A Critical Review of Epidemiologic Evidence for Work-Related Musculoskeletal Disorders of the Neck, Upper Extremity, and Low Back

Chapter 6. Low-Back Musculoskeletal Disorders: Evidence for
Work-Relatedness

Summary

Over 40 recent articles provided evidence regarding the relationship between low-back disorder and the five physical workplace factors that were considered in this review. These included (1) heavy physical work, (2) lifting and forceful movements, (3) bending and twisting (awkward postures), (4) whole-body vibration (WBV), and (5) static work postures. Many of the studies addressed multiple work-related factors. All articles that addressed a particular workplace factor contributed to the information used to draw conclusions about that risk factor, regardless of whether results were positive or negative.

The review provided evidence for a positive relationship between back disorder and heavy physical work, although risk estimates were more moderate than for lifting/forceful movements, awkward postures, and WBV. This was perhaps due to subjective and imprecise characterization of exposures. Evidence for dose-response was equivocal for this risk factor.

There is strong evidence that low-back disorders are associated with work-related lifting and forceful movements. Of 18 epidemiologic studies that were reviewed, 13 were consistent in demonstrating positive relationships. Those using subjective measures of exposure showed a range of risk estimates from 1.2 to 5.2, and those using more objective assessments had odds ratios (ORs) ranging from 2.2 to 11. Studies using objective measures to examine specific lifting activities generally demonstrated risk estimates above three and found dose-response relationships between exposures and outcomes. For the most part, higher ORs were observed in high-exposure populations (e.g., one high-risk group averaged 226 lifts per hour with a mean load weight of 88 newtons [N]) . Most of the investigations reviewed for this document adjusted for potential covariates in analyses; nevertheless, some of the relatively high ORs that were observed were unlikely to be caused by confounding or other effects of lifestyle covariates. Several studies suggested that both lifting and awkward postures were important contributors to the risk of low-back disorder. The observed relationships are consistent with biomechanical and other laboratory evidence regarding the effects of lifting and dynamic motion on back tissues.

The review provided evidence that work-related awkward postures are associated with low-back disorders. Results were consistent in showing positive associations, with several risk estimates

above three. Exposure-response relationships were demonstrated. Many of the studies adjusted for potential covariates and a few examined the simultaneous effects of other work-related physical factors. Again, it appeared that lifting and awkward postures both contribute to risk of low-back disorder.

There is strong evidence of an association between exposure to WBV and low-back disorder. Of 19 studies reviewed for this document, 15 studies were consistent in demonstrating positive associations, with risk estimates ranging from 1.2 to 5.7 for those using subjective exposure measures, and from 1.4 to 39.5 for those using objective assessment methods. Most of the studies that examined relationships in high-exposure groups using detailed quantitative exposure measures found strong positive associations and exposure-response relationships between WBV and low back disorders. These relationships were observed after adjusting for covariates.

Both experimental and epidemiologic evidence suggest that WBV may act in combination with other work-related factors, such as prolonged sitting, lifting, and awkward postures, to cause increased risk of back disorder. It is possible that effects of WBV may depend on the source of exposure (type of vehicle).

With regard to static work postures and low-back disorder, results from the studies that were reviewed provided insufficient evidence that a relationship exists. Few investigations examined effects of static work postures, and exposure characterizations were limited.

Introduction

Low-back pain (LBP) is common in the general population: lifetime prevalence has been estimated at nearly 70% for industrialized countries; sciatic conditions may occur in one quarter of those experiencing back problems [Andersson 1981]. Studies of workers’ compensation data have suggested that LBP represents a significant portion of morbidity in working populations: data from a national insurer indicate that back claims account for 16% of all workers’ compensation claims and 33% of total claims costs [Snook 1982; Webster and Snook 1994b]. Studies have demonstrated that back disorder rates vary substantially by industry, occupation, and by job within given industries or facilities [see Bigos et al. 1986a; Riihimaki et al. 1989a; Schibye et al. 1995; Skovron et al. 1994].

Back disorder is multifactorial in origin and may be associated with both occupational and non work-related factors and characteristics. The latter may include age, gender, cigarette smoking status, physical fitness level, anthropometric measures, lumbar mobility, strength, medical history, and structural abnormalities [Garg and Moore 1992]. Psychosocial factors, both work- and non work-related, have been associated with back disorders. These relationships are discussed at length in Chapter 7 and Appendix B.

The relationship of the disorder with employment can be complex: individuals may experience impairment or disability at work because of back disorders whether the latter was directly caused by job-related factors or not. The degree to which ability to work is impaired is often dependent on the physical demands of the job. Furthermore, when an individual experiences a back disorder at work, it may be a new occurrence or an exacerbation of an existing condition. Again, originally it may have been directly caused by work or by non work-related factors. Those suffering back pain may modify their work activities in an effort to prevent or lessen pain. Thus, the relationship between work exposure and disorder may be direct in some cases, but not in others.

When discussing causal factors for low-back disorders, it is important to distinguish among the various outcome measures, such as LBP, impairment, and disability. LBP can be defined as chronic or acute pain of the lumbosacral, buttock, or upper leg region. Sciatic pain refers to pain symptoms that radiate from the back region down one or both legs; lumbago refers to an acute episode of LBP. In many cases of LBP, specific clinical signs are absent. Low-back impairment is generally regarded as a loss of ability to perform physical activities. Low-back disability is defined as necessitating restricted duty or time away from the job. Although it is not clear which outcome measure is best suited for determining the causal relationship between low-back disorder and work-related risk factors, it is important to consider severity when evaluating the literature.

In addition to level of severity, outcomes may be defined in a number of other ways, ranging from subjective to objective. Information on symptoms can be collected by interview or questionnaire self-report. Back “incidents” or “reports” include conditions reported to medical authorities or on injury/illness logs; these may be symptoms or signs that an individual has determined need for medical or other attention. They may be due to acute symptoms, chronic pain, or injury related to a particular incident, and may be subjectively or objectively determined. Whether an incident is reported depends on the individual’s situation and inclinations. Other back disorders can be diagnosed using objective criteria—for example, various types of lumbar disc pathology.

There are many conditions in the low back which may cause back pain, including muscular or ligamentous strain, facet joint arthritis, or disc pressure on the annulus fibrosis, vertebral end-plate, or nerve roots. In most patients, the anatomical cause of LBP, regardless of its relationship to work exposures, cannot be determined with any degree of clinical certainty. Muscle strain is probably the most common type of work or non work back pain. While there is sometimes a relationship between pain and findings on magnetic resonance imaging (MRI) of disc abnormalities (such as a herniated disc and clinical findings of nerve compression), unfortunately, the most common form of back disorder is “non-specific symptoms,” which often cannot be diagnosed. It is important to include subjectively defined health outcomes in any consideration of work-related back disorders because they comprise such a large subset of the total. It may be too restrictive to define cases of back disorder using “objective” medical criteria. Therefore, in contrast to chapters for musculoskeletal disorders or other anatomic regions, this review of literature on the back used slightly different evaluation criteria. For consideration of back disorders, use of a subjective health outcome was not necessarily considered a study limitation. Furthermore, because back disorders were rarely defined by medical examination criteria, the evaluation criterion related to blinding of assessors (to health or exposure status) was also less relevant to a discussion of this literature.

In this review, epidemiologic studies of all forms of back disorder were included. The term “back disorder” is used to encompass all health outcomes related to the back. It should be pointed out that, in some studies, disorders of the low back were not distinguished from total back disorders. We assumed that a significant portion of these related to the low back, and articles using such a definition were included in our review.

The 42 epidemiologic studies discussed below were selected according to criteria that appear in the introduction of this document. Most (30) used a cross-sectional design, followed by prospective cohort (5), case-control (4), and retrospective cohort (2) designs. One study combined both cross-sectional and cohort analyses. Full descriptions of the studies appear in Table 6-6. Twenty-four investigations defined the health outcome only by report of symptoms on questionnaires or in interviews (for example, total back pain, LBP, and sciatica); used symptoms plus medical examination (back pain, low-back syndrome, sciatica, back insufficiency, lumbago, herniated lumbar disc, and lumbar disc pathology), 2 used sick leaves and medical disability retirements, and 6 used injury/illness reports. The last category included outcomes defined as “low-back complaints, injuries caused specifically by lifting or mechanical energy,” and “acute industrial back injury.” Clearly, the 42 studies used outcome definitions that correspond to several regions of the back and include disorders that may have been acute or chronic and subjectively or objectively determined.

In the studies included in this review, exposures were assessed primarily by questionnaire or interview (n=17), followed by observation or direct measurement (n=15) and by job title only (n=10). Study groups included general populations (Swedish, Dutch, U.S., Finnish, and English) and occupational groups (nurses, clerical employees, school lunch preparers, baggage handlers, and individuals working in construction, agriculture, maritime, petroleum, paper products, transportation, automobile, aircraft, steel, and machine manufacturing industries).

This review of epidemiologic studies of low- back disorder examined the following potential risk factors related to physical aspects of the workplace: (1) heavy physical work, (2) lifting and forceful movements, (3) bending and twisting (awkward postures), (4) WBV, and (5) static work postures. Psychosocial workplace factors were also included in a number of studies; these relationships are discussed separately in Chapter 7. Following are discussions of the evidence for each work-related physical risk factor.

Back to top

Heavy Physical Work

Definition

Heavy physical work has been defined as work that has high energy demands or requires some measure of physical strength. Some biomechanical studies interpret heavy work as jobs that impose large compressive forces on the spine [Marras et al. 1995]. In this review, the definition for heavy physical work includes these concepts, along with investigators’ perceptions of heavy physical workload, which range from heavy tiring tasks, manual materials handling tasks, and heavy, dynamic, or intense work. In several studies, evaluation of this risk factor was subjective on the part of participant or investigator, and in many cases, “heavy physical work” appeared to include other potential risk factors for back disorder, particularly lifting and awkward postures.

Studies Reporting on the Association Between LBP and Heavy Physical Work
Eighteen studies appeared to address the risk factor related to heavy physical work, although none of them fulfilled all four evaluation criteria (Table 6-1, Figure 6-1). In fact, most (78%) had acceptable participation rates, but only three defined health outcomes using both symptoms and medical exam criteria, and only two assessed exposure independent of self-report.

In nearly all of these studies, covariates were addressed in at least minimal fashion, such as restricting the study population as to gender and conducting age-stratified or adjusted analyses; in many, multivariate analyses were carried out. With regard to health outcome, while only three used medical exams, in addition to symptoms or injury reports, to arrive at case definitions, in many instances standard questionnaire instruments were used. The major study limitations, overall, were related to relatively poor ascertainment of exposure status.

Following are descriptions of seven studies that were most informative. Detailed descriptions for all 18 investigations can be found in Table 6-6.

Bergenudd and Nilsson [1988] followed a Swedish population-based cohort established in 1938.Back pain (total) presence and severity were self-assessed by questionnaire, as of 1983; exposures (light, moderate, or heavy physical work) were assessed based on questionnaires completed by the cohort from 1942 onward. Univariate results demonstrated that those with moderate or heavy physical demands in their jobs had more back pain than those with light physical demands (OR 1.83, 95% Confidence Interval [CI] 1.2-2.7). When stratified by gender, the relationship was slightly stronger for females (OR 2.03, 95% CI 1.1–3.7) than for males (OR 1.76, 95% CI 1.01–3.1). When prevalence was examined by exposure category, rates were 21.4%, 32.8%, and 31.3% for males (no trend was available for females, as none worked in the highest exposure category). Analyses were stratified by gender but did not account for other potential covariates. The longitudinal design ensured that exposures preceded health outcomes. Shortcomings included a relatively low response rate (67%), minimal exposure assessment, limited adjustment for covariates in analyses, and self-reporting of health symptoms.

