related disorders. To overcome

this problem the different diagnoses were

combined into a more general neck andshoulder disease entity in the analyses.A problem in epidemiological studies on

musculoskeletal disorders is the assessment ofexposure to work loads. In some studiesexposure has been approximated by jobtitle.'0 A job title, however, may comprisemany different tasks and indicate a huge vari-ation in exposure to work loads. To properlyidentify determinants more specific measure-

ments of exposure are needed. In many stud-ies information about exposure to variousfactors is collected by questionnaires. Animplicit assumption is then that subjects havebeen in this exposure for some time. Givenwell defined exposure categories and simple,concrete questions, a questionnaire may giverough but reasonable information on expo-sure conditions. In our study, dose-responserelations were obtained for several determi-nants, which strengthens the credibility of theexposure estimates. Inclusion criteria were setto ensure that the same exposure should havebeen present for at least some months. Thismay be considered a short time, but diseasein the neck and shoulders may develop fast in

some subjects if work conditions areadverse.'3 To capture the early stages of dis-ease, comparatively short durations of expo-sure had to be accepted.The advancement of automation and spe-

cialisation in work, usually with a generallyreduced physical workload and a reduction ofheavy lifting, have created work tasks withfast work rates and a high degree of repeti-tiveness. Small industries of the type thatgenerated many of the cases in this study,often have high demands on productivity tosurvive. They have generally become effectiveby adapting the work organisation to theequipment. The tasks have become impover-ished both physically and mentally, but thequantitative demands are high and based onpiece wages. The results clearly underlinethese organisational problems. Repetitivemovements demanding precision often alsocomprise lifting of small pieces from oneposition to another in the manufacturingindustries. Exposure to this kind of work loadis the most powerful physical risk in our studyand comparable with results from otherstudies.'-' 10 Work pace shows a pronounceddose-response relation with musculoskeletaldisorders in this study. Piece wages, commonfor workers in the manufacturing industries,have been reported by others6 to contribute tohigh work pace and to promote subjectivestrain.The increased risk of disease due to low

quality work content-that is, lack of stimula-tion and variation in the work tasks, lowopportunities for development, and low jobdiscretion-may be considered to reflectother aspects of the same work organisationprinciple that was creating physical risks.These characteristics of a work situation havein many studies turned out to be significantrisk factors for stress related cardiovasculardiseases, hence indicating that psychologicalmechanisms also may be mediating muscu-loskeletal disorders to some degree.Psychological demands and lack of possibilityto talk to work mates were associated withmuscle tension and emotional states in astudy by Theorell et al,'4 supporting theassumption of a relation between psychoso-cially adverse job conditions and symptomsfrom the locomotor system.Work role ambiguity-that is, uncertainty

about how to best perform the work anduncertainty about whether the person canmanage the tasks-is a powerful determinantfor disease in the neck and shoulder area.Work role ambiguity may have its origin inlack of feedback on performance and vagueor muddled instructions from the foremen orother superiors. Also, inappropriate trainingfor the job, or training that is not adapted tothe characteristics and background of theemployees, may cause uncertainties abouthow to manage the work.

Being a woman and being an immigrantwas highly associated with disease in the neckand shoulder area. Although these two char-acteristics are commonly associated with risk,only a few studies include sex and ethnic

265

on January 7, 2021 by guest. Protected by copyright.

http://oem.bm

j.com/

Occup E

nviron Med: first published as 10.1136/oem

.51.4.262 on 1 April 1994. D

ownloaded from

Ekberg, Bjorkqvtist, Maim, Bjerre-Kiely, Karlsson, Axelson

background in the analyses for possible deter-minants. Dimberg et all' found that women injobs with high strain on the neck and upperextremities were sicklisted twice as often asmen in the same type of job. Kvarnstrom'5found that occupational shoulder disordersoccurred 10 times more often in women thanin men in a large manufacturing industry.There was an over-representation of womenin the jobs with the highest risks, which maybe part of the explanation. Another suggestedexplanation was lower muscle strength inwomen. Others, however, have found littlesupport for physical strength protectingagainst musculoskeletal disorders.4 Exercise,which could be assumed to improve physicalstrength, seemed to be a risk factor in ourstudy.

Kvarnstr6m15 also compared Swedes andimmigrants in jobs with high risk for occupa-tional shoulder disorders and found a relativerisk for immigrants of 4-9 for assembly lineoperators and 3-3 for winders. He suggestedthat immigrants may have used faulty worktechniques owing to lack of education andinstruction, or had difficulties in changing jobwhen symptoms began to occur, due to thelanguage barrier. It is conceivable that similarreasons exist in our study, and also that beingan immigrant is associated with the experi-ence of work role ambiguity.

