Aims: To systematically review the evidence for the effectiveness of combining cognitive behavioural therapy (CBT) and exercise versus exercise alone in the management of patients with non-specific chronic low back pain (NSCLBP).

Methods: Electronic search of CINAHL, PUBMED, Sports Discuss, SCOPUS, AMED, MEDLINE, Cochrane Central Register of Controlled Trials, and EMBASE, between 1990 – July 2017; complemented by hand searching of citation lists and citation tracking. Two independent reviewers screened titles and abstracts from the retrieved search results. Studies were considered based on PRISMA guidelines. Data was extracted based on Cochrane the Handbook of Systematic Reviews guidelines; the Cochrane Data Collection Form for Intervention Reviews (RCTs only), was customized and utilized. Risk of bias assessment was undertaken utilizing the Cochrane Back Review Group recommendations employing two independent reviewers. Meta-analysis was used to produce a weighted average for primary outcome measures, namely pain and disability. Results: Four studies were included (n = 406 participants); all studies provided post intervention results on pain and disability. Meta-analysis showed no significant difference between groups for both pain and disability, post intervention in the short term (pain; SMD -0.02, 95% CI -0.23 to 0.19; disability; SMD 0.06, 95% CI -0.15 to 0.27); medium term (pain; SMD -0.01, 95% CI -0.24 to0.22; disability; SMD 0.00, 95% CI -0.23 to 0.23); and long term (pain; SMD 0.06, 95% CI -0.18 to 0.29; disability; SMD -0.06, 95% CI -0.39 to 0.27).

Conclusion: The findings from this review reveal that there is no significant difference between groups; that is, there is moderate level evidence that the addition of CBT to exercise for patients with NSCLBP does not improve pain and disability outcomes in either the short, medium, or long term; however, both interventions (CBT plus exercise and exercise alone) produce favourable outcomes.

Keywords: Behavioural therapy, Chronic low back pain, Exercise therapy, Meta-analysis, Systematic review

Low back pain (LBP) is a leading cause of pain and disability, with an estimated lifetime prevalence of 84% [1]. The condition is classified into three broad categories: non-specific low back pain (NSLBP); severe spinal involvement; and neurological involvement [1],[2]. NSLBP, comprises approximately 90% of all LBP cases and is described as pain with no known alaetiology. Consequently, diverse prognostic and treatment protocols are utilized [3], [4].

Individuals go through rapid amelioration of pain and disability within twelve weeks of inception of LBP, with varying suggestions for percentages; however, a view amongst some authors is that most individuals recover within this period [4]. In spite of this, in a subgroup of individuals with LBP, there is persistence after twelve weeks; pain in this subgroup, is known as chronic LBP [5], [6].

Improvement in quality and safety of health delivery is very essential; to this end, various Clinical Practice Guidelines (CPGs) meant to improve evidence based practice (EBP) are published annually [7]. Psychological therapies are recommended in guidelines for assessment and management of patients with non-specific chronic low back pain (NSCLBP) [4].

Current reviews of CBGs recommend psychological therapies such as cognitive behavioural therapy (CBT) for the management of NSCLBP; exercise therapy also features in recommendations of both national and international guidelines [4]. Furthermore, recent National Institute for Health and Care Excellence (NICE) guidelines also recommends a multi-model approach (e.g., combination of exercise therapy with CBT) for managing NSCLBP [8].

The evidence for the effectiveness of either CBT or exercise in managing NSCLBP has been investigated in recent systematic reviews; a systematic review by Richmond et al., [9], concluded that CBT produces favourable outcomes (small to moderate) for NSCLBP patients. On the other hand, a systematic review by Gomes-Neto et al., [10], also reported beneficial outcomes for NSCLBP patients, with the utilization of exercise therapy interventions.

There is a substantial body of clinical research to support these two interventions [9],[11] although there is a lack of consensus on the most effective form of exercise intervention [10]; however, the added benefits of combining CBT to exercise needs to be investigated.

Nonetheless, a key limitation to prevailing approaches to management is the possibility of addressing the pain and its consequences with multi-model treatment packages that do not address the needs of specific patient needs [4]. Systematic reviews have investigated the effectiveness of exercise therapy or CBT for the management of NSCLBP suggesting favourable outcomes [9], [10]; however, to the best of our knowledge, the comparative effectiveness of CBT plus exercise versus exercise alone is yet to be reviewed. Therefore, this systematic review was to answer the question:

• What is the effectiveness of combining CBT plus exercise versus exercise alone in the management of NSCLBP patients?

Types of studies included
This review included RCTs published in English with a clearly defined randomization process. Included RCTs investigated the comparative effectiveness of CBT plus exercise compared to exercise alone.

Types of participants included
Participants were considered for inclusion if they were classified as having NSCLBP; defined as pain that persists beyond twelve weeks with no associated pathoanatomical mechanism or specific condition [12], [13]. Adult, aged eighteen years and above were included. Participants with associated specific pathoanatomical conditions (e.g., fractures, ankylosing spondylitis, infection, spinal stenosis, radiculopathy), or an association with conditions such as pregnancy were excluded.

Interventions included
Studies defining CBT as a management approach involving a psychological intervention structured on utilizing cognitive and behavioural approaches established from evidenced based practice (EBP) were considered for inclusion in this review [14]. CBT approaches, encompassing models such as behaviour therapy (e.g., operant conditioning) and cognitive therapy, geared towards the promotion of good physical health in patients with NSCLBP were included [15].

RCTs utilizing exercise therapy interventions defined as, specific movements, supervised or recommended by a treating therapist through movement and activities aimed at relieving pain and promoting good physical health in patients with NSCLBP were considered; these included, mobility, flexibility and stability exercises; strengthening and stretching exercises; and aerobic exercises [16].

