Does manual therapy meaningfully change quantitative sensory testing and patient reported outcome measures in patients with musculoskeletal impairments related to the spine?: A ‘trustworthy’ systematic review and meta-analysis

ABSTRACT

Objectives

To perform a ‘trustworthy’ systematic review (SR) with meta-analysis on the potential mechanisms of manual therapy used to treat spinal impairments.

Design

SR with meta-analysis

Literature Search

Articles published between January 2010 and October 2022 from CENTRAL, CINAHL, MEDLINE, PubMed, ProQuest, and PEDro.

Methods

This SR included English-language randomized clinical trials (RCTs) involving manual therapy to treat spinal impairments in adults. The primary outcome was pressure pain thresholds (PPTs). To synthesize RCTs with high confidence in estimated effects using the GRADE, RCTs with questionable prospective, external, and internal validity, and high risk of bias (RoB) were excluded.

Results

Following title and abstract screening, 89 full-text RCTs were reviewed. Twenty-two studies included the criteria of interest. Sixteen were not prospectively registered, two contained discussion/conclusions judged to be inconsistent with the registry, and one was rated as having a high RoB. Three studies met the inclusion criteria; heterogeneous interventions and locations for PPT testing prevented synthesis into practice recommendations. The two studies with high confidence in estimated effects had small effect sizes, and one study had confidence intervals that crossed zero for the outcome measures of interest.

Discussion

Standardized PPT testing, as a potential measure of centrally mediated pain, could provide clues regarding the mechanisms of manual therapy or help identify/refine research questions.

Conclusion

High-quality RCTs could not be synthesized into strong conclusions secondary to the dissimilarity in research designs. Future research regarding quantitative sensory testing should develop RCTs with high confidence in estimated effects that can be translated into strong recommendations.

Introduction

Neck and low back pain (LBP) are common problems. In the US, neck pain is the 6^th leading cause, and LBP is the leading cause of disability [1]. Globally, in 2019, the prevalence and incidence of neck pain (NP) were 2,696.5 and 579.1 per 100,000 [2]. Furthermore, the point prevalence of the LBP was reported to be 7.5% in 2017, impacting an estimated 577.0 million people globally [3].

In patients suffering from NP, there are no overall differences in outcomes between surgical and conservative management [4]. Those who choose surgery over conservative management for lumbar symptoms can expect short-term symptom relief without any clinically meaningful benefit over conservative treatment during medium (6-week) and long-term (12-week) follow-up visits [5]. This suggests the observed treatment effects of surgery and conservative management at medium- and long-term follow-up may be through a shared treatment mechanism, as conservative management does not correct pathoanatomy.

In clinical practice guidelines, therapeutic exercise and manual therapy are considered foundational components in the conservative management of neck and LBP [6–9]. Additionally, international clinical practice guidelines for the management of nonspecific LBP in primary care from 15 countries recommend exercising 93% (14/15) of the time and spinal manipulation 81% (9/11) of the time [10].

Understanding treatment mechanisms is a priority of the National Institutes of Health (NIH) [11]. The proposed treatment goals for manual therapy are to reduce pain and improve joint mobility [12,13]. Manual therapy [14–17] and strengthening exercises [18] have been shown to produce local and remote hypoalgesia. What is unknown is whether the observed outcomes of manual therapy and exercise address pain through unique or shared mechanisms [19,20] that alter pressure pain thresholds (PPTs) locally, regionally, and/or distally. Previous systematic reviews (SRs) have suggested manual therapy produces local hypoalgesia [21], local but unclear remote hypoalgesia [22], and both local and remote hypoalgesia [19,20,23], while other SRs have suggested no hypoalgesic effect [17] or there is insufficient evidence of an analgesic effect [24]. Therefore, in the future, it is important to determine if manual therapy decreases pain more than exercise therapy alone to improve the precision of care in a patient-centered model. If present, this precision may speed recovery and lower the overall cost of conservative management [25]. Additionally, improving the accuracy of conservative spine pain management may improve recovery and prevent treatment progression to more costly and invasive alternatives such as surgery. Identifying if patients with spine pain respond to the potential specific treatment mechanism of manipulation and mobilization may be important to prevent overutilization, over-treatment, and minimize the use of low-value care [26].

A systematic review (SR) by Millan et al. [22] examined manual therapy interventions’ effect on experimentally induced pain. Although experimentally induced pain models may help to control variables relevant to the patient’s experience [27], the large variability of the pain models used in research, INCLUDING the cold pressor model [28], intramuscular injection of hypertonic saline solution [29,30], heat/capsaicin sensitization, and intradermal capsaicin models [31], may add unwarranted variability to study findings. It is unknown if these pain models create nociceptive symptoms, peripheral neuropathic symptoms, or both [32]. Additionally, SRs that include asymptomatic participants’ responses to manual therapy may not be generalizable to clinical conditions [21,23]. A previous SR exploring mechanical quantitative sensory testing (QST) in individuals with nonspecific LBP found that PPTs were significantly lower in remote body regions in these individuals than in healthy controls [33]. This suggests central sensitization in this subgroup of patients [33]. However, interpreting these findings is difficult as this SR included studies with a high risk of bias (RoB) and large confidence intervals, significantly hindering the confidence in estimated effects [33]. It is unknown if PPT is a strict measure of centrally mediated pain or if it is modifiable through a specific treatment mechanism when manual therapy is applied.