Burdorf and Zondervan [1990] carried out a cross-sectional study comparing 33 male workers who operated cranes with age-matched workers from the same Dutch steel plant who did not operate cranes. Symptoms of LBP and sciatica were assessed by questionnaire. Exposure was assessed by job title (crane operators were noted to experience frequent twisting, bending, stooping, static sedentary postures, and WBV) and by questionnaire (exposures to sedentary postures, WBV, heavy physical work, and frequent lifting were assessed for both current and past jobs). Crane operators were significantly more likely to experience LBP (OR 3.6, 95% CI 1.2–10.6). Among crane operators alone, the OR for heavy work was 4.0 (95% CI 0.76–21.2) after controlling for age, height, and weight. It was determined that this heavy work occurred in past and not in current jobs. Among crane operators alone, the OR for frequent lifting was 5.2 (95% CI 1.1–25.5). The frequent lifting in crane operators was also determined to be from jobs held in the past. Among workers who were not crane operators, history of frequent lifting was not associated with LBP (OR 0.70, 95% CI 0.14–3.5). Among crane operators, univariate ORs for WBV and prolonged sedentary postures were 0.66 (95% CI 0.14–3.1) and 0.49 (95% CI 0.11–2.2), respectively. In multivariate analyses controlled for age, height, weight, and current crane work, most of the associations with specific work-related factors were substantially reduced. The high prevalence of LBP in crane operators was explained only by current crane work. No measures of dose-response were examined. Limitations included a relatively low response rate for crane operators (67%)—with some suggestion that those with illness may have been under-represented (perhaps underestimating the OR)—and self-reporting of health outcomes and exposures. The investigators attempted to clarify the temporal relation between exposure and outcome by excluding cases of back pain with onset before the present job.

As part of a Finnish population-based health survey, Heliövaara et al. [1991] conducted a cross-sectional analysis of chronic low-back syndrome, sciatica, and LBP. Health outcomes were determined by interview and examination; work-related exposure information was obtained by a self-administered questionnaire, which included items related to lifting, carrying heavy objects, awkward postures, WBV, repeated movements, and paced work. The total number of factors was designated the “sum index of occupational physical stress.” Mental work stress measures were also included. A dose-response was observed for sciatica and the physical stress score (with an OR of 1.9, 95% CI 0.8–4.8 for the highest score) and for low-back syndrome and physical stress (OR 2.5, 95% CI 1.4–4.7), after adjusting for a number of covariates. The study did not address temporal relationships, and exposure information was derived from self-reports. Strengths included a high response rate, objective measure of health outcomes, and multivariate adjustment for covariates.

Johansson and Rubenowitz [1994] examined low-back symptoms cross sectionally in 450 blue-and white-collar workers employed in eight Swedish metal companies. The exposed group included assemblers, truck drivers, welders, smiths, and operators of several types of machines (lathes, punch presses, and milling). Outcome information was obtained by questionnaire. Exposure data were also obtained by questionnaire and included information on occupational, psychosocial, and physical workloads, including sitting, carrying, pushing, pulling, lifting, work postures, and repetitive movements. Questionnaire items related to carrying, pushing, pulling, and lifting were combined to produce an index of manual materials handling. The prevalence of work-related LBP was significantly higher in blue-collar employees than in white-collar workers (RR 1.8, p<0.05). In both white and blue-collar workers, work-related LBP was not significantly associated with either heavy or light materials handling, or bent or twisted work postures, after adjustment for age and gender. LBP was significantly associated with extreme work postures (blue-collar workers only) and monotonous working movements (white-collar workers only). In these analyses, relationships were presented as partial correlations; thus, a comparison of risk estimates was not possible. Limitations of the study included the cross-sectional design, collection of outcome and exposure data by self-report, and potential problems with multiple comparisons, as many independent variables were examined in analyses. Many of the exposed group (blue-collar workers) were engaged in machine operation tasks with perhaps limited opportunity for exposure to work with heavy physical demands. Also, heavy physical work and lifting were combined into a single index. Strengths included consideration of age and gender as covariates and inclusion of both physical and psychosocial workplace measures.

Svensson and Andersson [1989] examined LBP in a population-based cross-sectional study of employed Swedish women. Information on LBP and sciatica was obtained by questionnaire, as were exposure-related items. Physical exposures included lifting, bending, twisting, other work postures, sitting, standing, monotony, and physical activity at work. Lifetime incidence rates (IRs) varied by occupation, with ranges from 61%–83% in younger age groups and 53%–75% in older groups. A posteriori, the authors noted that, for these women, the highest lifetime incidence of LBP was not found in the jobs with the highest physical demands. The measure for “physical activity at work” was also not significantly associated with LBP in univariate analyses. Bending forward (RR 1.3), lifting (RR 1.2), and standing (RR 1.3) were associated with lifetime incidence of LBP in univariate analyses (p<0.05). None of the measures of physical workplace factors were associated with lifetime incidence of LBP in multivariate analyses.

A cross-sectional study of LBP in Finnish nurses was conducted [Videman et al. 1984]. LBP and sciatica were ascertained by questionnaire; exposure information was also self-reported and included items related to both physical loading factors at work and to work history. Exposures were reclassified as “heavy,” “intermediate,” and “light,” based on questionnaire responses. The derivation of this classification was not clear, but it may have been a combination of responses to questions on lifting, bending, rotation, standing, walking, and sitting. A dose-response was observed between prevalence of previous LBP and workload category in younger women (77%, 79%, and 83% for light, intermediate, and heavy categories). The trend was not observed in older age groups, nor for sciatica in any age group. LBP and sciatica rates were slightly higher for nurse aides than for qualified nurses, although the differences were not statistically significant. The authors suggested that aides had higher rates of back pain because of heavier workload, including patient handling and lifting. Lack of consistency of LBP OR across exposure and age groups suggested that a healthy worker effect was operating and that injured workers might be leaving the field, a phenomenon that the cross-sectional study design could not address.

Videman et al. [1990] carried out a cross-sectional study of 86 males who died in a Helsinki hospital to determine degree of lumbar spinal pathology. Disc degeneration and other pathologies were assessed in the cadaver specimens by discography and radiography. Subjects’ symptoms and work exposures—heavy physical work, sedentary work, driving, and mixed—were determined by interview of family members. In comparison to those with mixed work exposures, those with sedentary and heavy work had increased risk of symmetric disc degeneration with ORs of 24.6 (95% CI 1.5–409) and 2.8 (95% CI 0.3–23.7), respectively). Similar relationships were seen for vertebral end-plate defects and facet joint osteoarthrosis. Risk of vertebral osteophytosis was highest for those in the heavy work category (OR 12.1, 95% CI 1.4–107). For most pathologic changes, sedentary work appeared to have a stronger relationship than heavy work. Back pain symptoms were consistently higher in those with any form of spinal pathology, although the difference was significant only for anular ruptures. Results of this study were notable in that anular rupture, a classic pathologic condition of the disc, was not associated with exposure. This study was unusual in design in that it examined a combination of spinal pathological outcomes, symptoms, and workplace factors. However, participation in the study was dependent on obtaining information from family members; participation rates were not stated. While recall bias is often a problem in studies of the deceased, in this case, it should have been non differential, if present.

Strength of Association

The most informative studies were generally those that carried out exposure assessments which ranked physical workload based on questionnaire report. In a prospective study of back injury reports, Bigos et al. [1991b] found no associations with physical job characteristics (although the authors stated that the study population had low overall exposures). This study described the biomechanical methods that were used to directly assess spinal loads associated with jobs, but no results related to these measures were presented. Svensson and Andersson [1989] appear to have examined a measure for physical activity at work and its relationship to LBP in Swedish women. No associations were observed. In a population-based study, Bergenudd and Nilsson [1988] observed significantly more back pain in those with heavier physical work (OR 1.8 for moderate/heavy versus light work, p<0.01). ORs were slightly higher for females (OR 2.0) than for males (OR 1.8). Leigh and Sheetz [1989] found that back symptoms were associated with self-reporting that “job requires a lot of physical effort” (OR 1.5, 95% CI 1.0–2.2). Masset and Malchaire [1994] observed that LBP was not associated with overall physical workload in a group of Belgian steelworkers, although LBP was related to heavy shoulder efforts. In a study of blue-and white-collar workers, Johansson and Rubenowitz [1994] found higher LBP rates in blue-collar workers (RR 1.8, p<0.05). However, in more detailed analyses of exposure, back pain was not associated with indices for heavy or light materials handling after adjustment for age and gender (with partial correlation coefficients of less than 0.10). Burdorf and Zondervan’s 1990 study of crane operators demonstrated increased risk of LBP with exposure to heavy work (OR4.0, 95% CI 0.8–21.2) after controlling for age, height, and weight. Two studies used indices of physical stress to create questionnaire responses related to lifting, carrying heavy objects, awkward postures, repeated movements, and others. Heli”vaara et al. [1991] found that both low-back syndrome and sciatica were associated with physical stress scores, with ORs of 2.5 (p<0.05) and 1.9 (not significant) for the highest scores, respectively. A study of Finnish nurses classified exposures as “heavy,” “intermediate,” and “light” based on questionnaire response scores [Videman et al. 1984]; prevalence of LBP was slightly higher in the heavy category than in the light (RR 1.1, not significant) for younger women only. Sciatica was also examined, and no relationships were found.

The other studies that examined heavy physical work as a risk factor for back disorder classified exposure in a simpler manner, either by job title alone or by grouping jobs based on prior knowledge of the work or questionnaire responses. Burdorf et al. [1991] found that heavy physical work was associated with back pain in concrete workers in univariate, but not multivariate models (no risk estimate was reported). Hildebrandt [1995] found that individuals in jobs described as “heavy non-sedentary” were more likely to experience back pain than those in sedentary jobs (OR 1.2, p<0.05). In a cadaver study of lumbar disc pathology, Videman et al. [1990] found that those with jobs involving heavy physical work had increased risk of disc pathology in comparison to those with mixed work exposures (e.g., an OR of 2.8, 95% CI 0.3–23.7, for symmetric disc degeneration and an OR of 12.1, 95% CI 1.4–107, for vertebral osteophytosis). For most pathologic changes, sedentary work had a stronger relationship than heavy work.

Finally, several studies examined back disorder rates by job title or occupation alone. Hildebrandt et al. [1996] observed differences in back symptom rates by unit and task group in “nonsedentary” steel workers. The reference group also had high symptom rates; comparisons between the two groups did not yield significant differences. In multivariate analyses, Riihimaki et al. [1989b] found no significant difference in sciatic pain for carpenters and office workers (OR 1.0, 95% CI 0.8–1.3). Partridge and Duthie [1968] found that dock workers had slightly higher LBP rates than civil servants (RR 1.2, not significant). In a similar study, strand [1987] classified pulp mill jobs as heavy and the referent group of clerical jobs as light; mill workers were 2.3 times more likely to experience back pain than clerical staff (p=0.002). Clemmer et al. [1991] found that floor hands, roustabouts, and derrick hands had the highest rates for low-back strains and impact injuries, with RRs of 2.2 and 4.3 (no significance testing was done) in comparison to control room operators and maintenance professionals, those with the lowest rates. A study of hospital employees that matched cases with controls by department found that those on the day shift had an OR of 2.2 (p<0.005) in comparison to those working other shifts [Ryden et al. 1989]. In the last two studies, the authors determined a posteriori that job titles (or shifts) that were observed to have high back disorder rates were those requiring the heaviest physical effort.

Although in all 18 of these studies the authors stated that “heavy physical effort or work” was at least one of the risk factors of interest, the actual estimates of these exposures varied from assumptions based on job title to self-reported scores based on self-reported work activities. In no case were measured physical loads used as independent variables. Study populations included individuals working in health care, office work, manufacturing, construction, and general populations, all with varying degrees of physical work requirements. Some studies created physical “stress” indices that included more than one risk factor. Since most estimates of physical load were subjective, they tended to reflect the relative requirements of the jobs and individuals included in each study. Health outcomes also varied.

In summary, the strength of the relationship between back disorder and heavy physical work in some of the studies with more quantitatively defined exposures ranged from none [Bigos et al. 1991b; Johannsson and Rubenowitz, 1994; Masset and Malchaire 1994; Svensson and Andersson 1989; Videman et al. 1984] to ORs of 1.9 (not significant) for sciatica and 2.5 (p<0.05) for low-back syndrome [Heli”vaara et al. 1991], 1.5 (95% CI 1.0–2.2) [Leigh and Sheetz 1989], 1.8 (95% CI 1.2–2.7) [Bergenudd and Nilsson 1988], and 4.0 (p<0.05) for LBP [Burdorf and Zondervan 1990]. In another study, which used a scoring system and focused on a subject group of nurses, the RR was 1.1 (not significant) for the high-exposure category [Videman et al. 1984].

