AMERICAN JOURNAL OF INDUSTRIAL MEDICINE (2009)
Review
Risk Factors for Work-Related Musculoskeletal Disorders: A Systematic Review of Recent Longitudinal Studies Bruno R. da Costa,
PT, MSc
1,2
and Edgar Ramos Vieira,
PT, PhD
1,3,4
Objective This systematic review was designed and conducted in an effort to evaluate the evidence currently available for the many suggested risk factors for work-related musculoskeletal disorders. Methods To identify pertinent literature we searched four electronic databases (Cinahl, Embase, Medline, and The Cochrane Library). The search strategies combined terms for musculoskeletal disorders, work, and risk factors. Only case–control or cohort studies were included. Results A total of 1,761 non-duplicated articles were identified and screened, and 63 studies were reviewed and integrated in this article. The risk factors identified for the development of work-related musculoskeletal disorders were divided and organized according to the affected body part, type of risk factor (biomechanical, psychosocial, or individual) and level of evidence (strong, reasonable, or insufficient evidence). Conclusions Risk factors with at least reasonable evidence of a causal relationship for the development of work-related musculoskeletal disorders include: heavy physical work, smoking, high body mass index, high psychosocial work demands, and the presence of co-morbidities. The most commonly reported biomechanical risk factors with at least reasonable evidence for causing WMSD include excessive repetition, awkward postures, and heavy lifting. Additional high methodological quality studies are needed to further understand and provide stronger evidence of the causal relationship between risk factors and work-related musculoskeletal disorders. The information provided in this article may be useful to healthcare providers, researchers, and ergonomists interested on risk identification and design of interventions to reduce the rates of work-related musculoskeletal disorders. Am. J. Ind. Med. 2009. ß 2009 Wiley-Liss, Inc. KEY WORDS: musculoskeletal disorders; risk factors; longitudinal; work
INTRODUCTION 1 Faculties of Rehabilitation Medicine and Nursing, University of Alberta, Edmonton, Canada 2 Graduate Program in Physical Therapy, Pontif|¤ cia Universidade Cato¤ lica, Belo Horizonte, Brazil 3 Graduate Program in Rehabilitation Science, University Nine of July, Sao Paulo, Brazil 4 EWI Work International, Edmonton, Canada *Correspondence to: Edgar Ramos Vieira, 6-10, University Terrace, University of Alberta, 8303-112 Street, Edmonton, Alberta, CanadaT6G 2T4. E-mail:
[email protected]
Accepted 24 July 2009 DOI 10.1002/ajim.20750. Published online in Wiley InterScience (www.interscience.wiley.com)
2009 Wiley-Liss, Inc.
Musculoskeletal disorders are injuries or dysfunctions affecting muscles, bones, nerves, tendons, ligaments, joints, cartilages, and spinal discs. Musculoskeletal disorders include sprains, strains, tears, soreness, pain, carpal tunnel syndrome, hernias, and connective tissue injuries of the structures previously mentioned. According to the National Institute for Occupational Safety and Health (NIOSH), several epidemiological studies have demonstrated evidence of a causal relationship between physical exertion at work and
2
da Costa and Vieira
work-related musculoskeletal disorders (WMSD) [Bernard et al., 1997]. Several factors have been associated with WMSD such as repetitive motion, excessive force, awkward and/or sustained postures, prolonged sitting and standing. In the US alone, more than 600,000 workers have WMSD resulting in days away from work each year [United Electrical Radio and Machine Workers of America, 1999]. Moreover, WMSD is the most expensive form of work disability [Thiehoff, 2002; Picavet and Schouten, 2003; Guo et al., 2004]. It was estimated that the cost of WMSD was approximately 215 billion dollars in 1995 in the US; 26 billion Canadian dollars in 1998 in Canada, and 38 billion Euros in 2002 in Germany [Coyte et al., 1998; Praemer et al., 1999; Thiehoff, 2002]. Each type of WMSD (e.g., low back WMSD, carpal tunnel syndrome, tendinitis, epycondilitis, etc.) has different risk factors. It is difficult for a single study to satisfy all criteria to determine a causal relationship between risk factors and WMSD. Thus, it is important to integrate information derived from different studies of factors contributing to different types of WMSD. This way, risk factors and the level of evidence of their relationship with each WMSD can be evaluated. A previous thorough review published by NIOSH in 1997 included studies of risk factors for WMSD affecting the neck, the upper limbs, and the low back [Bernard et al., 1997]. The NIOSH report provides pivotal information on WMSD risk factors thanks to its extensive search of the literature and indepth discussion regarding potential causal relationships between exposure to risk factors and WMSD. However, a substantive amount of epidemiological studies have been published since the release of the NIOSH report. Other reviews have also investigated a potential causal relationship between risk factors and WMSD affecting specific body parts such as the low back and the upper limbs [Hoogendoorn et al., 1999, 2000a; National Research Council and Institute of Medicine (NRC/IOM), 2001]. But no reviews of high methodological quality studies of risk factors for multiple body parts were found in the published literature after the NIOSH review. Given the gap identified, our systematic review was designed and conducted in an effort to evaluate the evidence currently available in high methodological quality studies published after the NIOSH review (Bernard et al., 1997) of the numerous suggested risk factors for WMSD affecting different body parts. As it was done in the NIOSH report, we investigated and reported potential causal relationships between exposure to risk factors and development of WMSD. We also integrated the results of the previous body part-specific reviews by comparing those findings with our findings.
