Objectives Investigate the relationships between the ability/inability to perform five physical test exercises and the presence or absence of low back pain (LBP).
Setting Regional Australian council training facility.
Participants Consecutive participants recruited during 39 back education classes (8–26 participants per class) for workers in general office/administration, parks/gardens maintenance, roads maintenance, library, child care and management. Total sample (n=539) was reduced through non-consent and insufficient demographic data to n=422. Age 38.6±15.3 years, range 18–64 years, 67.1% male.
Methods Cross-sectional, exploratory, observational investigation. LBP presence was ascertained from a three-response option questionnaire: 0=none/rarely (no) 1=sometimes (some), 2=mostly/always (most). Statistical correlation was performed with the number of the five test exercises the individual successfully performed: (1) extension in lying: 3 s; (2) ‘toilet squat’; feet flat, feet touched: 3 s; (3) full squat then stand up: 5 times; (4) supine sit-up, knees flexed: 10 times; and (5) leg extension, supine bilateral: 10 times.
Results For the group ‘no-some’, 94.3% completed 4–5 test exercises, while for group ‘With’, 95.7% completed 0–1 test exercises. The relationship between LBP presence and number of exercises performed was highly significant (χ2 (10)=300.61, p<0.001). Furthermore, multinomial logistic regression predicting LBP (0=no, 1=some, 2=most) from the number of exercises completed, substantially improved the model fit (initial-2LL=348.246, final-2LL=73.620, χ2 (2)=274.626, p<0.001). As the number of exercises performed increased, the odds of reporting ‘some LBP’ or ‘most LBP’ dropped substantially (ORs of 0.34 and 0.17, respectively).
Conclusion The ability to complete/not complete five test exercises correlated statistically and significantly with a higher LBP absence/presence in a general working population. Training individuals to complete such exercises could facilitate reductions in LBP incidence; however, causality cannot be inferred. Randomised trials are recommended to establish the potential efficacy of exercise-based approaches, considering these five selected exercises, for predicting and managing LBP.
- low back pain
- preventive medicine
- functional exercises
- rehabilitation medicine
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
Strengths and limitations of this study
The sample diversity with continuity and subsequent homogeneity enabled generalisation to be inferred.
This was a cross-sectional, exploratory, observational investigation.
It is representative of a general working population as it contained diverse occupations, ages, both genders and a consecutive sample from regional council workers during an educational workshop.
The sample size was sufficient to ensure adequate power.
The functional exercises were not tailored for either dose or specificity for age or gender.
Low back pain (LBP) is among the world’s most prevalent occupational disorders in working populations1 and major global public health concerns2 3 and worsening due to increasing age and populations.4 It affects 12% of the world’s population at any given time2 5 with lifetime prevalence at 84% and chronicity around 23%.2 When disability-adjusted life years are considered, LBP is a leading global cause of disease burden.5 6 LBP is distinctive in that limited progress has occurred in identifying effective prevention strategies and treatments7 8 and remains nearly impossible to provide absolute certainty of a specific nociceptive cause, and only a small proportion has a recognised pathological cause.3 6 This is despite established recognition and identification of factors that predispose or correlate to future LBP.6 9 10 Predicting problematic LBP has several promising protocols including questionnaire-based biopsychosocial screening methods11–13 and movement patterns or maladaptive postures.14 There are, however, few or no validated physiological or physical predictive screening tests15 including measures of disuse or changed levels of physical conditioning.16
The LBP economic burden leads to reduced efficiency and productivity by individuals, organisations and the community compounding direct or indirect costs to private, professional and governmental medical care stakeholders, wages compensation, worker recruitment and training and productivity losses.5 17 These factors are further inflated by social consequences to individuals, families, communities and general society.18 19 Despite many recognised risk factors that predispose individuals to LBP,10 20 business process trends in work settings coupled with recent technology advancement have seen occupational and social changes that influence the requirements or personal choices to adopt static postures.21 In contrast, manual workers have gained both advantages and disadvantages, with occupational postures and loads in areas such as maintenance and building having remained consistent.14 22