Objectives

The objective of this SR was to identify if QST thresholds related to PPTs meaningfully change in response to spinal manual therapy interventions and determine if those potential changes are related to meaningful changes in patient-reported outcome measures (PROMs) in patients with musculoskeletal impairments related to the spine.

Methods

Protocol and registration

The protocol for this SR followed the trustworthy living SR project protocol published online on 9 September 2022 [34]. The protocol for this SR was reviewed by an institutional review board (IRB) and was deemed exempt from IRB oversight (University of Hartford, IRB-22-08-066). In addition, the protocol for this SR was prospectively registered through the International Prospective Register of Systematic Reviews (PROSPERO, CRD42022354953) on 9 January 2022 [35].

Design

This SR was reported in agreement with the PRISMA 2020 statement and flow diagram [36].

Eligibility criteria

This SR included randomized clinical trials (RCTs) published in English that meet the PICOTS criteria below [37]. The RCTs included Patients 18 years of age or older with subacute or chronic spinal musculoskeletal impairments consistent with an alteration in normal structure or function or an increase in pain or discomfort in the integument, muscles, bone, or joints of the body of an individual, limiting the function of the musculoskeletal system [38]. Subacute refers to pain present between 6 and 12 weeks [39], and chronic refers to pain present for 12 weeks or longer [40]. These timeframes were chosen secondary to the possibility that central sensitization may be a normal physiologic phenomenon and subacute and chronic symptoms cannot be differentiated from a physiological pain perspective [33]. Spinal refers to the joints of the cervical, thoracic, or lumbar regions. Manual therapy Interventions included mobilization and/or manipulation used to treat the spine. Mobilization refers to a treatment involving the therapist applying a sustained or oscillatory (at variable speeds and amplitudes) mechanical input to a joint to decrease pain and/or increase the range of motion. Manipulation refers to a treatment involving the clinician applying a high-velocity, low-amplitude thrust to a joint to reduce pain and/or increase the range of motion. The manual therapy interventions were Compared to placebo, no treatment, other forms of conservative care, or in addition to other forms of conservative care. The primary Outcomes included QST consisting of local, regional, and/or remote PPTs, as this outcome is most frequently identified in the literature [41]. Prospectively identified secondary outcomes of interest included measures of pain (Visual Analogue Scale [VAS], Numeric Pain Rating Scale [NPRS]), region-specific outcome measures, measures of the patient’s perceived improvement such as the Global Rating of Change (GRoC) [42] or Single Assessment Numeric Evaluation (SANE) [43], and measures of positive (Self-Efficacy) and negative (Fear-Avoidance, Kinesiophobia) psychological beliefs. Time of follow-up was [44]: Immediate = Closest to immediately following the intervention; Short-term = Closest to 1 month; Intermediate-term = Closest to 6 months; and Long-term = Closest to 12 months or longer. The types of Studies only included RCTs. Given that this SR assessed clinical outcomes, RCTs involving experimentally induced pain were excluded. Additionally, studies were excluded if they were: 1) pilot studies (secondary to being underpowered and non-definitive by design); 2) non-randomized trials (secondary to the heterogeneity of research design and journals not requiring prospective registration of these types of studies); 3) not published in a peer-reviewed journal; and 4) research not involving musculoskeletal interventions.

Information sources

The databases include Cochrane Central Register of Controlled Trials (CENTRAL); Cumulative Index to Nursing and Allied Health Literature (CINAHL); EBSCOhost MEDLINE (Medical Literature Analysis and Retrieval System Online); PubMed; PEDro, and ProQuest Nursing and Allied Health.

Search strategy

The Peer Review of Electronic Search Strategies (PRESS) checklist [45] was used by a professional librarian with expertise in SR search strategies to create the search strategies for each bibliographic database as described by Furlan et al. [46] and Lefevre et al. [47]. The search parameters included RCTs from 1 January 2010, to 1 October 2022. The specific search strategies used for this living SR can be found in Appendix 1. The search strategy was executed on 1 October 2022.

Study records

Data management

A Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) diagram is included in Figure 1. Title screening was performed in EndNote, and results were then imported to Covidence (Covidence systematic review software, Veritas Health Innovation, Melbourne, Australia; www.covidence.org) for further review. Two blinded authors (DF and SS) screened all abstracts and performed a full-text review using Covidence. A third reviewer was used to resolve conflicts if needed.

Figure 1.

Study Selection.

