Research Review By Dr. Shawn Thistle©


Download MP3

Date Posted:

June 2011

Review Title:

Occupational Tasks & Low Back Pain: 3 Systematic Reviews

Studies Included:

  1. Wai EK, Roffey DM, Bishop P et al. Causal assessment of occupational bending or twisting and low back pain: results of a systematic review. Spine J 2010; 76-88.
  2. Roffey DM , Wai EK, Bishop P et al. Causal assessment of awkward occupational postures and low back pain: results of a systematic review. Spine J 2010; 89-99.
  3. Wai EK, Roffey DM, Bishop P et al. Causal assessment of occupational carrying and low back pain: results of a systematic review. Spine J 2010; 628-638.

Background Information:

It is well known that low back pain (LBP) is a common and costly condition (how many reviews have started with this statement in one form or another?). Health care costs for LBP abound to be sure, but much of the financial burden of this condition is shouldered by industry in the form of time off work, lost wages, and compensation claims. However, the exact contribution of various tasks like lifting and twisting, carrying, or sustaining awkward postures to the genesis of LBP has not been fully studied.

In musculosketelal practice, we question all of our back pain patients about a number of things including mechanism of injury, pain pattern, history of injury, and so on. Personally, I always phrase a question to my LBP patients as follows: “Tell me specifically about what you do with your back on a daily basis – at work, play, and leisure?” I emphasize that they are specific – I want to know everything.

What I am trying to glean from their answer is essentially a ‘loading profile’ for their spine. In other words, what kind of loads, postures, positions, and tasks do they endure, how many times per day or hour, for how long, and so on. This information, combined with their history of sports participation and prior injury gives me a pretty clear picture of what their spine has ‘been through’.

Often, I think many manual therapists are more interested in the recent events leading to back pain, or the history of severe injury, rather than the daily ‘loading profile’ that our patients endure. I am certainly not saying that recent events and prior injuries are not important, they certainly are. I simply think we should be well aware of the entire story (so to speak).

To lessen the incidence and consequences of work-related LBP, it is necessary to improve our understanding of the etiology of LBP as it pertains to suspected occupational risk factors such as specific physical activities that workers are engaged in. That was the purpose of these three systematic reviews. The results of each will be summarized below.

Pertinent Results:

Systematic Review #1 - Occupational Bending & Twisting:

In this systematic review, 35 articles fulfilled the inclusion criteria (involving ~44300 participants). Before we get to the results, it is worth noting that bending and twisting was investigated in the following categories:
  1. Bending (trunk flexion < 20°)
  2. Bending (trunk flexion 20-45°)
  3. Bending (trunk flexion > 45°)
  4. Twisting (> 30°)
  5. Twisting (other)
Collectively, the 35 studies provided 243 estimates of the association between categories of bending and twisting listed above and LBP outcomes. Of these 243 estimates, 107 (or 44%) were reported as being statistically significant. Of these 107 - 61 (57%) were classified as ‘‘weak,’’ 20 (19%) were classified as ‘‘moderate,’’ and 26 (24%) were classified as ‘‘strong.’’ No difference was noted in the proportion of estimates considered statistically significant for high quality (30%) versus low-quality studies (32%).

Overall, the results indicate that there is conflicting evidence available to support a causal relationship between bending and LBP for the criteria association, dose-response and biological plausibility, and strong evidence against the criterion temporality. Based on the evidence reviewed, it was not possible to establish a clear causal relationship between occupational bending and LBP, therefore it is not likely that occupational bending or twisting is independently causative of LBP in workers.

