Journal of Biomechanics xxx (xxxx) xxx

Contents lists available at ScienceDirect

Journal of Biomechanics journal homepage: www.elsevier .com/locate / jb iomech

www.JBiomech.com

Review

No consensus on causality of spine postures or physical exposure and low back pain: A systematic review of systematic reviews

https://doi.org/10.1016/j.jbiomech.2019.08.006 0021-9290/� 2019 Published by Elsevier Ltd.

⇑ Corresponding authors at: School of Behavioural and Health Sciences, Australian Catholic University, Daniel Mannix Building, 17 Young St, Fitzroy, VIC 3065, Aus T.V. Swain). Deakin University, Institute for Physical Activity and Nutrition, School of Exercise and Nutrition Sciences, 221 Burwood Highway, Burwood, Victo Australia (D.L. Belavy).

E-mail addresses: [email protected] (C.T.V. Swain), [email protected] (D.L. Belavy).

Please cite this article as: C. T. V. Swain, F. Pan, P. J. Owen et al., No consensus on causality of spine postures or physical exposure and low back systematic review of systematic reviews, Journal of Biomechanics, https://doi.org/10.1016/j.jbiomech.2019.08.006

Christopher T.V. Swain a,⇑, Fumin Pan b, Patrick J. Owen c, Hendrik Schmidt b, Daniel L. Belavy c,⇑ a School of Behavioural and Health Sciences, Australian Catholic University, Melbourne, Australia b Julius Wolff Institut, Charité – Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Germany cDeakin University, Institute for Physical Activity and Nutrition, School of Exercise and Nutrition Sciences, Geelong, Victoria, Australia

a r t i c l e i n f o a b s t r a c t

Article history: Accepted 9 August 2019 Available online xxxx

Keywords: Low back pain Spine Posture Risk factors Umbrella review Occupation Physical activity

Specific spinal postures and physical activities have been linked to low back pain (LBP) but previous reviews have produced contrasting outcomes. This umbrella review examined (1) what relationship, if any, is evident between specific spinal postures or physical activities and LBP; (2) the quality of existing systematic reviews in this area; and (3) the extent to which previous systematic reviews demonstrate causality. Five electronic databases and reference lists of relevant articles were searched from January 1990 to

June 2018. Systematic reviews and meta-analyses on spine posture or physical exposure and LBP symp- toms (self-report) or outcomes (e.g. work absence, medical consultation) were included. The AMSTAR and the Bradford Hill Criteria were utilised to critically appraise the quality of included systematic reviews and to determine the extent to which these reviews demonstrated causality. Two independent reviewers screened 4285 publications with 41 reviews included in the final review.

Both positive and null associations between spine posture, prolonged standing, sitting, bending and twisting, awkward postures, whole body vibration, and components of heavy physical work were reported. Results from meta-analyses were more consistently in favour of an association, whereas sys- tematic reviews that included only prospective studies were less able to provide consistent conclusions. Evidence that these factors precede first time LBP or have a dose response relationship with LBP outcomes was mixed. Despite the availability of many reviews, there is no consensus regarding causality of physical exposure

to LBP. Association has been documented but does not provide a causal explanation for LBP. � 2019 Published by Elsevier Ltd.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 2. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00

2.1. Search strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 2.2. Inclusion and exclusion criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 2.3. Data extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 2.4. Quality assessment of reviews . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00

3. Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00

3.1. Description of studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 3.2. Risk of bias. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 3.3. Physical exposures and LBP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00

3.3.1. Spinal curvature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 3.3.2. Standing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00

tralia (C. ria 3125,

pain: A

2 C.T.V. Swain et al. / Journal of Biomechanics xxx (xxxx) xxx

3.3.3. Sitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 3.3.4. Bending and twisting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 3.3.5. Awkward postures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 3.3.6. Heavy physical work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 3.3.7. Whole body vibration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00

4. Discussion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 Declaration of Competing Interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 Appendix A. Supplementary material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00

1. Introduction

Early attempts at understanding low back pain (LBP) focussed largely on spine biomechanics. Studies on human cadaver spines and in vivo animal experiments demonstrated that repetitive movements, even small in magnitude, result in physical disruption to structures in and around the spine, preceding inflammation (Solomonow, 2012), injury (Adams and Hutton, 1985), and degen- eration (Osti et al., 1990). As these findings aligned with observa- tions from large epidemiological studies that documented a high prevalence of LBP in occupations that involved specific spine mechanical exposures (Marras et al., 1995), it was deduced that specific physical exposures represented important risk factors for LBP.

