Does Motor Control Exercise Restore Normal Morphology of Lumbar Multifidus Muscle in People with Low Back Pain? – A Systematic Review

Introduction

Low back pain (LBP), defined as pain or discomfort between the twelfth ribs and buttocks,^1,2 is the leading cause of disability worldwide.³ It affects up to 84% of people at least once in their lifetime. The prevalence of LBP is anticipated to increase with an aging global population.⁴ Since LBP can lead to tremendous medical burdens and work disability, the overall cost of LBP is expected to increase over time.⁴ Although LBP is ubiquitous, approximately 85% of LBP cases have unclear etiology.⁵ Biomechanical research suggests that the occurrence/maintenance of LBP may be related to the suboptimal motor control of deep trunk muscles.⁶ Specifically, lumbar multifidus muscle (LMM) is a major paraspinal muscle that provides intersegmental control of the spine^7–9 and withstands the compressive loading of the lumbar spine.¹⁰ Therefore, structural/functional deficits of LMM may be related to the onset or maintenance of chronic LBP (CLBP).

Compared to asymptomatic individuals, some patients with acute or chronic LBP demonstrate morphometric and/or functional changes in LMM (eg, reduced cross-sectional area (CSA),^11–15 increased intramuscular fatty infiltration,^14,16–18 decreased resting thickness,¹⁹ and percentage thickness changes during maximum voluntary isometric contraction¹⁹ or contralateral arm lift).^20,21 However, no significant relation between CSA/fatty infiltration of LMM and LBP has also been reported.²² Although LMM atrophy may be specific to the location and the side of symptoms,²³ prolonged immobilization may also result in general LMM atrophy.¹⁰ Given the close association between LMM and LBP, one rehabilitation approach is to improve the function and morphology of LMM. Of various physiotherapy interventions, motor control exercise (MCE) is thought to be able to restore LMM morphology and function in patients with LBP.^2,24 Multiple studies have investigated the effectiveness of MCE in restoring normal LMM morphometry^25,26 or decreasing LBP among patients with CLBP.^27–30 Some found that MCE increased LMM sizes in these patients^2,30,31 Although a recent Cochrane review found low- to moderate-quality evidence to support MCE in inducing clinically meaningful pain reduction in patients with CLBP as compared to different kinds of controls including sham intervention and education,³² no review has summarized the effectiveness of MCE in concomitantly restoring LMM morphology and reducing LBP. Further, temporal relations between post-MCE changes in LMM morphology and changes in pain intensity/LBP-related disability among patients with LBP have not been summarized. Therefore, this systematic review aimed to summarize the evidence regarding (1) the effectiveness of MCE in restoring normal LMM morphometry and decreasing LBP; and (2) whether the post-treatment changes in morphology were associated with changes in pain and/or function of patients with LBP.

Methods

Identification and Selection

This review conforms to the Preferred Reporting Items for Systematic reviews and Meta-Analysis (PRISMA) guidelines and is registered with PROSPERO (CRD42019120978).³³ A systematic search was conducted in CINAHL, MEDLINE, Cochrane Central Register of Controlled Trials, the Physiotherapy Evidence Database (PEDro), EMBASE and SPORTDiscus from inception to 30 September 2020. Non-English publications were excluded. The search keywords and Medical Subject Headings included were related to LBP, lumbar multifidus, physiotherapy, or rehabilitation (Appendix 1). Studies were included if they: (1) were randomized controlled trials (RCTs); (2) involved people with LBP regardless of chronicity; and (3) compared effects of MCE with another intervention/control groups(s) on at least one morphological/morphometric change of LMM (eg, CSA, resting/contracted thickness, percent thickness change during contraction, intramuscular fatty infiltration) (see Appendix 2 for details). Studies involving surgical interventions or cross-sectional comparisons between asymptomatic and symptomatic individuals, review articles, conference proceedings, theses, animal studies and grey literature were excluded. The reference lists of systematic reviews related to LMM morphology/morphometry were reviewed to identify relevant primary studies. The reference lists of the included studies were tracked backward, while forward citation tracing was performed using Web of Science. The corresponding authors of the included studies were contacted to identify additional relevant publications.

