Objective To summarise the evidence for non-pharmacological management of low back pain (LBP) in athletes, a common problem in sport that can negatively impact performance and contribute to early retirement.
Data sources Five databases (EMBASE, Medline, CINAHL, Web of Science, Scopus) were searched from inception to September 2020. The main outcomes of interest were pain, disability and return to sport (RTS).
Results Among 1629 references, 14 randomised controlled trials (RCTs) involving 541 athletes were included. The trials had biases across multiple domains including performance, attrition and reporting. Treatments included exercise, biomechanical modifications and manual therapy. There were no trials evaluating the efficacy of surgery or injections. Exercise was the most frequently investigated treatment; no RTS data were reported for any exercise intervention. There was a reduction in pain and disability reported after all treatments.
Conclusions While several treatments for LBP in athletes improved pain and function, it was unclear what the most effective treatments were, and for whom. Exercise approaches generally reduced pain and improved function in athletes with LBP, but the effect on RTS is unknown. No conclusions regarding the value of manual therapy (massage, spinal manipulation) or biomechanical modifications alone could be drawn because of insufficient evidence. High-quality RCTs are urgently needed to determine the effect of commonly used interventions in treating LBP in athletes.
- lower back
Statistics from Altmetric.com
The prevalence of low back pain (LBP) is highly variable across different sports (18%–65%)1 yet is consistently reported as a common reason for activity avoidance. While LBP is influenced by a dynamic interaction among biological, psychological and social factors,2–4 prevention and management strategies are likely to be different between athletes and the general population. In athletes, a previous episode of LBP, high training volume, periods of load increase and years of exposure increase the risk of LBP, impair performance and may contribute to early retirement.5 Although some risk factors for LBP in the general population may overlap (eg, cumulative load exposure in manual workers,6 socioeconomic status), most do not (eg, decreased levels of physical activity, smoking, obesity).7
Athletes are performance-focused and often willing to do whatever it takes, even sacrificing their health, to win. This dedication can mean athletes are vulnerable to making uninformed choices about their care8 9 and/or full symptom resolution before returning to sport (RTS)10 may be expected. A biomedical approach to treatment is routinely offered, with a focus on imaging, passive therapies, pharmacological treatments, spinal injections and surgical interventions—approaches associated with high costs, insufficient evidence of effectiveness and potential increase in disability.3 8 11 12
The IOC recommended that a treatment strategy for pain in elite athletes should (1) address all contributors to pain and (2) employ therapies providing maximum benefit and minimal harm.3 However, it is unclear if current management of LBP in sport aligns with these recommendations. Understanding what constitutes best practice for managing LBP in sport will help athletes and clinicians make informed decisions.
We aimed to conduct a systematic review to answer the question “What is the evidence for common non-pharmacological treatments for managing LBP in athletes?”
Definition of LBP
We use the term ‘low back pain’ to recognise that 90% of the condition cannot be attributed to a specific cause.13 The term ‘low back injury’ was used in some studies included in our review. We defined LBP as pain localised below the costal margin and above the inferior gluteal folds, with or without leg involvement.14
We followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses15 recommendations when reporting this prospectively registered systematic review (PROSPERO database identifier: CRD42018087229).
Eligible studies met the following inclusion criteria: (1) English language; (2) published from database inception to September 2020; (3) investigated any treatment for LBP in athletes aged ≥18 years; (4) study design was randomised controlled trial (RCT).
We excluded the following: (1) guidelines, letters, editorials, commentaries, unpublished manuscripts, dissertations, government reports, books or book chapters, conference proceedings, meeting abstracts, lectures and addresses, consensus development statements or guideline statements; (2) qualitative studies, systematic reviews, pilot studies aimed at demonstrating the feasibility of an RCT; (3) non-randomised controlled trials and observational studies.
Eligibility assessment was performed independently by two reviewers (FW and JT) using Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia, available at www.covidence.org). Studies were excluded based on title and abstract and the reason for exclusion was recorded. The reviewers independently examined the full text of the remaining studies to determine inclusion, resolving disagreements by discussion. If consensus was not reached, a third reviewer resolved any disagreements.
We searched the Medline, Embase, CINAHL, Web of Science and Scopus electronic databases from inception to September 2020. All search strategies were completed by an experienced librarian (DM) and are summarised in online supplemental material (see online supplemental file). The reference lists of selected articles were screened for other relevant articles. Grey literature searching included reference lists of included studies and conference proceedings of the following organisations: American College of Sports Medicine from 2011 to 2020, American Physical Therapy Association from 2012 to 2020 and World Confederation for Physical Therapy from 2011 to 2020.