LBP is a multifactorial condition and evidence does not always support a clear relationship between physical exposure, spine injury, and LBP. For example, signs of mechanical damage do not always correlate with symptoms (Brinjikji et al., 2015), sedentary populations report a high prevalence of LBP (Heneweer et al., 2009), and depression and emotional distress can predict both the onset of first time LBP (Jarvik et al., 2005), as well as the con- sumption of health care services related to LBP (Traeger et al., 2016). While such findings do not dismiss a role of physical expo- sure in the aetiology of LBP in some populations, they do present questions regarding their collective importance.

Several systematic reviews and meta-analyses have been per- formed with the intent of objectively synthesising the evidence regarding physical exposures and LBP. However, these have pro- duced conflicting, and at times, controversial findings. For exam- ple, while some systematic reviews support strong associations (Hoogendoorn et al., 1999), others, including a series of reviews summarised by Kwon et al. (2011), do not support a causal associ- ation between occupational exposures and LBP. Notably, these reviews received criticism for their scope and methods (Kuijer et al., 2011; McGill, 2011), as well as interpretation of individual (Andersen et al., 2011) and collective (Takala, 2010) findings. Therefore, to advance the understanding regarding specific physi- cal exposures as risk factors for LBP, an umbrella review was con- sidered. The aims of this review were to examine: (1) what relationship, if any, is evident between specific physical exposures and LBP; (2) what is the quality of existing systematic reviews in this area; and (3) to what extent do existing reviews demonstrate causality.

2. Methods

This review was conducted in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (Moher et al., 2009), and followed a methodology proposed by Smith et al. (2011) for conducting a systematic review of systematic reviews within the healthcare field. This review was prospectively registered on PROSPERO (CRD42018110739).

Please cite this article as: C. T. V. Swain, F. Pan, P. J. Owen et al., No consensus systematic review of systematic reviews, Journal of Biomechanics, https://doi.

2.1. Search strategy

Using a combination of controlled MeSH and free-text terms (Appendix 1), relevant publications were identified through systematic searches of the following five electronic databases: MEDLINE, SportDiscuss, EMBASE, CINAHL, and the Cochrane Data- base of Systematic Reviews. In addition, reference lists of included systematic reviews were manually searched.

2.2. Inclusion and exclusion criteria

Only systematic reviews and meta-analyses published in peer- reviewed journals (i.e. grey literature excluded) in English between January 1990 and June 2018 were included. For a review to be con- sidered systematic, the authors must have defined a strategy to: (a) search for studies, (b) appraise studies and (c) synthesise studies. Population groups of interest were adults (aged � 18 yr). The expo- sure was restricted to postural curvature, static posture (sitting and standing) or dynamic/occupational movements (e.g., bending, twisting, lifting). To ensure consistency in the type of biomechan- ical exposure, general physical activity, activities in leisure time and sports, or other athletic activities were excluded. The primary outcomes included self-reported LBP symptoms or LBP-specific outcomes (e.g., activity limitation, work absence, care-seeking, medication use). Studies within these systematic reviews included cross-sectional analyses, prospective cohorts and randomised con- trolled trials.

2.3. Data extraction

Data of the publication (e.g. authors, title, year), protocol of the systematic review (e.g. population, exposure and outcomes), results of the systematic review (e.g. number of studies included, qualitative synthesis of findings, quantitative estimates [meta- analyses only], observed limitations, and suggestions for future research) were extracted by two independent assessors (CTVS and FP). In addition, to determine the extent to which existing reviews have demonstrated causality, summary findings pertinent to the Bradford Hill Criteria (including strength of association [i.e. effect size], consistency [i.e. the association is evident in multiple settings], temporality [i.e. evidence that the exposure precedes LBP], biological gradient [i.e. dose-response], as well as experiment [i.e. association is supported by experimental or intervention stud- ies]) were extracted (Hill, 1965; Kwon et al., 2011). Any discrepan- cies were discussed between the two independent assessors, until consensus was reached. Prior to data extraction, the method was piloted and reviewed on five relevant systematic reviews.