Outcome measures utilized
Primary outcome measures considered for this review included disability and pain. Assessment tools with good validity and reliability were considered for both primary outcomes; back-specific disability assessment tools such as the Roland Morris Disability Questionnaire (RMDQ) and the Oswestry Disability Index (ODI) were considered. RMDQ, has a good test-retest reliability (r = 0.83 to 0.91) and internal consistency (Cronbach’s α = 0.84 to 0.96) [17] . The ODI, has a good internal consistency (Cronbach’s α = 0.71 to 0.87) and test-retest reliability (r = 0.83 to 0.99) [18].

Assessment tools with good validity and reliability, considered for measuring pain in LBP patients include the Numeric Rating Scale (NRS) and Visual Analogue Scale (VAS). The NRS, has good test-retest reliability (r = 0.96), and has excellent correlation with the VAS (r = 0.90); with VAS also having a good test-retest reliability (r = 0.94) [19], [20].

Secondary outcome measures included; quality of life (QoL), depression, cost-effectiveness, kinesiophobia, and function. Measurement of secondary outcome measures was considered based on the utilization of assessment tools with good reliability and validity; secondary outcome measures included quality of life (QoL), depression, function, cost effectiveness and kinesiophobia.

Identification of studies
Based on relevant mapped terms to subject headings (MeSH) (e.g., Cognitive Behavioural Therapy), search expressions (e.g., behavior therapy, exercise therapy), and keywords (e.g., Exercise), the search was developed; including RCTs published in English between 1990 to July 2017. The search was computer-based, involving eight databases; CINAHL, PUBMED, Sports Discuss, SCOPUS, AMED, MEDLINE, Cochrane Central Register of Controlled Trials, and EMBASE.

Unpublished and grey literature were also considered using similar search terms in the listed databases: www.guidelines.gov, www.opengrey.eu, www.clinicaltrials.com, and www.controlled-trials.com. Finally, a thorough examination of reference lists from retrieved studies was performed per the criteria of inclusion.

Selection process
Two independent (PA and EM) reviewers screened titles and abstracts from the search results. Studies were removed if they did not meet criteria for inclusion. Full texts were assessed to ascertain whether they meet the inclusion criteria. Consensus was used to address any disparities; a third reviewer (PH) was utilized if necessary to address any disparities. Studies were considered based on guidelines from the, Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) group [21].

Data extraction
Data was extracted based on Cochrane Handbook of Systematic Reviews guidelines [22]. Extraction of data was based on major headings such as; Study type, Title of study, Methods, Population (criteria for inclusion and exclusion, mean age, gender), outcome (outcome measures), and notes. For the purpose of extraction, the Cochrane Data Collection Form for Intervention Reviews (RCTs only) was customized and utilized [22]. Extracted literature was critically appraised utilizing the critical appraisal skills program (CASP) checklist for RCTs [23].

Risk of bias assessment in included studies
Risk of bias assessment was undertaken to ascertain the quality of methodology of selected studies utilizing the Cochrane Back Review Group recommendations [24]. Two independent reviewers (PA and EM) conducted this to determine a consensus. The criteria for assessing risk of bias was as follows:

• Random generation of sequence.
• Treatment allocation concealment.
• Outcome reporting selectivity.
• Blinding of patients, outcome assessors, and therapists.
• Insufficiency of outcome data.
Possibility of other sources of bias [16].

Cochrane Back Review Group guidelines contains twelve (12) questions which address these criteria; attrition bias, performance bias, selection bias and detection bias; answers to enquiries were; ‘yes’, ‘no’, and ‘unclear’ [24]. A low risk of bias was assigned to assessed studies if a ‘yes’ answer was obtained for six out of twelve enquiries without major flaws, example being a 20% or higher rate of dropout in a specific study group [24].

Assessing bias in publication
Assessing publication bias was undertaken by two methods namely; publication bias between studies, assessed utilizing funnel plots; and publication bias within studies assessed by examining whether outcomes reported were as expected for each period of follow-up [25]. Attempts were made to contact authors of published studies to provide additional information or clarify, if information provided was not sufficient [26].

Synthesis of data
Summary of findings (SoF) was generated using the grade pro guideline tool for development (GRADEpro GDT) software to determine the quality of findings [22]. Forests plots and meta-analysis were conducted using the review manager (RevMan 5) [27]. Meta-analysis was used to produce a weighted average, increase the power over results of individual studies, and improve the effect size estimates [28]. Due to the inherent heterogeneity in LBP within available literature, the random effects model was used [11].

The measure of primary effect for measures of outcomes was the standardized mean difference (SMD) [28]. Assessment of statistical heterogeneity was undertaken with the use of the I2 statistic [21]. Assessment of the I2 statistic was on a percentage basis:

• High possibility of heterogeneity: 75–100%
• Substantial possibility of heterogeneity: 50–90%
• Moderate possibility of heterogeneity: 30–60%
Substantial heterogeneity not present: 0–40% [22].

Selection and description of studies for this review
Results from the initial search produced 1,458 results (n = 1,458); subsequent removal of duplicates (n = 631), and titles and abstracts screening removed 789 studies (Figure 1)[21]. Thirty-eight (n = 38) full-text articles of potential studies relevant to the review were subsequently retrieved. No extra articles were identified through reference lists of retrieved articles.

Thirteen (n = 13) articles were excluded because they were not RCTs [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41]. Four articles (n = 4) were also excluded based on the fact that they were systematic reviews [42], [43], [44], [45]. A further ten (n = 10) articles were excluded because the inclusion criteria did not include patients with NSCLBP [46], [47], [48], [49], [50], [51], [52], [53], [54], [55]. Six (n = 6) RCTs were also excluded based on the fact that they did not answer the specific question under review [56], [57], [58], [59], [60], [61].