Using a standardised data extraction form, the following were extracted from all included studies by two authors (FW and S-JM): author names, year of publication, country, number of participants, sex and age of participants, type of sport, definition of LBP, treatment and control interventions, outcome measures, duration of follow-up, results and overall findings. The data extraction form was pilot tested, and discrepancies in data extraction were resolved by consensus.
The primary outcomes of interest were pain, disability and RTS. Secondary outcomes included quality of life and functional outcomes, such as trunk muscle function.
Risk of bias assessment in individual studies
Two reviewers (FW and S-JM) independently assessed the risk of bias in included RCTs using the Cochrane Risk of Bias tool,16 which assesses six bias domains: sequence generation; allocation concealment; blinding of participants and personnel, blinding of outcome assessment; incomplete outcome data; selective outcome reporting; other sources of bias (eg, clarity of diagnosis, compliance to intervention, features of study design). The risk of bias in each domain, and overall risk of bias for each outcome within a study, were judged as ‘low’, ‘high’ or ‘unclear’ risk.16 Trials lacking comprehensive information regarding outcome assessment were judged as unclear risk, and all others low risk, for the domain of detection bias.
For all included trials, we present study and participant characteristics and results in online supplemental tables 1 and 2. A meta-analysis was performed where possible. This was performed, for exercise, for the Oswestry Disability Index (ODI) and Visual Analogue Scale (VAS) using follow-up scores. Data from comparable groups in RCTs were pooled using a random-effects model, considering the extent of heterogeneity of effect size across pooled studies. If SDs were missing for continuous data, other measures of variability (eg, 95% CI, SEs, T values, p values and F values) that allowed for the calculation of SD using Review Manager (RevMan) V.5.3 (The Nordic Cochrane Centre, The Cochrane Collaboration, 2014) calculator tool were used. Some studies were not included in a meta-analysis for common outcomes. This was because of insufficient reporting of data such as missing measures of variability (SD or CI) which could not be provided by authors on request, or due to heterogeneity in the interventions such as inclusion of co-interventions in either or both arms of the study.
Clinical heterogeneity was assessed subjectively based on information about participants, interventions and outcome measurements of each study. Heterogeneity was further assessed by visual inspection of effect estimate similarities, the overlap of the CIs on the forest plots and consideration of the statistical test output of the χ² test (p<0.01 was interpreted as significant heterogeneity) and the I² statistic (I² of 75% to 100% was interpreted as considerable heterogeneity).16 Where heterogeneity was high, a sensitivity analysis was conducted to explore the influence of individual RCTs by sequentially removing trials from analysis to explore individual effects. Where it was inappropriate to pool data, data were presented in tables and qualitatively synthesised.
We accessed 1605 studies (title and abstract) following deletion of duplicates (n=24). After screening, 47 full-text articles were obtained for further review, excluding 33 articles and including 14 articles for full review17–30 (figure 1).
Trials were conducted across 10 countries (1 low, 2 upper-middle and 7 high income) and involved participants from 12 sports. Interventions included exercise (seven studies), biomechanical modifications (two studies) and manual therapy (five studies) (online supplemental table 1).
There were 541 participants (363 (67%) men; 108 (20%) women; sex was not reported for 70 (13%) athletes) with a mean age of 30.1 years (range 15–72 years; online supplemental table 1). The athletes’ competitive level ranged from recreational to international. The smallest trial included 16 participants; the largest trial included 110 participants; one trial did not report the number of participants.19 Length of follow-up ranged from 24 hours to 6 months.
The two most common outcomes were pain measured with VAS18–20 22–28 (n=9 trials) and disability measured with ODI18 22 24 25 30 (n=5). Other outcomes included muscle strength19 20 22 30(n=4), endurance25 (n=1) or cross-sectional area21 (n=1); quality of life measured with the 36-item Short Form Health Survey (SF-36) score22 (n=1); pain measured with the McGill Pain Questionnaire20 26 (n=2) or pressure pain threshold23 (n=1); disability measured with the Roland Morris Disability Questionnaire (RMDQ)17 (n=1); functional parameters such as distance walked, 1 min stand-ups, stair climb counts or golf swing biomechanics,27 28 mobility18 19 (n=2) or balance18 (n=1); physiological parameters such as tissue blood flow23 (n=1); and player availability21 (n=1).
Risk of bias assessment
Risk of bias16 assessment for items was frequently ‘unclear’, reflecting poor reporting in studies (figure 2 and online supplemental figure 1). In nine trials, it was deemed impossible to blind participants to an intervention, or no attempt was made to do so; these studies were at high risk for performance bias. Two trials17 19 did not report participant numbers in the results and were at high risk of attrition bias. Ten trials were at unclear risk of reporting bias. Two trials19 28 were at high risk for reporting bias.
