The mechanisms of manual therapy: a living review of systematic, narrative, and scoping reviews

Key points

1. 0162 reviews (35 systematic, 24 narrative, 4 scoping) were synthesized across 8 mechanistic domains
2. 02Neurological effects were most consistently supported: local pressure pain threshold increased in 12 reviews versus control or sham
3. 03Neurovascular effects included sympathoexcitation in most reviews and increased skin conductance in 12 reviews, though skin temperature showed no consistent change
4. 04Biomechanical changes (joint position, disc physiology, soft-tissue viscoelastic properties) were supported only by critically low quality reviews and did not clearly correlate with pain improvement
5. 05The clinical relevance of these mechanistic findings has not been well established; it is uncertain which mechanisms actually mediate the therapeutic response

How it was conducted

Design

Living review of systematic, narrative, and scoping reviews (review of reviews); registered prospectively with PROSPERO (CRD42023444839)

Databases searched

Medline, CINAHL, Cochrane Library, PEDro; initial search to October 3, 2023 with updated search to September 23, 2024

Included reviews

62 reviews: 35 systematic reviews, 24 narrative reviews, 4 scoping reviews

Populations covered

Asymptomatic humans (n=37 reviews), symptomatic humans (n=43), unspecified humans (n=7), animal models (n=6)

MT techniques covered

Manipulation/thrust (n=41 reviews), mobilization/non-thrust (n=23), soft tissue mobilization or massage (n=19)

Quality appraisal

AMSTAR-2 for methodological quality; ROBIS for risk of bias; applied to 39 systematic and scoping reviews

What they found

• 62 reviews included after screening 442 reviews post-duplicate removal; reviewer agreement was moderate to strong (title kappa=0.77, 95% CI 0.69-0.84; abstract kappa=0.83, 95% CI 0.78-0.89; full-text kappa=0.98, 95% CI 0.94-0.99)
• Of 39 reviews appraised for quality: 23 were critically low quality, 12 low quality, 4 moderate quality; 14 were high risk of bias and 25 low risk of bias
• Neurovascular: 12 reviews favored sympathoexcitation; increased skin conductance reported in 12 reviews; no change in skin temperature in 14 reviews; alpha-amylase decrease (sympathoinhibitory signal) across 3/3 reviews
• Neurological: local pressure pain threshold increased versus control or sham in 12 reviews; 4 reviews reported no effect on thermal pain threshold; improved conditioned pain modulation and reduced temporal summation supported by 1 review
• Neurotransmitter/neuropeptide: spinal manipulation decreased Substance P (SMD -0.48, 95% CI -0.87 to -0.10, low quality evidence); decreased oxytocin (SMD -2.61, 95% CI -3.5 to -1.72, low quality); decreased neurotensin (SMD -1.8, 95% CI -2.56 to -1.04, low quality); little to no change in norepinephrine or epinephrine across 7 reviews
• Neuroimmune: trends toward decreased pro-inflammatory cytokines (IL-1beta, TNF-alpha) and increased anti-inflammatory cytokines (IL-2, IL-10) across reviews, but moderate quality evidence for influencing IL concentration from only one systematic review
• Neuroendocrine: all reviews investigated cortisol with variable direction of change; larger response than control in 2 reviews; no significant difference from sham or control in 5 reviews
• Biomechanical: 14 critically low quality reviews; increased disc diffusion with 2 studies translating to improved clinical outcomes; no correlation found between joint position changes and pain or impairment in 1 review

Limitations

• Overall quality was predominantly critically low to low, with only 4 moderate-quality reviews, limiting confidence in conclusions
• The review did not differentiate mechanisms by MT technique type (manipulation vs. mobilization vs. soft tissue), despite these potentially having distinct effects
• Significant heterogeneity across included reviews in MT technique, mechanistic outcome measures, and study populations limits synthesis and comparison
• Whether identified mechanistic changes are specific to MT or are non-specific responses to any intervention remains unclear, as effect sizes often diminish when compared to active control or sham

Why it matters

For patients

Patients can understand that hands-on therapy does more than stretch or move joints, it influences the nervous system, immune system, and pain pathways, though exactly which of these effects explains symptom relief is still not known.

For clinicians

Clinicians should recognise that MT produces multisystem effects beyond the local tissue targeted, but should not assume mechanistic changes (such as immune or hormonal shifts) automatically translate to clinical benefit, as this translation is poorly established.

For readers

This living review provides the most current synthesis of MT mechanism research across 8 biological domains, useful for researchers prioritising future high-quality translational studies linking mechanisms to clinical outcomes.