Smokers were strongly over-representedamong the cases in this study. To be asmoker was significantly associated with dis-ease in the neck and shoulders. In a crosssectional study, Tsai et al 16 also foundsmoking to be one factor for increased risk ofmusculoskeletal injury (relative risk = 1-23).The effect of smoking on muscles, tendons,and nerves is not clear. It has been suggestedfor low back pain that the nicotine inducedvasoconstriction of small nutrient vessels mayreduce the blood flow to the intervertebraldiscs, thereby rendering them more vulner-able to injury.16 Similar mechanisms couldoperate for the neck and shoulders.

Control group A represents a group of peo-ple with no musculoskeletal symptoms. Asexpected, a comparison with this group ingeneral gives higher risk estimates than com-parisons with the other control group whichcomprised subjects with musculoskeletalsymptoms in other parts of the body.Although the 90% CIs for most of the factorsoverlap, it seems that some determinants areparticularly strong when compared withgroup A. These factors would be those thatare of greatest interest to focus on in preven-tive work. Light lifting, high work pace, workrole ambiguity, and a low quality of workseem to be those determinants that should begiven primary attention.

To conclude, musculoskeletal disease inthe neck and shoulders seems to have amultifactorial origin. The significant determi-nants constitute an aggregate of complexorganizational and physical factors. Repetitiveprecision movements and light lifting oftenmake a work situation that also has certainorganizational features, such as a high workpace, impoverished work content, minimaltraining, and lack of feedback on perfor-mance. Improvements of just one of thedeterminants may not necessarily lead toimproved health of the workers, as theremaining determinants may preserve thestrenuous situation. Rather, our results indi-cate that preventive efforts should focus onthe entire complex that constitutes the worksituation.

This study was supported by the Swedish Work EnvironmentFund, which is gratefully acknowledged.

1 Tola S, Riihimiki H, Videman T, Viikari-Juntura E,Hanninen K. Neck and shoulder symptoms among menin machine operating, dynamic physical work andsedentary work. Scand J Work Environ Health 1988;14:299-305.

2 Ohlsson K, Attewall R, Skerfving S. Self-reported symp-toms in the neck and upper limbs of female assemblyworkers. Scandal Work Environ Health 1989;15:75-80.

3 Linton S. Risk factors for neck and back pain in a workingpopulation in Sweden. Work and Stress 1990;4:41-9.

4 Westgaard RH, Jansen T. Individual and work related fac-tors associated with symptoms of musculoskeletal com-plaints. II Different risk factors among sewing machineoperators. BrJ IndMed 1992;49:154-162.

5 Ekberg K, Bjorkqvist B, Maim P, Bjerre-Kiely B, KarlssonM, Axelson 0. Cross-sectional study of risk factors forsymptoms in the neck-shoulder area. Ergonomics 1994(in press).

6 Brisson C, Vinet A, VWzina M, Gingras S. Effect of dura-tion of employment in piecework on severe disabilityamong female garment workers. Scand Jf Work EnvironHealth 1989;15:329-34.

7 Flodmark BT, Aase G. Musculoskeletal symptoms andtype A behaviour in blue collar workers. Br J Ind Med1992;49:683-7.

8 Dimberg L, Olafsson A, Stefansson E, et al. Sicknessabsenteeism in an engineering industry-an analysiswith special reference to absence for neck and upperextremity symptoms. ScandI Soc Med 1989;17:77-84.

9 Berg M, Sanden A, Torell G, Jirvholm B. Persistenceof musculoskeletal symptoms: a longitudinal study.Ergonomics 1988;31:1281-5.

10 Hagberg M, Wegman DH. Prevalence rates and oddsratios of shoulder-neck diseases in different occupa-tional groups. BrJInd Med 1987;44:602-10.

11 Waris P, Kuorinka I, Kurppa K, et al. Epidemiologicscreening of occupational neck and upper limb disor-ders: methods and criteria. ScandJ Work Environ Health1979;5:25-38.

12 Kourinka I, Jonsson B, Kilbom A, et al. StandardisedNordic questionnaires for the analysis of musculoskele-tal symptoms. Apple Ergonomics 1987;3:233-7.

13 Maeda K, Horiguchi S, Hosokawa M. History of the stud-ies on occupational cervicobrachial disorder in Japanand remaining problems. Jv Human Ergol 1982;11:17-29.

14 Theorell T, Harms-Ringdahl K, Ahlberg-Hulthn G,Westin B. Psychosocial job factors and symptoms fromthe locomotor system-a multicausal analysis. Scand JRehabi Med 1991;23:165-73.

15 Kvarnstrom S. Occurrence of musculoskeletal disorders ina manufacturing industry, with special attention tooccupational shoulder disorders. Scand J Rehabil Med1983;(suppl 8):6-35.

16 Tsai S, Gilstrap EL, Cowles SR, Waddell LC, Ross CE.Personal and job characteristics of musculoskeletalinjuries in an industrial population. J Occup Med1992;34:606-12.

266

on January 7, 2021 by guest. Protected by copyright.

http://oem.bm

j.com/

Occup E

nviron Med: first published as 10.1136/oem

.51.4.262 on 1 April 1994. D

ownloaded from