The direction of contemporary research on LBP prevention and recurrence has focused on non-modifiable factors and long-term exposures. These include: medical investigative relationships such as radiological23 24 or physiological findings25–27 that have produced mixed results even from the same study28 and biopsychosocial considerations29–32 or a mixture of these.9–11 In contrast, modifiable factors33 34 including movement patterns,14 35 physiological loads36 and exercise capacity37 38 receive limited attention yet they significantly influence LBP morbidity and symptomology,1 2 being recognised as potentially able to prevent LBP.11 20
LBP disorders are multifactorial with individual symptomology influenced by various pathoanatomical, physical, neurophysiological, psychological and social contributors.3 14 36 Consequently, voluntary activities that involve lumbopelvic-specific exercises are effective in primary and secondary LBP prevention.39 Such exercises improve fitness and occupational status by diminishing disability and problem severity35 40 and may counter selective atrophy of type II fibres found in the presence of pathological changes.41 42 However, muscle recruitment remains predominantly neural based during rehabilitation with psychological adaptations derived from improved motivation and pain tolerance.43 The conundrum remains that LBP reduces functional capacity, fitness and general health status (GHS), including depression,44 while low capacity from pathology, injury, GHS or sedentary lifestyle increases the risk of LBP.45 The need to consider modifiable factors is supported by recent research46 that confirmed the relationship between dynamic physical tests, self-reported LBP and reduced function.38 47
Existing research has a knowledge gap for modifiable factors demonstrating a need for observational studies in representative working populations.3 6 11 Addressing this gap will assist in identifying the relationship between LBP symptoms and individual physical functional movement capabilities. A representative group, with strong indicators of generalisability, is council workers. The group includes diversity of gender, age and occupations with variance in manual and sustained loads48 and stationary and sedentary postures.49 Cross-sectional analysis of these groups is a starting point in implied generalisation and provides insight into the capacities and abilities that may lead to the presence or risk of LBP.50 51
This observational study investigated council workers, as an implied representative general working population sample and evaluated whether the ability, or not, to perform five back-related exercises could determine or predict the presence or absence of LBP. We hypothesised that the test exercises would demonstrate the ability of the lumbar spine to: move in a controlled manner through normal range as a complex multisegmental functional activity with coordinated biomechanical and neuromuscular components and be stabilised, as part of the lumbo-pelvic-hip complex, through motor control of the integrated muscular system.36 52 Consequently, the ability to perform the exercises would correlate with lower self-reported LBP.
Once established, analysis of the findings might indicate what movements, or lack thereof, might be associated with the presence and/or absence of LBP for individuals in different occupational and physical activity settings. The outcomes might contribute to understanding the relevance of functional movement and exercises in relation to LBP and provide a direction for future prospective studies. Such studies could identify specific functional movements for specific tasks or risk groups, then provide structured exercise regimens that might reduce LBP and its predisposition.
A cross-sectional, exploratory, observational investigation was initiated over a period of 28 months in a population of employees with the Sunshine Coast Regional Council in Queensland, Australia. Workers from a convenience sample were consecutively recruited during 39 annual back educational programme classes of 2-hours duration. The first two classes provided a pilot study (n=33) to estimate effect size, and ‘Bootstrap analysis’ ensured the effect size had reasonable confidence. Standard power estimation calculations on the range of anticipated effect sizes provided minimum sample size goals. The participants were recruited from a range of occupations, ages and work locations to provide participants that reasonably reflected the population of interest. This representative population minimised selection bias; however, potential bias remained from non-response, the volunteer consent requirement and ascertainment bias. Most participants were classified in to 21 occupational categories with an additional ‘Other’ category for miscellaneous non-specified occupations. Class participant numbers ranged from 8 to 26, with a total sample of n=539. Only participants who consented were included. Data were excluded if there was insufficient demographic information. Consequently, the sample was reduced to a total of n=422, age 38.6±15.3 years, range 18–64 years, 67.3% male (see table 1). Males were predominant in manual occupational roles including maintenance and construction, while females were predominant in carer and resource management including child care, community services, library services and records roles.