2.4. Quality assessment of reviews

The second iteration of the assessment of multiple reviews (AMSTAR) tool (Shea et al., 2017) was utilised by two independent assessors (CTVS and FP) to critically appraise the quality of the

on causality of spine postures or physical exposure and low back pain: A org/10.1016/j.jbiomech.2019.08.006

Fig. 1. PRISMA chart for eligible study selection process.

Table 1 Summary of review conclusions: physical exposures and LBP.

Effect Estimates Association Temporality Dose – Response AMSTAR Exposure Range of effect estimates

from meta-analyses Number of reviews in favour/ total number of reviews Range of

Percentages §

Spinal curvature 0.01(�0.00, 0.11)y to �0.66(�0.91, �0.40)y 3/5 2/2 Nil 62–92 Standing 1.31(1.10, 1.56)� 4/8 2/3 1/3 54–92 Sitting Nil 0/7 0/3 0/2 31–85 Bending and twisting 1.68(1.41, 2.01)� 7/10 0/1 0/3 8–85 Awkward postures 1.14(1.08, 1.21)� to 2.03(1.26, 2.49)� 4/5 0/1 0/1 15–75 Lifting, carrying, manual

materials handling, and pushing or pulling

1.11(1.05, 1.18)� to 2.11(1.73, 2.57)� 11/16 2/5 3/6 8–88

High physical workloads or occupational specific demands

1.31(0.96, 1.33)� 10/12 1/1 1/1 8–85

Whole body vibration 1.39(1.24,1.55)� to 2.3(1.8,2.9) � 10/12 0/2 3/6 8–85

Key: y SMD (95% CI). � OR (95% CI). § Number of items scored as ‘partial’ or ‘yes’/total number of items.

C.T.V. Swain et al. / Journal of Biomechanics xxx (xxxx) xxx 3

systematic reviews included within this review. AMSTAR is a vali- dated instrument that contains 16 items relating to the registration, design, methods, interpretation, and reporting standards of system- atic reviews. Disagreements were resolved via discussion and a third reviewer (PJO) was available for adjudication if necessary.

3. Results

Search results are summarised in Fig. 1. The literature search returned a total of 6050 articles. Following duplication removal (n = 1717), removal of non-English publications (n = 48), a review of title and abstracts (n = 4285), 78 full texts were screened, with 41 publications included in the final review. A list of publications and reasons for exclusion following full text screening is included in Appendix 2.

3.1. Description of studies

Of the included reviews, 30 were systematic only, and 11 had a relevant meta-analysis component. The exposures examined included spinal curvature, standing, sitting, awkward spine pos- tures, bending and twisting movement of the spine, heavy physical work and whole-body vibration. Characteristics and a summary of review findings are presented in Appendix 3. An overview of review conclusions is presented in Table 1.

Please cite this article as: C. T. V. Swain, F. Pan, P. J. Owen et al., No consensus systematic review of systematic reviews, Journal of Biomechanics, https://doi.

3.2. Risk of bias

Risk of bias results are presented in Appendix 4. The systematic reviews partially or completely satisfied a median (range) 8(1–12) of 13 criteria and the meta-analyses satisfied a median (range) 13 (9–13) of 16 criteria. Higher scores were observed in more recent reviews. The individual items the reviews scored the worst on included the consideration of funding sources (0% of included reviews satisfied this criteria), explicitly stating a review was pre- registered (5%), performing data extraction in duplicate (49%), and providing a list of excluded studies with specific reasons for exclu- sion (49%). In addition, only four (36%) of the meta-analyses exam- ined the extent of publication bias present, with some evidence of publication bias in favour of smaller studies with significant find- ings identified (Burstrom et al., 2015; Coenen et al., 2018). Of note, although not an item on the AMSTAR, one review included only pri- mary studies that reported statistically significant outcomes (da Costa and Vieira, 2010). This review, as well as the three that scored the lowest on the AMSTAR (Burdorf and Sorock, 1997; Jansen and Burdorf, 2003; Jin et al., 2000), generally produced conclusions in favour of an association between physical exposures and LBP.