Included studies
Four studies met the inclusion criteria, thus were included for review [62], [63], [64], [65]. The four studies included a total of four hundred and six participants (n = 406). Mean ages with standard deviations (SD) amongst other study characteristics are reported in Table 1. The studies employed the NRS (n = 2) [63], [64], [65], and VAS (n = 2) [62, 64], were tools used to assess pain, while the RMDQ (n = 4); to assess disability in all four studies. Other outcome measures included the SF – 36 to measure QoL [63], [64], [65], the Beck Depression Inventory to measure depression [62], the Tampa Scale for Kinesiophobia (TSK) to measure kinesiophobia [65], and the Pain Specific Function Scale [63] to measure function. Intervention periods ranged from six [65] to twelve weeks [64], with at least more than one session every week [62], [63], [64], [65].

Risk of bias assessment in included studies
Table 2 illustrates information on the risk of bias in included studies; three out of the four studies had an overall low risk of bias as they had more than six ‘yes’ answers based on the Cochrane Back Review Group recommendations [62], [63], [64], [65]. However, the overall risk of bias was high in one study, with three ‘unclear’ answers and four ‘no’ answers [64].

Description of interventions
All four studies incorporated CBT principles plus exercise therapy in one intervention group, and exercise alone in the other intervention group. Table 3 describes the interventions, with results for the treatment effects between groups, and their corresponding 95% CI at baseline, post intervention, and follow-up (6 months and 12 months).

Reporting bias
Due to similarities in outcome measures in included studies (n = 4), a meta-analysis was deemed appropriate. However, after the initial meta-analysis, Khan et al., [64], was excluded because it demonstrated a high risk of bias, publication bias, and also increased the statistical heterogeneity (Pain; I2 =0–79%; Disability; I2 =0–91%); thus the meta-analysis covered the three remaining included studies.A meta-analysis is recommended if both clinical and statistical heterogeneity are deemed to be at a minimal level [66], [67], [68].

Table 4 illustrates the summary of findings (SoF) generated from GRADEPro for studies included in meta-analysis; the quality grading revealed a high quality in all included studies (n = 3).

Pain
Three RCTs [62], [63], [65], with 346 participants included in meta-analysis, reported post intervention effect on pain (0 – 10 scale). A combination of these high quality RCTs revealed that there was no difference between CBT plus exercise and exercise alone post intervention (SMD = -0.02; 95% CI, -0.23 to 0.19) (Figure 2). Statistical heterogeneity between studies was low (I2 = 0%; p-value = 0.92), indicative of a high homogeneity among included studies.

Medium (six months) and long term (twelve months) follow-up was only assessed by two studies; thus a meta-analysis was performed for two studies [62], [63] (Figure 3). Medium term follow-up assessment for pain comprising 286 participants revealed, no difference between study groups (SMD = -0.01; 95% CI, -0.24 to 0.22) (Figure 3). Statistical heterogeneity was low (I2 = 0%; p-value = 0.91), signifying a high homogeneity among studies.

Long term follow-up assessment for pain comprising of 286 participants from two studies, demonstrated no difference between study groups (SMD = 0.06; 95% CI, -0.18 to 0.29) (Figure 4). Statistical heterogeneity was low (I2 = 0%; p-value = 0.56), indicative of high homogeneity among included studies.

Khan et al.[64]; reported a significant improvement in both the CBT plus exercise group (pain: 2.66;SD = 1.39; P-value < 0.0001), and the exercise alone group (pain: 5.25; SD = 1.19; P-value < 0.0001); however, between group analysis for treatment effects was not reported.

Disability
Three RCTs [62], [63], [65], with 346 participants included in the meta-analysis reported post intervention effect on disability. A combination of these high quality RCTs revealed no difference between study groups (SMD = 0.06; 95% CI, -0.15 to 0.27) (Figure 5). Statistical heterogeneity was low (I2 = 0%; p-value = 0.94), signifying highly homogenous studies.

Medium term follow-up assessment for disability comprising 286 participants revealed no statistical difference between CBT plus exercise and exercise alone (SMD = 0.00; 95% CI, -0.23 to 0.23) (Figure 6). Statistical heterogeneity was low (I2 = 0%; p-value = 0.36), signifying a high homogeneity among included studies.

Long term follow-up for disability comprising 286 participants, demonstrated no difference between study groups (SMD = -0.06;95% CI, -0.39 to 0.27) (Figure 7). Statistical heterogeneity was moderate (I2 = 48%; p-value = 0.16), indicative of a possibility of a slight variation among included studies.

Furthermore, Khan et al.,[64], reported a significant improvement in both the CBT plus exercise group (disability: 5.33, SD = 2.67; P-value < 0.0001) and the exercise alone group (disability: 9.88, SD = 1.84; P-value < 0.0001) however, between group analysis for treatment effects was not reported.

Due to variations in reporting secondary outcomes for this review, a narrative synthesis was adopted aimed at describing the findings from included studies.

Quality of life (QoL)
Two studies assessed QoL as part of their outcome measures [63], [65] (Table 1). QoL was assessed using the Short-Form Health Survey questionnaire (SF-36) in both studies. Macedo et al.,[63], found no difference between groups for the physical component score (PCS) for SF-36 post intervention, although there was an improvement in QoL in both CBT plus exercise group (51.6; SD, 13.4) and exercise alone group (51.6; SD, 12.0), with the mean adjusted treatment effect showing a small effect ( = -0.2; 95% CI, -3.7 to 3.2; p-value = 0.89) [63].

Furthermore, the mental component score (MCS) post intervention also showed a similar increase in both CBT plus exercise group (55.8; SD, 13.0), and exercise alone group (56.0; SD, 10.9); with the mean adjusted treatment effect also showing a small effect ( = 2.3; 95% CI, -0.7 to 5.3; p-value = 0.14) [63]. Similar findings were observed at both medium and long term as illustrated in Table 3.

