The McKenzie Method delivered by credentialed therapists for chronic low back pain with directional preference: systematic review with meta-analysis

ABSTRACT

Objective

To determine the effectiveness of the McKenzie Method compared to any conservative interventions on pain and disability in patients with chronic low back pain (LBP) with directional preference (DP).

Methods

We searched six electronic databases up to September 2022. Eligible randomized controlled trials were those assessing the McKenzie Method delivered by credentialed therapists for chronic LBP with DP. Two reviewers independently selected studies, extracted data, assessed risk of bias with the revised Cochrane Risk of Bias 2.0 tool and certainty of evidence with the Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework.

Results

Five trials (n = 743) were included. There was low-certainty evidence that the McKenzie Method, compared to all other interventions combined, produced clinically important reductions in short-term pain (mean difference [MD] −1.11 points on a 10-point scale; 95% CI −1.83 to −0.40) and in intermediate-term disability (standardized mean difference [SMD] −0.53; 95% CI −0.97 to −0.09). Low-to-moderate certainty evidence showed that the McKenzie Method also resulted in clinically important improvements in short-term pain (MD −1.53; 95% CI −2.51 to −0.54) and disability (SMD −0.50; 95% CI −0.74 to −0.25) when compared specifically to other exercise approaches, and in intermediate-term pain (MD −2.10; 95% CI −2.94 to −1.26) and disability (SMD −1.01; 95% CI −1.58 to −0.43) as well as long-term disability (SMD −0,59; 95% CI −1.14 to −0.03) when compared to minimal intervention. Low-certainty evidence showed usually small, clinically unimportant effects in comparison to manual therapy.

Conclusion

We found low-to-moderate certainty evidence that the McKenzie Method was superior to all other interventions combined for up to 6 months for pain and up to 12 months for disability, with clinically important differences versus exercise in the short term and versus minimal interventions in the intermediate term. The only clinically important long-term effect was on disability compared to minimal intervention.

Introduction

Low back pain (LBP) is the leading cause of disability globally and results in substantial socioeconomic impact. [1,2] Individuals with chronic/recurrent LBP drive most of the costs associated with this condition. [3] Exercise therapy has been recommended as first-line treatment for chronic LBP. [4–7] A recent Cochrane review confirmed exercises are effective for pain and disability reduction, but effect sizes are modest. [8] As LBP is a heterogeneous condition, individualized interventions targeting more homogeneous samples with specific characteristics have been recommended. [9,10]

The McKenzie Method is a classification system widely used by therapists [11–14] to assess and manage spinal and extremity problems, with a focus on self-treatment. Its structured assessment enables the therapist to classify LBP into diagnostic syndromes and to offer a specific treatment based on the classification. Exercise and education are further individualized taking into consideration each patient’s needs and goals. The main classification of the McKenzie Method is comprised of patients who present a directional preference (DP) on a repeated-movement assessment. [15] Directional preference is a direction of repeated spinal movement or sustained position that leads to a lasting reduction in pain intensity or to a lasting change in its location (from a distal to a proximal site, toward the spine – the pain centralization phenomena). Around 60% of chronic LBP patients exhibit a DP when assessed by therapists who are properly trained (credentialed) in the McKenzie Method. [16] This approach advocates that patients with a DP should be taught to self-treat frequently throughout the day using these movements/positions that reduce or centralize their pain. With increased confidence in their ability for pain control, fear is reduced and physical function improves. [16]

Two recent systematic reviews have produced conflicting conclusions about the effectiveness of the McKenzie Method for chronic LBP. One review concluded that the McKenzie approach was superior to all other interventions combined (exercises and manual therapy plus exercises) to reduce pain and disability. [17] However, another review found no difference in comparison to active interventions (exercises) or passive interventions (manual therapies and electrophysical agents). [18] Notably, both reviews accepted trials where therapists were not sufficiently trained, i.e. were not credentialed in the McKenzie Method. To achieve credentialed-level status, a therapist must attend 4 post-grad courses, totaling 112 hours of training, and must successfully pass a credentialing examination offered by the McKenzie Institute International. A recent systematic review of inter-tester reliability studies showed that only therapists with sufficient training (credentialed level) in the McKenzie Method can assess and classify LBP patients into main syndromes (following the McKenzie approach) with sufficient inter-tester reliability. [19] Therapists with no or partial training (non-credentialed) showed poor inter-tester reliability and might be unable to properly follow the core principles of this approach. Additionally, a recent meta-analysis with meta-regression indicated that trials where therapists followed the core principles of the McKenzie Method (patients assessed and classified, treatment based on classification) produced clinically relevant greater reductions in pain and disability compared to trials where therapists did not follow these core principles. [20]

In addition to accepting studies with untrained or insufficiently trained McKenzie therapists, all previous systematic reviews included trials that assessed broad and heterogeneous LBP samples. No meta-analysis to date has focused on the more homogeneous subset of patients with chronic LBP with a DP.