3.3. Physical exposures and LBP

3.3.1. Spinal curvature Three meta-analyses (Chun et al., 2017; Laird et al., 2014; Sadler

et al., 2017), as well as two systematic reviews (Christensen and

on causality of spine postures or physical exposure and low back pain: A org/10.1016/j.jbiomech.2019.08.006

4 C.T.V. Swain et al. / Journal of Biomechanics xxx (xxxx) xxx

Hartvigsen, 2008; Coenen et al., 2017) included evidence regarding the association between spinal curves (i.e. sagittal lumbar spine curves when relaxed or standing) and LBP. Of these, two meta- analyses identified a significantly flatter lumbar spine in persons with LBP (standardised mean difference, SMD[95% confidence intervals, CI]: �0.66[�0.91, �0.40]) (Chun et al., 2017), or persons that went on to develop first time LBP (odds ratio, OR[95%CI]: 0.73 [0.55, 0.98]) (Sadler et al., 2017). Further, one review identified evi- dence that both increases in lumbar flexion as well as an increase in lumbar curvature were associated with symptom development following prolonged standing (Coenen et al., 2017). In contrast, one meta-analysis of studies using non-invasive assessment meth- ods identified no difference in lumbar lordosis in persons with and without LBP (SMD[95% CI]: 0.01[�0.00, 0.11]) (Laird et al., 2014). The final systematic review concluded that as only 10/29 studies identified a positive association, there was insufficient evidence to support any interaction (Christensen and Hartvigsen, 2008). Based on these findings, there is support for an association and some support for temporality, but the findings are not consistent.

3.3.2. Standing Eight reviews examined the interaction between standing expo-

sure and LBP outcomes (Bakker et al., 2009; Burdorf and Sorock, 1997; Coenen et al., 2017; Coenen et al., 2018; Heneweer et al., 2011; Hoogendoorn et al., 1999; Roffey et al., 2010d; Taylor et al., 2014). In favour of a relationship, one meta-analysis identi- fied substantial (>4h/d) occupational standing was associated with LBP symptoms (OR[95%CI]: 1.31[1.10, 1.56]) (Coenen et al., 2018). The OR were higher in high-quality studies (1.38[1.16, 1.64]), but no dose response relationship was identified. In a review of labora- tory based studies, 19/19 studies identified an association between standing and low back symptom development in at least a sub- group of participants (Coenen et al., 2017). In this review, data pooling did identify a dose response relationship, with symptom development after 71 min of standing in general populations, and after 42 min in subgroups of individuals classified as pain develop- ers. Evidence for temporality was demonstrated by one review that found standing for more than two hours preceded first time LBP in women and recurrent LBP in both men and women (Taylor et al., 2014). In contrast, four reviews concluded that evidence did not support an association (Bakker et al., 2009; Burdorf and Sorock, 1997; Hoogendoorn et al., 1999; Roffey et al., 2010d). Of these, one review highlighted limited evidence for a dose response rela- tionship (Roffey et al., 2010d), and two contained contrasting evi- dence that standing precedes LBP development (Bakker et al., 2009; Roffey et al., 2010d). In summary, there was evidence in sup- port of, as well as in opposition to, an association, dose-response, or temporal relationship between standing and LBP.

3.3.3. Sitting Seven reviews provided information regarding the relationship

between sitting and LBP (Bakker et al., 2009; Chen et al., 2009; Hartvigsen et al., 2000; Hoogendoorn et al., 1999; Lis et al., 2007; Roffey et al., 2010c; Taylor et al., 2014). All concluded that there was either no, or limited evidence, for an association, a dose- response relationship, or temporality. Among the primary studies included in the reviews, one identified a negative (protective) effect of prolonged sitting (Taylor et al., 2014). Reviews do not sup- port sitting on its own as a factor related to LBP.

3.3.4. Bending and twisting One meta-analysis (Lotters et al., 2003) and nine systematic

reviews, provided information regarding bending and twisting movements or flexed postures as risk factors for LBP (Bakker et al., 2009; Burdorf and Sorock, 1997; Coenen et al., 2017; Heneweer et al., 2011; Hoogendoorn et al., 1999; Jansen and

Please cite this article as: C. T. V. Swain, F. Pan, P. J. Owen et al., No consensus systematic review of systematic reviews, Journal of Biomechanics, https://doi.

