Introduction

Chronic low back pain (cLBP) is one of the global leading causes of disability and healthcare expenditure1,2,3. First-ever episodes of LBP have an incidence of 15%, and 80% of subjects experience at least one episode of activity-limiting LBP within one year4. The prevalence of cLBP is increasing not only because of population ageing and obesity but also as a consequence of psychosocial and economic strains5,6,7. Thus, considerable efforts have been put in place to identify the most effective way to manage this condition8,9,10,11. Recent guidelines suggest non-pharmacologic treatment as first-line therapy, accompanied by pharmacologic management when symptoms cannot be sufficiently controlled12,13,14.

Physiotherapy has emerged as an effective and non-invasive approach for the management of cLBP, with the goal to improve pain and disability by acting on muscular strength and flexibility, range of motion, and muscular imbalance15,16,17. Furthermore, education and lifestyle modifications aim to provide patients with the tools to prevent future episodes of cLBP18,19,20,21. Different physiotherapeutic regimes have been developed and investigated in this setting22,23. In particular, different forms of exercise, manual therapy, physical agent modalities, and education, or a combination of these in a multidisciplinary approach have been efficiently applied in the setting of cLBP24,25. Available guidelines also highlight a discrepancy regarding the most effective physiotherapeutic management, and clear directions in this respect are lacking13,26,27. The available literature has focused on one particular type of physiotherapy at a time or has directly compared a limited number of similar approaches28,29. The lack of a widely accepted classification of the different physiotherapeutic management options has obviously made direct comparisons difficult. In particular, available classifications have failed to group physiotherapeutic approaches in a way that would allow to highlight possible outcome differences in terms of pain management and function improvement30,31.

This investigation compared the efficacy of the different physiotherapeutic and non-conventional approaches in the setting of nonspecific and/or mechanic cLBP. A Bayesian network meta-analysis of level I studies was conducted for this purpose.

Methods

Eligibility criteria

All the randomised controlled trials (RCTs) which compared the efficacy of conventional and non-conventional physiotherapy programs in patients with cLBP were accessed. According to the authors´ language capabilities, articles in English, German, Italian, French, and Spanish were eligible. Only RCTs with level I of evidence, according to the Oxford Centre of Evidence-Based Medicine32, were considered. Reviews, opinions, letters, and editorials were not considered. Animals, in vitro, biomechanics, computational, and cadaveric studies were not eligible. Studies reporting on non-specific33 or mechanical34, cLPB were included. The pain was defined as chronic when symptoms persisted for a minimum of three months7. Studies including patients with radiculopathy and/or neurologic symptoms were excluded from this analysis. Only studies which analysed patient-reported outcome measures (PROMs) were considered. Missing quantitative data under the outcomes of interest warranted the exclusion of the study.

Search strategy

This study was conducted according to the 2015 PRISMA Extension Statement for Reporting of Systematic Reviews Incorporating Network Meta-Analyses of Health Care Interventions35. The following algorithm was established:

  • P (Problem): cLBP;

  • I (Intervention): Physiotherapy;

  • C (Comparison): different modalities of physiotherapy;

  • O (Outcomes): pain and disability.

In January 2024, the following databases were accessed: PubMed, Web of Science, and Embase. No time constraint was set for the search. The search was restricted to only RCTs. The medical subject headings (MeSH) used in PubMed are shown in the appendix. No additional filters were used in the database search.

Selection and data collection

Two authors (A.K., L.S.) performed the database search. Disagreements were settled by a third author (N.M.) with long experience on systematic reviews. All the resulting titles were screened by hand and, if suitable, the abstract was accessed. If the abstract matched the topic, the full text was accessed. If the full text was not accessible or available, the article was not considered for inclusion. A cross reference of the bibliography of the full text was also conducted to identify additional studies. All pdf of full texts were saved in a dedicated folder shared between the authors in a private cloud. Duplicates were deleted. Study selection and collection lasted three months and the search was updated at each revision phase (last update January, 28 2024).