Magalhaes et al., [65] , also reported that though improvements occurred in both CBT plus exercise (PCS: 75.3, SD 33.4, P<0.001; MSC: 84.4, SD = 31.2, P<0.001), and exercise alone groups (PCS: 68.1, SD = 41.1, P<0.001; MCS: 78.1, SD = 38.1, P<0.002), however, there was no statistical difference between groups post intervention (PCS: 6.0; 95% CI, -30.5 to 18.5, p-value = 0.415; MCS: 4.5; 95% CI, -31.5 to 22.5, p-value = 0.388).

Function
Function was assessed by only one RCT [63] (Table 3). Function was assessed with the use of the Patient-Specific Functional Scale (PSFS); mean adjusted treatment effects were small post intervention ( = 0.2; 95% CI, -0.5 to 0.9, p-value = 0.53), with both interventions showing similarities (CBT plus exercise: 5.5; SD = 2.4; Exercise alone: 5.9; SD = 2.1) [63].

Medium term and long-term assessments demonstrated no difference between study groups, although both reported improvement in function (Medium term: CBT plus exercise; 5.7; SD = 2.4; Exercise alone; 5.7; SD = 2.3; Long term: CBT plus exercise; 6.1; SD = 2.3; Exercise alone; 5.9; SD = 2.2), with small mean adjusted treatment effect (Medium term: = -0.2; 95% CI, -0.9 to 0.5, p-value = 0.53 Long-term: = -0.4; 95% CI, -1.1 to 0.3, p-value = 0.25) [63].

Depression
Assessment of depression was undertaken by only one study [62] Table 1. The Beck Depression Inventory (BDI) was utilized in assessing depression; between group (CBT plus exercise and exercise alone) mean difference post intervention revealed a statistically significant but not clinically meaningful difference ( = 2.17; 95% CI, 0.18 to 4.17, p<0.05), post intervention [62]. The minimal clinically important difference (MCID) recommended by NICE and developed by the National Collaborating Centre for Mental Health (NCCMH) is a difference greater than or equal to 3.0 BDI points [69].

Furthermore, medium, and long-term assessment results also revealed statistically significant but not clinically meaningful mean differences (medium term: = 0.49; 95% CI, -1.54 to 2.51, p<0.05; long-term: = 1.05; 95% CI, -0.97 to 3.07, p<0.05) between CBT plus exercise and exercise alone groups [62].

Kinesiophobia
Kinesiophobia was assessed by only one study [65]. The Tampa Scale for Kinesiophobia was the outcome tool utilized for measurement. Post intervention mean difference between groups was not significant ( = 4.0; 95% CI, -1.7 to 9.7; p-value = 0.321), although there was significant improvement in individual groups (CBT plus exercise: = 36.6, SD = 8.1, p-value < 0.007; Exercise: = 35.6, SD = 9.5, p-value< 0.001).

Summary of findings
The aim of this review was to ascertain the comparative effectiveness of CBT plus exercise versus exercise alone in the management of patients with NSCLBP. Overall, the meta-analysis showed no significant difference between the CBT plus exercise group in comparison to the group receiving exercise alone, for primary outcomes (pain and disability) and secondary outcomes (QoL, kinesiophobia, function). However, although there was a significant difference between groups for depression in one study [62], favouring CBT plus exercise, this was not clinically meaningful; mean difference in BDI values was below 3.0 which is considered as not clinically meaningful according to NICE recommendations [69].

Statistical heterogeneity post intervention, medium term, and long term was predominantly low (I2 = 0%) (except for disability at long term: I2 = 48%), for pain and disability; therefore, our findings, suggesting that combining CBT with exercise is no better than exercise alone for the management of NSCLBP post intervention, medium term, and long term, can be considered to be robust. Furthermore, Khan et al.,[64], also reported no significant difference between groups although between group analysis was not reported; however, a high risk of bias was ascertained from the evidence.

Strengths and weaknesses of included studies
Khan et al., [64], was excluded from the meta-analysis due to possible high risk of bias, low methodological quality, high heterogeneity, and potential reporting bias. The three studies included in the meta-analysis, overall showed a low risk of bias, with a high quality in the overall grading, and a predominantly low heterogeneity (I2 = 0% for all except long term outcome for disability; I2 = 48%); this signifies a strong internal validity and reliability [70].

Treating therapists, and participants were not blinded in all three studies, thus possibly increasing the performance bias; however, it is difficult to achieve this blinding in active physiotherapy studies [11]. Macedo et al.,[63], and Smeets et al.,[62], performed their trial analysis based on intention to treat; however, Magalhaes et al.,[65], did not perform trial analysis based on intention to treat as there were drop-outs in both groups (CBT plus exercise = 3; Exercise alone = 3), thus possiblyweakening the internal validity of the study [71].

Again, study characteristics were varied based on treatment delivery (format, content, dosage), participants employment status, economic status, duration of pain, and associated symptoms such as leg pain; patients whose characteristics indicated psychopathology were also excluded in two studies [62],[65].

Furthermore, none of the included studies incorporated a patient subgrouping process (e.g., risk assessment tool; STarT Back); to stratify patients based on the risk of chronicity potentially enabling a more targeted intervention process. Guidelines such as NICE recommend stratification as part of comprehensive management approach for NSCLBP [8]; consequently, systematic stratification and identification of high risk groups who are potentially likely to respond to a multimodal approach (including exercise and CBT) is an area for future research.