Therefore, the purpose of this meta-analysis was to determine the effectiveness of the McKenzie Method delivered by credentialed therapists compared to any conservative interventions for improving pain and disability in patients with chronic LBP with a DP.

Methods

Protocol and registration

The protocol of this systematic review was pre-registered on PROSPERO (CRD42021266570). We adhered to the Preferred Reported Items for Systematic Reviews and Meta-Analysis (PRISMA) recommendations [21] and followed the recommendations from the Cochrane Collaboration [22].

Inclusion criteria

We only included randomized controlled trials. Below we describe our eligibility criteria following the PICO framework.

Population

Our population of interest was adults (≥18 years of age) with chronic (≥3 months duration) nonspecific LBP with or without leg symptoms, with a DP determined by the therapist through repeated movement testing. If the authors did not indicate whether patients’ symptoms were chronic, or if they accepted chronic and (sub)acute populations, we included the study if the mean (or median) symptom duration was ≥3 months. Additionally, we accepted studies with mixed populations (chronic and [sub] acute) where it was not possible to calculate mean or median symptom duration, if over 85% of the participants had recurrent episodes of back pain. Patients were considered to have recurrent episodes if they had had 2 or more episodes, including the current one. If a trial also included patients without a DP, the authors were contacted and only data of patients with a DP was included.

Intervention and comparators

We considered as the intervention any session of the McKenzie Method delivered by credentialed McKenzie therapists (i.e. therapists who attended 4 post-grad courses encompassing approximately 112 hours of training and successfully passed a credentialing examination). All therapists in the intervention group should have been credentialed by the McKenzie Institute International prior to the start of the intervention. We accepted studies with co-interventions as long as they were equally offered to both intervention and control groups. We considered as comparators any conservative interventions such as other exercise therapies (e.g. motor control exercises, stabilization exercises), manual therapies (e.g. mobilization, manipulation), or minimal intervention (e.g. advice, educational booklet, usual care, placebo).

Outcomes

The primary outcomes of interest were pain intensity and disability, and the secondary outcome was patients’ global impression of recovery.

Exclusion criteria

We excluded studies assessing prevention of chronic LBP (with asymptomatic patients at baseline). We also excluded studies of patients with serious spinal pathologies (e.g. malignancy, fracture), or specific conditions like lumbar radiculopathy (with 2 or more neurological deficits), lumbar spinal stenosis, inflammatory disease, and pregnancy, among others.

Information sources

We searched the following databases from inception to 3 September 2022: MEDLINE (via PubMed), EMBASE, Web of Science, LILACS, PEDro and Cochrane Central Register of Controlled Trials (CENTRAL). Also, we manually searched the reference list from the eligible trials and previous systematic reviews on the topic. There were no language restrictions in our search.

Search strategy

Our search strategies (Appendix 1) were composed by subject headings and free-text terms for the population, intervention, and study type, applying the necessary adaptations for the specifications of each database, with the help of a professional librarian. We did not include outcomes of interest to enhance the comprehensiveness of the search.

Study selection

We combined search results in a reference management system (Zotero 6.0.8) which automatically removed duplicate records. Two reviewers (V.H. and M.A.Z.M.) independently used the Rayyan system [23] to manually screen titles and abstracts to identify relevant trials according to the predetermined inclusion and exclusion criteria. Full-text versions were obtained and evaluated independently. Disagreements were discussed between the two reviewers. If no consensus could be reached, a third reviewer (P.K.Z.) made the final decision about trial inclusion.

Data extraction

The same two reviewers independently extracted and recorded data from the included trials on an Excel spreadsheet. Extracted information included sample sizes, characteristics of participants, clinical settings, interventions provided, co-interventions, tools used to assess primary and secondary outcomes, time of assessment, results from the outcomes and data to assess risk of bias. Any disagreement was discussed between the two reviewers. If no consensus could be reached, the same third reviewer made the final decision. We performed pilot testing of the extraction of data with two studies. When data were incomplete or unclear, we contacted the authors for clarification.

Risk of bias

We used the revised Cochrane Risk of Bias 2 tool (RoB2) to judge risk of bias in each trial for each result of interest, as recommended by the Cochrane Collaboration. [24] We were interested in the effect of assignment to intervention. Two reviewers (V.H. and M.A.Z.M.) independently made the risk of bias judgments. We resolved disagreements through discussion or arbitration by a third reviewer (P.K.Z.). We assessed risk of bias in five domains: (1) the randomization process; (2) deviations from intended interventions; (3) missing outcome data; (4) measurement of the outcome; and (5) selection of the reported result. For each domain, we made a judgment of high risk of bias, low risk of bias, or some concerns. Finally, we reached an overall risk of bias judgment at least as severe as the most severe domain risk.