The test exercises were selected based on having significant elements of lumbo-pelvic-hip function and being recognised for reducing symptomology or risk of LBP. The five selected exercises were chosen to represent a balanced variation of functions required for normal daily activities.35 Three exercises previously investigated, ‘repeated sit-ups’, ‘repeated squats’ and ‘extension in lying’ (EIL),38 showed a positive correlation with LBP prevention and were consequently included. The sustained squat and leg extension exercises, respectively, require functional movement36 52 and a predominantly isometric abdominal coactivation,53 which occur or simulate daily occupational and sports activities.54 Other exercises were considered but excluded, such as active spine flexion, which has shown poor correlation with LBP.55
All participants were volunteers and performed five functional movement exercises during an educational session with other attendees, supervised by the session leader, a sports physiotherapist certified in McKenzie Manual Diagnostic Therapy. The instructions for exercise justification, instructions, completion and reliability are detailed in table 2. Intraobserver reliability for screening tests movement instruction is recognised as being moderate to high.56
During the educational sessions each participant completed a self-report questionnaire: ‘How often do you have low back pain?’ with three response options: ‘rarely/none’, ‘sometimes’ or ‘always/mostly’, with the time frame and symptoms interpreted within their life context. This three-point scale is condensed from the WHO’s five points: ‘never’, ‘rarely’, ‘sometimes’, ‘often’ and ‘very often’.57 The central three-point response provides an ‘intermediate’ option, which is critical from psychological and statistical perspectives. Psychologically, three cognitive perspectives facilitate response accuracy by reducing cognitive load,58 59 which improves precision and consistency.60 Statistically, responses were coded on a 0–1–2 scale61 62: 0=rarely/none (no LBP), 1=sometimes (some LBP) and 2=always/mostly (most LBP).
Statistical analysis was performed using SPSS V.23.0 for Windows with significance set at p<0.05. Following preliminary data screening to ensure data quality (eg, no aberrant values), an initial cross-tabulation of LBP (0=none, 1=some, 2=most) and number of exercises was performed to explore whether self-reported LBP was related to the number of exercises completed. A χ2 test evaluated whether the null hypothesis (that the number of exercises completed would be consistent across LBP groups) was tenable or able to be rejected. Standard power calculations on the effect sizes verified that the minimum sample size was exceeded.
A multinomial logistic regression was performed, exploring whether the number of exercises (EX_SUM) predicted LBP (categorised as 0, 1 and 2) to test the null hypothesis that the probability or odds of being classified into LBP groups are not different because of number of exercises performed and, if rejected, to quantify the change in odds or probability of LBP as it relates to number of exercises performed. This test also allowed us to evaluate whether participant gender interacted with EX_SUM, or whether there were non-linear effects present. Regression diagnostics for this analysis (eg, residuals and influence) were examined to ensure no aberrant cases were inappropriately influencing the analysis.63 None were identified.
Finally, if the null hypothesis from the prior multinomial logistic regression was rejected, we performed a second multinomial logistic regression on LBP entering each exercise as a predictor (rather than simply the count of number of exercises completed) to examine whether all exercises were uniquely predictive or whether some subset of exercises were more predictive than others. All five exercises were entered simultaneously, allowing for examination of unique effects of each variable controlling for all other variables in the equation. Regression diagnostics were examined, and no aberrant cases were identified.63
Patient and public involvement
The research question and outcome measures were developed over a 3-year period during delivery of a work site back care education programme to a regional council in Queensland, Australia. This involved both formal and informal work-related discussions with attendees and management enabling the programme and exercise selection to be progressively modified. This procedure informed programme progression, specifically the exercises and their relation to the presence or not of LBP, and ensured the priorities of exercise simplicity for the identification and prevention of LBP. The experience gained by this process refined the programme and the selected preferences guiding the statistical relation between the exercises and the presence or not of LBP. The results of each session were disseminated immediately to each participant, and after the initial 3 years of the programme and pilot statistical analysis, the statistical relation was discussed with the council management as part of the programme feedback.