METHODS Literature Search and Data Management We searched four electronic databases (Cinahl, Embase, Medline, and The Cochrane Library) from January 1997 to
March 2008 in order to identify studies evaluating risk factors for WMSD. Our strategy was to combine terms for musculoskeletal disorders, work, and risk factors. The terms included free text-words and subject headings specific to each database (Appendix 1). We based our procedure on guidelines for effective searching [Gallagher et al., 2002; Gillespie and Gillespie, 2003]. We also used information from the ‘‘help’’ section of each examined database, and consulted one medical librarian. All the titles identified were merged into RefWorks Web Based Bibliography Management Software (http://www.refworks.com/).
Inclusion and Exclusion Criteria In this review, injuries due to slips, trips, falls, or similar accidents were not included. To be eligible the articles had to:
(1) (2) (3)
(4) (5) (6)
be a case–control or cohort study of risk factors for WMSD precipitation in the adult population; be published in or after 1997, the year the NIOSH report was released [Bernard et al., 1997]; report statistically significant association (i.e., odds ratio with confidence interval that does not include ‘‘1.0’’) between risk factor and WMSD; be peer reviewed; be published in English; be published in full text.
We did not include case series and cross-sectional studies as these designs are not ideal to identify determinants of a health condition [Walker-Bone and Cooper, 2005; Carroll et al., 2008]. Case–control and cohort studies are more appropriate to evaluate causality. They were defined according to Domholdt [2005].
Screening Process Both authors screened independently and in duplicate the titles and abstracts of all identified citations and retrieved the full texts of all potentially eligible studies when available. The reference lists of all retrieved studies were screened for additional relevant articles. Both authors applied independently and in duplicate the eligibility criteria to all potentially eligible studies identified. Both authors were blinded to each other’s search process and results prior to the agreement analyses. Kappa statistics was used to calculate the chance-adjusted between-reviewer agreement on the selection of articles by applying the inclusion/exclusion criteria defined for screening titles and abstracts. Disagreements regarding study eligibility were resolved afterwards through discussions until consensus was reached regarding the inclusion or exclusion of each article.