Selection process

RCTs registered prospectively with the registry matching the published manuscript regarding the study’s primary aim and outcome measure were screened consistent with the methodology of Riley et al. [48]. The assessment of these variables by two blinded reviewers is reliable [48]. Additionally, RCTs needed to meet criterion 1 of the PEDro score (external validity), have good to excellent internal validity on the PEDro score (PEDro ≥6), have a moderate-to-low-RoB on revised Cochrane RoB 2 tool for randomized trials, and have a moderate to high methodological quality on the GRADE [48]. These selection criteria were developed to prevent downgrading the strength of recommendations when using the GRADE Evidence to Recommendation Framework [49] while also identifying areas where further high-quality, low RoB research is needed. When available, official PEDro scores were used. If official PEDro scores were unavailable, two blinded authors (DF and SS) independently scored the RCTs on the PEDro. The RoB 2 and GRADE were assessed using the same methodology. A third reviewer was used to resolve conflicts. The details of the data collection process can be found in Appendix 2.

Data items

RCTs meeting the inclusion and exclusion criteria were assessed as previously discussed to ensure they had: established external validity on the PEDro; moderate-to-high internal validity on the PEDro; moderate-to-low RoB on the RoB 2; and moderate-to-high quality on the GRADE criteria. The data items included QST, PROMs for pain (VAS, NPRS), and region-specific PROMs. While initially planned for inclusion, no other secondary outcomes were reported in the selected RCTs. The time points of interest were previously discussed in the PICOTS section [44].

Data syntheses

While initially planned for meta-analysis, the results are reported descriptively due to the heterogeneity of outcome measures utilized in the included studies. The clinically meaningful differences for pain intensity and region-specific PROMs were established by taking the most conservative thresholds for the tools identified based on the literature [50], given the wide variability of these values [51]. The most conservative threshold for PPT has been reported to be 150 kilopascals (kPa) [52].

Confidence in cumulative evidence

The strength of the final recommendation was established by ensuring confidence in the estimates of effect. This systematic review included only prospectively registered RCTs published in peer-reviewed journals consistent with the trial registry. Without verifying prospective intent, assessments of the external validity, internal validity, and risk bias of RCT are not possible. Finally, studies where external validity could not be determined (PEDro criterion 1), rated as low in methodological quality (PEDro <6), or rated as high in bias using the Cochrane collaboration RoB 2 tool were excluded as they decreased the confidence in estimated effects. The strength of the recommendation was planned to be established based on the GRADE evidence to recommendation framework [49].

Protocol deviations

The number of the identified studies (three) did not allow for a statistical determination of heterogeneity and a meta-analysis. We, therefore, assessed the individual studies descriptively based on their clinical and methodological heterogeneity to determine what variables were comparable between the RCTs. Due to clinical and methodological heterogeneity, it was impossible to synthesize the identified studies into practice recommendations using the GRADE criteria. Consequently, we discussed the individual studies from the context of confidence in their estimated effects. Confidence in the estimated effect was determined by examining p-values (statistical significance), estimated effects (differences larger than the minimally detectable change [MDC], minimal clinically important difference [MCID], and/or at least a moderate effect size), and precision (the size of the reported confidence interval and if the confidence intervals overlapped).

Results

Study selection (flow of studies)

The study selection process is illustrated in Figure 1. After title screening in EndNote, four hundred and fifty-one studies were imported into Covidence for screening, three duplicates were removed, and 448 abstracts were screened. After abstract screening, 89 full-text studies were reviewed. Thirty had the wrong population, 28 used the wrong outcome measures, and nine used the wrong intervention.

Twenty-two studies were screened for prospective registration, PEDro quality scores, and the Cochrane collaboration RoB 2 assessment (Figure 2). Five studies were ruled out for not being registered. Eleven of the 17 registered studies were removed for not being prospectively registered. Two of the six prospectively registered studies were removed secondary to the discussion and conclusion not matching the primary outcome. This left four studies that were screened for quality and RoB. Complete citations for these studies are reported in Table 1, and the studies excluded with reasons are reported in Table 2.

Figure 2.

Prospective registration, quality, and risk of bias screening.

Table 1.

Studies included for quality and risk of bias assessment.

Authors	Year	Title	Journal	Volume	Issue	Pages
Carrasco-Martinez et al. [53]	2019	Short-term effectiveness of the flexion-distraction technique in comparison with high-velocity vertebral manipulation in patients suffering from low-back pain	Complement Ther Med	44	n/a	61–67
de Oliveira et al. [54]	2020	Directed vertebral manipulation is not better than generic vertebral manipulation in patients with chronic low back pain: a randomized trial	J Physiother	66	3	174–179
Rodriguez-Sanz et al. [55]	2020	Does the Addition of Manual Therapy Approach to a Cervical Exercise Program Improve Clinical Outcomes for Patients with Chronic Neck Pain in Short- and Mid-Term? A Randomized Controlled Trial	Int J Environ Res Public Health	17	18	n/a
Valera-Calero et al. [56]	2019	Endocrine response after cervical manipulation and mobilization in people with chronic mechanical neck pain: a randomized controlled trial	Eur J Phys Rehabil Med	55	6	792–805

Table 2.