For descriptive purposes, a cross-tabulation of LBP (0=none, 1=some, 2=most) and the number of exercises accomplished is presented in table 3. Most participants reporting no LBP could complete most exercises. For individuals with no LBP, 85.5% could complete at least four exercises. Exercise completion dropped significantly for participants with ‘some’ LBP. In this group, only 22.9% were able to complete four or more exercises, and for participants with ‘most’ LBP, only 10.5% were able to complete four or more exercises. Analysing participants in each category who failed to complete more than one exercise, the pattern is reversed. Only 2.9% of those with no LBP had trouble completing more than one exercise, while 23.7% of those with ‘some LBP’ and 74.3% of those with ‘most LBP’ were unable to complete more than one. A Pearson χ2 test was performed demonstrating a significant relationship between the variables of ‘LBP’ and ‘number of exercises performed’ (Χ2 (10)=300.61, p<0.001).
A multinomial logistic regression predicting LBP (0, 1, 2, with 0 being the reference group) from the count of exercises that could be completed (EX_SUM, ranging from 0 to 5), showed a strong effect (initial-2LL=348.246, final-2LL=73.620, Χ2 (2)=274.626, p<0.001; table 4).
As presented in table 5, as EX_SUM increased incrementally, the odds of reporting some LBP or most LBP reduced substantially: OR=0.34 (95% CI 0.27 to 0.44) and 0.17 (95% CI 0.12 to 0.23), for LBP=1 and 2, respectively. No curvilinear effect was present nor any gender effect.
A second multinomial logistic regression with the five exercise variables entered individually, rather than entering the total number accomplished, evaluated whether tests were individually predictive of LBP. As shown in table 6, overall, the effect was similarly strong63 (initial-2LL=429.93, final-2LL=147.40, Χ2 (2)=282.53, p<0.001).
As table 7 presents, most exercises were individually predictive of LBP (when LBP=1, EIL was not uniquely predictive with all other variables in the equation). All others were statistically significant (p<0.002) with ORs ranging in magnitude from 0.21 to 0.38. For ‘Most’ LBP (LBP=2), all exercises were significant independent predictors of LBP (all p<0.017), with ORs ranging from 0.09 to 0.35.
Sensitivity for the first analysis (per cent of participants with LBP correctly classified into LBP category) was 82.3%, and specificity (percent of participants with no LBP classified as such) was 85.6%. The positive predictive value (true positives divided by true and false positives) was 89.1%, and negative predictive value (true negatives divided by true and false negatives) was 77.1%. Sensitivity for the second analysis was 79.5%, and specificity was 87.9%. Positive predictive value was 90.4%, and negative predictive value was 74.9%.
We also took into consideration a simple analysis relating the presence or absence of LBP to exercises. This approach, combining two groups of LBP (some and mostly) into one category potentially reduces the goodness of the analysis by combining two different groups into one heterogeneous group. If the two groups were distinct, this would increase error variance and decrease the power and informativeness of the analyses. Ancillary binary logistic regression analyses therefore tested the null hypothesis that the two LBP groups were similar. Results of this analysis, which predicted LBP (ie, some vs mostly) showed that EX_SUM was significantly related to this outcome (initial-2LL=339.05, final-2LL=284.96, Χ2 (1)=54.09, p<0.001), leading us to reject the null hypothesis and assert that these two groups are significantly distinct.