Dichotomous estimates of physical workload yielded ORs of 1.2 [Hildebrandt 1995], 2.8-12.1 [Videman et al. 1990], and no association (results were observed in univariate but not multivarate analyses, with no risk estimates reported) [Burdorf et al. 1991]. Exposures based on job title alone yielded estimates from none [Hildebrandt et al. 1996], non significant ORs of 1.0 and 1.2 [Partridge and Duthie 1968; Riihimaki et al. 1989b], to significant ORs of 2.2–4.3 [strand 1987; Clemmer et al. 1991; Ryden et al. 1989]. Half of the studies had positive point estimates for this risk factor but were low to moderate in magnitude. In five studies that found no association between back disorder and heavy physical work, no details were given. Two of the highest significant ORs were based on exposed groups in the oil and steel industries [Burdorf and Zondervan 1990; Clemmer et al. 1991]. For these, true exposure to heavy physical work was probably more likely than for some of the other study populations. For many of the investigations, exposure estimates were subjectively assessed. In many cases, study groups had potentially low exposures or exposure to heavy physical work in combination with other risk factors.

Temporal Relationship

Fourteen of the 18 reviewed studies had a cross-sectional design that could not directly address this issue. Three mentioned potential problems related to this study design. strand [1987] suggested that exposure misclassification occurred in her study of paper mill workers (some individuals were transferred to clerical jobs—the unexposed group—after experiencing a back injury in the mill). In the Videman et al. 1984 study of nurses, lack of consistency of LBP OR by age and exposure group suggested that injured workers were leaving the field. A study of cadavers carried out by Videman et al. [1990] seemed to have potential for problems with temporal relationships, as exposure information for past periods depended on recall of study participants’ activities by family members.

Two cross-sectional studies attempted to clarify temporal relationships by excluding from analysis the cases with disorder onset prior to current job [Burdorf et al. 1991; Burdorf and Zondervan 1990]. Both showed results suggesting a positive relationship between exposure and back disorders. Three studies had cohort designs in which temporal relationships between outcome and exposure could be determined [Bergenudd and Nilsson 1988; Bigos et al. 1991b; Clemmer et al. 1991]: in one, no association was observed, in another, a modest increase in risk was seen. In the third, exposure (assessed a posteriori by job title) was significantly associated with back injuries. A case-control study conducted using hospital personnel records appeared free from recall bias and showed a significant association between low-back injury and working the day shift (assessed a posteriori as having the heaviest workload) [Ryden et al. 1989].

Although the majority of studies were limited by their cross-sectional designs, results were similar for these and other studies with designs that could assess temporal relationships.

For most studies, the data are compatible with a temporal relationship in which exposure preceded disorder.

Consistency in Association

Half of the 18 studies examined demonstrated no significant association between exposure and outcome. All of those which showed significant associations (n=9) were positive in direction, (one OR of 1.2, two ORs between 1.5 and 2, and six ORs between 2.2 and 12.1).

Study groups included males working in industrial environments, office workers, health care employees—female, for the most part—and population-based groups that included both genders and many occupations. That some consistency in results was noted among these diverse groups, particularly after adjustment for covariates, suggests that the observed associations have validity and can be generalized across working populations.

Coherence of Evidence

Information derived from a large number of laboratory and field studies using a wide variety of approaches provides a plausible explanation for associations between LBP and physically demanding jobs [Waters et al. 1993]. Research conducted in the 1950s demonstrated that disc degeneration occurs earlier in life among workers who perform heavy physical work than among those who perform lighter work. Similar findings are reported in more recent investigations [Videman et al. 1990]. The stresses induced at the low back during manual materials handling are due to a combination of the weight lifted, and the person’s method of handling the load. The internal reaction forces needed to equilibrate the body segment weights and external forces such as weight of the load being lifted are supplied by muscle contraction, ligaments, and body joints. Injury to the supporting tissues can occur when the forces from the load, body position, and movements of the trunk create compressive, shear, or rotational forces that exceed the capacities of the discs and supporting tissues needed to counteract the load moments. Rowe [1985] hypothesized that disc and facet degeneration and ligament strain are responsible for the potentially high rates of LBP disability in those whose jobs demand heavy physical activity.

The Videman et al. [1990] cross-sectional study of cadavers addressed two aspects of the causal chain linking exposure to heavy physical work and back disorder. First, the study demonstrated an association between subjective health outcome measures and more objective measures: back pain symptoms (assessed from family members) were consistently higher in those with signs of spinal pathology. Second, the study demonstrated an association between objective measures of disorder and heavy work exposures: individuals whose jobs included heavy work exposures showed increased risk of symmetric disc degeneration, vertebral osteophytosis, and facet joint osteoarthrosis. Significant relationships were also found for back pain and disability. We agree with the conclusion of Videman et al. [1990] that states that “back injury and sedentary or heavy (but not mixed) work contributed to the development of pathologic findings in the spine. The severity of back pain was related to the heaviness of work. Work-related factors may be responsible for the development of pathologic changes and for increased episodes of LBP and disability.”

Another important contribution to the coherence of evidence is that the Bureau of Labor Statistics Annual Survey of Injuries and Illnesses has demonstrated significant elevations in overexertion injuries and disorders in industries which are associated with heavy work, such as nursing and personal care and air transportation. Some broad population surveys such as the National Health Interview Survey (NHIS) from 1988 and the 1990 Ontario Health Survey (OHS) found increased back pain or long-term back problems with exposure to factors such as lifting, pulling, and physical pushing [Guo et al. 1995; Liira et al. 1996]. In the NHIS, the two occupations with the highest significant rates of work-related LBP were male construction laborers (with a prevalence ratio [PR] of 2.1) and female nursing aides, orderlies, and attendants (PR 2.8) [Guo et al. 1995]. In the OHS, the number of simultaneous physical exposures was directly related to risk increase after adjustment for covariates. For the highest exposure index level, the adjusted OR was 3.18 (95% CI 1.72–5.8), which occurred in 3% of the population [Liira et al. 1996]. It is important to point out that truly heavy work probably occurs in only a tiny proportion of all jobs in most industries and in only a minority of many high-risk industries, which is why misclassification of exposures is likely in population-based studies.

Exposure-Response Relationships

Only a few studies examined exposure in sufficient detail to assess exposure-response relationships with low-back disorders. Results were mixed. Heliövaara et al. [1991] observed an exposure-response between sciatica and physical stress score; the Videman et al. [1984] results demonstrated a dose-response between LBP prevalence and workload categories in younger nurses, but not in older groups, or for sciatica in any age group. In strand’s 1987 “high exposure group” (pulp mill workers), duration of employment was associated with back pain. Bergenudd and Nilsson [1988] and Johansson and Rubenowitz [1994] observed no exposure-response relationships between back disorders and their exposure measures. On the whole, evidence of exposure-response is equivocal, based on the paucity of information available.

Conclusions: Heavy Physical Work

The reviewed epidemiologic investigations provided evidence that low-back disorders are associated with heavy physical work. Despite the fact that studies defined disorders and assessed exposures in many ways, all studies which demonstrated significant associations between exposure and outcome were positive in direction and showed low to moderate increased risk. Exposures were assessed subjectively, for the most part; and in some cases, classification schemes were crude. This study limitation may have led to misclassification of exposure status to the extent that it caused a dampening effect on risk estimates, where non differential misclassification caused bias toward a null value for the measure of association. This may account for the moderate ORs that were observed. A few studies were able to examine dose-response relationships between outcomes and exposure; these results were equivocal. Most studies utilized cross-sectional study designs; however, five of six studies which used specific methodologies to address temporality showed positive associations between exposure and outcome. Many studies addressed potential effects of covariates by restriction in selection of study participants, stratification, or multivariate adjustment in statistical analyses.

In many studies, “heavy physical work” exposure appeared to include other work-related physical factors (particularly lifting and awkward postures).

Back to top

Lifting and Forceful Movements

Definition

Lifting is defined as moving or bringing something from a lower level to a higher one. The concept encompasses stresses resulting from work done in transferring objects from one plane to another as well as the effects of varying techniques of patient handling and transfer. Forceful movements include movement of objects in other ways, such as pulling, pushing, or other efforts. Several studies included in this review used indices of physical workload that combined lifting/forceful movements with other work-related risk factors (particularly heavy physical work and awkward postures). Some studies had definitions for lifting which include criteria for number of lifts per day or average amount of weight lifted.

Studies Reporting on the Association Between LBP and Lifting and Forceful Movements

Eighteen studies examined relationships between back disorders and lifting or forceful movements. Only one, Punnett et al. 1991 case-control study of back pain in auto workers, fulfilled the four evaluation criteria (Table 6-2, Figure 6-2). The majority (66%) had adequate participation rates; four defined outcomes using both symptoms and medical exam criteria. Blinding of investigators with regard to case/exposure status was not mentioned in most, but it could be confirmed in two papers and inferred (by study methodology ) in two others. Seven studies used an exposure assessment that included observation or direct measurement; an additional nine obtained exposure information by self-report on questionnaire or interview. Only two relied on job title alone to characterize exposure.

Thirteen investigations were cross-sectional in design; three were case-control, and two were prospective. Eleven defined the health outcome by symptom report on interview or questionnaire.

Descriptions of seven studies which provided the most information regarding the relationship between low-back disorder and lifting and forceful movements follow. Detailed descriptions for all 18 investigations can be found in Table 6-6.

The Punnett et al. [1991] case-control study examined the relationship between back pain and occupational exposures in auto assembly workers. Back pain cases (n=95) were determined by symptoms at interview and medical examination; controls included those free of back pain. For all participants (or proxies in the same jobs), jobs were videotaped and work cycles were reviewed using a posture analysis system. Exposures included time spent in various awkward postures. Peak biomechanical forces were estimated for up to nine postures where a load weighing at least10 lb was held in the hands. In multivariate analyses that adjusted for a number of covariates (age, gender, length of employment, recreational activities, and medical history), time in non-neutral postures (mild or severe flexion and bending) was strongly associated with back disorder (OR 8.09, 95% CI 1.4–44). Lifting was also associated with back disorder (OR 2.16, 95% CI 1.0–4.7). When the subset with physical medical findings was examined, associations were more pronounced. Although few study subjects were unexposed to all of the postures studied, a strong increase in risk was observed with both intensity and duration of exposure. It was not possible to determine the relative contributions of different awkward postures because all were highly correlated. Only participants’ current jobs (for referents), or job when symptoms started (for cases) were analyzed; the study design thus assumed a short-term relationship between exposure and outcome (although length of time in job was also included in the models). The authors attempted to ensure that exposure preceded disease by identifying time of onset and measuring exposures in the job held just prior. The strong associations, after adjustment for covariates, are notable.

Burdorf et al. [1991] examined back pain symptoms in a cross-sectional study of male concrete fabrication workers and a referent group of maintenance workers. Back pain symptoms were assessed by questionnaire. Exposures were measured using the Ovako Working Posture Analysis System, which assessed postures for the back and lower limbs along with lifting load. Information on exposures in previous jobs was also collected. Concrete workers experienced significantly more back symptoms than referents (OR 2.8, 95% CI 1.3–6.0). Univariate results showed associations between back pain and both posture index and WBV in current job (correlations were presented). Lifting was not found to be associated with back pain (and exposure was found not to vary significantly across the six job categories examined in the study). In multivariate analyses adjusting for age, both posture index and WBV were significantly associated with back pain, with ORs of 1.23 (p=0.04) (for an ordinal scale of 6) and 3.1 (p=0.01) (dichotomous), respectively. These two measures were highly correlated and analyzed separately. Strengths of the study include use of a standard symptom questionnaire, high participation rates, an objective measure of exposure, and an attempt to clarify the temporal relation between exposure and outcome by excluding cases of back pain with onset before the present job.

Chaffin and Park [1973] carried out a prospective study of back complaints in 411 employees of four electronics manufacturing plants. The outcome included visits to the plant medical department because of back complaints over a one-year period. Exposure was assessed by evaluating 103 jobs with a range of manual lifting for lifting strength rating (LSR) and load weights. The LSR is a ratio of the maximum weight lifted on the job to the lifting strength, in the same load position, for a large/strong man. Results showed a strong increase in back complaint incidence with LSR for both males and females (with an approximate five-fold increase in risk comparing males in the highest and lowest LSR). A similar increase was observed for females, although there were no women in the highest exposure category. No dose-response was observed by frequency of lifts (a relatively high risk of back complaints was observed for the lowest exposure category). Covariates (age, weight, and stature) were examined and found not to contribute to back complaints. The prospective study design helped increase the likelihood that exposure preceded disorder. Study limitations include lack of information on participation rates and an outcome consisting of incident reports. Time of true onset was not ascertained, and it is possible that symptom onset preceded or coincided with exposure assessment despite the longitudinal study design. The detailed exposure assessment addressed only lifting as a risk factor; presence of other risk factors related to back disorders was not identified.