Studies excluded with Reasons.

Authors	Year	Title	Journal	Volume	Issue	Pages	Reason
Alshami et al. [57]	2021	Effect of manual therapy with exercise in patients with chronic cervical radiculopathy: a randomized clinical trial	Trials	22	1	716	Retrospective registry
Bautista-Aguirre et al. [58]	2017	Effect of cervical vs. thoracic spinal manipulation on peripheral neural features and grip strength in subjects with chronic mechanical neck pain: a randomized controlled trial	Eur J Phys Rehabil Med	53	3	331–341	Retrospective registry
Behrangrad et al. [59]	2017	Comparison of ischemic compression and lumbopelvic manipulation as trigger point therapy for patellofemoral pain syndrome in young adults: A double-blind randomized clinical trial	J Bodyw Mov Ther	21	3	554–564	Retrospective registry
Casanova-Méndez et al. [60]	2014	Comparative short-term effects of two thoracic spinal manipulation techniques in subjects with chronic mechanical neck pain: a randomized controlled trial	Man Ther	19	4	331–337	Retrospective registry
de Oliveira et al. [61]	2013	Immediate effects of region-specific and non-region-specific spinal manipulative therapy in patients with chronic low back pain: a randomized controlled trial	Phys Ther	93	6	748–756	Retrospective registry
Fagundes Loss et al. [62]	2020	Immediate effects of a lumbar spine manipulation on pain sensitivity and postural control in individuals with nonspecific low back pain: a randomized controlled trial	Chiropr Man Therap	28	1	25	Discussion and conclusion did not match primary outcome
Galindez-Ibarbengoetxea et al. [63]	2018	Short-term effects of manipulative treatment versus a therapeutic home exercise protocol for chronic cervical pain: A randomized clinical trial	J Back Musculoskelet Rehabil	31	1	133–145	Unregistered
Galindez-Ibarbengoetxea et al. [64]	2018	Immediate Effects of Osteopathic Treatment Versus Therapeutic Exercise on Patients with Chronic Cervical Pain	Altern Ther Health Med	24	3	24–32	Unregistered
García-Pérez-Juana et al. [65]	2018	Changes in Cervicocephalic Kinesthetic Sensibility, Widespread Pressure Pain Sensitivity, and Neck Pain After Cervical Thrust Manipulation in Patients with Chronic Mechanical Neck Pain: A Randomized Clinical Trial	J Manipulative Physiol Ther	41	7	551–560	Retrospective registry
Kardouni et al. [66]	2015	Immediate changes in pressure pain sensitivity after thoracic spinal manipulative therapy in patients with subacromial impingement syndrome: A randomized controlled study	Man Ther	20	4	540–546	Unregistered
La Touche et al. [67]	2013	Does mobilization of the upper cervical spine affect pain sensitivity and autonomic nervous system function in patients with cervico-craniofacial pain?: A randomized-controlled trial	Clin J Pain	29	3	205–215	Unregistered
Lopez-Lopez et al. [68]	2015	Mobilization versus manipulations versus sustain apophyseal natural glide techniques and interaction with psychological factors for patients with chronic neck pain: Randomized controlled trial	Eur J Phys Rehabil Med	51	2	121–132	Retrospective registry
Martínez-Segura et al. [69]	2012	Immediate changes in widespread pressure pain sensitivity, neck pain, and cervical range of motion after cervical or thoracic thrust manipulation in patients with bilateral chronic mechanical neck pain: a randomized clinical trial.	J Orthop Sports Phys Ther.	42	9	806–814	Unregistered
Packer et al. [70]	2014	Effects of upper thoracic manipulation on pressure pain sensitivity in women with temporomandibular disorder: a randomized, double-blind, clinical trial.	Am J Phys Med Rehabil.	93	2	160–168	Retrospective registration
Arjona Retamal et al. [71]	2021	Effects of instrumental, manipulative, and soft tissue approaches for the suboccipital region in subjects with chronic mechanical neck pain. A randomized controlled trial	Int J Environ Res Public Health	18	16	8636	Discussion and conclusion did not match primary outcome
Salom-Moreno et al. [72]	2014	Immediate changes in neck pain intensity and widespread pressure pain sensitivity in patients with bilateral chronic mechanical neck pain: A randomized controlled trial of thoracic thrust manipulation vs non-thrust mobilization	J Manipulative Physiol Ther	37	5	312–319	Retrospective registry
Sarker et al. [73]	2019	Effect of Spinal Manipulation on Pain Sensitivity, Postural Sway, and Health-related Quality of Life among Patients with Non-specific Chronic Low Back Pain: A Randomised Control Trial	Journal of Clinical & Diagnostic Research	13	2	n/a	Retrospective registry
Snodgrass et al. [74]	2014	Dose optimization for spinal treatment effectiveness: A randomized controlled trial investigating the effects of high and low mobilization forces in patients with neck pain	J Orthop Sports Phys Ther	44	3	141–152	Retrospective registry