Previous research demonstrated a relationship between dynamic physical tests, self-reported LBP and reduced function.38 However, such research has been neglected in recent decades29–32 as focus shifted towards physiological and radiological findings9 10 and biopsychosocial attributes.3 6 7 11 Grönblad et al 38 showed three physical exercises (repetitive sit-ups, squats and EIL) had a positive correlation with LBP. Our current study builds on this research as it expands the number of test exercises. It also shows a higher statistical correlation between physical exercise tests and LBP than found previously. These findings with robust effect sizes, and the 95% CIs,63 demonstrate a substantial relationship. Our results indicate that for each increase in the exercise number accomplished, the odds of having some LBP were about one-third less than that of those participants accomplishing one fewer exercise. The authors feel these research findings are generalisable to settings other than those originally tested due to several factors. The council worker population included 21 distinct occupational categories across manual and sedentary requirements under sustained and moveable loads,48 49 field work in both outdoor and indoor settings and included a broad distribution of age groups and both genders that indicate the abilities and capacities of workers that present some of the highest potential risk of LBP.50 51
This study clearly showed that the presence of LBP is significantly statistically related to the ability to perform the chosen exercise tasks. All exercises were uniquely predictive of LBP (except EIL where LBP=1). The total number of exercises completed was strongly related to LBP. The relevance of a gender effect and potential curvilinear effects was tested as per the accepted recommendation63 and found to have no effect on the results. Effectively, those able to perform more exercises were substantially less likely to report LBP. Consequently, these exercises have the potential to be a part of the areas of recommended necessary investigation in future research3 11 in terms of the ability to provide a clinical diagnosis related to the potential or risk that an individual may develop LBP and perhaps even future impairment.
The ability to perform repeated squatting has been demonstrated to be inversely related to LBP as the balance between hip and lumbar spine mobility must be met, that is, better squatting ability is associated with reduced LBP.47 These researchers found females more susceptible to LBP if they had lower physical performance capacity, a finding not evident in our study. Furthermore, excess/prolonged squatting has a negative effect through increased LBP.64 Similarly, EIL is beneficial and facilitates lumbar lordosis maintenance.65 There is a direct link between a reduced lordosis and LBP.66 Lordosis maintenance is essential for disc symptomology centralisation for LBP management and preventative exercise strategies.65 67 The exercise alone was not predictive of LBP.
Back endurance testing is a statistically accurate LBP screening test as poor performance in static back endurance correlates to higher incidence.68 69 However, EIL is a passive test using the arms as the prime mover. It is possible that individuals with excessive lumbar extensor activation and substitution during this test may confound the results. Furthermore, some studies have indicated that trunk muscle strength measures in isolation are unrelated to LBP symptoms and functional ability.
Exercise therapy is an efficient, cost-effective LBP management strategy,70 71 but there is no evidence to support any single exercise. Coordinated muscle activity around the lumbopelvic region is considered vital for mechanical spinal stability.36 72 Several rehabilitative ‘stabilisation exercise’ approaches emphasise retraining functional movement patterns, rather than focusing on specific muscles.35 73 74 The tests we chose activate and challenge the global muscles of the abdomen and trunk, the ‘abdominal brace’ mechanism75 and their ability to act and interact in a synergistic and functional manner. We screened functional test performance where the aim was assessing participants’ functional status regardless or not of LBP and its known or potential cause. As LBP increases in industrial societies with no clear cause, it is important to consider risk factors of physical workload and awkward posture7 as well as preventative strategies that may play a key role in reducing healthcare system demands and societal support. The exercise tests we used primarily address abdominal and lumbopelvic muscles and their coordination with lower limb muscle activity and maintenance of balance. This coordination was recently defined as ‘integral’ in understanding lumbar stability as a complex integrated model.36 Personal efficiency in physical self-test completion can act as a screening methodology for individuals at risk of LBP. It is, however, important that the method of test performance is considered, for example, there is no relation demonstrated between sit-up performance and LBP when the feet are held.76 This action preferences hip flexor activity over abdominal participation. Alternative actions that preference abdominal muscles, for example, partial curl-up, are more highly correlated to LBP.77 78 Our results provide guidance for future work that may contribute to a comprehensive screening, prevention and management approach to LBP.