A case-control study of prolapsed lumbar disc was carried out using a hospital population-based design [Kelsey et al. 1984]. Cases (n=232) included individuals diagnosed with prolapsed lumbar disc; an equal number of controls matched on sex, age, and medical service were selected. Exposure was assessed using a detailed occupational history that was not described but presumably was obtained by interview. An association with work-related lifting without twisting the body was observed at the highest lifting level (25 lb or more) (OR 3.8, 95% CI 0.7–20.1). Twisting without lifting was associated with disc prolapse (OR 3.0, 95% CI 0.9–10.2); a combination of both risk factors had an OR of 3.1 (95% CI 1.3–7.5). The highest risk was observed for simultaneous lifting and twisting with straight knees (OR 6.1, 95% CI 1.3–27.9). Despite the fact that exposures were self-reported, these associations were notably strong. The potential existed for differential recall bias for cases and controls because study subjects were interviewed about work-related factors after case status was established. Interviewers may not have been blinded to case/control status.

In Liles et al. [1984] prospective study of 453 individuals working in jobs with manual material handling requirements, incidence of back injuries was examined with regard to lifting. The study group included those who lifted frequently (at least 25 lifts per day of not less than 4.53 kg, with exposure of at least two hours per day). The outcome included reported or recorded lifting injuries to the back. Lifting exposures were assessed until job change (up to a two-year period) using the Job Severity Index (JSI). The JSI is a measure of the physical stress level associated with lifting jobs and is a function of the ratio of job demands to the lifting capacities of the person performing the job. Information on weight, frequency of lifting, and task geometry is collected through comprehensive task analysis. When the study group (working in 101 jobs from 28 plants) was classified into 10 equal categories according to JSI, a dose-response relationship with injury was observed (RR 4.5, 95% CI 1.02–19.9 for total injuries, comparing category 10 to category 1). Study limitations included no statement relating to response rate or participant selection, no adjustment for confounders, and no statistical testing. The outcome definition specified that the back injury be lifting- related, which increased the likelihood that the outcome would be related to the exposure measured. The prospective design assured that measured exposures preceded injury onset. Other strengths included objective assessment of exposure.

Using an unusual cross-sectional study design, Marras et al. [1993, 1995] examined the relationship between low-back disorders and spinal loading during occupational lifting. A total of 403 jobs from 48 diverse manufacturing companies were assessed for risk of low-back disorder using plant medical department injury reports. Jobs were ranked into three categories according to risk, then assessed for position, velocity, and acceleration of the lumbar spine during lifting motions in manual materials handling using electrogoniometric techniques. Those in high-risk jobs averaged 226 lifts per hour, with an average load weight of 88.4 N. A combination of five factors distinguished between high- and low-risk jobs: lifting frequency, load moment, trunk lateral velocity, trunk twisting velocity, and trunk sagittal angle. The highest combination of exposure measures produced an OR of 10.7 (95% CI 4.9–23.6 in comparison to the lowest combined measures). In univariate analyses, the most powerful single variable was maximum moment (a combination of both weight of the object and distance from the body), which yielded a significant OR of 3.3 between low- and high-risk groups [Marras et al. 1995]. The study design was unusual in that the unit of analysis appeared to be the job rather than the individual. Neither participation rates nor total number of participants was stated. No information appeared regarding the proportions of individuals within jobs who were recruited for measurement of lifting motions. However, the unit of analysis was job, and each was characterized by measurement of at least one study subject. Effects of covariates were not addressed (multivariate analyses appeared to include only biomechanical variables). The study results emphasized the multifactorial etiology of back disorders, including contributions of lifting frequency, loads, and trunk motions and postures. The study design did not allow for examination of temporal relationships.

Walsh et al. [1989] examined the relationship between self-reported LBP and work-related factors in a population-based cross-sectional study of 436 English residents. LBP was ascertained by interview, as was lifetime occupational history (including exposures to standing, walking, sitting, driving, lifting, and using vibrating machinery). Exposures were ascertained either as of the birthday prior to onset of symptoms or by lifetime occupational history prior to onset of symptoms. Using the most recent job (as of the birthday prior to symptoms), driving was associated with symptoms in males (RR 1.7, 95% CI 1.0–2.9), as was lifting or moving weights of 25 kg or more (RR 2.0, 95% CI 1.3–3.1), when all exposures were considered in multivariate analyses. For women, lifting (RR 2.0, 95% CI 1.1–3.7) was associated with symptoms. When lifetime exposures were considered, lifting remained significantly associated for males (RR 1.5, 95% CI 1.0–2.4). Both sitting (RR 1.7, 95% CI 1.1–2.6) and use of vibrating machinery (RR 5.7, 95% CI 1.1–29.3, based on one case) were associated with symptoms in females. The multivariate analyses stratified on sex and adjusted for age and simultaneous work exposures. While information on symptoms and exposures was obtained cross sectionally, the authors attempted to construct a retrospective cohort design by gathering data on lifetime work exposures and back symptoms. While in the design lifetime exposures were cumulated only prior to disorder onset, it would not be expected that participants could recall these relationships accurately. Temporal relationships were unclear.

Strength of Association

The most informative studies included those that employed independent measures of exposure to assess lifting demands, as they provided the best contrast among levels of exposure and were subject to the least misclassification. A case-control study by Punnett et al. [1991] found an OR of 2.16 (95% CI 1.0–4.7) for the relationship between back pain (ascertained by symptoms and medical exam) and lifting, after adjusting for covariates (including awkward postures). In their 1973 investigation, Chaffin and Park found a strong increase in incidence of medical visits related to back problems with increased LSR (with an approximate five-fold increase in risk comparing males in the highest and lowest categories); they did not find a similar dose-response relationship for frequency of lifts. Marras et al. [1993, 1995] examined the relationship between low-back injury reports and spinal loading during lifting, and found an OR of 10.7 (95% CI 4.9–23.6) for simultaneous exposures to lifting frequency, load weight, two trunk velocities, and trunk sagittal angle. Both lifting and postures contributed to the high ORs. In Magora’s [1972, 1973] studies of LBP and occupational physical efforts, the highest LBP rate was observed in those who lifted rarely. When LBP was ranked by level of sudden maximal effort, the highest rate was seen for those who did it often, with a dose-response for three categories (10.9, 11.3, and 18.0, respectively, with a RR of 1.65 [95% CI 1.3–2.1]) when comparing lowest to highest). Liles et al. [1984] found a significant association between incidence of back injuries related to lifting and lifting exposures as assessed by JSI: the RR was 4.5 (95% CI 1.02–19.9) comparing the highest and lowest exposure categories. Burdorf et al. [1991] found no association between back pain symptoms and lifting load (the latter did not vary across the six job categories examined in the study). Huang et al. [1988] conducted detailed ergonomic evaluations of two school lunch preparation centers with differing rates of musculoskeletal (including back) disorders. The center with higher disorder rates had greater lifting and other work-related demands. Unfortunately, the study was ecologic in design and did not link exposures and outcomes to calculate risk estimates for the study groups, although several areas for ergonomic intervention were identified.

Other studies assessed exposures by self-report on interview or questionnaire. Johansson and Rubenowitz [1994] examined low-back symptoms by index of manual materials handling (which included lifting and other risk factors). In neither white- nor blue-collar workers was LBP significantly associated with the index. In Kelsey’s 1975 case-control study of herniated lumbar discs, cases and controls had similar histories of occupational lifting (RR 0.94, p=0.10). In a second case-control study of prolapsed lumbar disc, Kelsey et al. [1984] found that an association with work-related lifting without twisting was observed only at the highest lifting level (OR 3.8, 95% CI 0.7–20.1). A combination of both risk factors at moderate levels yielded an OR of 3.1 (95% CI 1.3–7.5). The highest risk was seen for simultaneous lifting and twisting with straight knees (OR 6.1, 95% CI 1.3–27.9). Svensson and Andersson [1989] found a significant association between lifetime incidence of LBP and lifting in univariate analyses (RR 1.2, p<0.05), but not in multivariate analyses. Holmstrom et al. [1992] found an association between one-year prevalence of LBP and an index of manual materials handling (OR 1.27, 95% CI 1.2–1.4), after adjusting for age. No association was observed in multivariate analyses. Toroptsova et al. [1995] found that LBP and lifting were related in univariate analyses (OR 1.4, p<0.05); no multivariate analyses were conducted. In the Walsh et al. [1989] examination of LBP and work-related factors, LBP was associated with lifting (in jobs just prior to injury) (RR 2.0, 95% CI 1.1–3.7), when age, sex, and all exposures were considered in multivariate analyses. When lifetime exposures were considered, lifting remained significantly associated for males (RR 1.5, 95% CI 1.0–2.4). In Burdorf and Zondervan’s 1990 study, an OR of 5.2 (95% CI 1.1–25.5) was observed for LBP and frequent lifting among crane operators. No relationship was seen for the referent group of non crane operators from the same plant (OR 0.70, 95% CI 0.14–3.5).

In a study that determined exposure status on the basis of job title, Videman et al. [1984] found slightly higher rates (not significant) of LBP in nursing aides than in qualified nurses. The authors stated that aides had higher workloads related to patient handling and lifting. Knibbe and Friele [1996] found that LBP rates were higher for registered nurses than for nursing aides, whom they stated had more lifting responsibilities (OR 1.2, p=0.04). After adjusting for hours worked, however, aides had the higher rate (RR 1.3, no statistical testing done). Undeutsch et al. [1982] examined back pain in baggage handlers, a group characterized by frequent bending, lifting, and carrying of loads. Although no exposures were estimated for this group, symptoms were significantly associated with length of employment after adjusting for age (p=0.035).

In the studies using more quantitative exposure assessments, strengths of association for the relationships between low-back disorder and lifting included estimates including a negative relationship [Magora 1972], no association [Burdorf et al. 1991], and several positive associations with ORs in the 2.2–10.0 range. One study found a positive relationship between sudden maximal efforts and LBP (OR 1.7) [Magora 1973]. Punnett et al. [1991] found a point estimate of 2.16 after adjusting for other covariates; Chaffin and Park [1973] found a strong relationship (OR 5) for LSR (but not lifting frequency); Marras et al. [1993, 1995] found that the highest risk of injury was related to lifting in combination with posture-related risk factors (OR 10.7). Liles et al. [1984] observed an OR of 4.5 for back injuries and the highest JSI. The investigation of school lunch preparers did not calculate risk estimates [Huang et al. 1988].

Studies that used subjective measures of exposure found point estimates including none [Johansson and Rubenowitz 1994; Kelsey 1975a,b; Videman et al. 1984] to a range including 1.3, 1.4, 2.0, 3.8, and 5.2 [Burdorf and Zondervan 1990; Holmstrom et al. 1992; Kelsey et al. 1984; Knibbe and Freile 1996; Toroptsova et al. 1995; Undeutsch et al. 1982; Walsh et al. 1989]. Although the Kelsey et al. [1984] exposure estimates were based on self-report, they showed important relationships between lifting and posture in multivariate analyses. While the OR for lifting alone was 3.8 (for the highest lifting level), the OR rose to 6.1 when postures related to twisting and bent knees were included in the model.

In summary, the articles reviewed provide evidence of a strong positive association between low-back disorder and lifting. Results from these and other studies emphasized the importance of awkward postures in the risk of low-back disorder.

Temporal Relationship

Two prospective studies assessed exposures prior to identification of back disorders. Both demonstrated positive associations between exposure and back disorder. Thirteen of the 18 studies were cross-sectional analyses. In two of these, investigators excluded cases of LBP with onset prior to the current job to increase the likelihood that exposure preceded disorder. A third cross-sectional study truncated self-reported exposures on the birthday preceding disorder onset. One case-control study truncated exposures prior to disorder onset. Of the four cross-sectional and case-control studies which attempted to address temporality, three found positive relationships between lifting and back disorder.

Consistency in Association

Although the 18 studies used varying designs, outcomes, and exposure assessment methods, they were fairly consistent in demonstrating a relationship between lifting and low-back disorder when objective measures of exposure were used to evaluate populations with high exposures. Results were less consistent when subjective exposure measures were utilized.

A NIOSH review of earlier publications related to patient lifting demonstrated results consistent with this review [Jensen 1990]. A comprehensive literature search evaluated all studies published between 1967 and 1987 that contained original research on nursing personnel and back problems. Of 90 studies, six were identified which distinguished between two or more groups of nurses with differing frequencies of patient handling and reported on back problems for each group. A weighted analysis of results from the six reports demonstrated an overall increase in back problems of 3.7 in those in the higher lifting frequency category.