Three of the four studies had an official PEDro score, and one was assigned a PEDro score through individual blinded assessment and consensus (Table 3). All four studies met criterion 1 of external validity based on inclusion and exclusion criteria and had PEDro scores of 8. The Cochrane RoB 2 assessment revealed that the clinical trials by Carrasco-Martinez et al. [53], de Oliveira et al. [54], and Valera-Calero et al. [56] had a ‘moderate RoB.’ The study by Rodriguez-Sanz et al. [55] had ‘some concerns’ related to the randomization process and was assessed to have a ‘high risk’ of bias pertaining to missing outcome data. These shortcomings resulted in this study being rated as having a ‘high risk’ of bias. It was therefore eliminated from the inclusion of further analysis secondary to low confidence in the estimated effect.

Table 3.

Quality and risk of risk of bias assessment.

Authors	Official PEDro Scale	PEDro Criterion 1	PEDro Scores	RoB 2 Randomization Process	RoB 2 Deviations from the intended interventions	RoB 2 Missing outcome data	RoB 2 Measurement of the outcome	RoB 2 Selection of the reported result	RoB 2 Overall Risk
Carrasco-Martinez et al. [53]	No	Yes	8	Low risk	Low risk	Low risk	Low risk	Some concerns	Some concerns
de Oliveira et al. [54]	Yes	Yes	8	Low risk	Low risk	Low risk	Low risk	Some concerns	Some concerns
Rodriguez-Sanz et al. [55]	Yes	Yes	8	Some Concerns	Low risk	High risk	Low risk	Some concerns	High risk
Valera-Calero et al. [56]	Yes	Yes	8	Low risk	Low risk	Low risk	Low risk	Some concerns	Some concerns

Note: PEDro = Physiotherapy Evidence Database; RoB 2 = Cochrane risk of bias tool.

Study characteristics

The key characteristics, results, clinical heterogeneity, description of the interventions, and methodological heterogeneity of individual studies can be found in Tables 4 through 5. The papers by Carrasco-Martinez et al. [53] and de Oliveira et al. [54] were similar in the duration of symptoms (chronic, defined as greater than three months) and region (lumbar). There were otherwise dissimilarities between the interventions and the comparators, including the location for PPT testing, how the PPT was measured, the duration of follow-up, how pain was assessed based on duration, the patient-reported outcome measures, and perceived improvement. No measures were utilized for psychological beliefs.

Table 4.

Clinical heterogeneity.

Authors	Region	Duration of Symptoms	Age, years	Gender	Setting	Follow-Up
Carrasco-Martinez et al. [53] (n = 150)	Lumbar	Chronic (>3 months)	FDT 43.37 (12.93)* HVLS-SM	FDT Male 40.0 (53.3)* Female 35.0 (46.7)* HVLS-SM Male 31.0.0 (41.3)* Female 44.0 (58.7)*	Physiotherapy clinic in Andalusia, Spain	Immediate
de Oliveira et al. [54] (n = 143)	Lumbar	Chronic (>3 months)	Symptomatic spinal level = 45 (13)* Mid-thoracic spine = 45 (14)*	Symptomatic spinal level Male = 17 (23)* Female = 57 (77)* Mid-thoracic spine Male = 16 (2)* Female = 58 (78)*	Public outpatient physiotherapy clinic in the state of Sao Paulo, Brazil.	4-Week
Valera-Calero et al. [56] (n = 83)	Cervical C5–6	Chronic (≥3 months)	Manipulation 35.64 (8.11)* Mobilization 37.25 (10.54) * Sham 36.96 (8.89)*	Manipulation 12 men, 16 women Mobilization 10 men, 18 women Sham 10 men, 17 women	Research Unit of the University of Alcala de Henares (Spain)	1-week

Note: FDT = Flexion distraction technique; HVLA-SM = High velocity and low amplitude spinal manipulation; *Mean (Standard Deviation).

=Comparable; =Not Comparable.

Table 5.

Description of Interventions.