Due to the conflicting nature of the evidence, it is possible that different subtypes of occupational bending and twisting may contribute to LBP. Further, the following trends did emerge in their literature analysis:
  • Among the subgroup comparisons, the strongest association was noted for higher degrees of bending (> 45°) and more severe or disabling types of LBP.
  • Similarly, a number of studies suggest that twisting is also causally associated with more disabling LBP (severe or “chronic”), despite a lack of evidence that twisting is associated with other types of disabling LBP (sick leave). The evidence regarding causation from twisting was conflicting.
  • Although numerous theories exist regarding how bending and twisting may lead to LBP, very few studies actually discuss or investigate the biological plausibility of their results.
Systematic Review #2 - Awkward Occupational Postures & LBP:

In this systematic review, 27 studies (involving ~70000 participants) fulfilled the inclusion criteria but only 8 were considered to be of high methodological quality. 7 were prospective cohort studies, 16 were cross-sectional studies and 4 were case-control studies. Seven studies did not report on their statistical analysis to adjust for LBP confounders. Keeping the relatively low quality of this literature in mind, the following trends emerged:
  • Collectively, these studies reported on 111 estimates on the association between awkward occupational postures and LBP – of these, 48% (n=53) were considered statistically significant – of these 53, 66% were classified as ‘weak’, 17% as ‘moderate’, 7% as ‘strong’, 6% as ‘protective’, and 4% could not be classified due to insufficient information.
  • Overall, lower quality studies reported more statistically significant estimates than higher quality studies (57% vs. 35%, respectively).
  • A total of 8 studies (involving ~6400 patients) were considered high quality and used multivariate analyses: 3 were case-control studies, 1 was cross-sectional, and 4 were prospective cohort in design.
  • From these 8 studies, only 2 reported significant associations in their risk estimates, and within these 2 studies, only 19% of the unique risk estimates were considered significant.
  • Therefore, overall - there was strong evidence for consistency of no association between awkward occupational postures and LBP.
  • 2 studies assessed dose response, with one study demonstrating a non-significant dose-response trend. This evidence is conflicting.
  • Three studies were able to assess temporality, but all demonstrated non-significant risk estimates.
  • Biological plausibility was discussed by two studies. This evidence is also understudied and conflicting.
  • There was no available evidence to assess the experiment criterion for causality.
  • Of the 6 possible subgroup comparisons conducted by these authors – awkward posture and severe LBP had a moderate association and moderate biological plausibility, but the strength of the association was weak and none of the other causal criteria were evaluated.
Although it is within reason that repeated awkward postures may contribute to LBP, the results of this systematic review indicate that occupational posture does not fulfill any objective criteria for causation of LBP, and therefore is not likely an independent cause of LBP in workers.

Systematic Review #3 - Occupational Carrying & LBP:

In this systematic review, 22 studies satisfied eligibility criteria (involving ~28000 participants) – 9 were considered to be of high quality. 6 were case-control, 6 cross sectional, and 8 prospective cohort. Pertinent results of this review include:
  • Collectively, these 22 studies reported 109 estimates on the association between occupational carrying and LBP – of these, 24% (n=26) were considered statistically significant – of these 26, 58% were classified as ‘weak’, 31% as ‘moderate’ and 12% as ‘strong’
  • There was a significant difference in the number of significant estimates from high quality studies (only 1 of 38 or 2%) versus low quality studies (25 of 71 or 35%).
  • Overall, there was strong and consistent evidence against an association and temporal relationship between carrying and LBP from high-quality studies.
The results of this particular review are in contrast with previous literature. The authors cite methodological issues and low overall quality in these existing studies.

I think it is worth mentioning that carrying is often associated with other physical movements and cannot always be considered an isolated task (after all, you have to bend or lift or twist or ‘something’ before you carry something, right?). This may explain, to some degree, the lack of association revealed in this review when carrying alone was evaluated.

Clinical Application & Conclusions:

The results of these 3 systematic reviews may seem a bit surprising. Overall, the associations between occupational tasks and LBP seem to be very weak. However, it seems reasonable (to most of us, anyway) that occupational twisting, bending, carrying or awkward postures (especially those that are repetitive) could contribute to LBP. So what gives?