Risk Factors for WMSD
Evaluation Process
RESULTS
The methodological quality of the eligible studies was evaluated in relation to whether they fulfilled the criteria for inference of causality or not, and to how likely misleading factors (chance, bias, and confounding factors) affected the results. The criteria we used to evaluate a potential causal relationship included: strength of association, consistency between studies, temporality, dose–response relationship, and coherence [Bernard et al., 1997]. These criteria were established in accordance with the work of Hill [1965], the work of Lilienfeld and Lilienfeld [1980], and the NIOSH report [1997]. To fulfill the criterion ‘‘strength of association,’’ relative risks or odds ratio defining the association between exposure and outcome had to present a ratio of at least ‘‘3.0’’ and a confidence interval that did not cross ‘‘1.0’’ in at least one study. Risk factors were considered ‘‘consistent’’ if more than one study reported their association with the outcome of interest. Only risk factors identified in prospective cohorts were considered to have fulfilled the criterion ‘‘temporality.’’ To satisfy the ‘‘dose–response relationship’’ criterion, at least one study had to show evidence of a dose–response curve of the observed association (evidence that disorder is dose-dependent and aggravates as exposure increases or vice-versa). ‘‘Coherence’’ means that an association is consistent with the natural history and biology of the investigated disorder. The five major misleading factors that we detected were: poorly defined case population, use of conventional stepwise regression, no assessment of variables’ co-linearity, insufficient sample sizes (e.g., 4 weeks and pain within past year (compared to never had pain) HR: 3.3 (1.9^5.8) Previous LBP >1year prior to baseline measurements HR: 1.8 (1.2^2.7) Previous LBP pain within the past year prior to baseline measurements HR: 1.9 (1.4^2.7) Total duration of previous LBP pain 1^4 weeks HR: 1.8 (1.2^2.7) >4 weeks HR: 2.3 (1.6^3.3) >4 weeks and pain within past year (compared to never had pain) HR: 2.6 (1.8^3.9) Repetition, posture
Previous neck/shoulder pain >1year prior to baseline measurements HR: 1.6 (1.1^2.3) Previous neck/shoulder pain within the past year prior to baseline measurements HR: 2.8 (2.0^3.9)
Job demands Reference category OR: 1.0 High job demands OR: 1.5 (1.3^1.8) Level of distress Low (reference) OR: 1.0 Medium OR: 1.4 (1.1^1.7) High OR: 1.8 (1.4^2.5) Physical examination Pain pressure threshold Reference category (high threshold) OR: 1.0 Low pain pressure threshold: 1.3 (1.1^1.5) Co-morbidities
Psychosocial factor
High physical exposure OR: 1.5 (1.2^1.9)
Reference category OR: 1.0
Prospective cohort
587 female nurses from two hospitals in the Questionnaire to investigate physical risk factors in south of England the workplace, psychosocial factors at work, psychological factors, health-related factors, and other individual factors
(Continued )
Co-linearity not assessed Method of regression not reported
Enrollment response rate of 59%
Follow-up response rate of 45% Co-linearity not assessed. Unclear case definition.
8
Low back WMSD
Low back WMSD
Krause et al. [1998]
Eriksen et al. [1999]
Low back WMSD
Type of WMSD
Gheldof et al. [2007]
Refs.
TABLE I. (Continued
Risk factor evaluated and main findings
Study design
Years of professional driving 0^5 years OR: 6.07 (4.05^9.11) >5^15 years (reference) OR: 1.00 >15 years OR: 0.49 (0.33^0.71) Regular driving hours per week 20^30 hr OR: 0.37 (0.15^0.93) 31^40 hr (reference) OR: 1.00 Overtime driving per week (10 hr) Diesel bus (reference) OR: 1.00 Cable car OR: 3.04 (1.85^5.00) Age (year) 25^44 (reference) OR: 1.00 45^60 OR: 0.63 (0.45^0.88) 61^65 OR: 0.28 (0.08^0.98) Psychosocial factor High psychological demands OR: 1.50 (1.13^1.99) High job dissatisfaction OR: 1.56 (1.09^2.23) Heavy physical work
Negative affectivity OR: 1.06 (1.01^1.11) Posture
Prospective cohort
Prospective cohort
Assist a patient to mobilize using a walking stick, Zim mer frame,orcrutches,never(reference) HR:1.0,5/ shift HR: 1.6 (1.1^2.3) Move a patient around in a wheelchair,bed,hoist,trolley, commode, etc., never (reference) HR: 1.0, 5/shift HR: 1.6 (1.1^2.4) Wash/dress a patient while they are on a chair/commode, never (reference) HR: 1.0, 5/shift HR: 1.7 (1.1^2.8) Psychosocial factor Prospective cohort