Authors	Description of Manual Interventions	Description of Comparison	Description of Comparison
Carrasco-Martinez et al. [53] (n = 150)	‘Patients were placed in the lateral decubitus position, keeping upper-level flexion and lumbar spine rotation until the L3-L4 level. The lower limb was passively placed in a triple flexion position. The foot rested on the popliteal space of the opposite inferior limb. In this position, spinal manipulation consisted of a high-velocity, low-amplitude spinal manipulation (HVLA-SM) applied on the inferior level in an anterior and downwards direction.’ & ‘90-second ischemic compression on the four trigger points of the lumbar quadrate muscle, and also on two trigger points of the iliacus-psoas muscle.’ (n = 75)	‘Patients were placed in the decubitus prone position on the table, with their umbilicus placed between the dorsal zone of the table and the lumbar body. In this way, anterior superior iliac spines remain placed in the inferior zone of the table. A Velcro strap was placed around the ankles. The therapist then used the heel of the hand to contact the upper spinal apophysis at the level to be treated. A modern table with an engine placed under the caudal body was used, which imparts a rhythmic up-and-down movement of the lower limbs in a pumping motion. In each downward movement, the therapist pressed on the L3 lumbar spine vertebra, following the direction toward the table and the patient’s head. In total, there were ten pumping movements, a one-minute pause, a repetition of the previous ten pumping movements and then another pause, and so on until a total of five minutes were completed.’ & ‘90-second ischemic compression on the four trigger points of the lumbar quadrate muscle, and also on two trigger points of the iliacus-psoas muscle.’ (n = 75)
de Oliveira et al. [54] (n = 143)	Manipulation at symptomatic spinal level (n = 71) This was identified using central pressure posterior (CPA) to anterior and unilateral pressure posterior to anterior (UPA). Manipulative technique performed was not described	Manipulation at mid-thoracic spine (n = 72) Technique was applied to the T5–6 level Manipulative technique performed was not described
Valera-Calero et al. [56] (n = 83)	Manipulation (n = 28) “ … mid-range, left rotational force to C5-C6, with right side bending and left rotation. The patient was positioned supine with the neck in a neutral relaxed position. The proximal or middle phalanx of the physical therapist’s right index finger contacted over the posterolateral aspect of the right C5 articular pillar using a chin hold. An upward and forward gliding thrust, parallel to the zygapophysial joint plane and in the direction of the patient’s left eye was applied.	Mobilization (n = 28) ‘The cervical mobilization technique consisted of a grade III posteroanterior joint oscillatory mobilization technique to the articular pillar of C5/6 on the subject’s symptomatic side.’ ‘The patient was positioned prone with the neck in a neutral, relaxed position, and the physical therapist stood at the end of the plinth.’ ‘The cervical mobilization technique was delivered one time in one treatment session. The treatment involved three, 1-minute applications with a 1-minute interval between each’	Sham (n = 27) ‘The intervention consisted of the same degree of head rotation and manual support to the head and neck as in the cervical manipulation technique used in this study, but with no thrusting force. A rapid application of motion was created only through the drop action of the head-piece cam mechanism with associated sharp sound.’ ‘The sham manipulation technique was delivered one time in one treatment session. Total duration of the technique was about 15 seconds, including the positioning of the therapist and patient and the delivery of the technique.’

Note: =Comparable; =Not Comparable.

Carrasco-Martinez et al. [53] demonstrated a statistically significant (p < .001) difference between lumbar participants treated with the flexion-distraction technique versus the manipulative technique, with those in the flexion-distraction group showing a greater increase in PPT levels (Table 6). Confidence intervals were small across comparisons. Comparable mean differences were observed on either side of the spine at each site, and small effect sizes (ETA²) were observed across all comparisons and sites. Similar findings were observed for changes in pain on the VAS, with statistically significant differences between the flexion-distraction and manipulation groups (p < .001), small confidence intervals, and small effect sizes (Table 6). Compared to the manipulation group, a greater reduction in pain levels was observed for those in the flexion-distraction group.

Table 6.

Methodological heterogeneity.