Below I mention a few limitations of these reviews that we should all keep in mind. Sometimes, the very way that research is conducted limits its ability to detect the very ‘thing’ it is searching for. This is certainly true but we must be open to accepting that concepts we take for granted as being ‘true’ may in fact not be. Keeping that in mind, there might be other factors at play here. For example, I think you would all agree that there is no such thing as a homogenous population of LBP patients. Similarly, people in general (those without LBP) are not homogenous. Every person responds differently to treatment and has different ‘capacity’ for loading, repetition of tasks, and so on.

Recent literature suggests that different subgroups of LBP patients do exist, each with unique characteristics and likelihoods of positively responding to a variety of treatment options. In an occupational environment, some people may be able to sustain a particular job, whereas others can’t without sustaining injury. Factors like these are tough to evaluate in observational research which may alter the results of these systematic reviews.

Further, studies such as those included here attempt to give us an overview of population trends. We can use this information to guide us to some degree as some patterns do exist, but we must remember that each individual patient is just that, an individual. One thing I always mention to my students that summarizes this idea is: “The epidemiology of a condition doesn’t matter much to the individual sitting in front of you.”

Due to the social and economic burden of LBP in the workplace, it is important for us as evidence-informed clinicians to be aware of the state of the current literature. Keeping their limitations in mind, these 3 reviews suggest that occupation tasks, at this time, cannot be directly associated to LBP from an associative, causative, or temporal standpoint. More research is required.

Study Methods:

All three systematic reviews aimed to identify, evaluate, and summarize the literature pertaining to the relationship between the occupational task in question and LBP, using the Bradford Hill criteria.

In each paper, the literature was searched using Medline, Embase, CINAHL, Cochrane Library, the Occupational Safety and Health (OSH-ROM) database, gray literature (i.e. studies not published in peer-reviewed journals), hand-searching occupational health journals, reference lists of included studies, conference proceedings, websites and content experts.

Study quality was evaluated using a modified version of the Newcastle-Ottawa Scale. Levels of evidence supporting specific Bradford-Hill criteria were evaluated for different categories of tasks and types of LBP outcomes.

Inclusion criteria:
  1. Published in English or French
  2. Related to occupational exposure
  3. Related to LBP
  4. Related to etiology or causation
  5. Related to occupational task of discussion in the particular systematic review
Exclusion criteria:
  1. No specific population, exposure, and outcome (i.e. too broad)
  2. Nonscientific studies (i.e. commentaries, letters to the editor)
  3. Literature reviews
  4. Related only to treatment of LBP (i.e. does not address a specific risk factor)
  5. Health services research only (i.e. costs of injuries)
  6. Basic sciences, biomechanics studies, and cadaver studies
  7. Less than 30 exposed subjects
  8. Whole-body vibration, psychosocial or environmental risk factors only
  9. Neck pain, thoracic pain, whole-spine pain, sciatica because of disc herniation, or other nonspecific back pain
For each analysis, the following Bradford-Hill criteria were evaluated:
  1. Association (including strength of significant associations)
  2. Dose-response
  3. Experiment
  4. Temporality
  5. Biological plausibility

Study Strengths / Weaknesses:

In all three reviews, the potential weaknesses are very similar. Inherently, systematic reviews can be limited by the quality of the included evidence. Reporting flaws are common in the literature. Frequently noted reporting weaknesses included failure to adopt common operational definitions of LBP, failure to report basic data about the study population (i.e. age, gender), failure to describe the type of statistical methods used (i.e. univariate vs. multivariate), failure to adjust for known confounders, and a failure to disclose which variables were adjusted for in multivariate analyses.

Many studies also report associations as “non-significant” without showing their data, which makes data pooling impossible. Further, because the authors chose relative strict inclusion/exclusion criteria, it is possible that some worthy studies may have been overlooked.  Lastly, although the Bradford-Hill criteria are commonly accepted and utilized, they are sometimes difficult to fulfill, particularly in the area of causation as noted. That being said, the cross-sectional nature of many of the included studies would make causation difficult to ascertain anyway, since cross-sectional studies measure the risk factor and the outcome at the same time.