Burdorf, 2003; Jin et al., 2000; Ribeiro et al., 2012; Wai et al., 2010a). In the meta-analysis, the OR(95% CI) for bending and twist- ing and LBP was 1.68(1.41, 2.01) (Lotters et al., 2003). Individuals with high exposure 1.31(0.92, 1.87) had slightly higher odds than individuals with low exposure 1.14(0.85, 1.52), however, neither of these were statistically significant on their own. In the eight remaining systematic reviews, five identified strong or mostly pos- itive evidence for an association (Burdorf and Sorock, 1997; Heneweer et al., 2011; Hoogendoorn et al., 1999; Jansen and Burdorf, 2003; Jin et al., 2000), one identified evidence that axial rotation increases symptoms following prolonged standing (Coenen et al., 2017), one identified conflicting evidence (Bakker et al., 2009), and two did not identify convincing evidence (Ribeiro et al., 2012; Wai et al., 2010a). Two systematic reviews commented on dose response, stating that there was either limited or no clear evidence in support of a relationship (Ribeiro et al., 2012; Wai et al., 2010a). One of these also stated that there was strong evidence against temporality (Wai et al., 2010a), although, a review that included only prospective cohort studies with people free of LBP at baseline did identify conflicting evidence (Bakker et al., 2009). Collectively, while there is evidence from a meta- analysis of a relationship, the findings from systematic reviews were inconsistent, the evidence for a dose response relationship is limited, and temporality is conflicting.

3.3.5. Awkward postures One meta-analysis (Griffith et al., 2012) and four systematic

reviews examined the association between awkward, flexed, or non-neutral postures and LBP (da Costa and Vieira, 2010; Lis et al., 2007; Nelson and Hughes, 2009; Roffey et al., 2010a). In the meta-analysis, the effects (OR[95% CI]) ranged from 1.14 (1.08, 1.21) to 2.03(1.26, 2.49) (Griffith et al., 2012). Three system- atic reviews concluded that there was reasonable evidence for awkward postures and LBP (da Costa and Vieira, 2010), that sitting combined with awkward postures increased risk for LBP (Lis et al., 2007), or that non-neutral postures combined with lifting increased LBP risk (Nelson and Hughes, 2009). In contrast, one review concluded there was strong evidence against an association or temporality (Roffey et al., 2010a).

3.3.6. Heavy physical work There were 22 reviews that examined exposures including

heavy physical work, lifting, manual materials handling, carrying, pushing or pulling, and specific occupational demands. Of these, three of three meta-analyses identified significant positive associ- ations between these exposures and LBP. Specifically, Coenen et al. (2014) identified an OR(95% CI) of 1.11(1.05, 1.18) per 10 kg lifted and 1.09(1.03, 1.15) per 10 lifts/d. Griffith et al. (2012) identified OR (95%CI) ranging from 1.40(1.30, 1.62) to 2.11(1.73, 2.57) for lift- ing or heavy physical work forces. Lotters et al. (2003) identified increased OR(95%CI) for manual materials handling: 1.51(1.31, 1.74), including higher odds for individuals with high exposure (1.61[1.26, 2.05]) compared to low exposure (1.27[1.00, 1.62]); however, the odds for heavy physical work were not significant 1.31(0.96, 1.33). In addition, eight systematic reviews concluded evidence in favour of an association between these components of heavy physical work and LBP was strong (Heneweer et al., 2011), reasonable (da Costa and Vieira, 2010), consistent (Burdorf and Sorock, 1997; Ferreira et al., 2013; Nelson and Hughes, 2009), or present (Jansen and Burdorf, 2003; Janwantanakul et al., 2012; Kuiper et al., 1999). Of note, heavy physical work was associated with LBP when controlling for heritability (Ferreira et al., 2013), and lifting as a risk factor may be more important in men than women (Hooftman et al., 2004). Three further reviews that examined occupational specific demands (e.g. related to nursing) provided conclusions in favour of these

on causality of spine postures or physical exposure and low back pain: A org/10.1016/j.jbiomech.2019.08.006

C.T.V. Swain et al. / Journal of Biomechanics xxx (xxxx) xxx 5

associations (Lagerstrom et al., 1998; Sherehiy et al., 2004; Yassi and Lockhart, 2013).

In contrast, the conclusions of five systematic reviews did not support an association, describing the evidence as conflicting (Bakker et al., 2009; Roffey et al., 2010b; Wai et al., 2010c), or strongly against (Roffey et al., 2010e; Wai et al., 2010b). Similarly, some reviews identified evidence for a dose response relationship (Burdorf and Sorock, 1997; Ferreira et al., 2013) although others described this as minimal and mostly non-significant (Roffey et al., 2010b, e; Wai et al., 2010b, c). Three reviews identified evi- dence that lifting (Bakker et al., 2009; Taylor et al., 2014) and high job strain (Janwantanakul et al., 2012) can precede episodes of LBP, whereas four reviews interprete