Two reviewers (SMP and SBB) independently screened the titles and abstracts based on the selection criteria. Potential full-text articles were retrieved and reviewed. Disagreements in the study inclusion at each stage were resolved by discussion. Any unresolved disagreements were decided by a third reviewer (AW). The inter-rater agreement at each screening stage was analyzed by Kappa coefficients (κ). The agreement was interpreted as none to slight (κ=0.01–0.20), fair (κ=0.21–0.40), moderate (κ=0.41–0.60), good agreement (κ=0.61–0.80), or almost perfect (κ=0.81–1.00).³⁴

Data Extraction

The two reviewers independently extracted authors’ names, year of publication, case definition, sample size, patients’ characteristics, intervention details, outcome measures, measurement methods, attrition rate, and pre- and post-treatment results using a standardized extraction form. The primary outcome measures included LMM morphometry (eg, resting, and contracted LMM thickness, percent thickness change during contraction, volume, CSA, and intramuscular fatty infiltration, etc.) and pain. The LMM morphometric data (eg, CSA, volume, resting thickness, contracted thickness, percent thickness changes) at each lumbar level on both sides were extracted from each included study, whenever possible. Percent thickness change was calculated from [(thickness contracted – thickness rest)/thickness rest x 100].³⁵ Greater percent LMM thickness change during contraction as measured by ultrasonography was thought to be an indirect measure of LMM contraction.^36,37 The LMM CSA was commonly used to estimate the muscle atrophy/weakness.²² Increased muscle CSA signified muscle hypertrophy.^38,39 Secondary outcome measures included correlations between changes in LMM morphology and LBP intensity/LBP-related disability.

Risk of Bias Assessment

The two reviewers (SMP and SBB) independently assessed the Risk of Bias (RoB) using the Cochrane collaboration RoB Tool (RoB 2.0).⁴⁰ Any disagreements regarding the scores were resolved by the third reviewer (AW). Each item was scored as low, some concern, or high risk of bias according to the Cochrane handbook descriptions.

The GRADE Approach

The two authors (SMP and SBB) independently assessed the quality of evidence of the primary outcomes using the GRADE as per GRADE handbook of grading quality of evidence and strength of recommendations. The assessment was based on the study design, risk of bias, inconsistency, indirectness, imprecision, and other considerations.⁴¹ The quality of evidence was rated at four levels: high, moderate, low, and very low. GRADE was assessed using http://gradepro.org.

Data Synthesis

A meta-analysis was planned to pool relevant data from the included studies. However, given the high clinical heterogeneity among studies (ie, different muscle measurement methods, such as ultrasonography, computerized tomography (CT) or magnetic resonance imaging, and diverse treatments) a qualitative analysis was conducted.

Since some included studies did not report within- or between-group treatment effects, secondary-analyses were conducted using Review Manager (RevMan 5.3) to compare within- and between-group differences, as well as the corresponding mean differences (MD) and 95% confidence intervals (CI) in primary outcomes using methods (ie, calculating mean change in each group by subtracting post-intervention mean from baseline mean or calculating mean differences between two groups using post-intervention measurements) recommended in the Cochrane handbook for Systematic Reviews of Intervention.⁴² To facilitate the comparisons of LMM volume, CSA and pain intensity among studies, the measurement unit in cm³, cm² and cm were converted into mm³, mm² and mm, respectively. Minimal clinically important difference (MCID) for pain, which means the smallest change in pain that a patient considers clinically meaningful, was set at 20mm on visual analogue scale (VAS).⁴³ Minimal detectable change at 95% confidence (MDC₉₅) was used to indicate the post-treatment change in scores that exceeded the measurement error (ie, true change). For patients with LBP, the MDC₉₅ for LMM CSA, resting and contracted thickness was 100mm²,⁴⁴ 3.6mm,³⁵ and 1.8mm,³⁵ respectively. The MDC₉₅ for percent thickness change during contraction was 15.7%.³⁵

Results

Study Selection

The search yielded 4114 citations. Nine RCTs were included from 41 screened full-text articles (Figure 1). The 2 reviewers demonstrated good agreements in selecting relevant papers at the first (κ=0.68) and second stages of screening (κ=0.76) (Appendix 3).

Figure 1 A flow diagram of the literature search.