Study strengths and limitations
The strengths of this study include the cross-sectional nature, the sample including both genders, diverse age groups and occupations but within one organisation and geographical region. This enabled continuity, and a degree of homogeneity in the otherwise varied sample, that strengthened the statistical findings with respect to general working populations. The sample had adequate power and representation of the constructs under consideration. The findings were statistically substantial in the effect size and the determined relationship between the physical tests and the presence of LBP. Causality, however, cannot be inferred from this study.
Other exercise tests may have similar utility. In choosing the exercise tests, we did not consider exercise dose and specificity for age and gender and these may be confounding factors. However, the statistical findings showed that the exercises chosen were relevant and that neither gender nor age influenced the results.
Other potential limitations were the use of a self-assessed diagnosis as participants were not physically examined and the reported LBP was their interpretation of the area ‘above the buttocks to the region of waist’. Additionally, that participant self-reported gender was the only potential moderator or confounding variable included in the data. As noted above, gender itself was not a significant predictor in any analysis (p>0.80) and thus not included in analyses reported. We were unable to test for a significant interaction between gender and exercises (eg, EX_SUM) due to quasicomplete separation in the data. However, a trend appeared where the effects for males were slightly stronger. This might represent a direction for future research within larger samples or simply a sample artefact.
Determining the exercises indicative of LBP is imperative for diagnosis and setting discharge goals; the next step is to determine which intervention regimen/s could improve the ability to harmoniously perform and maintain the exercises in an optimised and scalable manner. This would require a prospective, longitudinal study with symptomatic/non-symptomatic LBP patients. Challenges in assessing efficacy are test standardisation plus gender variation in repetitions number or degree of movement as males are generally stronger and females more flexible. Furthermore, all measurements at baseline and follow-up must be accurate and sensitive. Consequently, a combination of physical tests and patient-reported outcomes are needed, where many currently preferred tools may not be sufficiently sensitive.27
Furthermore, this study had limited demographic variables. Consequently, future research may consider moderating factors aside from gender. Perhaps age is a differential consideration. However, the very strong analyses effects observed and that our lack of explicitly modelling these hidden variables would have biased the results towards the null, it is unlikely that unobserved variables are true confounders but might clarify and increase the sensitivity of some effects if modelled. As an observational study, however, it was not possible to indicate whether gradually training individuals to complete these five exercises could facilitate reductions in LBP. From the authors’ clinical management protocol, it may be speculated that this appears possible.
In a group of 422 predominantly male, Australian Council workers presenting in a mixed general working population, the ability to complete or not-complete five simple functional exercises showed a significant and meaningful clinical correlation with the presence or absence of LBP. Those able to perform more exercises were significantly less likely to report the presence of LBP, either sometimes or most of the time. Conversely, those unable to perform any or one exercise were more likely to report the presence of LBP most of the time. These findings could be useful for diagnostic purposes, and we hypothesised that training pain-free individuals to be able to complete the five exercises on a regular basis could facilitate prevention of LBP in a general working population. Furthermore, that a graded introduction of these or similar exercises as part of a supervised rehabilitation programme, for individuals recovering from an episode of LBP, may facilitate overall recovery and reduce recurrence. A prospective trial to investigate this hypothesis is to be initiated.
The authors wish to thank the participants and the management for their contributions and for enabling the programme and the recording of the findings to be completed and statistically analysed.
Contributors CPG provided the concept and design of the study, acquisition of participants and data, assisted data analysis and interpretation, drafting of manuscript and critical revising. HRM and JH provided drafting and critical revising of manuscript. JO provided manuscript design, participants and data analysis and interpretation, drafting of manuscript and critical revising. LL provided data interpretation, drafting and critical revising of manuscript. MM provided concept and design input, assisted data analysis and interpretation, drafting of manuscript and critical revising.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent Not required.
Ethics approval Ethics approval was from the Educational Committee of the Sunshine Coast Council with data collected under James Cook University, H1673 , and the University of the Sunshine Coast HREC:S04/48/MC and HREC:08/10.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Extra data can be accessed via the Dryad data repository at http://datadryad.org/ with the doi: 10.5061/dryad.9g8q52g.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.