Coherence of Evidence

Lifting and manual materials handling have been studied as risk factors for low back disorder for decades. Studies of workers’ compensation claims have shown that manual material handling tasks, including lifting, are associated with back pain in 25%-70% of injuries [Cust et al. 1972; Horal 1969; Snook and Ciriello 1991]. Data from the 1994 Bureau of Labor Statistics annual Survey of Occupational Injuries and Illnesses demonstrated that the industry with the highest rate of time-loss injuries due to overexertion was nursing and personal care facilities (where employees are required to engage in frequent patient handling and lifting).

During lifting, three types of stress are transmitted through the spinal tissues of the low back: compressive force, shear force, and torsional force [Waters et al. 1993]. It has been suggested that disc compression is believed to be responsible for vertebral end-plate fracture, disc herniation, and resulting nerve root irritation [Chaffin and Andersson 1984]. In early biomechanical assessments, models showed that large moments are created in the trunk area during manual lifting. Static evaluations of the trunk demonstrated that lifting results in large compressive forces on the spine.

More recently, biomechanical investigations have focused on spine loading and disc tolerances associated with asymmetric loading of the trunk. In laboratory experiments, dynamic trunk motion components of lifting have been associated with greater spine loading. Increased trunk motion during lifting activities has been associated with increased trunk muscle activity and intra-abdominal measures, among other changes [Marras et al. 1995]. Some laboratory studies have shown that lateral shear forces make trunk motions more vulnerable to injury than in a compressive loading situation. There is also in vitro evidence that the viscoelastic properties of the spine may cause increased strain during increased speed of motion [Marras et al. 1995].

Current models for lifting-related musculoskeletal injury stress that biomechanical considerations comprise only part of the assessment of risk [Waters et al. 1993]. Other criteria include physiologic measures of metabolic stress and muscle fatigue and psychophysical considerations (the worker’s perception of his/her lifting capacity, a combination of perceived biomechanical and physiologic attributes of the job). All three criteria are important in assessing risk across the full spectrum of job and individual worker variability.

Exposure-Response Relationships

Eight studies examined exposure-response relationships in some form. Of these, four found dose-response relationships between low-back disorder and objective measures of lifting [Chaffin and Park 1973; Liles et al. 1984; Marras et al. 1995; Punnett et al. 1991]; another found a dose-response between disorder and sudden maximal efforts [Magora 1973]. A study of baggage handlers found an association between back disorder and length of employment [Undeutsch et al. 1982]. Two studies found no dose-response relationship (using a posture analysis assessment and a manual materials handling index) [Burdorf et al. 1991; Johansson and Rubenowitz 1994].

The majority of studies which examined exposure-response relationships, and in particular those that utilized quantitative exposure measures, demonstrated these trends.

Conclusions: Lifting and Forceful Movements

There is strong evidence that low-back disorders are associated with work-related lifting and forceful movements. The five studies reviewed for this chapter which showed no association between lifting and back disorder used subjective measures of exposure, poorly described exposure assessment methodology, or showed little differentiation of exposure within the study group. The remaining 13 studies were consistent in demonstrating positive relationships, where those using subjective measures of exposure showed a range of risk estimates from 1.2 to 5.2, and those using more objective assessments had ORs ranging from 2.2 to 11. Studies using objective measures to examine specific lifting activities generally demonstrated risk estimates above three and found dose-response relationships between exposures and outcomes. For the most part, higher ORs were observed in high-exposure populations (e.g., one high-risk group averaged 226 lifts per hour with a mean load weight of 88 N. Evidence from other studies and reviews has also suggested that groups with high- frequency exposure to lifting of heavy loads, such as nursing staff, are at high risk of back disorder.

Most of the investigations reviewed for this document adjusted for potential covariates in analyses: two-thirds of the studies showing positive associations examined effects of age and gender. Nevertheless, some of the relatively high ORs that were observed were unlikely to be caused by confounding or other effects of lifestyle covariates. Several studies suggested that both lifting and awkward postures were important contributors to the risk of low-back disorder. The observed relationships are consistent with biomechanical and other laboratory evidence regarding the effects of lifting and dynamic motion on back tissues.

Back to top

Bending and Twisting (Awkward Postures)

Definition

Bending is defined as flexion of the trunk, usually in the forward or lateral direction. Twisting refers to trunk rotation or torsion. Awkward postures include non-neutral trunk postures (related to bending and twisting) in extreme positions or at extreme angles. Several studies focus on substantial changes from non-neutral postures. Risk is likely related to speed or changes and degree or deviation from non-neutral position. For the purposes of this review, awkward postures also included kneeling, squatting, and stooping. In most of the studies included in this review, awkward postures were measured concurrently with other work-related risk factors for back disorder.

Studies Reporting on the Association Between LBP and Awkward Postures

Twelve studies examined the relationship between low back disorder and bending, twisting, and awkward postures (Table 6-3, Figure 6-3). Most (nine) also examined the effects of occupational lifting. See the previous discussion of lifting and forceful movements. Nine studies were cross-sectional in design, two case-control, and one prospective.

Participation rates were adequate for 83% of the investigations (Table 6-3). Four studies assessed postures using objective measures (however, in the study by Magora [1972], details on their observation methods were not reported; the rest estimated exposures from interview or questionnaire responses). Health outcomes included low-back and sciatic pain symptoms, lumbar-disc prolapse, and back injury reports. In four investigations, outcomes were defined using both symptoms and medical examination criteria. Only one investigation, the Punnett et al. [1991] case-control study of back pain in auto workers, fulfilled the four evaluation criteria (Table 6-3, Figure 6-3).

Several other studies, while not meeting all of the four criteria, are particularly notable because they used objective measures of exposure assessment [Burdorf et al. 1991; Marras et al. 1993, 1995] or met more than one of the criteria [Holmstrom et al. 1992; Kelsey et al. 1984]. As discussed earlier, the physical examination criterion may be less important in low-back disorders because of the paucity of specific physical findings in most cases of low-back disorders.

Descriptions of five studies which offered the most information regarding the effects of bending, twisting, and awkward postures follow. Please note that there is some overlap with studies that examined lifting effects. Detailed descriptions of the 12 studies appear in Table 6-6.

The Punnett et al. [1991] case-control study examined the relationship between back pain and occupational exposures in auto assembly workers. Back pain cases (n=95) were determined by symptoms at interview and medical examination; controls included those free of back pain. For all participants or proxies in the same jobs, jobs were videotaped and work cycles were reviewed using a posture analysis system. Exposures included time spent in various awkward postures. Peak biomechanical forces were estimated for up to nine postures where a load weighing at least 10 lb was held in the hands. In multivariate analyses that adjusted for a number of covariates (age, gender, length of employment, recreational activity and medical history), time in non-neutral postures mild or severe flexion and bending were strongly associated with back disorder (OR 8.0, 95% CI 1.4–44). In the same model, lifting was also associated (OR 2.16, 95% CI 1.0–4.7). When the subset with physical medical findings was examined, associations were more pronounced. Although few study subjects were unexposed to all of the postures studied, a strong increase in risk was observed with both intensity and duration of exposure. It was not possible to determine the relative contributions of different awkward postures because all were highly correlated. Only participants’ current jobs (for referents) or jobs when symptoms started (for cases) were analyzed; the study design thus assumed a short-term relationship between exposure and outcome. Although length of time in job was also included in the models, the authors attempted to ensure that exposure preceded disease by identifying time of onset and measuring exposures in the job held just prior. The strong associations, after adjustment for covariates, are notable.

Burdorf et al. [1991] examined back pain symptoms in a cross-sectional study of male concrete fabrication workers and a referent group of maintenance workers. Back pain symptoms were assessed by questionnaire. Exposures were measured using the Ovako Working Posture Analysis System, which assessed postures for the back and lower limbs, along with lifting load. Information on exposures in previous jobs was also collected. Concrete workers experienced significantly more back symptoms than referents (OR 2.8, 95% CI 1.3–6.0).

Univariate results showed associations between back pain and both posture index and WBV in current job. Correlations were presented showing lifting was not found to be associated with back pain or to vary significantly across the six job categories examined in the study. In multivariate analyses adjusting for age, both posture index and WBV were significantly associated with back pain, with ORs of 1.23 (p=0.04) (for an ordinal scale of 6) and 3.1 (p=0.001) (dichotomous), respectively. Those in the highest posture index category were steel benders, who spent an average of 47% of their time in bent back postures (compared to 12% for the lowest exposed group). The posture index and WBV measures were highly correlated and analyzed separately. Strengths of the study included use of a standardized symptom questionnaire, high participation rates and objective measure of exposure, and an attempt to clarify the temporal relation between exposure and outcome by excluding cases of back pain with onset before the present job.

Using an unusual cross-sectional study design, Marras et al. [1993, 1995] examined the relationship between low-back disorders and spinal loading during occupational lifting. A total of 403 jobs from 48 diverse manufacturing companies were assessed for risk of low-back disorder using plant medical department injury reports. Jobs were ranked into three categories according to risk then assessed for position, velocity, and acceleration of the lumbar spine during lifting motions in manual materials handling using electrogoniometric techniques. A combination of five factors distinguished between high- and low-risk jobs: lifting frequency, load moment, trunk lateral velocity, trunk twisting velocity, and trunk sagittal angle. The highest combination of exposure measures produced an OR of 10.7 (95% CI 4.9–23.6) (in comparison to the lowest combined measures). The study design was unusual in that the unit of analysis appeared to be job rather than individual. Neither participation rate nor total number of participants was stated. No information appeared regarding the proportions of individuals within jobs who were recruited for measurement of lifting motions. However, the unit of analysis was job, and each was characterized by measurement of at least one study subject. Effects of other covariates were not addressed (multivariate models appeared to include only biomechanical variables). The study results emphasize the multifactorial etiology of back disorders, including contributions of lifting frequency, loads, and trunk motions and postures. The study design did not allow for examination of temporal relationships.

A case-control study of prolapsed lumbar disc was carried out using a hospital population-based design [Kelsey et al. 1984]. Cases (n=232) included individuals diagnosed with prolapsed lumbar disc; an equal number of controls matched on sex, age, and medical service were selected. Exposure was assessed using a detailed occupational history (not described, but presumably obtained by interview). An association with work-related lifting, without twisting the body, was observed at the highest lifting level (OR 3.8, 95% CI 0.7–20.1). Twisting without lifting was associated with disc prolapse (OR 3.0, 95% CI 0.9–10.2); a combination of both risk factors had an OR of 3.1 (95% CI 1.3–7.5). The highest risk was observed for simultaneous lifting and twisting with straight knees (OR 6.1, 95% CI 1.3–27.9). Despite the fact that exposures were self-reported, these associations were notably strong. The potential existed for differential recall bias for cases and controls, because study subjects were interviewed about work-related factors after case status was established. Interviewers may not have been blinded to case/control status.

Holmström et al. [1992] examined the relationship between LBP and work task activities in across-sectional study of male construction workers. One-year prevalence of LBP was ascertained by questionnaire. A sample of workers was clinically examined. Exposure relative to lifting, handling, and work postures was obtained by self-report. After adjustment for age, the index for manual material handling, which included lifting, was associated with LBP with a RR of 1.27 (95% CI 1.2–1.4). Stooping and kneeling postures showed a dose-response relationship with LBP, particularly severe LBP (with ORs 1.3, 1.8, and 2.6 in comparison to those with no stooping; ORs 2.4, 2.6, and 3.5 in comparisons to those with no kneeling, respectively). No association was observed with sitting. In multiple regression analyses, LBP was associated with stooping (p<0.001) and kneeling (p<0.01). While the authors attempted to adjust for some covariates (age, gender, and psychosocial factors) in analyses, they did not appear to examine simultaneous effects of physical work-related factors in a single model. The cross-sectional design could not ascertain the temporal relationships between exposure and disorder.