Characteristics of the Included Studies

The 9 included RCTs were published between 1996 and 2020, involving 451 participants (410 chronic, 41 acute LBP). The mean ages of participants ranged from 31⁴⁵ to 50.8⁴⁶ years. The effectiveness of MCE (focusing on the activation of deep trunk muscles in different positions)^{2,24,29,30,45–49} in restoring normal LMM morphology or decreasing LBP were compared with McKenzie exercise,²⁹ general exercise,^2,30 general physiotherapy (eg, transcutaneous electrical nerve stimulation (TENS), therapeutic ultrasound therapy, infrared radiation, and traction),^2,46–48 massage,⁴⁶ high-load lifting exercise,²⁴ general strengthening plus aerobic exercises⁴⁹ and analgesics^45,46 (Appendix 4). The number of MCE sessions ranged from 12 to 36. Table 1 summarizes the characteristics of these studies. These RCTs had either 2,^{2,24,29,30,45,48,49} 3,⁴⁷ or 4 treatment arms.⁴⁶ Five studies involved a combination of one or two treatments with MCE in at least one arm^2,45–47,49 (eg, MCE plus massage,⁴⁶ MCE plus TENS,⁴⁶ MCE plus general physiotherapy,^2,47 MCE plus manual therapy⁴⁹ and MCE plus analgesics^45,46).

Table 1 Characteristics of Included Studies

Ultrasonography,^{2,24,29,30,45,46} CT-scans^47,48 or magnetic resonance imaging⁴⁹ were used to image LMM morphology in the included studies. Most studies measured bilateral CSA,^2,45,49 resting thickness,^2,24,29,30 and contracted thickness^29,46 from ultrasound and magnetic resonance images. Other studies measured CSA from CT images.^47,48 Although the current study aimed to extract morphometric data from each vertebral level, only one included study reported the CSA of LMM from each of the 5 lumbar levels (L1 to L5).⁴⁹ Similarly, only 1 included study reported the LMM volume of each lumbar level from L1 to L5.⁴⁹ Although LMM morphometry on the painful side might differ from non-painful side,^14,50 most of the included studies did not specify the side of measurements. These studies reported the post-treatment morphometric changes in LMM in terms of percentage or actual dimensions. Given the diverse treatment combinations and LMM morphometry measurement methods in the included studies, the planned meta-analysis was not conducted.

Risk of Bias

Risk of bias assessment for individual trials is presented in Figure 2. Seven studies^{2,24,29,30,45,46,49} were considered to have a low risk of bias, while two^47,48 were deemed to have a high risk of bias.

Figure 2 Risk of bias assessment according to Cochrane Collaboration’s tool (RoB 2.0) for randomized controlled trial.

Effects of MCE on LMM Morphology

The quality of evidence and details of the effectiveness of MCE in restoring normal LMM morphology are presented in Appendix 5 and Table 2, respectively.

Table 2 Effect of Motor Control Exercise on the Morphometry of Lumbar Multifidus Muscle

Volume of LMM

Only one study⁴⁹ with low risk of bias investigated the effects of MCE plus manual therapy on the volume of LMM.

Within-Group Comparisons

Low-quality evidence suggested that 10 sessions of MCE plus manual therapy did not significantly increase the volume of LMM in comparison to general strengthening plus aerobic exercises.⁴⁹

Between-Group Comparisons

Low-quality evidence suggested that 10 sessions of MCE plus manual therapy were not significantly better than general strengthening plus aerobic exercises in increasing LMM volume.⁴⁹

CSA of LMM

Three studies^2,45,49 with low and two^47,48 with high risk of bias investigated the effects of MCE on LMM CSA.

Within-Group Comparisons

Very low- to low-quality evidence substantiated that 12 sessions or more MCE with or without adjunct treatments (eg, resistance training, TENS, massage, manual therapy) significantly increased CSA of LMM at multiple lumbar levels.^2,45,47–49 Similarly, there was very low- to low-quality evidence that 36 sessions of MCE caused post-treatment increases in LMM CSA by 121 mm², which exceeded MDC₉₅⁴⁸ (Table 2).