Studies did not include a range of psychological measures including, kinesiophobia, self-efficacy, and depression, which are highlighted as measures in CBT interventions [72]. Also, none of the included studies assessed cost effectiveness of combining CBT with exercise; although, systematic reviews by Lin et al., [73], and Kamper et al., [74], have concluded that management utilizing either CBT or exercise for pain and disability is cost-effective, its inclusion would help better inform clinical decisions.

Implications for clinical practice
The results of a research have the potential to determine the direction of clinical practice, allocation of funding, and the determination of areas that may necessitate quality improvement. The results confirm the favorable outcomes of guideline recommended therapies (CBT and exercise) for the management of NSCLBP. Importantly, the results showed no significant difference between groups receiving CBT plus exercise and exercise alone for measures of pain and disability (short or long term).

These results have important implications in relation to service delivery planning for patients with NSCLBP and clinicians managing patients with NSCLBP. The addition of CBT in this review to exercise-based therapy did not improve outcomes of pain and disability suggesting that focus for the interventions should be based around exercise primarily although certainly a multimodal approach may need to be considered for some patients.The studies did not (for example) look at subgrouping of patients based on psychological profiling and it may be that specific subgroups of patients with higher levels of psychological distress do better with a multimodal approach combining exercise and CBT therapies.

The other important fact to highlight is that this review focused on pain and disability as outcome measures and few of the included studies incorporated psychological outcome measures such as depression and kinesiophobia. It might be argued that the added value of CBT to exercise therapy would be reflected in these outcomes and further research is required in this area.The need to situate the evidence in a contextual framework while taking into consideration individual patients needs is essential. Consequently, the results of this review; a confirmation of guideline recommended approaches, and an outcome revealing no significant difference between groups; will serve as a valuable resource in clinical practice. Furthermore, relevant questions and recommendations from this review will serve as an opportunity for further research to update existing evidence.

The results of this review are relevant to the physiotherapy community. The fact that all included studies confirm guideline-recommended therapies as having a favorable outcome for NSCLBP is an affirmation of the relevance of practicing in an EBP framework. Furthermore, the results from this review have the potential to draw the attention of physiotherapists to institutionalizing stratified care interventions, geared towards specific patient needs.

Limitations of this review
This review was conducted with a comprehensive literature search, with two independent reviewers undertaking a full text screening for including and extracting data, thus minimizing bias; however only studies published in English were included in this review possibly introducing a language bias.

How these results add to the literature base
A systematic review with meta-analysis by Kamper et al., [74], investigated combined treatments in a multidisciplinary biopsychosocial framework, compared to usual care (which entailed health professional prescribed treatment), reported that the combined biopsychosocial framework had favourable outcomes; however, effect sizes were small.

Furthermore, a systematic review with narrative synthesis by George, [75], comparing a biopsychosocial framework (entailing neurobiological conditioning, graded exposure, and empowerment of the patient), to CBT, usual care or no treatment, also concluded favourable outcomes for the multimodal biopsychosocial framework.

This systematic review is to the best of our knowledge, the first to investigate the comparative effectiveness of CBT plus exercise versus exercise alone in the management of NSCLBP. Results from this review suggest that although combining CBT with exercise or exercise alone are effective management approaches for managing NSCLBP patients, combining CBT plus exercise is not more effective than exercise alone on patient outcomes.

Although we are confident our findings are robust, the following recommendations are based on observations made from the study:

• Firstly, future RCTs that seek to answer this question should have a clear stratification procedure based on validated tools such as the STarT Back to clearly ascertain patients risk levels to be able to make conclusions on effectiveness within the subgroups.
• Secondly, therapists should consider patients preferences within available resources when deciding on which approach to adopt to management.
• Finally, although guidelines recommend a CBT approach combined with exercise for NSCLBP, perhaps more clarity in necessary to ascertain which subgroup of NSCLBP patients would benefit from a combined therapy.