Paper-Descriptives	Description of Manual Interventions	Description of Comparison	Pressure Pain Thresholds Location	Pressure Pain Thresholds
Carrasco-Martinez et al. [53] (n = 150)	Manipulation & Ischemic compression (n = 75)	Lumbar flexion distraction technique & Ischemic compression (n = 75)	Regional on Trigger Points	Quadratus Lumborum Site 1 P <0.001 for all comparisons Mean (CI) = 0.579 (0.457 to 0.701) Manip (L) Mean (CI) = 1.109 (0.987 to 1.231) Flex-distract (L) Mean (CI) = 0.530 (0.398 to 0.661) Manip (R) Mean (CI) = 1.1161 (1.030 to 1.292) Flex-distract (R) Mean Difference = 0.530 (L); 0.631 (R) Effect size (ETA2) = 0.195 (L); 0.228 (R) Quadratus Lumborum Site 2 P <0.001 for all comparisons Mean (CI) = 0.436 (0.322 to 0.549) Manip (L) Mean (CI) = 1.004 (0.891 to 1.118) Flex-distract (L) Mean (CI) = 0.401 (0.288 to 0.513) Manip (R) Mean (CI) = 0.997 (0.884 to 1.109) Flex-distract (R) Mean Difference = 0.569 (L); 0.596 (R) Effect size (ETA2) = 0.249 (L); 0.268 (R) kg/cm
de Oliveira et al. [54] (n=143)	Manipulation at symptomatic spinal level (n=71	Manipulation at mid-thoracic spine (n=72)	Regional (Lumbar)-Remote (Tibialis Anterior)	Lumbar P-value =Not reported Mean (SD)=128 (331) Symptom level Mean (SD)=122 (232) Mid-thoracic Mean Difference (CI)=6 (−88 to 101) kPa Effect size= Not reported P=0.652 Mean (SD)=2.68 (0.13) Manip Mean (SD)=2.93 (0.13) Mob Mean Difference [CI] (Manip v. Mob)=-0.25 [−0.69 to 0.20]
Valera-Calero et al. [55] (n=83)	Manipulation (n=28)	Mobilization (n=28)-Sham (n=27)	Local-C5/6 zygapophyseal joint	Mean Difference [CI] (Manip v. sham)= −0.06 [−0.51 to 0.39] Mean Difference [CI] (Mob v. sham)=0.19 [−0.26 to 0.64] Effect size (ETA2)=0.015 kg/cm
Authors	Pain	Patient Reported Outcomes	Perceived improvement	Psychological beliefs
Carrasco-Martinez et al. [53]	VAS P<0.001 Mean (CI)= −1.566 (−1.786 to −1.345) Manip Mean (CI)= −2.908 (−3.129 to −2.687) Flex-distract MeanDifference=−1.342 Effect size= Not reported	Oswestry P<0.001 Mean (CI)= −6.393 (−7.439 to −5.347) Manip Mean (CI)= −11.474 (−12.520 to −10.427) Flex-distract MeanDifference=−5.081 Effect size (ETA2)=0.239	n/a	n/a
de Oliveira et al. [54]	NPRS P value=Not reported Mean (SD)=3.3 (2.2) Symptom level Mean (SD)=3.1 (2.3) Mid-thoracic Mean Difference (CI)=0.0 (−0.9 to 0.9) kPa Effect size= Not reported	Roland Morris P-value =Not reported Mean (SD)=3.4 (5.1) Symptom level Mean (SD)=3.4 (5.6) Mid-thoracic Mean Difference (CI)=0.1 (−1.7 to 1.5) kPa Effect size= Not reported	Global Perceived Effect P value =Not reported Mean (SD)=2.4 (2.1) Symptom level Mean (SD)=3.0 (1.7) Mid-thoracic Mean Difference (CI)= −0.1 (−1.0 to 0.8) kPa Effect size= Not reported	n/a
Valera-Calero et al. [55]	VAS P<0.001 Mean (SD)=5.21 (0.30) Manip Mean (SD)=4.33 (0.30) Mob Mean (SD)=6.06 (0.30) sham Mean Difference [CI] (Manip v. Mob)=-0.88 [−0.15 to 1.91] Mean Difference [CI] (Manip v. sham)= −0.86 [−1.90 to 0.18] Mean Difference [CI] (Mob v. sham)=-1.74 [−2.78 to −0.70] Effect size (ETA2)=0.07	Neck Disability Index P=0.12 Mean (SD)=13.57 (2.39) Manip Mean (SD)=13.57 (2.39) Manip Mean (SD)=25.15 (2.43) sham Mean Difference [CI] (Manip v. Mob)=-1.46 [−9.72 to −6.79] Mean Difference [CI] (Manip v. sham)= −11.58 [−19.90 to −3.25] Mean Difference [CI] (Mob v. sham)=-10.11 [−18.44 to −1.79] Effect size (ETA2)=0.10	n/a	n/a

Note: =Comparable; =Not Comparable; SD= Standard Deviation; CI = Confidence Interval; L = Left; R = Right; SD = Standard Deviation; kPa = Kilopascals; Manip = Manipulation; Mob = Mobilization; kg/cm = Kilograms per centimeter.

de Oliveira et al. [54] did not report p-values or calculated effect sizes for their between-group comparisons of PPT changes in the lumbaP=0.12r and tibialis anterior regions between manipulation performed at the symptomatic and mid-thoracic levels (Table 4). In both locations, the authors reported no between-group differences for PPT in the groups treated at the symptomatic level compared to the mid-thoracic level; however, the confidence levels associated with the mean differences in these comparisons were large (Table 6). Similarly, the difference between the groups on pain levels (NPRS) did not have p-values or effect sizes reported. The mean difference was 0.5, with a narrow confidence interval crossing zero, indicating a non-significant effect (Table 6).