Between-Group Comparisons

Low-quality evidence supported that MCE along with analgesics induced significantly greater increases in LMM CSA than analgesic alone among patients with acute LBP.⁴⁵ Likewise, there was very low-quality evidence that 18 or more sessions of MCE or MCE plus general physiotherapy caused significantly greater increases in LMM CSA than general physiotherapy alone in patients with CLBP.^47,48 However, only 36 sessions of MCE induced significantly greater increase in LMM CSA that exceeded MDC₉₅ (by 120 mm²) than general physiotherapy in patients with CLBP (Table 2).⁴⁸ However, there was low-quality evidence that 12 sessions of MCE plus general physiotherapy/MCE plus manual therapy were not significantly different from 12 sessions of general exercise plus general physiotherapy/general strengthening plus aerobic exercises in altering LMM CSA.^2,49

Resting LMM Thickness

Four studies^2,24,29,30 examined changes in the resting LMM thickness at the L4-5 level among patients with CLBP. The treatments ranged from 2 to 3 days/week, with 30–60 minutes each for 4 to 8 weeks. All four studies demonstrated a low risk of bias.^2,24,29,30

Within-Group Comparisons

Very low- to low-quality evidence suggested that 12 to 18 sessions of MCE with/without adjunct treatment, general exercises, high-load lifting, McKenzie exercise, or general exercises plus general physiotherapy significantly increased resting LMM thickness.^2,24,29,30 Although these post-MCE increases in the resting LMM thickness ranged from 1.1mm to 1.8mm, they did not exceed MDC₉₅.^2,29,30 (Table 2).

Between-Group Comparisons

There was very low- to low-quality evidence that 12 to 18 sessions of MCE or MCE plus general physiotherapy were not significantly better than other treatments (eg, general exercises,³⁰ high load lifting exercise,²⁴ McKenzie exercise,²⁹ general exercise plus physiotherapy,² in increasing LMM resting thickness (Table 2).

Contracted LMM Thickness

Two studies with low risk of bias^29,46 evaluated the effects of 16 to 18 sessions of MCE on the contracted thickness of LMM at the L4-5 level in patients with CLBP.

Within-Group Comparisons

Low-quality evidence suggested that MCE with/without adjunct treatment significantly increased the contracted thickness of LMM ranging from 0.3mm to 2.8mm.^29,46 However, only 18 sessions of MCE caused significant increases in contracted thickness of left LMM that exceeded MDC₉₅ (Table 2).²⁹

Between-Group Comparisons

There was low-quality evidence that MCE was comparable to McKenzie exercise in increasing LMM contracted thickness.²⁹ Low-quality evidence suggested that although MCE plus TENS caused significantly greater increases in contracted LMM thickness than MCE plus massage or analgesic alone, the differences did not exceed MDC₉₅ (Table 2).⁴⁶

Effects of MCE on Percent LMM Thickness Changes During Contraction and LMM Fatty Infiltration

Despite the comprehensive search, no RCT investigated the effects of intervention on percent LMM thickness changes during contraction or LMM fatty infiltration.

Effects of MCE on LBP Intensity of the Included Studies

Of the 9 included RCTs, 7 trials reported post-treatment decreases in LBP intensity (Table 3). Seven included studies^{2,24,29,30,45,48,49} used VAS to measure LBP intensity, which comprises a 10cm straight line with the two endpoints indicating no pain (0cm) and maximum pain (10cm), respectively.⁵¹

Table 3 Effect of Motor Control Exercise on Low Back Pain

Within-Group Comparisons

There was very low- to low-quality evidence that 4 to 24 weeks of MCE,³⁰ McKenzie exercise,²⁹ general exercises,³⁰ high-load lifting exercises,²⁴ MCE plus manual therapy,⁴⁹ general strengthening plus aerobic exercises,⁴⁹ and general physiotherapy^2,48 significantly decreased pain. The average pain reduction following MCE alone ranged from 2.8mm to 18.5mm on VAS, which were smaller than MCID.^24,29,30 There was very low- to low-quality evidence that combining MCE or general exercises with general physiotherapy,² MCE on a gymnastic ball or general physiotherapy alone⁴⁸ significantly reduced CLBP intensity by 33mm to 46mm on VAS, which exceeded the MCID for pain using VAS (>20mm) (Table 3).⁴³ Similarly, low-quality evidence supported that MCE with analgesics and nonsteroidal anti-inflammatory drugs significantly reduced acute LBP, although the extent of pain reduction was not reported.⁴⁵