Seven trials17 18 21 22 24 27 30 investigated various forms of exercise. Exercise interventions included core stability training, Swiss ball, Thera-band, unstable surfaces, periodised resistance training, Pilates and backward or ‘retro’ walking. Outcomes related to pain and disability most commonly reported were VAS18 22 24 27 and ODI.18 22 24 30
VAS was used to assess LBP in four trials.18 22 24 27 Three of these studies were included in meta-analysis; one did not provide sufficient data.18 Overall, the total pain reduction for exercise was 1.65 points, 95% CI −2.74 to −0.55, p=0.003, I2=91% (VAS mean difference) compared with control interventions (usual care or another form of exercise). Two out of four trials compared one exercise mode to another, although in all cases, pain improved with either intervention. Removing Kumar et al 27 from the meta-analysis reduced I2 to 10% and the mean difference to 1.04 points (95% CI −1.04 to −0.67) (figure 3A).
Core stabilisation exercise (final score 2.4/10) was superior to conventional strengthening exercise (final score 3.3/10) for improving pain in cricket players (p=0.002),24 although both interventions improved pain. Compared with a basic lumbar extensor strength programme, a dynamic stabilisation training programme resulted in greater improvement in VAS (2.1 times more) and greater rate of improvement of other outcome measures in hockey players.27 When backward walking was added to a conventional exercise programme (body-weight trunk resistance exercises), it resulted in a significant (p=0.003) improvement of pain to 1/10. Conventional exercise resulted in an improvement to 3/10.18
Back-related disability was reported in five trials using the ODI18 22 24 30 and the RMDQ.17 Four of these studies were included in a meta-analysis; one did not provide sufficient data.18 Overall, resistance exercise significantly reduced disability (disability (ODI and RMDQ) standardised mean difference −2.6 points (95% CI −5.14 to −0.04)) compared with control. Heterogeneity was very high (I2=96%) (figure 3B) and there was no significant reduction in this on sensitivity analysis.
Core stabilisation exercise (final ODI 19.8±4.9) was superior to conventional strength exercise (final ODI 22.2±5.4) for improving disability in cricket players (p<0.001) with improvement with both exercise types.24 Resistance exercise (Thera-band) (final ODI 17.2±1.78) was superior to Swiss ball exercise (final ODI 18±0.8) in hockey players (p<0.001), although both types of exercise training resulted in improvements.30 A periodised resistance training programme was compared by Jackson et al 22 to a group who continued with ‘regular recreational activity only’ (no resistance training allowed) and resulted in significant improvements in disability in middle-aged athletes.
In judo athletes, a trunk exercise programme on either a stable or unstable surface (Swiss ball) demonstrated a significant decrease in RMDQ (total score=24) (final score 4.9±3.0 for the unstable surface group and 3.2±3.0 in the stable surface group) (p<0.001), with no significant between-group differences.17
Outcomes related to quality of life (QoL) (measured with SF-36) were also reported,22 and there were significant improvements in the mental and physical composite scores in the SF-36 in the intervention group of periodised resistance training compared (mental final score 57.3±5.4 and physical final score 55.4±5.1 in middle-aged athletes).
An ‘unstable shoe’ intervention in golfers reduced LBP in a laboratory test to 21.97 mm on a 100 mm continuous VAS compared with 37.83 mm in those not wearing the shoe (perceived LBP between groups t=−2.02, p=0.05). There was, however, no difference in LBP on the golf course.28 Recreational cyclists with LBP while cycling adjusted their bike seats to achieve anterior angles of inclination of 10 or 15 degrees.29 At 6-month follow-up, 72% reported they had no LBP, and 20% reported a ‘major reduction’ in pain. The authors did not report separate groups’ results, only that there were no between-group differences.
Five trials19 20 23 25 26 (n=187, one trial did not report sample size) investigated the effects of manual therapy on LBP. There were short-term beneficial effects of massage19 23 and spinal manipulation.20 25 Pain and disability outcomes were measured using VAS19 20 23 and ODI.25 Most trials included co-interventions.