Strength of Association

The more informative studies included the Punnett et al’s [1991] case-control investigation, which fulfilled the four evaluation criteria, plus several others that used independent exposure assessments. In the Punnett et al. study, multivariate analyses that adjusted for covariates demonstrated that time in non-neutral postures was strongly associated with back disorders (OR 8.09, 95% CI 1.4–44). In the same model, the OR for lifting was 2.2. Burdorf et al. [1991] found associations between posture index and back symptoms in both univariate and multivariate analyses: in multivariate analyses adjusting for age, the OR for posture index was 1.23 (p=0.04), for an ordinal scale of six levels. Posture index was highly correlated with WBV. However, the Kelsey et al’s [1984] case-control study of prolapsed lumbar discs found that twisting without lifting had an OR of 3.0 (95% CI 0.9–10.2); in combination, the two had an OR of 3.1 (95% CI 1.3–7.5). The highest risk was observed for a combination of lifting, twisting, and straight knees (OR 6.1, 95% CI 1.3–27.9). In the Marras et al. [1993, 1995] cross-sectional study, back injuries were associated with spinal loading during lifting, which included simultaneous exposures to lifting frequency, load weight, trunk lateral velocity, trunk twisting velocity, and trunk sagittal angle. An OR of 10.7 (95% CI 4.9–23.6) was observed for the highest combination of exposure measures. Univariate ORs were 1.73 (95% CI 1.38–2.15) for trunk lateral velocity, 1.66 (95% CI 1.34–2.05) for trunk twisting velocity, and 1.60 (95% CI 1.31–193) for maximum sagittal flexion when comparing the high-and low-risk groups [Marras et al. 1993].

The other studies showed a range of point estimates. In univariate analyses, Magora [1972, 1973] found that for bending, the highest rate of LBP was observed for the rarely/never category. For twisting and reaching, the highest LBP rate was in the sometimes category. Johansson and Rubenowitz [1994] found no associations between low-back symptoms and bent or twisted work postures in blue- and white-collar workers. After adjustment for age and gender, however, extreme work postures were significantly associated with the outcome in blue-collar workers. Relationships were presented as partial correlations, thus preventing calculation of risk estimates. Riihimäki et al. [1994] observed that occupational exposure to twisted and bent postures were associated with incidence of sciatic pain in univariate but not multivariate analyses. No risk estimates were provided. In Svensson and Andersson’s 1989 study of LBP in Swedish women, bending forward was associated with lifetime incidence in univariate (RR 1.3, p<0.05) but not multivariate analyses. The Masset and Malchaire [1994] univariate analyses demonstrated that trunk torsions were associated with LBP in steel workers (OR 1.55, p<0.05); no associations were shown in multivariate analyses. Toroptsova et al. [1995] demonstrated that LBP in the past year was associated with bending (OR 1.7, p<0.01) in univariate analyses (multivariate analyses were not conducted). Riihimäki et al. [1989a] observed a dose-response for sciatic pain and self-reported twisted or bent postures; the OR for the highest exposure category was 1.5 [95% CI 1.2–1.9]. Holmstrom et al. [1992] observed that stooping and kneeling postures were associated with LBP, particularly severe disorder, with ORs of 2.6 and 3.5 (p<0.05), respectively.

In summary, three of the four studies using more quantitative exposure assessments showed elevated risk estimates for the relationship between low-back disorder and bending, twisting, or awkward postures, with ORs ranging from 1.23 (for a scaled variable) to 8.09; the highest risk estimate, an OR of 10.7, was based on combined exposure to lifting and posture risk factors. Most of these were based on multivariate analyses that adjusted for covariates (usually age and gender). The remaining studies demonstrate risk estimates ranging from no association (in one study), 1.3–1.7 in univariate but not multivariate analyses, to a high of 3.5 in another study. Studies utilized a number of definitions for awkward postures, as noted.

Temporal Relationship

One prospective study assessed exposures prior to identification of back disorders. Results demonstrated positive associations in univariate but not multivariate analyses. [Riihimäki et al. 1994]. Nine of 12 studies were cross-sectional in design. In one of these, investigators excluded cases of LBP with onset prior to the current job to increase the likelihood that exposure preceded disorder. [Burdorf et al. 1991]. No association between exposure and back disorder was observed. One case-control study examined only exposures experienced in the job just prior to disorder onset [Punnett et al. 1991]. A strong association between exposure to awkward postures and back pain was observed.

Consistency in Association

Although the 12 studies used varying designs, outcomes, and exposure assessment methods, the studies using quantitative exposure measures were fairly consistent in demonstrating a moderate relationship between awkward postures and low-back disorder.

Coherence of Evidence

Nine of the 12 studies which examined posture effects also studied effects of lifting. Therefore, a discussion of coherence of evidence for the former relationship is similar to that found in the section on lifting and forceful movements. Forward flexion can generate compressive forces on the structures of the low back similar to lifting a heavy object. Similarly, rapid twisting can generate shear or rotational forces on the low back [Marras et al. 1995].

Exposure-Response Relationships

Six studies examined dose-response relationships between posture and low-back disorder. In one, no dose-response relationship was found between LBP and estimates for bending and twisting/reaching. In the other five studies, relationships were demonstrated between back injury and spinal loading score, LBP and posture index, sciatic pain and awkward postures, LBP and stooping, and low-back symptoms and kneeling.

Conclusions: Awkward Postures

The investigations that were reviewed provided evidence that low-back disorders are associated with work-related awkward postures. Results were consistent in showing increased risk of back disorder with exposure, despite the fact that studies defined disorders and assessed exposures in many ways. Several studies found risk estimates above three and dose-response relationships between exposures and outcomes. Many of the studies adjusted for potential covariates in their analyses, and a few examined the simultaneous effects of other work-related risk factors in analyses. Several studies suggested that both lifting and awkward postures were important contributors to risk of low back disorder.

Back to top

Whole Body Vibration (WBV)

Definition

WBV refers to mechanical energy oscillations which are transferred to the body as a whole (in contrast to specific body regions), usually through a supporting system such as a seat or platform. Typical exposures include driving automobiles and trucks, and operating industrial vehicles.

Studies Reporting on the Association Between LBP and Whole Body Vibration

Nineteen investigations addressed WBV as a risk factor for back disorder. Fifteen study designs were cross-sectional, two were cohort, one was case-control, and one had both cross-sectional and cohort components.

None of the 19 studies fulfilled all of the four evaluation criteria (Table 6-4, Figure 6-4). Participation rates were over 70% for 13 investigations. Seven used independent measures of exposure for estimation of WBV; in 10 studies, exposure information was obtained by questionnaire or interview. In two studies, exposure to WBV was based on job title alone. Health outcomes included symptom report of LBP, sciatica, or lumbago, sick leaves or disability retirements related to back disorders, and medically confirmed herniated lumbar disc.

Five of the nine studies which met two or more of the evaluation criteria used similar methodologies and offered the most information regarding the association between WBV and back disorder. Detailed descriptions for all 19 investigations can be found in Table 6-6.

Bovenzi and Betta [1994] examined the relationship between WBV and back disorder in a cross-sectional study of male tractor drivers. The unexposed group included male revenue inspectors and administration workers with no vibration exposure. Outcomes included various types of back symptoms reported by questionnaire. Vibration measures were obtained from a representative sample of tractors and linked to individual information on number of hours driven yearly (obtained by questionnaire). Self-reported exposures to postural loads were also obtained. In comparison to referents, tractor drivers demonstrated an OR of 3.22 (95% CI 2.1–5.2) for lifetime LBP. For LBP in the past year, the OR was 2.39 (95% CI 1.6–3.7). For LBP in the past year, ORs ranged from 2.31 to 3.04 by exposure levels for total vibration dose, equivalent vibration magnitude, and duration of exposure, after adjustment for covariates. In multivariate analyses, chronic LBP showed a dose-response relationship with total vibration dose (OR 2.00, 95% CI 1.2–3.4, for the highest category), equivalent vibration magnitude (OR 1.78, 95% CI 1.04–3.0, for the highest category), and duration of exposure (OR 2.13, 95% CI 1.2–3.8, for the highest category). Exposure-response relationships were observed for postural load categories, with ORs of 4.56 (95% CI 2.6–8.0) for LBP in the past year and 2.30 (95% CI 1.2–4.5) for chronic LBP (for the highest exposure categories). Multivariate analyses adjusted for age, body mass index, education, sports activity, car driving, marital status, mental stress, climatic conditions, back trauma and postural load (or vibration dose, depending upon the exposure examined).

Bovenzi and Zadini [1992] used a similar cross-sectional study design to examine low back symptoms in male bus drivers. Referents included maintenance employees who worked for the same company. Back pain symptoms were assessed by questionnaire. WBV was measured for a sample of buses used over the relevant time period. Cumulative vibration exposures were calculated using this information, along with questionnaire items related to work duration, hours, and previous exposures. In comparison to referents, bus drivers demonstrated an OR of 2.80 (95% CI 1.6–5.0) for lifetime LBP; the OR for LBP in the past year was 2.57 (95% CI 1.5–4.4). In multivariate analyses, the ORs for LBP in the previous year were 1.67, 3.46, and 2.63 for three total vibration dose categories. Similar trends were observed for other measures of vibration (equivalent vibration magnitude and total duration of exposure), and after exclusion of those with exposure in previous jobs. Statistically significantly increasing trends were observed for nearly all types of back symptoms by exposure level (to all three measures of vibration) after adjustment for covariates. Multivariate analyses adjusted for age, awkward postures, duration of exposure, body mass index, mental workload, education, smoking, sports activities, and previous exposures.

Three studies of WBV effects were conducted by the same group of Dutch investigators. The first examined back pain and WBV exposures cross sectionally in male helicopter pilots [Bongers et al. 1990]. A referent group of non flying Air Force officers (with characteristics similar to pilots) was also included. Information on back symptoms was obtained by questionnaire. Vibration measures were assessed in two helicopters of each type used by the study group. Individual exposures were calculated by matching this with questionnaire items related to hours of flying time and types of helicopters flown. Information on exposure to bent/twisted postures was also obtained by questionnaire. In comparison to controls, ORs for pilots were elevated for a number of back symptoms: 9.0 (95% CI 4.9–16.4) for LBP and 3.3 (95% CI 1.3–8.5) for sciatica. All of the above were adjusted for age, height, weight, climate, bent and twisted postures, and feeling tense at work. In multivariate analyses, ORs for LBP were 13.8, 7.5, 6.0, and 13.4 for four categories for total flight time (in comparison to controls). ORs for LBP by total vibration dose were 12.0, 5.6, 6.6, and 39.5. By hours of flight time per day, ORs were 5.6, 10.3, and 14.4 for LBP. Although there was some concern that pilots with back pain may have dropped out of employment, risk estimates were high (particularly in analyses by exposure level). Transient back pain appeared to increase with daily exposure time, while chronic back pain appeared more associated with total flight time and total vibration dose.

In a second study by the same group, WBV exposures were examined in male tractor drivers and a referent group of inspectors and maintenance technicians [Boshuizen et al. 1990a,b]. Two investigations were conducted using the same population: a 1986 cross-sectional study of a cohort identified in 1975, and a cohort analysis of sick leaves and disability retirements due to back disorder through the same time period. For the cross-sectional analyses, information on back symptoms was obtained by questionnaire. Vibration was measured for a sample of vehicles and linked with questionnaire information related to types of vehicles driven, hours, and previous employment. Information regarding exposure to awkward postures was also collected. Results from the cohort analysis showed an incidence density ratio of 1.47 (95% CI 1.04–2.1) for a comparison of sick leaves due to back disorders in exposed and referent groups. An increase in sick leaves for disc disorders by vibration dose was observed, with an OR of 7.2 (95% CI 0.92–179) for the highest category. Cross-sectional study results demonstrated increases in LBP symptom prevalence by vibration dose category. Multivariate ORs increased by vibration dose (an OR of 2.8, 95% CI 1.6–5.0, for the highest category) and years of exposure (an OR of 3.6, 95% CI 1.2–11, for the highest category) after adjustment for duration of exposure, age, height, smoking, awkward postures, and mental workload.

Boshuizen et al. [1992] also conducted a cross-sectional study of back pain in fork-lift truck and freight container tractor drivers exposed to WBV. Referents included other employees working for the same shipping company, but with no vibration exposure. Back pain symptoms were assessed by questionnaire. Exposures were estimated by measurement of vibration in a sample of vehicles, combined with questionnaire responses. Cumulative exposures were calculated, truncating at time of symptom onset. Prevalence of back pain was higher in the exposed group than in referents: the RR for back pain was 1.4 (p<0.05); RRs for LBP and lumbago were 1.4 (p<0.05) and 2.4 (p<0.05), respectively, after adjusting for age. Differences in LBP were observed only in younger age groups after multivariate adjustment for mental stress, years of lifting, awkward postures, height, smoking, and hours of sitting. There was no association between total vibration dose and back pain (OR 0.99, 95% CI 0.85–1.2) or lumbago (OR 1.14, 95% CI 0.91–1.4). Only vibration in the 5 years immediately preceding symptom onset was significantly associated with back pain (OR 2.4, 95% CI 1.3–4.2) and lumbago (OR 3.1, 95% CI 1.2–7.9). It appeared that a healthy worker selection effect was operating, as differences in back pain were observed only for those in younger age groups.