Method of investigation
Study limitations
562 workers who responded to a questionnaire seeking employed individuals sent to the general population of Ullensaker (Norway)
1463 transit vehicle operators employed by the San Francisco Municipal Railway
Questionnairetoinvestigatephysical risk factorsinthe Enrollment response rate of workplace, psychological factors, health-related 67% factors, and other individual factors
Method of Medical examination and questionnaire with regression not questions regarding the following domains: reported physical risk factors in the workplace,psychosocial factors at work, health-related factors, and other individual factors
Questionnaire and clinical examination to investigate Response rate Sample of 812 individuals that was of 63% physical risk factors in the workplace,psychosocial predominantly composed of male industrial factors at work, health-related factors, and other workers from Belgian and Dutch individual factors metallurgical/steel companies
Study population
9
Low back WMSD
Low back WMSD
van Poppel et al. [1998]
Kerr et al. [2001]
Psychosocial factor Lower workplace social environment score (cases lower) OR: 2.6 (1.30^5.42) Highereducationrelativeto othersinsimilarjobs (cases lower) OR: 2.2 (1.05^4.92) Higher job satisfaction (cases higher) OR: 1.7 (1.15^2.48)
Prior compensation claim for LBP OR: 2.2 (1.07^4.43)
Compensation claim
Decrement of 4.8 units (cases lower) OR: 2.0 (1.20^3.58)
Back pain more than twice in the past year OR: 9.8 (2.8^34.4) Psychological factor Job dissatisfaction OR: 1.2 (1.01^1.4) Posture Riding forklift truck OR: 0.7 (0.5^0.99) BMI (kg/m2) Case^control
Heavy physical work smoking Heavy lifting, much standing, and smoking OR: 4.04 (1.15^14.21) Co-morbidities Musculoskeletal pain in the previous year (except for LBP) OR: 1.61 (1.03^2.52) Physical activity Performing physical activity less than once a week OR: 1.55 (1.03^2.33) Co-morbidities Prospective cohort
Heavy lifting and much standing OR: 2.30 (1.21^4.34)
Questionnaire to investigate physical risk factors in the workplace, psychosocial factors at work, health-related factors, and other individual factors
Co-linearity not assessed
(Continued )
‘‘Cases’’ response Questionnaires and work-site assessment to Assembly and non-assembly workers of a rate of 56% investigate physical risk factors in the workplace, modern automobile production complex psychosocial factors at work, health-related that manufactures passengercars and light factors, and other individual factors trucks ‘‘Controls’’ Cases: 137 workers reporting a new episode of response rate low back pain to one of the on-site of 43% occupational nursing stations Unclear case Controls: The main control group consisted of definition 179 workers randomly selected (unmatched) from the same study base as the case subjects. A second control group was also recruited,with 65 controls matched to cases by the actual job performed (not job title) Co-linearity not assessed
312 workers involved with manual material handling from the cargo department of a major Dutch airline company at Schiphol Airport
Co-linearity not assessed
10
Risk factor evaluated and main findings
Never/occasionally performing monotonous work (reference) OR: 1.0 Performing monotonous work at least half of the time OR: 1.8 (1.1^2.8) Co-morbidities Any pain lasting 24 hr or longer in the month prior to baseline OR: 1.5 (1.1^2.1) Temperature in the workplace Working in hot conditions OR: 1.7 (1.1^2.6)
Non-smoker (reference) OR: 1.0 Ex-smoker OR: 2.3 (1.3^4.3) Physical activity Jogging Not at all or only little (reference) OR: 1.0 Moderately or actively OR: 3.9 (1.4^10.7) Psychosocial factor Job satisfaction Rather or very satisfied OR: 1.0 Rather or very dissatisfied OR: 2.8 (1.2^6.7) Psychosocial factor
Higher coworker support (cases higher) OR: 1.6 (1.07^2.32) Higher perceived exertion at work (cases higher) OR: 3.0 (1.79^5.36) Heavy physical work (biomechanical measures) Higher peak lumbar shear (cases higher) OR: 1.7 (1.02^2.86) Higher cumulative lumbar disc compression (cases higher) OR: 2.0 (1.22^3.59) Higher peak hand force (cases higher) OR: 1.9 (1.21^3.10) Low back WMSD Smoking (sciatic pain)
Type of WMSD
Harkness et al. Low back WMSD [2003a]
Miranda et al. [2002a]
Refs.