Data categorisation

Categorization was carried out by three authors (M.N., B.M., F.C.) assessing therapeutic interventions reported in the articles identified. Two independent authors involved in Physical and Rehabilitation Medicine (PRM) used their expertise and referred to recent guidelines and/or systematic reviews regarding the topic of cLBP re-educational techniques to divide treatment protocols into 11 categories: Therapeutic Exercise (TE), Adapted Physical Exercise (APE), Adaptive Training Exercise/Complementary Medicine (CM), Manual Therapy (MT), Physical Agent modalities (PA), Education, Cognitive Re-education (CR), Multidisciplinarity, Kinesiotaping (KT), Sham Therapy (ST), No Intervention. It is important to highlight that most of these categories (TE, APE, MT, PA, Education, CR, Multidisciplinarity, KT and ST) were considered as physiotherapeutic approaches performed by a physiotherapist. Physiotherapy “is services provided by physiotherapists to individuals and populations to develop, maintain and restore maximum movement and functional ability throughout the lifespan. The service is provided in circumstances where movement and function are threatened by ageing, injury, pain, diseases, disorders, conditions and/or environmental factors and with the understanding that functional movement is central to what it means to be healthy36. Instead, Adaptive Training Exercise/Complementary Medicine are usually performed by professionals different from the physiotherapist”. We decided to include the RCTs focused on these techniques because the results (in terms of improvement of the LBP) have been widely demonstrated in the published peer-reviewed literature. The first step was to consider interventions regarding exercise, which can be defined as "a series of specific movements with the aim of training or developing the body by a routine practice or as physical training to promote good physical health"36. Many different types of treatments can fall under the term exercise therapy (ET), each with its own design, duration, frequency, intensity, and mode of delivery. ET aims to increase muscle strength and function, to improve joint range of motion, and consequently reduce pain and increase mobility29. A key distinction has to be made between TE and APE. The former involves movement prescribed to correct impairments, restore muscular and skeletal function, and/or maintain a state of well-being, while APE involves exercise adaptations that could facilitate physical activity across a wide range of disabling conditions37. When LBP is caused by suboptimal postures that place excessive or damaging loads upon the spine APE is applied through postural techniques such as McKenzie, Souchard, or Pilates. In addition, active and passive movements can be differentiated according to the degree of activity expressed by the patient in performing the exercise. Another distinction involved MT: spinal manipulation differs from mobilisation because it is performed through the application of high-velocity impulses and thrusts administered beyond the normal joints’ range of motion (ROM), sometimes producing audible sounds. Physical agents are sources of energy that can be applied on the body surface with therapeutic purposes to improve the quality of life of the patient. They include heat, electrical current, vibration, laser, and ultrasounds, all of which are widely used for the treatment of chronic low back pain38. Various techniques derived from Eastern Medicine, such as Shiatsu, Tai-Chi, Qi Gong, and Yoga have been included in the Complementary Medicine category. The educational category consists of studies in which the main techniques were advice to the patients and the Back School, a technique developed in Sweden in 1969 consisting of patient education and exercises aimed at optimizing functional recovery. Another category became necessary for CR, a technique widely used in neurological disorders; CR can be effectively applied to cLBP to help patients become more aware of their condition and their pain, improve confidence to engage with normal activities of daily living, and reach their life goals and ultimately engage in a healthy lifestyle39. A final category involving a purely re-educational intervention is that regarding KT, a technique that uses of a thin functional elastic bandage applied to the patient's skin with the goal to reduce pain and increase blood flow and muscle performance while reducing muscle stiffness40. Multidisciplinarity was used when two or more techniques were used at the same time without one of them being predominant. Lastly, two more self-explanatory categories were needed to completely divide screened papers: Sham Therapy (ST) and No Intervention.