Summary measures

For pain intensity, we used mean difference (MD) since all the scales used in the original studies were similar. We converted MDs to a common 0–10 scale. For disability and global impression of recovery, we used standardized mean difference (SMD) since studies used different scales and instruments. All results included pooled mean with a 95% CI and were classified at three levels: small effect size (MD < 1 point or SMD < 0.5), medium effect size (MD 1 to 2 points or SMD 0.5 to 0.8) or large effect size (MD > 2 points or SMD > 0.8). [25–29] Medium and large effect sizes were considered clinically important.[4,25–27,29–31]

Synthesis of results

We assessed our primary outcomes at the following time periods: short term (<3 months after randomization), intermediate term (3 to <12 months after randomization), and long term (≥12 months after randomization). In trials where multiple time points were available within a time period, we chose the time point closest to 1 month (short term), 6 months (intermediate term), and 1 year (long term). We assessed the secondary outcome only in the short term (<3 months after randomization, closest to 1 month). If authors assessed back pain intensity and leg pain intensity or back+leg pain intensity, we always chose the back pain intensity for our analysis.

When population, intervention, comparators, and outcomes were sufficiently similar across studies, we pooled MD or SMD using random-effects models with between-study variance estimated by the DerSimonian and Laird method. Statistical significance was set at p < 0.05. We calculated MD and SMD using either endpoint or change from baseline results, depending on what each author presented, with a preference for adjusted MD when available. We assessed statistical heterogeneity through the I² statistic and the Cochran test. Random-effects models were used irrespective of the statistical heterogeneity found. In trials with more than one comparator, we considered each comparison independently, and we divided the number of participants allocated to the group treated with the McKenzie Method by the number of comparison groups to avoid double counting. Funnel plots to evaluate potential publication biases were not inspected since we retrieved a small number of studies. Statistical analysis was performed using the package ‘meta’ and ‘metafor’ for R language (version 4.1.2) in the RStudio software (version 2021.09.1, Boston, MA).

Sensitivity and subgroup analysis

We explored the robustness of effect estimates in a sensitivity analysis by excluding studies at overall high risk of bias. For subgroup analyses, we grouped the studies according to type of control intervention: exercises (e.g. motor control, strengthening); manual therapy (e.g. mobilization, manipulation); and minimal interventions (e.g. advice, medication, modalities). We performed meta-analysis for these subgroups irrespective of the heterogeneity found.

Certainty in the body of evidence

Two reviewers (V.H. and M.A.Z.M.) independently used the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach to assess the overall certainty in the body of evidence for each result of interest. [32–35] A third reviewer (P.K.Z.) made the final decision if consensus could not be reached. The domains evaluated by GRADE include study design, risk of bias, inconsistency, indirectness, imprecision and publication bias. The GRADE approach involves downgrading the certainty in the evidence from high certainty to moderate, to low, or to very low, based on the cumulative assessment of specific criteria for each domain. The specific criteria we used to judge each GRADE domain are described in Appendix 2.

Results

Study selection

Our literature search resulted in a total of 7884 results (Figure 1). After excluding duplicates, two reviewers independently screened 4942 titles and abstracts and selected 56 articles for full-text review. Five studies (described in seven articles) met our eligibility criteria and were included in our meta-analysis. [36–42] Authors of all these five studies responded to our requests for additional information regarding sample characteristics and outcome data. A list of all excluded full-text studies with reasons for exclusion can be found in Appendix 3.

Figure 1.

Flow diagram of search strategy and results.

Study characteristics

The characteristics of the included studies are described in Table 1. In total, the five trials enrolled 743 participants (ranging from 18 to 307 per trial). Patients were recruited mostly from outpatient physical therapy clinics and primary care centers, with mean ages ranging from 37 to 48 years and with average pain duration ranging from 15 to 104 weeks. The duration of treatment regimens varied from 2 to 15 weeks, and the number of sessions varied from 1 to 12. The McKenzie Method was compared with other conservative interventions for pain and disability in five studies in the short term (<3 months),[36,37,39–42] in two studies in the intermediate term (3 to <12 months)[40–42] and three studies in the long term (≥12 months),[38,40–42] and with other conservative interventions for global impression of recovery (our secondary outcome) in three studies in the short term.[36,37,42]

Table 1.

Characteristics of included trials.