Chinese massage was associated with a reduction in pain of 1.04 points (VAS 0–10) after 4 weeks (p<0.001) compared with ‘simple massage therapy’ (between-group difference −0.64, 95% CI −1.04 to −0.24; both improved pain).26 Female weightlifters with LBP reported significant improvement in pain (VAS 0–10) with sport massage therapy (final score 2.62±1.31) and lumbopelvic stability training (final score 3.18±1.64).23 Sport massage therapy had greater therapeutic benefits than lumbopelvic stability training (p=0.001)23 compared with pre-treatment and between treatment periods in a randomised cross-over trial. Massage plus application of ‘thermal magnets’ improved pain, strength and mobility in 96.5% of an unreported number of athletes with LBP compared with 87.4% who received ‘conventional’ treatment (not described).19 These studies were not combined in a meta-analysis because of the presence and heterogeneity of co-interventions in each study.
In male collegiate athletes with acute LBP, spinal manipulations combined with icing and stretching improved pain by an average of 2 points (VAS 0–10) 24 hours after one treatment session,20 compared with lying in a prone position of comfort for the same duration of treatment. Collegiate athletes with chronic LBP had a single spinal manipulation with or without a 24-hour application of Kinesio Tape, and both groups improved similarly in pain and function (p<0.001).25 These studies were not combined in a meta-analysis because of the presence of co-interventions.
Grading of quality of evidence
Due to the very small body of evidence, the heterogeneity of interventions, populations and outcomes, and that this was a systematic review and not a clinical guideline, it was not feasible and would not add value to apply an assessment system.
We identified 14 RCTs dealing with treatment of LBP in athletes, with biases across multiple domains. Exercise was the most frequently investigated intervention. While exercise in general appears to improve pain and function, it is uncertain if exercise interventions result in faster or higher rates of RTS. It is also uncertain which types of exercise are most beneficial and for whom. No conclusions regarding the value of manual therapy (massage, spinal manipulation) or biomechanical modifications alone could be drawn because of insufficient evidence. Meta-analysis was limited because of insufficient reporting in some studies, meaning that estimates of effect should be interpreted with caution due to the small number of studies which were eligible for inclusion. We could not identify any RCTs evaluating the effect of spinal injections or spinal surgery in athletes with LBP.
Darlow and O’Sullivan called for treatment of LBP in athletes to be guided and “governed by strong evidence and rationale”.8 Our review revealed (1) a striking shortage of high-quality studies evaluating the effectiveness of any treatment, and importantly, (2) no RCTs dealing with effectiveness of injections and surgeries, making comparisons and treatment decisions challenging at best. In addition, there is a general lack of data on important outcomes such as return to sport after any intervention.
Previous reviews have focused on unidimensional or patho-anatomical aspects of management of LBP in sport.31–35 To our knowledge, the current review is the first to systematically explore evidence for all types of treatment. Advice to remain active along with education and self-management, considered first-line strategies in the treatment of LBP in the general population,36 were not featured in the trials included in our review. Cognitive-behavioural therapy, considered first-line treatment for chronic LBP in the general population,37 was not mentioned or investigated. Current treatment of LBP in sport does not appear to align with the recommendations outlined within the IOC Consensus statement on pain.3 Evidence-based practice for managing LBP in adults has clear recommendations for acute and chronic management,35 but there is a paucity of literature on return to function and/or RTS.
All exercise approaches reduced pain and improved function in athletes with LBP. Any exercise appears to be better than rest, and targeted, dynamic, functional (sport-specific) exercise may be the most beneficial. Core stability exercises appear to be no more effective than other forms of exercise in the long term (≥12 months), a finding consistent with the literature in the general population.38–40 Consistent activity, with an appropriate rate of increase, may reduce LBP as long as other factors including sleep, mood and interpersonal relationships are being addressed41; however, these factors were not identified as treatment targets in any of the studies reviewed. People with LBP have unique movement patterns identified as provocative42; therefore, tailoring exercise type and dose may promote better results.43
There was insufficient evidence to support biomechanical modifications alone for managing LBP in sport. Concomitant interventions or comparator groups were not described, and it was difficult to distinguish pain relief due to the intervention versus discontinuation of sporting activity itself. Only one trial had a control group,28 with no description of participant characteristics.
Spinal manipulation is an effective adjunct for managing persistent LBP in general44 and is generally recommended in guidelines for both acute and persistent LBP.36 There was, however, insufficient evidence to support manual therapy (massage and spinal manipulation) as stand-alone interventions for managing LBP in athletes. Poor-quality trials suggested there may be short-term benefits of massage with the addition of herbal ointment or thermal magnetic therapy, and benefits from spinal manipulation. Two trials did not report long-term follow-up.20 23
It is unclear which interventions are the most effective treatments for LBP in athletes. Due to their singular focus on top performance, elite athletes may be vulnerable to a ‘risk minimisation’ approach (rest and early advanced imaging), and early use of aggressive treatments (injections and surgery), particularly when substantial financial resources and pressures exist.8 Clinicians have a responsibility to provide guidance on evidence-based treatments rather than what is ‘available’.