Evaluation of the Causal Relationship Between Back Disorder and Whole Body Vibration

Strength of Association

Recent studies that included quantitative exposure assessments provided the most information regarding the relationship between WBV and back disorder [Bongers et al. 1988; Boshuizen et al.1990a, b; Bovenzi and Betta 1994; Bovenzi and Zadini 1992]. (Two other recent studies also described quantitative exposure assessments, but no results relating to these were presented [Burdorf et al. 1993; Magnusson et al. 1996]). In all five, ORs were calculated by levels of vibration exposure, expressed in several ways (usually including magnitude and duration of exposure). In the five studies, overall ORs comparing back pain in exposed and referent groups ranged from 1.4 [Boshuizen et al. 1992] to 9.5 [Bongers et al. 1990]. Analyses conducted by exposure level demonstrated stronger relationships. In Bovenzi and Betta’s 1994 study of tractor drivers, ORs for lifetime LBP were 3.79 for total vibration dose, 3.42 for equivalent vibration magnitude, and 4.51 for duration of exposure (for the highest exposure levels). For LBP in the previous year, ORs were 2.36, 2.29, and 2.74 for the highest levels of the same three exposure measures. In Bovenzi and Zadini’s 1992 study of urban bus drivers, the highest ORs for LBP were observed for intermediate rather than the highest exposure categories: 3.46 for total vibration dose, 3.77 for equivalent vibration magnitude, and 3.08 for total duration of WBV exposure. The Bongers et al. [1990] investigation of back pain in helicopter pilots demonstrated that the highest ORs for LBP were found in the highest categories for total flight time (OR 13.4, 95% CI 5.7–32), total vibration dose (OR 39.5, 95% CI 10.8–156) and hours of flight time per day (OR 14.4, 95% CI 5.4–38.4). A study of tractor drivers demonstrated LBP ORs of 2.8 (95% CI 1.6–5.0) for the highest total vibration dose and 3.6 (95% CI 1.2–11) for the highest exposure duration category [Boshuizen et al. 1990a]. In the same population, the OR for all sick leaves due to back disorder was 1.47, comparing exposed (95% CI 1.04–2.1) and referent groups [Boshuizen et al. 1990b]. For sick leaves related to intervertebral disc disorders, the highest OR was observed for the highest exposure category (OR 7.2, 95% CI 0.92–179). The Boshuizen et al. [1992] study of forklift truck and freight container tractor drivers showed no association between back pain and total vibration dose (OR 0.99, 95% CI 0.85–1.2) but did show an association for vibration in the preceding five years (OR 2.4, 95% CI 1.3–4.2). In this study the increase in LBP prevalence in the exposed group was only significant for those in younger age groups (an OR of 5.6 for those age 25-34) in multivariate analyses. In all five of these cross-sectional studies, ORs were calculated by vibration exposure category after adjusting for a number of covariates, as mentioned in the detailed study descriptions, above.

Other studies assessed both exposure and low-back disorder by interview or questionnaire. Burdorf and Zondervan [1990] observed no association between WBV exposure and LBP in crane operators in univariate analyses (OR 0.66, 95% CI 0.14–3.1); no associations were observed in multivariate analyses. Toroptsova et al. [1995] also found no association between LBP and vibration in their study (no definition for vibration was provided, but WBV was suggested). In the Riihimaki et al. 1994 prospective study, sciatic pain was associated with vibration in univariate but not multivariate models (no risk estimates were provided). While the definition for “vibration” was not clear, the authors suggested it could be interpreted as low-level WBV. The Masset and Malchaire [1994] cross-sectional study found that LBP was associated with vehicle driving (OR 1.2, p<0.001) in univariate analyses. Similar results were observed in multivariate analyses (OR 1.2, p<.005). Riihimaki et al. [1989a] observed an OR of 1.3 (95% CI 1.1–1.7) for longshoremen and earthmovers in comparison to a referent group with no vibration exposure. In the same study, no association was seen for annual car driving (OR 1.1, 95% CI 0.9–1.4). Walsh et al. [1989] found that driving (on job held prior to symptoms) was significantly associated with low-back symptoms in males (RR 1.7, 95% CI 1.0–2.9) after adjusting for age and other job exposures in multivariate analyses. Burdorf et al. [1991] found that WBV was significantly associated with back pain (OR 3.1, p=0.001) in multivariate analyses that adjusted for age. The Kelsey [1975a] case-control study found a significant association between herniated lumbar disc and time driving (OR 2.75, p=0.02), and more specifically, working as a truck driver (OR 4.7, p<0.02). Burdorf et al. [1993] investigation demonstrated an OR of 3.29 (95% CI 1.5–7.1) for crane operators and 2.51 (95% CI 1.5–5.4) for vibration-exposed straddle-carrier drivers after adjusting for a number of covariates. In a study of Danish salespeople, annual driving distance was associated with low-back symptoms [Skov et al. 1996]. A dose-response relationship was observed in multivariate analyses, with an OR of 2.79 (95% CI 1.5–5.1) for the highest category.

Four studies assessed exposures primarily by job title. Magnusson et al. [1996] observed an OR of 1.79 (95% CI 1.2–2.8) for bus and truck drivers in comparison to an unexposed referent group. In a study of crane operators, the exposed group demonstrated ORs of 2.00 (95% CI 1.1–3.7) for all intervertebral disc disorders and 2.95 (95% CI 1.2–7.3) for disc degeneration after adjustment for age and shift [Bongers et al. 1988]. An examination of risk estimates of disc degeneration by years of exposure showed the highest OR (5.73) in the highest exposure category. In the Johanning [1991] study of subway train operators, an OR of 3.9 (95% CI 1.7–8.6) was observed for sciatica. While not a primary focus of the Magora [1972, 1973] studies of LBP in eight selected occupations, it was observed that bus drivers had back pain rates similar to those of the comparison group of bankers (RR 1.19, 95% CI 0.8–1.7).

Thus, four out of five studies using quantitative exposure assessments demonstrated positive associations between back disorder outcomes and vibration exposures, with ORs ranging from 1.4 to 39.5. The fifth cross-sectional study found no overall association between exposure and back disorder but found associations in selected subgroups (which suggested that the study population was biased, as noted above). In all of these studies, risk estimates by exposure category were calculated after adjustment for many covariates.

In the remaining studies, risk estimates varied, including no association (n=3), ORs of 1.2, 1.7, and 2.8 for driving, an OR of 1.8 for truck or bus driving, an OR of 4.7 for truck driving, an OR of 1.3 for machine operation, ORs of 2.0, 2.95 and 5.73 for crane operation, an OR of 3.1 for WBV, and an OR of 3.9 for subway train operation.

In summary, the evidence from these investigations suggests a positive association between WBV and back disorder. Relationships were particularly strong for high-exposure groups where exposures were assessed using observational or measurement approaches.

Temporal Relationship

Three studies had prospective designs in which temporal relationships between outcome and exposure could be determined [Bongers et al. 1988; Boshuizen et al. 1990b; Riihimaki et al. 1994]. In two of these, clear positive relationships between back disorder and exposure were demonstrated [Bongers et al. 1988; Boshuizen et al. 1990b]. Twelve studies had a cross-sectional design that could not directly address temporality. However, three attempted to clarify relationships by excluding from analysis the cases with disorder onset prior to current job [Burdorf et al. 1991, 1993; Burdorf and Zondervan 1990]. A fourth cross-sectional study truncated self-reported exposures on the birthday preceding disorder onset [Walsh et al. 1989]. In these four investigations, positive relationships between back disorder and WBV were also observed.

Consistency in Association

Results with regard to the relationship between low back disorder and WBV were most consistent in the studies using observational or measurement approaches to exposure assessment. The strength of association was more variable in studies using job titles or questionnaires to assess exposures. The variability in the associations does not appear to be related to confounding exposures, since most studies adjusted for age, gender and at least several other confounders. Studies using more quantitative exposure measures were fairly consistent in showing the higher risk estimates.

In addition to the epidemiologic investigations that were reviewed for this document, many more were conducted in the 1960s though the 1980s. Others have summarized this evidence in earlier reviews. Hulshof and Veldhuijzen van Zanten [1987] concluded that, although studies varied in methodologies and quality, most showed a strong tendency toward a positive association between WBV exposure and LBP. Seidel and Heide [1986] stated that the literature they reviewed indicated an increased risk of spine disorders after intense long-term exposure to WBV. Bongers and Boshuizen [1990] conducted a meta-analysis of studies published through 1990 that examined the relationship between WBV and several back disorders. The overall OR for WBV exposure and degenerative changes of the spine was 1.5; the summary OR for LBP was also 1.5. These conclusions are consistent with the positive associations observed in the evidence reviewed above (although the studies published in the 1990s have tended to report larger ORs).

Other evidence for the relationship is provided by surveillance data. The U.S. population-based National Health Interview Survey, carried out in 1988, found that males employed as truck drivers and tractor equipment operators had a RR of 2.0 for back pain in comparison to all male workers [Guo et al. 1995].

Coherence of Evidence

Laboratory studies have shown that exposure to WBV causes spine changes that may be related to back pain. These include fatigue of the paraspinal muscles and ligaments, lumbar disc flattening, disc fiber strain, intradiscal pressure increases, disc herniation, and microfractures in vertebral end-plates [Wilder and Pope 1996]. Studies of acute effects have shown that the vertebral end-plate is the structure that is most sensitive to high WBV exposure, followed by the intervertebral disc [Wikstr”m et al. 1994]. Experimental investigations have demonstrated that high exposures to vibration cause injuries such as degeneration and fracturing of the vertebral end-plate. With regard to intervertebral discs, several studies have suggested that vibration causes creep, an increase in intradiscal pressure resulting from compressive loading. Pressure peaks may cause ruptures in the superficial structure of the disc and changes in the nutritional balance that lead to degeneration. Thus, prolonged vibration exposure may cause spine pathology through mechanical damage and/or changes in tissue metabolism.

In addition to pathology of the vertebrae and intervertebral discs, vibration exposure has been shown to cause changes in electromyographic (EMG) activity in muscles of the lower back [Wikstr”m et al. 1994]. For example, EMG experiments have demonstrated that lower back muscle exhaustion increases during WBV exposure in truck driving. Decreased stability of the lower back may result from slower muscle response, perhaps increasing the risk of injuring other structures.

Laboratory investigations have shown that other work-related factors, including prolonged sitting,lifting, and awkward postures, may act in combination with WBV to cause back disorder [Dupuis 1994; Wikstr”m et al. 1994; Wilder and Pope 1996].

Exposure-Response Relationships

Five of six studies which carried out quantitative exposure assessment demonstrated exposure-response relationships between WBV and back disorder. Bovenzi and Betta [1994] observed a dose-response between chronic LBP and total vibration dose, equivalent vibration magnitude, and duration of exposure. Bovenzi and Zadini [1992] found statistically significantly increasing trends for nearly all types of back symptoms by exposure level, after adjustment for covariates. Bongers et al. [1990] demonstrated increased ORs for sciatic pain and transient back pain with increasing hours of daily flight time. In their cohort of tractor drivers, Boshuizen et al. [1990b] observed an increase in risk of sick leaves for disc disorder by total vibration dose level.

In other studies, Bongers et al. [1988] found an increase in risk of disc degeneration by years of exposure to crane operation; Skov et al. [1996] found an increase in low-back symptoms with annual driving distance. Johanning [1991] found no association between years of employment as a subway train operator and back pain symptoms.

The majority of studies which examined back disorders by exposure level demonstrated dose-response relationships.

Conclusions: Whole Body Vibration

There is strong evidence of a positive association between exposure to WBV and back disorder. Of the 19 studies reviewed for this chapter, four demonstrated no association between WBV and back pain. Possible explanations for these results included use of subjective exposure assessments that perhaps resulted in misclassification of exposure status and, in one cross-sectional study, operation of a healthy worker selection effect (where those with higher exposures dropped out of the study group). The remaining 15 studies were consistent in demonstrating positive associations, with risk estimates ranging from 1.2 to 5.7 for those using subjective exposure measures, and from 1.4 to 39.5 for those using objective assessment methods. Most of the studies that examined relationships in high-exposure groups using detailed quantitative exposure measures found strong positive associations and exposure-response relationships between WBV and back pain. These relationships were observed after adjusting for age and gender, along with several other covariates (which, depending on the study, may have included smoking status, anthropometric measures, recreational activity, and physical and psychosocial work-related factors). This evidence is supported by results observed in many earlier epidemiologic investigations that have been summarized in other reviews.