Between-Group Comparisons

There was low-quality evidence that MCE alone caused significantly greater CLBP reduction than general exercise alone,³⁰ or McKenzie exercise alone.²⁹ However, there was no evidence that MCE with or without adjunct treatments was significantly better than high load lift exercise,²⁴ general physiotherapy,⁴⁸ general strengthening plus aerobic exercises,⁴⁹ general exercise plus general physiotherapy,² or drug alone⁴⁵ in reducing acute or chronic LBP. Given the high clinical heterogeneity among studies, meta-analysis was not conducted.

Temporal Relations Between Changes in LMM Morphology and Changes in LBP Intensity or LBP-Related Disability

Only two included RCTs with low risk of bias investigated the correlations between changes in LMM morphology and the corresponding changes in LBP intensity among patients with acute (n=41)⁴⁵ or CLBP (n=65).²⁴ There was no evidence that post-treatment increases in LMM resting thickness²⁴ or CSA⁴⁵ were related to LBP reduction (Table 4). Likewise, no evidence suggested that post-treatment increases in LMM CSA were related to changes in Roland Morris Disability Index scores in patients with acute LBP⁴⁵ (Table 4).

Table 4 Correlation Between Post-Treatment Change in Lumbar Multifidus Muscle (LMM) Morphology and the Corresponding Changes in Low Back Pain (LBP) Intensity or LBP-Related Disability

Protocol Deviations from PROSPERO Registration

Although the original protocol planned to summarize evidence regarding the effectiveness of various physiotherapy interventions in restoring normal LMM morphology and reducing pain in patients with LBP, the search yielded diverse treatments. Since the initial review question was too broad and MCE was the most commonly studied LBP treatment, we narrowed it down to a more specific research objective. Therefore, the current review focused on the effectiveness of MCE in restoring normal LMM morphology and decreasing pain in people with low back pain.

Discussion

This is the first systematic review to summarize the evidence regarding the effects of MCE on LMM morphology, LBP, and the correlations between changes in LMM morphology and LBP intensity or LBP-related disability. Our findings suggest that MCE may be little or no better than other interventions in changing LMM morphology or decreasing pain intensity. Similarly, there is no correlation between changes in LMM morphology and LBP or LBP-related disability.

The weak effects of post-MCE changes in LMM morphology (eg, thickness or CSA) may be related to insufficient exercise dosages (ie, frequency, intensity, type, and duration of MCE). Sokunbi Oluwaleke et al found that thrice weekly MCE for 6 weeks caused significantly greater increases in LMM CSA than once weekly MCE.⁵² Exercise-induced skeletal muscle hypertrophy usually occurs after exercising for at least 6-weeks.⁵³ Previous research has shown that muscle strengthening at 2–3 sessions per week yielded significantly greater CSAs of quadriceps and elbow flexors than exercising once-weekly.⁵⁴ Our findings suggest that the number of treatment sessions rather than exercise types might elicit post-treatment LMM morphological changes. However, there was conflicting evidence regarding whether these post-treatment changes in CSA exceeded the measurement error. Future studies should investigate the dose–response relationship between MCE intervention frequency/duration/intensity and the corresponding changes in LMM morphometry at different lumbar levels to determine optimal treatment dosage.

Interestingly, MCE^29,45 and high-load lifting exercises²⁴ appear to selectively increase the resting thickness²⁴ and contracted thickness^24,29 of LMM on the painful side to reduce asymmetry, which is not uncommon among patients with acute¹² /chronic LBP.^14,50 However, since most of the studies had small sample sizes and short-treatment durations, future large-scale prospective studies with longer follow-ups are warranted to determine the long-term effect of MCE or high-load lifting exercises on restoring LMM symmetry among patients with acute/chronic LBP and to identify the mechanisms underlying the selective muscle hypertrophy.