Study	Participants	Inclusion Criteria	Intervention	Outcomes
Halliday 201637,38	N = 70 Mean age = 48 years Median symptom duration = 32 weeks Setting: Physiotherapy department of Concord Repatriation General Hospital, Sydney, Australia	Greater than 3-month history of LBP and a directional preference observed with a mechanical assessment based on the McKenzie method. The area of pain could be localized between the twelfth rib and the buttock crease. Patients reporting referred pain into one or both legs with or without sensory and/or motor changes were also included	McKenzie: Treatment involved mechanical therapy including patient and therapist generated forces utilizing repeated or sustained end range loading strategies in loaded or unloaded postures according to the patients’ directional preference. Force and loading strategy were guided by symptom response (aiming to reduce, centralize and abolish symptoms). Postural education including provision of a standard McKenzie lumbar roll when an extension preference was identified. All participants received a copy of ‘Treat Your Own Back’ book to supplement treatment and self-management Motor control exercises: promoting independent contraction of the deep stabilizing muscles such as transversus abdominis and multifidus, facilitated by pelvic floor contraction while controlling breathing by maintaining resting tidal volumes. Initially in static tasks, progressing to dynamic tasks. Patients were instructed to practice exercises daily at home for 30-minutes. Patients were encouraged to attend twice a week for the first 4 weeks and once per week for the remaining 4 weeks even if their symptoms resolved during the treatment episode Both interventions: no more than 12 sessions over 8 weeks	Pain (VAS) Disability (PSFS) Global Improvement (GPE) Outcomes assessed at 8 weeks and 1 year after randomization)
Long 200436	N = 229 Mean age = 42 years Mean symptom duration of current episode = 15 weeks† Setting: consecutive patients with LBP presenting to outpatient departments and physical therapy clinics in Canada, United States, Germany, England and Kuwait	Low back pain, with or without leg symptoms, with or without one neurological sign, age 18–65 years, demonstrating a directional preference during the mechanical assessment	McKenzie ‘matched’: Subjects were taught unidirectional end-range lumbar exercises matching the direction of their directional preference identified during baseline assessment Exercise ‘opposite’: Subjects were also taught unidirectional end-range exercises, but in a direction opposite to their directional identified during baseline assessment Exercise ‘multidirectional’: Subjects were taught commonly prescribed multidirectional, midrange lumbar exercises, and stretches for the hip and thigh muscles Both the ‘opposite’ and ‘multidirectional’ groups were provided education consistent with LBP clinical guidelines, including advice aimed at minimizing fear avoidance behavior and to remain active. Members of the ‘matched’ group were likewise instructed to remain active but also to avoid activities and positions that increase intensity or radiation of symptoms. All interventions: 3 to 6 visits over the 2-week study period	Back Pain (VAS) Disability (RMDQ) Global Impression of Recovery (SRI) Outcomes assessed at 2 weeks after randomization
Miller 200539	N = 18* Mean age = 47 years Mean symptom duration = 104 weeks Settings: LBP patients referred by physician to outpatient physical therapy clinic in United States	Chronic LBP for >7 weeks, 18 years of age or older	McKenzie: Treatment may have included posture correction, end-range spinal repeated movements, or use of manual techniques to reduce or eliminate signs and symptoms following the principles of McKenzie approach for derangement classification (directional preference exercises). Ten to 15 min of daily home exercise. Six weeks of treatment. Spinal Stabilization Exercises: Six-week specific lumbar stabilizing program with exercises focused on strengthening the lumbar multifidus and transversus abdominis muscles through performance of a lower abdominal contraction (the patient moved the umbilicus toward the spine while the spine was maintained in a neutral alignment). Once the patient was able to initiate an isolated co-contraction of these muscles, they were progressed to holding the contraction while performing a progressive exercise program consisting of movements of the extremities in multiple positions and during functional tasks such as sitting, standing, and walking. Ten to 15 min of daily home exercise. Six weeks of treatment.	Pain (VAS) Disability (FSQ) Outcomes assessed at 6 weeks after randomization
Kilpikoski 200940,41	N = 119* Mean age = 44 years* Mean symptom duration: NA Recurrent symptoms: 93% (111/119)* Setting: referred by occupational health care centers in Finland	LBP with or without pain radiating to 1 or both legs, age 18–65 years, employed, acute or chronic duration, first or recurrent episodes, centralization present at initial assessment	McKenzie: consisted of an educational component, supported with the ‘“Treat Your Own Back”’ book, and an active therapy component with provided instructions in exercises repeated several times a day according to the principles of the approach (10–15 repetitions every 1–2 h with or without a sustained end-range position on a regular basis according to symptom response). On occasions, if improvements were not sustained or were too slow in coming, patient-generated forces were supplemented by clinician-generated forces: therapist’s over-pressure or mobilization or both within same treatment direction principle of management Orthopedic Manual Therapy (OMT): patients underwent spinal manipulation if indicated, specific mobilization, and muscle-stretching techniques. Furthermore, patients were taught to perform self-mobilization and stretching exercises at home once a day. Usually, 3–5 individually selected home-exercises were prescribed to actively mobilize the low back, with 2–3 sets of 15–20 repetitions for each exercise, and lumbar stabilization exercises with 10 repetitions of 10 sec, and stretching exercises to be performed once a day for 45–60 sec Advice: 45–60 min counseling from a physiotherapist concerning the good prognosis for LBP and concerning pain tolerance, medication, and early return to work. Patients were told to avoid bed rest and continue their routine as actively as possible, including exercise activities, within the limits permitted by their back pain. They were also instructed to contact their physicians if their symptoms worsened. For support, a 2-page educational booklet ‘The Back Book’ was also supplied Number of visits was one for subjects in the advice-only group, and ranged from 3 to 7 in the OMT and McKenzie groups (mean 6 treatments in each group)	Low Back Pain (VAS) Disability (RMDQ) Outcomes assessed after treatment (McKenzie and OMT groups only), and at 3, 6 and 12 months after randomization (McKenzie, OMT and Advice groups)
Petersen 201142	N = 307* Mean age = 37 years Mean symptom duration: 95 weeks Setting: primary care specialist center in Copenhagen, Denmark	Persistent low back pain with or without leg pain, >6 weeks; able to speak and understand Danish; centralization or peripheralization of symptoms; 18–60 years of age	McKenzie: directional-preference exercises; manual vertebral mobilizations were not allowed. Educational booklet ‘Treat Your Own Back’ and lumbar roll for posture change at therapist discretion Spinal Manipulation: vertebral mobilization and high-velocity thrust as well as myofascial trigger point massage at therapist discretion. General mobilizing exercises, that is, self-manipulation, alternating lumbar flexion/extension movements, and stretching, were allowed but not specific exercises in the directional preference. An inclined wedged pillow for correction of the seated position was available to the patients if the chiropractor believed this to be indicated Both groups: given stabilization/stretching/mobilizing/strengthening home exercises at therapist discretion, and encouraged to continue exercising at home or at the gym post intervention. All patients received the booklet ‘The Back Bok”. Maximum of 15 treatments for no more than 12 weeks	Back Pain‡ (The Back and Leg Pain Questionnaire) Disability (RMDQ), Global Perceived Effect (GPE) Outcomes assessed after treatment (until 3 months), at 5 months and at 15 months after randomization