Laboratory studies have demonstrated WBV effects on the vertebrae, intervertebral discs, and supporting musculature. Both experimental and epidemiologic evidence suggests that WBV may act in combination with other work-related factors such as prolonged sitting, lifting, and awkward postures to cause increased risk of back disorder.

It is possible that effects of WBV may depend on the source of exposure. For example, in the studies reviewed for this document, ORs were particularly high for helicopter pilots. It was not possible to determine differences for other types of vehicles (automobiles, trucks, and agricultural, construction, and industrial vehicles).

Back to top

Static Work Postures

Definition

Static work postures include isometric positions where very little movement occurs, along with cramped or inactive postures that cause static loading on the muscles. In the studies reviewed, these included prolonged standing or sitting and sedentary work. In many cases, the exposure was defined subjectively and/or in combination with other work-related risk factors.

Studies Reporting on the Association Between LBP and Static Work Postures

Ten studies examined relationships between low back disorder and static work postures, which may have included prolonged sitting, standing, or sedentary work. For none was static work posture the primary occupational exposure of interest. Instead, it was often one of many variables examined in larger studies of several or many work-related risk factors. Nine of the studies were cross-sectional in design; one was a case-control study.

None of the investigations fulfilled the four research evaluation criteria (Table 6-5, Figure 6-5). Participation rates were acceptable for 60%. For four, case definitions included both symptoms and medical examination criteria. Health outcomes included symptom report of back pain, sciatica, or lumbago, back pain as ascertained by symptoms and medical exam, herniated lumbar disc, and lumbar disc pathology. One study claimed to assess job-related exposures by observation; the nine others obtained information on static work postures by self-report on interview or questionnaire.

Below are descriptions of four of the more informative studies. Detailed descriptions for all 10 investigations are found in Table 6-6).

Burdorf and Zondervan [1990] carried out a cross-sectional study comparing 33 male crane operators with non crane operators from the same Dutch steel plant, matched on age. Symptoms of LBP and sciatica were assessed by questionnaire. Activities in current and past jobs were assessed by questionnaire; exposures were rated according to level of heavy work, frequency of lifting, WBV, and prolonged sedentary posture. Crane operators were significantly more likely to experience LBP (OR 3.6, 95% CI 1.2–10.6). Among crane operators alone, the OR for heavy work was 4.0 (95% CI 0.76–21.2) after controlling for age, height, and weight. It was determined that this heavy work occurred in the past and not in current jobs. Among crane operators alone, the OR for frequent lifting was 5.2 (95% CI 1.1–25.5). The frequent lifting in crane operators was also determined to be from jobs held in the past. Among non crane operators, history of frequent lifting exposure was not associated with LBP (OR 0.70, 95% CI 0.14–3.5). Among crane operators, univariate ORs for WBV and prolonged sedentary postures were 0.66 (95% CI 0.14–3.1) and 0.49 (95% CI 0.11–2.2), respectively. In multivariate analyses controlled for age, height, weight, and current crane work, associations with specific work-related factors were substantially reduced; the high prevalence of LBP in crane operators was explained only by current crane work. No measures of dose-response were examined. Limitations included a low response rate for crane operators (67%), with some suggestion that those with illness may have been underrepresented (perhaps underestimating the OR), and self-report of health outcomes and exposures. The investigators excluded cases of LBP with onset before the present job to increase the likelihood that exposure preceded disease.

Kelsey [1975b] carried out a hospital population-based case-control study of herniated lumbar discs and their relationship to a number of workplace factors, including time spent sitting, chair type, lifting, pulling, pushing, and driving. Cases were defined by symptoms, medical evaluation, and radiology; exposures were ascertained by interview (over lifetime job history). Cases (n=223) and controls (n=494 unmatched controls) had similar histories of job-related lifting (RR 0.94, p=0.10). Findings indicated that sedentary work (sitting more than half the time at work) was associated with disc herniation, but only for the age group 35 years and older (RR 2.4, p=0.01). (The RR for those less than 35 was 0.81). Disc herniation was also associated with time spent driving (RR 2.75, p=0.02) and, more specifically, with working as a truck driver (RR 4.7, p<0.02), suggesting a relationship with WBV. The study design had several potential limitations, including possible unrepresentativeness of the study population (because the group was hospital-based). As exposure information was obtained retrospectively, cases may have over-reported exposures thought to be associated with back problems. Strengths include a well-defined outcome and consistent results in comparisons to the two control groups.

Svensson and Andersson [1989] examined LBP in a population-based cross-sectional study of employed Swedish women. Information on LBP and sciatica was obtained by questionnaire, as were exposure-related items. Physical exposures included lifting, bending, twisting, other work postures, sitting, standing, monotony, and physical activity at work. Lifetime IRs varied by occupation, with ranges from 61%–83% in younger age groups and 53%–75% in older groups. After the study was completed, the authors noted that for these women, the highest lifetime incidence of LBP was not found in jobs with the highest physical demands. The measure for “physical activity at work” was also not significantly associated with LBP in univariate analyses. Bending forward (RR 1.3), lifting (RR 1.2), and standing (RR 1.3) were associated with lifetime incidence of LBP in univariate analyses (p<0.05). Sitting was not (OR 0.84, p=0.10). None of the measures of physical workplace factors were associated with lifetime incidence of LBP in multivariate analyses.

Videman et al. [1990] studied 86 males who died in a Helsinki hospital to determine the degree of lumbar spinal pathology. Disc degeneration and other pathologies were determined in the cadaver specimens by discography and radiography. Subjects’ symptoms and work exposures (heavy physical work, sedentary work, driving, and mixed) were determined by interview of family members. In comparison to those with mixed work exposures, those with sedentary (OR 24.6, 95% CI 1.5–409) and heavy work (OR 2.8, 95% CI 0.3–23.7) had increased risk of symmetric disc degeneration. Similar relationships were seen for end-plate defects and facet joint osteoarthrosis. For most pathologic changes,

sedentary work appeared to have a stronger relationship than heavy work. Back pain symptoms were consistently higher in those with any form of spinal pathology, although the difference was significant only for anular ruptures. This study was unusual in design in that it examined a combination of spinal pathological outcomes, symptoms, and workplace factors. However, participation in the study was dependent on obtaining information from family members; participation rates were not stated. While recall bias is often a problem in studies of the deceased, in this case it should have been non differential, if present.

Strength of Association

The ten studies were approximately equal in terms of information they provided relating to static work postures. Burdorf and Zondervan [1990] observed an OR of 0.49 (95% CI 0.11–2.2) for the univariate relationship between prolonged sedentary postures and LBP in crane operators. Holmstrom et al. [1992] found no association between LBP and sitting (in univariate or multivariate analyses). In the Magora [1972, 1973] cross-sectional investigation, the highest LBP rates were observed for those in the “rarely” category for variables related to sedentary postures, sitting, and standing. No dose responses were observed. In the Toroptsova et al. [1995] study of machine manufacturing workers, sitting, standing, and static work postures were not associated with LBP history in univariate analyses. No details were provided. In multivariate analyses, Masset and Malchaire [1994] found a non significant association between LBP and seated posture (OR 1.5, p=0.09) in multivariate analyses. Svensson and Andersson’s 1989 study of Swedish women found that standing was associated with lifetime incidence of LBP in univariate analyses (OR 1.3, p<0.05), but not in multivariate models. Sitting was not associated in univariate analyses (OR 0.84, p=0.10). Walsh et al. [1989] found that low-back symptoms were associated with lifetime occupational exposure to sitting in females only (RR 1.7, 95% CI 1.1–2.6) in multivariate analyses that considered other work exposures. Kelsey’s 1975b case-control study demonstrated that sedentary work (sitting more than half the time at work) was associated with lumbar disc herniations, but only for those 35 and older (RR 2.4, p=0.01); the RR for those less than 35 was 0.81. In a study of salespeople, a dose-response was observed for sedentary work and low back symptoms. An OR of 2.45 (95% CI 1.2–4.9) was seen for the highest category after adjustment for covariates [Skov et al. 1996]. The Videman et al’s [1990] study of cadavers found that those with histories of either sedentary or heavy work exposure had increased risk of symmetric disc degeneration (OR 24.6, 95% CI 1.5–409 and OR of 2.8, 95% CI 0.3–23.7, respectively). Similar results were seen for other disc pathologies. For most pathologic changes, sedentary work appeared to have a stronger relationship than heavy work.

In summary, most (n=6) risk estimates for variables related to static work postures, including standing and sitting, were not significantly different from one. Others found small to moderate significant increases in risk: ORs of 1.3 for standing, 1.7 for sitting (females only), and 2.4 and 2.5 for sedentary work. The Videman et al. [1990] cadaver study found high risks of disc pathology in those with a history of sedentary work. Study quality was similar across the range of point estimates observed. Therefore, an estimate of the strength of association is difficult to determine. The magnitude cannot be estimated based on the available data.

Temporal Relationship

Eight of 10 studies were cross-sectional in design. Two of these attempted to use additional methodologies to increase the likelihood that exposure preceded disorder by excluding cases with onset prior to current job and truncating exposures prior to disorder onset. One found a positive relationship between prolonged sitting and LBP symptoms.

Consistency in Association

The studies showed poor consistency in estimation of the relationship between low- back disorder and static work postures, perhaps due to considerable differences in definition of exposure.

Coherence of Evidence

As mentioned elsewhere, LBP has been associated with mechanical forces causing an increased load on the lumbar spine [Waters et al. 1993]. Increased loading on the spine causes increased intervertebral disc pressures, which in turn, may be responsible for herniation and back pain. In laboratory experiments, disc pressure has been found to be substantially greater in unsupported sitting than in standing positions [Chaffin and Andersson 1984].

Studies reviewed for this document suggested relationships between back disorder and non work activities seemed to be consistent with the hypothesis that static work postures might be associated with back disorder. Kelsey [1975a] observed that, in addition to sedentary work, amount of time spent sitting on weekends was associated with herniated discs. The finding that sedentary work was associated with herniated discs only in older age groups suggested that duration of exposure may be important and that a threshold may exist. Toroptsova et al. [1995] observed that back pain was lower in those who engaged in sports activity, perhaps suggesting that greater muscle strength prevents back pain.

Several authors offered explanations for the lack of associations they observed. It was pointed out that perception of “sedentary” is subjective and that many jobs that investigators (or subjects) considered to include prolonged static postures may actually have allowed considerable movement throughout the day (such as office workers). Other “sedentary” groups (such as industrial sewing machine operators) may be forced by work schedules to maintain static postures for long periods. It is important to have a true range of exposure if differences in associated disorders are to be detected.

Exposure-Response Relationships

Three studies addressed dose-response relationships, two of which did not demonstrate any trends. Magora [1972, 1973] found the highest risk of LBP in the lowest exposure categories for sedentary postures, sitting, and standing. Videman et al. [1990] found a high rate of lumbar disc pathology in those with histories of sedentary and heavy work, with relationships stronger for sedentary work. A dose-response for LBP symptoms and sedentary work was observed by Skov et al. [1996].

Conclusions: Static Work Postures

Ten studies examined the relationship between low-back disorder and static work postures. In most cases, this exposure was not of primary interest but was one of many potential workplace risk factors that were included in analyses. Static work posture was defined in several ways, including sedentary work and work-related sitting and standing. Exposure information was ascertained by interview for nine of 10 studies. The strength of association could not be easily estimated because a large proportion of point estimates did not differ statistically significantly from unity. As a whole, the results from these studies provide inadequate evidence that a relationship exists between static work postures and low-back disorder.

Back to top

Role of Confounders

As mentioned above, back disorder is multifactorial in origin and may be associated with both occupational and non work-related factors and characteristics. The latter may include demographics, leisure time activities, history of back disorder, and structural characteristics of the back [Garg and Moore 1992]. The relative contributions of these covariates may be specific to particular anatomic areas and disorders. For example, a recent study of identical twins demonstrated that occupational and leisure time physical loading contributed more to disc degeneration of the upper than the lower lumbar region [Battie et al. 1995]. For both anatomic areas, age and twin effects (genetic influences and early shared environment) were the strongest identifiable predictors for this particular health outcome.

Psychosocial factors, both work- and non work-related, have been associated with back disorders. These relationships are discussed at length in Chapter 7 and Appendix B.

In the studies reviewed for this document, gender and age effects were addressed in most (86% and 74%, respectively). Approximately 40% addressed work-related psychosocial factors. In addition to these, many studies addressed other potential confounders in their analyses.

Back to top