Data items

Two authors (A.K., L.S.) independently performed data extraction. The following data at baseline were extracted: author and year of publication, journal of publication, men:women ratio, number of patients included with related mean age and BMI (kg/m2), mean length of symptoms duration prior to the physiotherapy, and the length of the follow-up. Data concerning the following patient-reported outcome measures (PROMs) were collected at baseline and at last follow-up: Visual Analog Scale (VAS) or numeric rating scale (NRS), Roland Morris Disability Questionnaire (RMQ)41 and Oswestry Disability Index (ODI)42. As VAS and NRS showed a high correlation, these were used interchangeably for the purpose of the present work43. Data were extracted in Microsoft Office Excel version 16.72 (Microsoft Corporation, Redmond, USA).

Assessment of the risk of bias and quality of the recommendations

The risk of bias was evaluated in accordance with the guidelines in the Cochrane Handbook for Systematic Reviews of Interventions44. Two reviewers (A.K. and L.S.) evaluated the risk of bias in the extracted studies independently. Disagreements were solved by a third senior author (N.M.). RCTs were evaluated using the risk of bias of the software Review Manager 5.3 (The Nordic Cochrane Collaboration, Copenhagen). The following endpoints were evaluated: selection, detection, performance, attrition, reporting, and other biases.

Synthesis methods

The statistical analyses were performed by the main author (F.M.) following the recommendations of the Cochrane Handbook for Systematic Reviews of Interventions45. Cohen’s Kappa (K) was used to quantify the inter-rater agreement among authors for full-text selection. The IBM SPSS version 25 was used. Cohen’s K was interpreted according to Altman’s definition46: K <0.2: poor, 0.2< K <0.4: fair, 0.41< K <0.60: moderate, 0.61< K <0.80: good, and K >0.81 excellent. For descriptive statistics, IBM SPSS version 25 was used. The mean and standard deviation were used. To assess baseline comparability, data distribution was analysed using the Shapiro-Wilk test. Analysis of variance (ANOVA) and the Kruskal-Wallis test were used for parametric and non-parametric data, with P values > 0.1 considered satisfactory. The network meta-analyses were performed using STATA SoftwareMP (version 14; StataCorporation, College Station, Texas, USA). The network meta-analyses were performed through the STATA routine for Bayesian hierarchical random-effects model analysis using the inverse variance method. The standardized mean difference (STD) was used for continuous data. The overall inconsistency was evaluated through the equation for global linearity via the Wald test. If PWald > 0.1, the null hypothesis could not be rejected, and the consistency assumption is accepted at the overall level of each treatment. Both confidence (CI) and percentile (PrI) intervals were set at 95% in each interval plot. Edge plots were performed to display direct and indirect comparisons and respective statistical weights. Interval plots were performed to rank treatments according to their estimated effect size. The funnel plots were performed to investigate the risk of bias related to each comparison. Greater plot asymmetries are associated with greater data variability, which indicates a greater risk of bias.

Ethical approval

This study complies with ethical standards.

Results

Study selection

2354 RCTs were retrieved. A total of 1156 studies were excluded because they were duplicates. Another 1006 articles did not fulfil the eligibility criteria and were therefore discarded. Reasons for non-inclusion include in detail: study design (N = 697), low level of evidence (N = 148), therapy protocols that could not be classified into one of the 11 therapeutic categories of interest (TE, APE, CM, MT, PA, CR, KT, ST, Education, Multidisciplinarity, or No Intervention) (N = 149), and language limitations (N = 12). After full-text evaluation, an additional 42 investigations were excluded because quantitative data on the outcomes of interest were not available. Finally, 150 RCTs were available for inclusion. The inter-examiner agreement between the authors was good (Cohen's K = 0.71) for full-text selection. The results of the literature search are shown in Figure 1.

Figure 1
figure 1

PRISMA flow chart of the literature search.

Full size image

Risk of bias assessment

The analysis of the risk of bias showed a low risk of selection bias because all included studies were RCTs. The allocation of patients to each treatment group was performed with a high degree of quality in most studies, resulting in a low to moderate risk of allocation bias. Moderate risk was present for the risk of detection and performance bias, which was attributed to the lack of information on the blinding of investigators and patients during treatment and follow-up. In some studies, information on study dropouts during study enrollment or analysis was incompletely reported, resulting in moderate attrition bias. The risk of reporting bias was found to be overwhelmingly moderate, and the risk of other biases was mostly low. In summary, the risk of bias graph indicates a moderate quality methodological assessment of RCTs (Figure 2).