Abbreviations: VAS, visual analog scale; RMDQ, Roland Morris Disability Questionnaire; GPE, Global Perceived Effect; SRI, Self-Rated Improvement; FSQ, Functional Status Questionnaire; OMT, Orthopedic Manual Therapy; NA, not available.

*Refers to sample of participants with a directional preference.

^†Calculated from the average of the 3 treatment groups; the authors excluded 22 participants who reported a symptom duration of >100 weeks for this measure.

^‡Data on back pain intensity were provided by authors upon request.

We did not include the short-term results (immediately after treatment according to our pre-specified criteria of <3 months) of the comparison of the McKenzie Method vs. advice from the three-armed Kilpikoski’s trial [41] in our analysis, since the authors did not measure outcomes at this time point for the advice group and only carried forward their baseline values.

Risk of bias

Overall risk of bias for pain in the short term was rated as some concerns in four [36,37,41,42], and as high risk in one study [39] (Figure 2). All five studies were rated as some concerns in the dimension ‘measurement of the outcome’, and as low risk of bias in ‘missing outcome data’ and in ‘deviations from the intended interventions’. Three studies were rated as some concerns in ‘selection of the reported result’ [39,41,42], and two studies were rated as low risk in this domain. [36,37] Two studies had low risk of bias in ‘randomization process’ [36,37], two had some concerns [41,42] and only one study was rated as high risk of bias in this domain. [39] For judgments about risk of bias of intermediate- and long-term outcomes, see Appendix 4.

Figure 2.

Risk of bias for each outcome in the short term. Green = low risk, yellow = some concerns, red = high risk. D1= randomization process, D2 = deviations from intended interventions, D3 = missing outcome data, D4 = measurement of the outcome, D5 = selection of the reported result. Judgements were consistent (identical) at different follow-ups from the same study (intermediate and long term) and among different comparators in three-armed trials.

Intervention effects

The summary of all the results is presented in Table 2 and detailed in what follows.

Table 2.

Summary of results and certainty of evidence in the evaluation of the effectiveness of the McKenzie method vs. comparators for chronic low back pain with a directional preference.

Outcome	Comparator	Follow-up	Number of participants (Studies)	Effect Size (95% CI)*	Certainty of Evidence
Pain Intensity	All other interventions	Short term	643 (5)	MD −1.11 (−1.83 to −0.40)	Low
		Intermediate	410 (2)	MD −0.97 (−1.70 to −0.08)	Low
		Long term	469 (3)	MD −0.36 (−0.79 to 0.08)	Moderate
	Exercise	Short term	281 (3)	MD −1.53 (−2.51 to −0.54)	Low
		Long term	63 (1)	MD 0.50 (−0.65 to 1.65)	Low
	Manual therapy	Short term	362 (2)	MD −0.51 (−0.94 to −0.07)	Low
		Intermediate	357 (2)	MD −0.49 (−0.89 to −0.09)	Low
		Long term	353 (2)	MD −0.47 (−0.95 to 0.01)	Low
	Minimal intervention	Intermediate	53 (1)	MD −2.10 (−2.94 to −1.26)	Low
		Long term	53 (1)	MD −0.60 (−1.56 to 0.36)	Very low
Disability	All other interventions	Short term	642 (5)	SMD −0.33 (−0.52 to −0.14)	Moderate
		Intermediate	409 (2)	SMD −0.53 (−0.97 to −0.09)	Low
		Long term	461 (3)	SMD −0.25 (−0.44 to −0.07)	Moderate
	Exercise	Short term	281 (3)	SMD −0.50 (−0.74 to −0.25)	Moderate
		Long term	63 (1)	SMD −0.12 (−0.61 to 0.38)	Low
	Manual therapy	Short term	361 (2)	SMD −0.16 (−0.40 to 0.08)	Low
		Intermediate	356 (2)	SMD −0.29 (−0.51 to −0.07)	Low
		Long term	345 (2)	SMD −0.23 (−0.44 to −0.02)	Low
	Minimal intervention	Intermediate	53 (1)	SMD −1.01 (−1.58 to −0.43)	Low
		Long term	53 (1)	SMD −0.59 (−1.14 to −0.03)	Low
Global Impression of Recovery	All other interventions	Short term	547 (4)	SMD 0.87 (0.14 to 1.59)	Moderate
	Exercise	Short term	265 (3)	SMD 1.13 (0.50 to 1.76)	Low
	Manual therapy	Short Term	282 (1)	SMD 0.15 (−0.09 to 0.38)	Low