TABLE I. (Continued
Prospective cohort
Prospective cohort
Study design
520 newly employed workers from 12 diverse occupational groups
2,984 workers from a large forestry company in Finland
Study population
Study limitations
Questionnairetoinvestigatephysical risk factorsinthe Follow-up response rate of workplace, psychosocial factors at work, 55% health-related factors, and other individual factors
Questionnaire, to investigate physical risk factors in Enrollment response rate of the workplace, physical activity, health-related 47% factors, and other individual factors
Method of investigation
11
Low back WMSD
Myers et al. [1999]
Eriksen et al. Low back WMSD [2004]
Peek-Asa et al. Low back WMSD [2004]
Low back WMSD
Hoogendoorn et al. [ 2000b]
Positioning patients in bed 5^9 times per shift OR: 1.63 (1.14^2.31)
No positioning of patients in bed (reference) OR: 1.0
45^54 years RR: 0.78 (0.66^0.92) Lifting High lifting intensity 40 RL: 2.51 (1.22^5.14) Co-morbidities History of wrist/hand/finger discomfort RL: 3.14 (1.41^6.99) Hand threshold limit value above the proposed action limit RL: 2.14 (1.01^4.54) Age
Age
Job change at follow up RR: 4.8 (1.5^15.4)
High work load HR: 3.2 (1.1^9.4)
Heavy physical work
Previous upper limb work-related MSD OR: 3.90 (1.31^11.62)
Co-morbidities
Risk factor evaluated and main findings
Study population
Method of investigation Questionnaire and clinical examination to investigate physical risk factors in the workplace, psychosocial factors at work, health-related factors, and other individual factors
Prospective cohort 501workers from 4 industrial and 3 clerical work sites
Prospective cohort 501workers from 4 industrial and 3 clerical work sites
Study limitations
Co-linearity not assessed Unclear case definition Method of regression not reported
Co-linearity not assessed Unclear case definition. Method of regression not reported
Method of regression not reported
(Continued )
Enrollment response rate Questionnaires and physical examination not reported to investigate physical risk factors in the workplace, psychosocial factors at work, health-related factors, and other individual factors
Co-linearity not assessed Response rate of 67% Unclear case definition Questionnaire and physical examination to Method of regression assess upper limbs physical risk factors in not reported the workplace, psychosocial factors at work, health-related factors, and other individual factors.Electro-diagnostic test to assess performance of the median and ulnar sensory nerves. Rating of jobs ergonomic exposure according to the American Congress of Governmental Industrial Hygienists’ threshold limit value Co-linearity not assessed
Interview, questionnaire and standardized Prospective cohort 790 workers from an automotive stamping plant and physical examination of the upper an automobile engine assembly limbs. During interview data was collected on psychosocial strain and plant, both in Detroit (USA) current andpastmedical and workhistories
Prospective cohort 364 assemblers in a trailer Investigation of health status and sick production company of Finland leaves via medical records; and on-site ergonomics observations and job titles used to assess exposure
Prospective cohort 166 blue-collar workers of shoe factories
Study design
TABLE II. Overview of Studies Investigating Risk Factors for Upper Limb Work-Related Musculoskeletal Disorders (WMSD)
14
Lassen et al. [2004]
Leclerc et al. [2001]
Refs.
Upper limb WMSD (elbow and wrist/ hand pain)
Upper limb WMSD (carpal tunnel syndrome,wrist tendinitis, and lateral epicondylitis)
Type of WMSD
TABLE II. (Continued)
Computer work Using the mouse 0^30 1.93 (1.12^3.34) Physical examination Shoulder discomfort at baseline OR: 1.84 (1.03^3.29) Worst discomfort rating at baseline OR: 1.21 (1.06^1.38) Epicondylitis
BMI (kg/m2)
>40 years old OR: 1.76 (1.04^2.98)
Risk factor evaluated and main findings
Study population
Method of investigation Follow-up response rate of 51% Method of regression not reported Co-linearity not assessed
Study limitations
Prospective cohort 5658 Danish engineering technical assistants and machine technicians
Questionnaire and clinical examination to investigate physical risk factors in the workplace, psychosocial factors at work, health-related factors, and other individual factors
Unclear case definition
Use of stepwise regression Prospective cohort 598 individuals from18 different Questionnaire with questions regarding the following domains: physical risk factors companies exposed to in the workplace, psychosocial factors at repetitive work in one of the work, health-related factors, and other following: assembly line work, clothing and shoe industry, food individual factors industry, packaging, and supermarket cashiers.The sample also included 326 subjects not exposed to repetitive work Co-linearity not assessed
Study design
15
Miranda et al. [2001]
Shoulder WMSD
Age (years)