Figure 2
figure 2

Cochrane risk of bias tool.

Full size image

Study characteristics and results of individual studies

Data from 12,773 patients were collected. The mean symptom duration was 61.2 ± 51.0 months and the mean follow-up was 4.3 ± 5.9 months. The mean age was 44.5 ± 9.4 years. The mean BMI was 25.8 ± 2.9 kg/m2. The generalities and demographics of the included studies are shown in Table 1.

Table 1 Generalities and patient baseline of the included studies.
Full size table

Pain

The Adapted Physical Exercise group evidenced the lowest pain score (SMD −1.61; 95% CI −5.48 to 2.27), followed by Multidisciplinary (SMD 1.30; 95% CI −2.08 to 4.67) and Adapted Training Exercise/Complementary Medicine (SMD 1.64; 95% CI −1.30 to 4.59). The equation for global linearity found no statistically significant inconsistency (PWald = 0.1). These results are shown in Figure 3.

Figure 3
figure 3

From left to right: edge, interval, and funnel plot of the comparison pain.

Full size image

RMQ

The Adapted Physical Exercise group evidenced the lowest RMQ score (SMD −4.58; 95% CI −18.78 to 9.62) followed by Therapeutic Exercises (SMD −1.07; 95% CI −15.25 to 13.12) and Multidisciplinary (SMD 0.66; 95% CI −11.53 to 12.85). The equation for global linearity found no statistically significant inconsistency (PWald = 0.2). These results are shown in Figure 4.

Figure 4
figure 4

From left to right: edge, interval, and funnel plot of the comparison RMQ.

Full size image

ODI

The Multidisciplinary group evidenced the lowest ODI score (SMD 6.59; 95% CI −10.29 to 23.47), followed by Adapted Physical Exercise (SMD 11.49; 95% CI −12.65 to 35.62) and Physical Agent modalities (SMD 13.29; 95% CI −9.63 to 36.21). The equation for global linearity found no statistically significant inconsistency (PWald = 0.08). These results are shown in Figure 5.

Figure 5
figure 5

From left to right: edge, interval, and funnel plot of the comparison ODI.

Full size image

Discussion

Within the considered physiotherapeutic and non-conventional approaches to manage nonspecific and/or mechanic cLBP, adapted physical exercise, physical agent modalities, and a multidisciplinary approach seemt to represent the most effective strategy in reducing pain and disability.

One of the main difficulties in comparing different types of physiotherapeutic management in cLBP is the lack of a comprehensive and widely accepted classification of the various available therapies. The present work is based on a novel, expert-based classification of the different types of physiotherapeutic and non-conventional approaches available for the management of cLBP. While different classifications have been proposed over time, none has been able to successfully highlight the different effectiveness of each kind of management in terms of disability and pain levels30,31. As opposed to the previously published works, the presented classification was able not only to include all the treatments available in the current literature but also to differentiate between the efficacy of different types of management. Hopefully, this classification will simplify comparisons between different types of regimens.

APE showed to be one of the most efficient physiotherapeutic strategy, and it is also one of the most investigated commonly management option in the literature. The results of the present work contrast with those of a recent network meta-analysis (NMA) that compared different types of exercise and physiotherapeutic management in the setting of cLBP196. While there is agreement that PE and MT are less effective than active therapy options, Owen et al.196 reported no-to-low evidence for the efficacy of Pilates and McKenzie regimens for the management of cLBP. Both therapeutic options fall in the same APE category in the present work. This allowed to aggregate data from different studies and achieve a higher numerosity for the analysed category. In turn, this might have led to stronger evidence supporting APE in the present work. In support of the role of APE in the setting of cLBP, a recent NMA by Fernandez-Rodriguez et al.28 showed that the most effective treatment protocol included, among others, at least one session of Pilates or strength exercise per week. Similar results were also obtained by Hayden et al.197, who compared APE schemes to other exercise and treatment types, and concluded that Pilates and McKenzie regimens promoted functional restoration and reduced pain intensity.