1. Balagué F, Mannion AF, Pellisé F, Cedraschi C. Non-specific low back pain. Lancet 2012 Feb 4;379(9814):482–91.   [CrossRef]   [PubMed]    
2. Moffett J, McLean S. The role of physiotherapy in the management of non-specific back pain and neck pain. Rheumatology (Oxford) 2006 Apr;45(4):371–8.   [CrossRef]   [PubMed]    
3. Koes BW, van Tulder M, Lin CW, Macedo LG, McAuley J, Maher C. An updated overview of clinical guidelines for the management of non-specific low back pain in primary care. Eur Spine J 2010 Dec;19(12):2075–94.   [CrossRef]   [PubMed]    
4. Maher C, Underwood M, Buchbinder R. Non-specific low back pain. Lancet 2016;389(10070):736–47.   [CrossRef]    
5. Costa Lda C, Maher CG, McAuley JH, et al. Prognosis for patients with chronic low back pain: Inception cohort study. BMJ 2009 Oct 6;339:b3829.   [CrossRef]   [PubMed]    
6. Kent PM, Keating JL. The epidemiology of low back pain in primary care. Chiropr Osteopat 2005 Jul 26;13:13.   [CrossRef]   [PubMed]    
7. Airaksinen O, Brox JI, Cedraschi C, et al. Chapter 4. European guidelines for the management of chronic nonspecific low back pain. Eur Spine J 2006 Mar;15 Suppl 2:S192–300.   [CrossRef]   [PubMed]    
8. Low back pain and sciatica in over 16s: Assessment and management (NG59). NICE. 2016;NG59(November):1–18. [Available at: https://www.nice.org.uk/guidance/ng59]    
9. Richmond H, Hall AM, Copsey B, et al. The effectiveness of cognitive behavioural treatment for non-specific low back Pain: A systematic review and meta-analysis. PLoS One 2015 Aug 5;10(8):e0134192.   [CrossRef]   [PubMed]    
10. Gomes-Neto M, Lopes JM, Conceição CS, et al. Stabilization exercise compared to general exercises or manual therapy for the management of low back pain: A systematic review and meta-analysis. Phys Ther Sport 2017 Jan;23:136–42.   [CrossRef]   [PubMed]    
11. Smith BE, Littlewood C, May S. An update of stabilisation exercises for low back pain: A systematic review with meta-analysis. BMC Musculoskelet Disord 2014 Dec 9;15:416.   [CrossRef]   [PubMed]    
12. Mehling WE, Ebell MH, Avins AL, Hecht FM. Clinical decision rule for primary care patient with acute low back pain at risk of developing chronic pain. Spine J 2015 Jul 1;15(7):1577–86.   [CrossRef]   [PubMed]    
13. Violante FS, Mattioli S, Bonfiglioli R. Low-back pain. In: Lotti M, Bleecker M, editors. Occupational Neurology, Volume 131. 1ed. UK: Elsevier; 2015. p. 397–410.    
14. Olatunji BO, Hollon SD. Preface: The current status of cognitive behavioral therapy for psychiatric disorders. Psychiatr Clin North Am 2010 Sep;33(3):xiii–xix.   [CrossRef]   [PubMed]    
15. Hanscom DA, Brox JI, Bunnage R. Defining the role of cognitive behavioral therapy in treating chronic low back pain: An overview. Global Spine J 2015 Dec;5(6):496–504.   [CrossRef]   [PubMed]    
16. Hayden JA, Cartwright JL, Riley RD, Vantulder MW; Chronic low back pain IPD meta-analysis group. Exercise therapy for chronic low back pain: Protocol for an individual participant data meta-analysis. Syst Rev 2012 Dec 21;1:64.   [CrossRef]   [PubMed]    
17. Smeets R, Köke A, Lin CW, Ferreira M, Demoulin C. Measures of function in low back pain/disorders: Low back pain rating scale (LBPRS), oswestry disability index (ODI), progressive isoinertial lifting evaluation (PILE), quebec back pain disability scale (QBPDS), and Roland-Morris disability questionnaire (RDQ). Arthritis Care Res (Hoboken) 2011 Nov;63 Suppl 11:S158–73.   [CrossRef]   [PubMed]    
18. Roland M, Fairbank J. The Roland-Morris disability questionnaire and the oswestry disability questionnaire. Spine (Phila Pa 1976) 2000;25(24):3115–24.   [PubMed]    
19. Lee JJ, Lee MK, Kim JE, et al. Pain relief scale is more highly correlated with numerical rating scale than with visual analogue scale in chronic pain patients. Pain Physician 2015 Mar–Apr;18(2):E195–200.   [PubMed]    
20. Ferraz MB, Quaresma MR, Aquino LR, Atra E, Tugwell P, Goldsmith CH. Reliability of pain scales in the assessment of literate and illiterate patients with rheumatoid arthritis. J Rheumatol 1990 Aug;17(8):1022–4.   [PubMed]    
21. Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA group. Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. Ann Intern Med 2009 Aug 18;151(4):264–9.   [CrossRef]   [PubMed]    
22. Higgins J, Green S. Cochrane handbook for systematic reviews of interventions. The Cochrane Collaboration 2011.    
23. Singh J. Critical appraisal skills programme. J Pharmacol Pharmacother 2013;4(1):76.    
24. Furlan A, Pennick R, Bombardier C, van Tulder M. 2009 updated method guidelines for systematic reviews in the Cochrane back review group. Spine (Phila Pa 1976) 2009 Aug 15;34(18):1929–41.   [CrossRef]   [PubMed]    
25. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997 Sep 13;315(7109):629–34.   [PubMed]    
26. Hayden JA, van Tulder MW, Malmivaara A, Koes BW. Exercise therapy for treatment of non-specific low back pain. Cochrane Database Syst Rev 2005 Jul 20;(3):CD000335.   [CrossRef]   [PubMed]    
27. Elamin MB, Flynn DN, Bassler D, et al. Choice of data extraction tools for systematic reviews depends on resources and review complexity. J Clin Epidemiol 2009 May;62(5):506–10.   [CrossRef]   [PubMed]    
28. Borenstein M, Hedges LV, Higgins JPT, Rothstein RH. Introduction to Meta-Analysis. 1ed. UK: John Wiley and Sons, Ltd; 2009.    
29. Turner JA, Clancy S, McQuade KJ, Cardenas DD. Effectiveness of behavioral therapy for chronic low back pain: A component analysis. J Consult Clin Psychol 1990 Oct;58(5):573–9.   [CrossRef]   [PubMed]    