Abbreviations: MD, Mean Difference; SMD, Standardized Mean Difference; CI, Confidence Interval.

*Bolded effect sizes indicate statistical significance. For outcomes pain intensity and disability, a negative effect size represents an effect in favor of the intervention group (McKenzie Method); for outcome global impression of recovery, a positive effect represents an effect in favor of the intervention group (McKenzie Method).

Also, a detailed description of all GRADE assessments can be found in Appendix 5.

Pain intensity

Short term

The McKenzie Method resulted in a moderate effect size for pain reduction in the short term compared to all other conservative interventions combined (MD −1.11; 95% CI −1.83 to −0.40; I² 74%; low certainty evidence). Compared to exercise, its effect was moderate (MD −1.53; 95% CI −2.51 to −0.54; I² 66%; low certainty), and compared to manual therapy, small (MD −0.51, 95% CI −0.94 to −0.07; I² 0%; low certainty) (Figure 3A). There was no comparison with minimal intervention at this time point. Sensitivity post-hoc analysis excluding Halliday et al. [37] removed all the remaining heterogeneity in the comparison to exercise (MD −2.01; 95% CI −2.60 to −1.41; I² 0%) (Appendix 6).

Figure 3.

Forest plots of effectiveness of McKenzie method vs comparators for pain in the A) short term, B) intermediate term and C) long term. *Adjusted between-group mean differences reported without adjusted means and standard deviations, ^† comparison group received multidirectional exercise, ^‡ comparison group received exercise opposite to directional preference, ^§ comparison group received orthopedic manual therapy, ǁ comparison group received advice.

Intermediate term

In the intermediate term, the McKenzie Method had a small effect for pain reduction compared to all other conservative interventions combined (MD −0.97; 95% CI −1.87 to −0.08; I² 83%; low certainty). Compared to manual therapy, its effect was small (MD −0.49; 95% CI −0.89 to −0.09; I² 0%; low certainty) and compared to minimal intervention (advice) it was large (MD −2.10; 95% CI −2.94 to −1.26; low certainty; one study) (Figure 3B). There was no comparison with exercise at this time point. No heterogeneity remained within subgroups.

Long term

The McKenzie Method’s effects in the long term for pain reduction were not significantly different compared to all other interventions combined (MD −0.36; 95% CI −0.79 to 0.08; I² 10%; moderate certainty). Long-term effects were not significantly different compared to exercise (MD 0.50; 95% CI −0.65 to 1.65; low certainty; one study), manual therapy (MD −0.47; 95% CI −0.95 to 0.01; I² 0%; low certainty) and minimal intervention (advice) (MD −0.60; 95% CI −1.56 to 0.36; very low certainty; one study) (Figure 3C).

Disability

Short term

In the short term, the McKenzie Method presented a small effect for disability reduction compared to all other interventions combined (SMD −0.33; 95% CI −0.52 to −0.14; I² 19%; moderate certainty). Compared to exercise, its effect was moderate (SMD −0.50; 95% CI −0.74 to −0.25; I² 0%; moderate certainty) but compared to manual therapy the difference was not significant (SMD −0.16; 95% CI −0.40 to 0.08; I² 12%; low certainty) (Figure 4A). There was no comparison with minimal intervention at this time point.

Figure 4.

Forest plots of effectiveness of McKenzie method vs comparators for disability in the (a) short term, (b) intermediate term and (c) long term. * comparison group received multidirectional exercise, ^† comparison group received exercise opposite to directional preference, ^‡ comparison group received orthopedic manual therapy, ^§ comparison group received advice.

Intermediate term

In the intermediate term, the McKenzie approach had a moderate effect for disability reduction compared to all other conservative interventions combined (SMD −0.53; 95% CI −0.97 to −0.09; I² 69%; low certainty). Compared to manual therapy, its effect was small (SMD −0.29; 95% CI −0.51 to −0.07; I² 5%; low certainty) and compared to minimal intervention (advice) it was large (SMD −1,01; 95% CI −1,58 to −0.43; low certainty; one study) (Figure 4B). There was no comparison with exercise at this time point. The heterogeneity in the comparison to other conservative interventions was largely explained by the comparators’ subgroups.