Recently, APE has gained popularity for the management of cLBP, and its use has been supported by a number of publications198,199,200,201,202,203. In addition to its efficacy, APE presents further advantages such as the possibility of individualizing the therapeutic regimen according to the specific needs and interests of the patients204,205. These characteristics can increase compliance with the management197 and, consequently, its efficacy. Furthermore, APE protocols have been applied safely in elderly and fragile cLBP patients, a particularly relevant group considering population aging205. In this setting, APE seems to be able not only to improve pain and function but also to reduce the fear of falling and increase balance205. Interestingly, while improving symptoms and function, APE does not seem to increase trunk muscle size55. This finding might be related to the short duration of the study (eight weeks)55, but might also indicate that the efficacy of APE does not only rely on muscule size. This, in turn, might explain why APE was more effective than other forms of exercise. Possible intervening mechanisms might be the focus of APE on functional improvement or balance, or the encouraging effects of APE on psychosocial outcomes206 and improvement of kinesiophobia207,208: further studies will be required to understand more clearly why this type of management is particularly effective in patients with cLBP.

This important finding can be explained considering that active physiotherapy involves the active participation of the patient in performing therapeutic exercises or activities that promote mobility, strength, and functional improvement17. It encourages patients to actively participate in their rehabilitation, fostering self-management and independence17. This translates into a greater awareness of patients of their means, in adapting their body to the surrounding environment. Patient do not feel that they have a disability that limits the activities of daily living, but, thanks to the Adapted Physical Exercise, subjects develop the means to differently tackle the required tasks.

The application of physical agents also proved to be an effective strategy for the management of cLBP. Passive physiotherapy refers to interventions where the patient receives treatment without actively engaging in physical movements, as happens during the application of the physical agents. It relies on external therapeutic interventions facilitated by the physiotherapist on the affected muscles, which often appear hypercontracted in case of pain. Passive stretch reduces stiffness (viscoelastic stress relaxation) and decreases stretch-induced pain16. This could represent the first step to consequently work on the functional use of these muscles, as it happens in APE. In other terms, passive treatment can help with immediate pain relief, but active treatment keeps the patient functional in the long term.

Lastly, considering the weight of psychosocial factors in the setting of cLBP209, it is not surprising that multimodal therapy was effective under the outcomes of interest considered. Furthermore, the available evidence supports the hypothesis that multimodal management exerts a positive influence in return to work210 and reduction of work absenteeism211. Heitz et al.212 identified several modifiable and non-modifiable risk factors for the development of persistent cLBP in patients with subacute and cLBP, 56 of them somatic and 61 of them psychosocial. These figures show clearly that focussing solely on the somatic aspects leaves out a vast number of psychological factors involved in the development of cLBP. These data and the evidence presented in the present work thus support the inclusion of psychologic management in the therapy of nonspecific cLBP. While similar positive findings around the employment of multimodal management in cLBP have been reported by different studies213,214,215,216,217, future research should focus on what type of psychological therapy is best used in what type of setting215.

This work does not come without limitations. The main one is represented by the heterogeneity in the inclusion criteria and therapeutic schemes in the available literature. Future studies should focus on adopting a uniform classification of different therapeutic options to allow easier comparability, and larger cohorts with sub-analysis of patients in different age ranges or with different symptom durations will be helpful to analyze whether different patient cohorts can benefit from different management options. Three trained physical therapists (M.N., B.M., F.C.) collectively performed data categorisation to reduce the risk of bias related to data classification. However, they often faced bias and lack of information and needed further clarifications from the authors of the included studies. The inter-rater agreement was not evaluated during the literature search, which also might impact negatively the quality of the results of the present Baysiean network meta-analysis.

Conclusion

Within the considered physiotherapeutic and non-conventional approaches to manage nonspecific and/or mechanic cLBP, adapted physical exercise, physical agent modalities, and a multidisciplinary approach might represent the most effective strategy in reducing pain and disability.