30. Nicholas MK, Wilson PH, Goyen J. Comparison of cognitive-behavioral group treatment and an alternative non-psychological treatment for chronic low back pain. Pain 1992 Mar;48(3):339-47.   [CrossRef]   [PubMed]    
31. Vlaeyen JW, Haazen IW, Schuerman JA, Kole-Snijders AM, van Eek H. Behavioural rehabilitation of chronic low back pain: Comparison of an operant treatment, an operant-cognitive treatment and an operant-respondent treatment. Br J Clin Psychol 1995 Feb;34 ( Pt 1):95–118.   [CrossRef]   [PubMed]    
32. Reid MC, Otis J, Barry LC, Kerns RD. Cognitive-behavioral therapy for chronic low back pain in older persons: A preliminary study. Pain Med 2003 Sep;4(3):223–30.   [PubMed]    
33. Monticone M, Giovanazzi E. Usefulness of a cognitive behavioural and rehabilitative approach to enhance long lasting benefit after lumbar spinal stenosis and degenerative spondylolisthesis surgery. A case report. Eur J Phys Rehabil Med 2008 Dec;44(4):467–71.   [PubMed]    
34. George SZ, Wittmer VT, Fillingim RB, Robinson ME. Comparison of graded exercise and graded exposure clinical outcomes for patients with chronic low back pain. J Orthop Sports Phys Ther 2010 Nov;40(11):694–704.   [CrossRef]   [PubMed]    
35. Hansen Z, Daykin A, Lamb SE. A cognitive-behavioural programme for the management of low back pain in primary care: A description and justification of the intervention used in the back skills training trial (BeST; ISRCTN 54717854). Physiotherapy 2010 Jun;96(2):87–94.   [CrossRef]   [PubMed]    
36. McDonough SM, Tully MA, O'Connor SR, et al. The back 2 activity trial: Education and advice versus education and advice plus a structured walking programme for chronic low back pain. BMC Musculoskelet Disord 2010 Jul 15;11:163.   [CrossRef]   [PubMed]    
37. Beissner K, Parker SJ, Henderson CR Jr, Pal A, Iannone L, Reid MC. A cognitive-behavioral plus exercise intervention for older adults with chronic back pain: Race/ethnicity effect? J Aging Phys Act 2012 Apr;20(2):246–65.   [PubMed]    
38. Stankovic A, Lazovic M, Kocic M, et al. Lumbar stabilization exercises in addition to strengthening and stretching exercises reduce pain and increase function in patients with chronic low back pain: Randomized clinical open-label study. Turk J Phys Med Rehab 2012;58(3):177–83.   [CrossRef]    
39. Bello AI, Quartey J, Lartey M. Efficacy of behavioural graded activity compared with conventional exercise therapy in chronic non-specific low back pain: Implication for direct health care cost. Ghana Med J 2015 Sep;49(3):173–80.   [PubMed]    
40. Ferrari S, Vanti C, Costa F, Fornari M. Can physical therapy centred on cognitive and behavioural principles improve pain self-efficacy in symptomatic lumbar isthmic spondylolisthesis? A case series. J Bodyw Mov Ther 2016 Jul;20(3):554–64.   [CrossRef]   [PubMed]    
41. van Wingerden JP. Behavioural therapy combined with physical training improves function and diminishes pain in chronic back pain. Arch Phys Med Rehabil 2017;95(10):5–6.    
42. Brunner E, De Herdt A, Minguet P, Baldew SS, Probst M. Can cognitive behavioural therapy based strategies be integrated into physiotherapy for the prevention of chronic low back pain? A systematic review. Disabil Rehabil 2013 Jan;35(1):1–10.   [CrossRef]   [PubMed]    
43. Ostelo RW, van Tulder MW, Vlaeyen JW, Linton SJ, Morley SJ, Assendelft WJ. Behavioural treatment for chronic low-back pain. Cochrane Database Syst Rev 2005 Jan 25;(1):CD002014.   [CrossRef]   [PubMed]    
44. Henschke N, Ostelo RW, van Tulder MW, et al. Behavioural treatment for chronic low-back pain. Cochrane Database Syst Rev 2010 Jul 7;(7):CD002014.   [CrossRef]   [PubMed]    
45. Rihn JA, Radcliff K, Norvell DC, et al. Comparative effectiveness of treatments for chronic low back pain: A multiple treatment comparison analysis. Clin Spine Surg 2017 Jun;30(5):204–25.   [CrossRef]   [PubMed]    
46. Grunnesjö MI, Bogefeldt JP, Svärdsudd KF, Blomberg SI. A randomized controlled clinical trial of stay-active care versus manual therapy in addition to stay-active care: Functional variables and pain. J Manipulative Physiol Ther 2004 Sep;27(7):431–41.   [CrossRef]   [PubMed]    
47. Carr JL, Klaber Moffett JA, et al. A randomized trial comparing a group exercise programme for back pain patients with individual physiotherapy in a severely deprived area. Disabil Rehabil 2005 Aug 19;27(16):929–37.   [CrossRef]   [PubMed]    
48. Steenstra IA, Anema JR, Bongers PM, de Vet HC, Knol DL, van Mechelen W. The effectiveness of graded activity for low back pain in occupational healthcare. Occup Environ Med 2006 Nov;63(11):718–25.   [CrossRef]   [PubMed]    
49. Ewert T, Limm H, Wessels T, et al. The comparative effectiveness of a multimodal program versus exercise alone for the secondary prevention of chronic low back pain and disability. PM R 2009 Sep;1(9):798–808.   [CrossRef]   [PubMed]    
50. Albaladejo C, Kovacs FM, Royuela A, Del Pino R, Zamora J. The efficacy of a short education program and a short physiotherapy program for treating low back pain in primary care: A cluster randomized trial. Spine (Phila Pa 1976) 2010 Mar 1;35(5):483–96.   [CrossRef]   [PubMed]    
51. Bosmans JE, Pool JJ, de Vet HC, van Tulder MW, Ostelo RW. Is behavioral graded activity cost-effective in comparison with manual therapy for patients with subacute neck pain? An economic evaluation alongside a randomized clinical trial. Spine (Phila Pa 1976) 2011 Aug 15;36(18):E1179–86.   [CrossRef]   [PubMed]    