The current review found low-quality evidence that there were no clinically important differences between MCE and other physiotherapy interventions in reducing CLBP. Our finding concurred with a prior meta-analysis⁵⁵ and a Cochrane review,³² which revealed low- to high-quality evidence that MCE and other interventions had comparable effects on reducing non-specific LBP. However, these findings contradict another meta-analysis of eight studies, which concluded that MCE was more effective than general exercises in decreasing pain in patients with CLBP.⁵⁶ The disparity might be ascribed to the differences in measurement scales used in studies to measure pain intensity, treatment duration and dosages, criteria used for exercise progression, and follow-up periods. The discrepancy in results might also be attributed to less patients (n=603) involved in Gomes-Neto et al's meta-analysis⁵⁶ as compared to that of Smith et al⁵⁵ (n=2258).

The current review found no evidence to support a significant correlation between changes in LMM morphology and changes in LBP or LBP-related disability.^24,45 These findings differed from that of cohort studies, which found that patients with improved LBP displayed improved LMM morphometry (eg, increased percent thickness change during contraction).^57,58 The discrepancy may be due to the fact that many prior studies only evaluated the immediate post-treatment changes in LMM morphology and LBP intensity without long-term follow-ups. It is plausible that post-treatment morphological changes may be transient or may take time to develop. Future RCTs should clarify the association between temporal changes in LMM morphometry and the corresponding changes in LBP/LBP-related disability at different follow-up time points.

Additionally, while multiple factors may affect the clinical outcomes of patients with CLBP (eg, depression, anxiety, fear avoidance, catastrophizing and sleep),^59–61 all included RCTs in the current review did not adjust for these confounders in their analyses, which might have affected the reported temporal relations. Future studies should conduct path analyses to determine if LMM morphology may mediate or moderate LBP intensity/LBP-related disability after considering other potential confounders. The findings may help refine assessments and treatments for patients with LBP and concomitant aberrant LMM morphology.

Multiple factors may affect the measured LMM morphometry. First, since LMM thickness is a 2-dimensional measurement, changes in resting/contracted thickness as measured by ultrasonography can be affected by multiple factors (eg, the tightness of surrounding tissues, line of force, etc.).⁶² Therefore, LMM CSA measurements may be better to reveal morphometric changes. Second, LMM morphology as measured by ultrasonography is user dependent. The assessors’ experiences may affect the measured results. Unfortunately, all included RCTs did not report the test–retest reliability of their LMM measurements. Although the current review used previously reported MDC₉₅ to determine whether the reported LMM morphometric changes exceed measurement errors, the actual measurement error in each study might differ. Third, changes in LMM CSA as measured on CT scans are not directly related to muscle function, although bigger CSAs are thought to be associated with greater muscle strength. Future studies should evaluate the effects of MCE on LMM function (eg, electromyographic activity) in addition to morphology.

Strengths and Limitations

This review had several strengths. Comprehensive literature searches in 6 databases, standardized screening, data extraction, and methodological quality assessments of the studies were performed to ensure proper extraction and evaluation of data. The study protocol was registered with PROSPERO, while the reporting of the review followed the PRISMA guideline to ensure credibility and comprehensiveness of data. Further, since this review only included RCTs, our conclusion was drawn based on studies with the highest level of evidence.

Our review had some limitations. First, given the heterogeneity of outcome measures, exercise intensity, and underreporting of the side of LMM morphology in the included studies, no meta-analysis was conducted. Future studies should standardize the reporting/definition of LMM morphology and interventions to enable meta-analyses. Second, the sample sizes of the RCTs were small, ranging from 30^29,47 to 122,⁴⁶ which might have limited the statistical power. Future research should estimate the sample size based on the effect sizes of existing studies to ensure sufficient power to detect post-treatment changes in LMM morphology. Third, only RCTs published in English were included. Future systematic reviews should include non-English publications to improve the generalizability of findings. Fourth, the mean age of participants in the RCTs ranged from 30.9⁴⁵ to 50.8⁴⁶ years. Our findings may not be generalized to younger/older patients with LBP.

Conclusions

There is preliminary evidence that MCE may change LMM morphology, although it may be dose dependent. Specifically, 36 or more sessions of MCE may increase LMM CSA in patients with CLBP. However, existing evidence does not support that MCE is more effective than other exercises in treating acute/chronic LBP. That said, future research is warranted to determine the effects of MCE on segmental or global morphometry (including intramuscular fatty infiltration) of LMM and clinical outcomes, as well as to quantify the causal relationships between changes in LMM morphology and LBP/LBP-related disability.