Long term

In the long term, the McKenzie Method had a small effect for disability reduction compared to all other conservative interventions combined (SMD −0.25; 95% CI −0.44 to −0.07; I² 0%; moderate certainty). Compared to manual therapy, its effect was small (SMD −0.23; 95% CI −0.44 to −0.02; I² 0%; low certainty) and compared to minimal intervention (advice) it was moderate (SMD −0,59; 95% CI −1.14 to −0.03; low certainty; one study). No significant difference was found compared to exercise (SMD −0.12; CI 95% −0.61 to 0.38; low certainty; one study) (Figure 4C).

Global impression of recovery

The McKenzie approach resulted in a large effect on patient’s global impression of recovery compared to all other conservative interventions combined in the short term (SMD 0.87; 95% CI 0.14 to 1.59; I² 93%; moderate certainty). Compared to exercise its effect was large (SMD 1.13; 95% CI 0.5 to 1.76; I² 81%; low certainty) but compared to manual therapy there was no significant difference (SMD 0.15; 95% CI −0.09 to 0.38; low certainty; one study) (Figure 5). No study compared the McKenzie Method to minimal intervention for this outcome. Heterogeneity source was not identifiable.

Figure 5.

Forest plot of effectiveness of McKenzie method vs comparators for global impression of recovery in the short term. * comparison group received exercise opposite to directional preference, ^† comparison group received multidirectional exercise.

Sensitivity analysis

The exclusion of the only (small) study at high risk of bias (Miller et al.) [39] did not change our estimates and conclusions (Appendix 7).

Discussion

We found low-to-moderate certainty evidence that the McKenzie Method, when delivered by credentialed therapists to patients with chronic LBP with a DP, produced significant improvement beyond that provided by all other conservative interventions combined in pain up to 6 months and in disability up to 12 months. Clinically important effects for pain and disability were observed in comparison to exercise approaches in the short term and to minimal intervention in the intermediate term. The McKenzie approach also resulted in a clinically important reduction in long-term disability compared to minimal intervention but was not superior to any comparator for long-term pain. The comparison with manual therapy showed at most small clinically unimportant effects at any time point. Our findings of favorable patients’ global impressions of recovery corroborate with our main findings in the short term. The McKenzie Method produced a clinically important increase in patients’ perceived recovery when compared to exercise but was not different from manual therapy.

Through selection of trials specifically focused on use of the McKenzie Method for the subset of patients with persistent LBP with a DP, appropriately assessed and treated by credentialed McKenzie therapists, we provide evidence of a larger benefit from its application than previous reviews. Lam et al. (2018) reviewed treatment of chronic LBP independent of the presence of DP with the McKenzie Method and found it to result in a small short-term effect on pain and disability when compared to other conservative interventions, while Sanchis-Sánchez et al. (2021) showed no benefit. In our review, in most comparisons where the McKenzie Method was superior in the short term, we observed moderate and clinically important effect sizes. These differences in effect magnitudes could be explained by the fact that we analyzed a more homogeneous subset (patients with DP) that might respond best to matched directional treatment. Moreover, Lam and Sanchis-Sánchez reviews accepted studies in which therapists were not credentialed in the McKenzie Method. Insufficient training might have resulted in lack of reliable classification [19] and in therapists being unable to follow the core principles of the McKenzie approach.

Our findings are consistent with those of Halliday et al. (2019), who showed that among trials reporting an evaluation of the McKenzie Method, those in which therapists followed the core principles of the McKenzie Method (patients assessed and classified, treatment individualized based on the classification, forces applied according to symptomatic response) produced greater reductions in pain and disability (of clinically relevant magnitude) than those trials in which therapists did not. [20]

Tagliaferri et al. (2022) recently compared different classification systems to generalized interventions for LBP and found the McKenzie Method to have small non-clinically important effects on intermediate-term disability, with no other significant results at other time points or outcomes. In a secondary analysis, they assessed the sub-classification of directional preference and found moderate effects only for short-term pain but questioned the clinical importance of this result. Most studies included in their analysis used therapists who were not credentialed in the McKenzie Method. [43]

We found the McKenzie Method to possibly result in clinically important short-term reductions in pain and disability compared to other exercise approaches. In line with these findings, Hayden et al. (2021) did a Cochrane network meta-analysis of 217 trials comparing many different forms of exercise for chronic LBP and concluded that the McKenzie Method, Pilates and functional restoration were superior to other exercise approaches to reduce short-term pain and disability. [44] What our study adds is that for the approximately 60% of chronic LBP patients who present a DP, [16] the McKenzie Method delivered by sufficiently trained therapists, when compared to other exercise regimens, produced larger, moderately sized and clinically important reductions in short-term pain and disability.