Pain is the result of the interaction of multiple biopsychosocial factors, yet current practice for managing LBP in athletes appears to be guided by a biomedical approach. No trials addressed the potential benefit of interventions involving psychological and social aspects of LBP. Athletes desiring ‘a quick fix’ to RTS may be vulnerable to accept treatment options that lack evidence of effectiveness and may carry risk of harm.
Until robust evidence is produced for athlete populations, the most responsible approach is likely to follow recommendations for LBP management in the general population, adopting a biopsychosocial approach and considering the athlete’s specific circumstances. Clinicians and athletes should employ shared decision-making regarding individual treatment goals—factoring in expectations regarding pain, disability, QoL and RTS when possible.
The body of literature for treating LBP in athletes is scarce and of low quality, which limits our ability to draw firm conclusions. Several trials include very small sample sizes, leading to inadequate numbers of observations and the potential for sparse data bias.45 There were no RCTs without biases to assess the effects of any intervention, and as a result, meta-analysis was conducted on those available. Most athletes were male (≥65%), LBP was inadequately defined across studies, exposure to sport was generally not measured and staging of the LBP episode was generally unclear. Confounders such as previous LBP were often not considered and adherence to interventions was generally poorly reported. There was a high risk of bias across studies. Importantly, there was a general lack of long-term follow-up (variable reporting of follow-up, recurrence rates and time to RTS), which makes overestimation of positive outcomes likely.
Differences between protocol and review
Our original question was “What is the optimal management (conservative, medical, surgical) for LBP in sport?” We realised while searching the literature that available evidence would not answer this question adequately, so we changed the query to “What is the evidence for commonly used treatments for managing LBP in athletes?” Given the paucity of RCTs, we initially broadened our search to non-randomised studies. Following suggestions from the peer-review process, we decided to exclude the non-randomised studies from this systematic review. An original secondary outcome was imaging, but imaging outcomes were so infrequently reported that we were unable to include them. We planned a subgroup analysis for rowers, but no RCTs dealing with management of LBP in rowers were identified.
Research implications and future directions
High-quality RCTs that assess the effects of commonly used interventions for treating LBP (with or without leg pain) compared with no treatment and other relevant treatments are urgently needed. Consensus on a core outcomes set for athletes is also required. Elite athletes often require profession-related outcome measures, such as RTS rate, career longevity and performance-based outcomes. We recommend that international athletic organisations collaborate on a structured programme for research into LBP in athletes (eg, forming expert groups and holding international conferences similar to the International Conference on Concussion in Sport.46
While several treatments for LBP in athletes improved pain and function, it was unclear what the most effective treatments are, and for whom. All exercise approaches reduced pain and improved function in athletes with LBP, but the effect on RTS is unknown. There was insufficient evidence to draw any conclusions regarding activity or biomechanical modifications, or manual therapy alone. High-quality RCTs are urgently needed to determine the effect of commonly used interventions in treating LBP in athletes.
What is already known
Low back pain (LBP) is a common problem in sport and can negatively impact performance and ultimately contribute to early athlete retirement.
There are no evidence-based or consensus-based guidelines for treating LBP in athletes. Outside of sport, clinical practice guidelines recommend education, self-management, and physical and psychological therapies for most people with LBP.
What are the new findings
There are very few randomised controlled trials (RCTs) evaluating the effectiveness of commonly used interventions for LBP in athletes.
The few trials that exist are generally of poor quality and have biases in multiple domains.
Exercise was the most frequently evaluated intervention and was generally associated with reduced pain and improved function in athletes, but there were no data on returning to sport.
It is unclear whether some types of exercise are more beneficial than others, and exercise interventions were frequently not clearly defined and described.
There are no RCTs investigating spinal injections or surgery to manage LBP in athletes.
Implications for clinical practice and further research
High-quality RCTs are urgently needed to determine the effectiveness of common interventions for managing LBP in athletes. Until such trials are completed, management of LBP in athletes should follow recommendations from evidence-based guidelines for the general population.
The authors would like to thank Dr Parham Rasoulinejad for his thorough review of this manuscript prior to submission.
Twitter @janesthornton, @clare_ardern, @DrLarissaTrease, @drkateackerman, @fionawilsonf
Contributors Study concept, design and management: JST, FW, AV, KW. Data analysis: JST, FW, S-JM, LT, KW. Data review and interpretation: all authors. Manuscript writing: all authors.
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.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
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.