52. Murtezani A, Hundozi H, Orovcanec N, Sllamniku S, Osmani T. A comparison of high intensity aerobic exercise and passive modalities for the treatment of workers with chronic low back pain: A randomized, controlled trial. Eur J Phys Rehabil Med 2011 Sep;47(3):359–66.   [PubMed]    
53. Brooks C, Kennedy S, Marshall PW. Specific trunk and general exercise elicit similar changes in anticipatory postural adjustments in patients with chronic low back pain: A randomized controlled trial. Spine (Phila Pa 1976) 2012 Dec 1;37(25):E1543–50.   [CrossRef]   [PubMed]    
54. Siemonsma PC, Stuive I, Roorda LD, et al. Cognitive treatment of illness perceptions in patients with chronic low back pain: A randomized controlled trial. Phys Ther 2013 Apr;93(4):435–48.   [CrossRef]   [PubMed]    
55. Linden M, Scherbe S, Cicholas B. Randomized controlled trial on the effectiveness of cognitive behavior group therapy in chronic back pain patients. J Back Musculoskelet Rehabil 2014;27(4):563-8.   [CrossRef]   [PubMed]    
56. Roche G, Ponthieux A, Parot-Shinkel E, et al. Comparison of a functional restoration program with active individual physical therapy for patients with chronic low back pain: A randomized controlled trial. Arch Phys Med Rehabil 2007 Oct;88(10):1229-35.   [CrossRef]   [PubMed]    
57. van der Roer N, van Tulder M, Barendse J, Knol D, van Mechelen W, de Vet H. Intensive group training protocol versus guideline physiotherapy for patients with chronic low back pain: A randomised controlled trial. Eur Spine J. 2008 Sep;17(9):1193–200.   [CrossRef]   [PubMed]    
58. Vibe Fersum K, O'Sullivan P, Skouen JS, Smith A, Kvåle A. Efficacy of classification-based cognitive functional therapy in patients with non-specific chronic low back pain: A randomized controlled trial. Eur J Pain 2013 Jul;17(6):916–28.   [CrossRef]   [PubMed]    
59. Johnson RE, Jones GT, Wiles NJ, et al. Active exercise, education, and cognitive behavioral therapy for persistent disabling low back pain: A randomized controlled trial. Spine (Phila Pa 1976). 2007 Jul 1;32(15):1578–85.   [CrossRef]   [PubMed]    
60. Sveinsdottir V, Eriksen HR, Reme SE. Assessing the role of cognitive behavioral therapy in the management of chronic nonspecific back pain. J Pain Res 2012;5:371–80.   [CrossRef]   [PubMed]    
61. Mannion AF, Müntener M, Taimela S, Dvorak J. A randomized clinical trial of three active therapies for chronic low back pain. Spine (Phila Pa 1976). 1999 Dec 1;24(23):2435–48.   [PubMed]    
62. Smeets RJ, Vlaeyen JW, Hidding A, Kester AD, van der Heijden GJ, Knottnerus JA. Chronic low back pain: physical training, graded activity with problem solving training, or both? The one-year post-treatment results of a randomized controlled trial. Pain 2008 Feb;134(3):263–76.   [CrossRef]   [PubMed]    
63. Macedo LG, Latimer J, Maher CG, et al. Effect of motor control exercises versus graded activity in patients with chronic nonspecific low back pain: A randomized controlled trial. Phys Ther. 2012 Mar;92(3):363–77.   [CrossRef]   [PubMed]    
64. Khan M, Akhter S, Soomro RR, Ali SS. The effectiveness of cognitive behavioral therapy (CBT) with general exercises versus general exercises alone in the management of chronic low back pain. Pak J Pharm Sci 2014 Jul;27(4 Suppl):1113–6.   [PubMed]    
65. Magalhães MO, Muzi LH, Comachio J, et al. The short-term effects of graded activity versus physiotherapy in patients with chronic low back pain: A randomized controlled trial. Man Ther 2015 Aug;20(4):603–9.   [CrossRef]   [PubMed]    
66. Akobeng AK. Principles of evidence based medicine. Arch Dis Child 2005;90(4):837–40.    
67. Alderson P, Green S. Cochrane Collaboration open learning material for reviewers. 2002. [Available from: http://training.cochrane.org/path/introduction-systematic-reviews-pathway]    
68. Ghersi D, Berlin J, Askie L. On behalf of the Cochrane prospective meta-analysis methods group. Chapter 19: Prospective meta-analysis. In: Higgins JPT, Green S, editors. Cochrane Handbook for Systematic Reviews of Interventions. Version 5.0.2 [updated September 2009]. The Cochrane Collaboration 2009.    
69. National Collaborating Centre for Mental Health. Depression: The Treatment and Management of Depression in Adults (Updated Edition). UK: British Psychological Society; 2010.   [PubMed]    
70. Viswanathan M, Ansari, Mohammed T, et al. Assessing the risk of bias of individual studies in systematic reviews of health care interventions. Methods guide for effectiveness and comparative effectiveness reviews. Rockville (MD): Agency for Healthcare Research and Quality. (US); 2008. AHRQ methods for effective health Care 2012.   [PubMed]    
71. Polit DF, Gillespie BM. Intention-to-treat in randomized controlled trials: Recommendations for a total trial strategy. Res Nurs Health 2010 Aug;33(4):355–68.   [CrossRef]   [PubMed]    
72. Ehde DM, Dillworth TM, Turner JA. Cognitive-behavioral therapy for individuals with chronic pain: Efficacy, innovations, and directions for research. Am Psychol 2014 Feb–Mar;69(2):153–66.   [CrossRef]   [PubMed]    
73. Lin CW, Haas M, Maher CG, Machado LA, van Tulder MW. Cost-effectiveness of guideline-endorsed treatments for low back pain: A systematic review. Eur Spine J. 2011 Jul;20(7):1024–38.   [CrossRef]   [PubMed]    
74. Kamper SJ, Apeldoorn AT, Chiarotto A, et al. Multidisciplinary biopsychosocial rehabilitation for chronic low back pain: Cochrane systematic review and meta-analysis. BMJ 2015 Feb 18;350:h444.   [CrossRef]   [PubMed]    
75. George SI. What is the Effectiveness of a Biopsychosocial Approach to Individual Physiotherapy Care for Chronic Low Back Pain? Internet J Allied Heal Sci Pract. 2008;6(1):1–10.