Owen et al. (2020) conducted another network meta-analysis of exercise therapies for chronic LBP, but contrary to our (and Hayden’s) findings, concluded that the McKenzie approach was no better than control interventions in the short term. [45] Different inclusion criteria led to different studies being included in each review. Most studies included in Owen’s analysis had therapists with no training in the McKenzie Method, whereas in Hayden’s review most studies had therapists who were credentialed or partially trained in this approach.

Our review has some strengths: we followed a pre-registered protocol, reported results according to PRISMA and followed Cochrane recommendations. There were no language restrictions in our literature search. We used the revised Cochrane Risk of Bias 2 tool to assess risk of bias and GRADE to assess the certainty of the body of evidence for each outcome during each period. Additionally, we evaluated benefits across short, medium, and long terms.

The studies included in our meta-analysis had some methodological limitations. One study [39] had outcomes judged to be at overall high risk of bias due to high risk of bias in the randomization process. Sensitivity analysis excluding this study did not affect our results and conclusions. A few of the meta-analyses in which we compared the McKenzie Method to other conservative interventions (taken together) presented important inconsistency, which was explained (or partially explained) by our preplanned subgroup analysis by comparator. Following our protocol, the expected heterogeneity was incorporated in the random-effects meta-analysis. We could not assess publication bias due to the small number of trials.

As a possible limitation of this review, we did not search in the gray literature. Also, we planned to use SMD for pain but decided to use MD instead to increase interpretability of findings. This change from our protocol did not affect our conclusions (data of post hoc sensitivity analysis using SMD for pain is available in Appendix 8). An additional limitation stems from the small number of studies and the methodological limitations of the trials, which resulted in increased risk of bias, inconsistency and imprecision in our estimates, reducing our certainty in the body of evidence. Finally, that we only assessed studies with credentialed McKenzie therapists and only analyzed the data of patients with a DP limits the generalizability of our findings. Nonetheless, since a large proportion (over 60%) of chronic LBP patients present a DP on physical examination, [16] a considerable number of patients could benefit from this approach.

New evidence likely will change our estimates. Given the large disease burden resulting from chronic LBP, we need large, high-quality multicentric studies that assess the effect over short, intermediate and long terms of the McKenzie Method delivered by adequately trained therapists in accordance with the Method’s core principles in patients with chronic LBP with a DP.

Conclusion

We found low-to-moderate certainty evidence from five trials that, in patients with chronic LBP with a DP, the McKenzie Method delivered by credentialed therapists was superior to all other conservative interventions combined for up to 6 months for pain and up to 12 months for disability. Clinically important improvements were observed in the following outcomes: short-term pain and intermediate-term disability compared to all other conservative interventions combined; short-term pain and disability compared to exercise; intermediate-term pain and disability compared to minimal intervention, and long-term disability compared to minimal intervention.

Biographies

Vicente Hennemann is a physical therapist and a clinical researcher with a Master’s in Epidemiology from the Federal University of Rio Grande do Sul, Brazil. He runs a full-time private practice where he helps patients with musculoskeletal disorders and chronic pain and is also a faculty member for the McKenzie Institute. His research interests include studying the effectiveness of interventions for musculoskeletal disorders, conducting systematic reviews, and investigating pain mechanisms.

Patrícia K. Ziegelmann is a full professor of the Statistics Department and the Postgraduate Program in Epidemiology at Universidade Federal do Rio Grande do Sul (Brazil). She is also a researcher at the Brazilian Institute for Health Technology Assessment (IATS) and a fellow of the National Council for Scientific and Technological Development (CNPq). She has extensive experience in the field of evidence synthesis, metanalysis, network metanalysis, and MAIC. She currently coordinates the PIT Cancer research group – Translational Integrative Research in Cancer, where she has been working with survival analysis based on real-world data.

Miriam A. Z. Marcolino is a physical therapist, and doctorate student at the Postgraduate Program in Epidemiology at Universidade Federal do Rio Grande do Sul (Brazil). She is also a researcher at the Brazilian Institute for Health Technology Assessment (IATS). She has experience in the Health Technology Assessment field, including systematic reviews with meta-analysis, health economic modeling, and real word data analysis.

Bruce B. Duncan is an internist and an epidemiologist. He is currently a full professor of social medicine at the Universidade Federal do Rio Grande do Sul (UFRGS). His early work (1984-1995) focused on spreading capacity in clinical epidemiology throughout Brazil. He is a co-founder and an active member of the faculty of the Postgraduate Studies Program in Epidemiology-UFRGS, a fellow of the Brazilian National Council for Scientific and Technological Development (CNPq) since 1984, and a member of the Brazilian Academy of Science. He is the editor of a major Brazilian primary care textbook and has published extensively on questions of primary care and the etiology, management, and burden of chronic non-communicable diseases.

Funding Statement

The author(s) reported there is no funding associated with the work featured in this article.

Disclosure statement

Vicente Hennemann is an instructor for the McKenzie Institute.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/10669817.2024.2408084