Valera-Calero et al. [56] tracked outcomes on PPT locally at the C5–6 zygapophyseal joint, performing between-group comparisons for manipulation, mobilization, and sham treatment groups. When examining the one-week outcomes, there were no statistically significant comparisons across manipulation versus mobilization, mean difference = −0.25 (−0.69 to 0.20), manipulation versus sham, mean difference = −0.06 (−0.51 to 0.39), or mobilization versus sham, mean difference = 0.19 (−0.26 to 0.64), comparisons, p = 0.652, ETA² = 0.015 (Table 6). When comparing the VAS for manipulation, M = 5.21 (0.30), mobilization, M = 4.33 (0.30), and sham, M = 6.06 (0.30) groups, statistically significant differences were observed across manipulation versus mobilization, mean difference = −0.88 (−0.15 to 1.91), manipulation versus sham, mean difference = −0.86 (−1.90 to 0.18), and mobilization versus sham, mean difference = −1.74 (−2.78 to −0.70) comparisons, p = 0.01, ETA² = 0.07. When comparing the manipulation, M = 5.21 (0.30), mobilization, M = 4.33 (0.30), and sham, M = 6.06 (0.30) groups, statistically significant differences were observed across manipulation versus mobilization, mean difference = −0.88 (−0.15 to 1.91), manipulation versus sham, mean difference = −0.86 (−1.90 to 0.18), and mobilization versus sham, mean difference = −1.74 (−2.78 to −0.70) comparisons, p = 0.01, ETA² = 0.07.

All three papers differed in selected outcome measures. Carrasco-Martinez et al. [53] was the only paper to report outcomes on the Oswestry Disability Index (ODI). Like the VAS findings, the between-groups comparison of the flexion-distraction group, M = −11.474 (−12.520 to −10.427), and the manipulation group, M = −6.393 (−7.439 to −5.347), indicated a greater improvement in scores for the flexion-distraction group, mean difference = −5.081, p < 0.001, ETA² = 0.239 (Table 6). de Oliveira et al. [54] examined patient-reported outcomes via the Roland Morris Disability Index and was the only included study that reported participants’ perceived improvement via the Global Perceived Effect. No p-values or calculated effect sizes were reported for the between-groups comparisons. There were minimal differences when comparing those treated at the symptomatic, M = 3.4 (5.1) and mid-thoracic, M = 3.4 (5.6) levels, mean difference = 0.1 (−1.7 to 1.50) kPa. Similar minimal differences were noted when examining the Global Perceived Effect on those treated at the symptomatic, M = 2.4 (2.1), and mid-thoracic, M = 3.0 (1.7), levels, mean difference = −0.1 (−1.0 to 0.8) kPa (Table 6). Valera-Calero et al. [56], the only identified study investigating persons with chronic cervical spine pain, examined outcomes on the Neck Disability Index (NDI). When examining the between-groups comparisons on the NDI (manipulation, M = 13.57 (2.39); mobilization, M = 15.04 (2.39); and sham, M = 25.15 (2.43), no statistically significant differences were observed between groups mean difference = −1.46 (−9.72 to −6.79), mean difference = −11.58 (−19.91 to 3.25), and mean difference = −10.11 (−18.44 to −1.79), respectively, p = 0.12, ETA² = 0.10 (Table 6).

Discussion

Summary of results

Confidence in estimated effects for the individual studies

This SR sought to identify if QST thresholds related to PPT testing meaningfully changed in response to manual therapy. Since PPT measures centrally mediated hyperalgesia in chronic pain [75], these changes, if present, would suggest that manual therapy can change this type of pain. In addition, this investigation intended to identify studies with high confidence in the estimated effects that could be used to confidently make strong practice recommendations based on the GRADE criteria. Unfortunately, most studies screened for inclusion did not meet these rigorous standards. Furthermore, the included studies were grossly dissimilar, not allowing for synthesis using the GRADE criteria. Comparing the results of this SR to previous reviews on the topic, it is evident that the high-quality, low RoB literature does not exist in large enough quantity to support these prior conclusions fully. As a result, these results cannot be confidently translated to clinical practice recommendations using the GRADE criteria.

Across all three studies, small effect sizes suggest the observed differences for PPT and pain may not be clinically meaningful. Carrasco-Martinez et al. [53] demonstrated statistically significant (p < .001) between-group differences between participants treated with the flexion-distraction technique versus the manipulative approach in the lumbar region. Although the confidence intervals were small across comparisons, the corresponding effect sizes were small. de Oliveira et al. [54] reported greater PPT in the groups treated at the mid-thoracic level than at the lumbar symptomatic level. However, the confidence intervals associated with the mean differences were large and crossed zero, indicating a non-significant effect. The between-group differences in pain measures had a mean difference of zero, with a narrow confidence interval crossing zero, indicating a non-significant effect. Valero-Calero et al. [56] examined the primary outcome of PPT. Their non-significant between-group comparisons are evident given the non-significant p-values, large confidence intervals crossing zero, and small effect sizes. The same study examined the outcomes of the VAS; the only significant findings were between the mobilization and sham treatment groups. This between-group comparison’s confidence interval was small and did not cross zero. However, the small effect size lessens the strength of confidence in this finding.

Patient-reported outcomes were reported across all three studies, including the ODI [53], Roland Morris [54], and NDI [56]. Only Carrasco-Martinez et al. [53] reported statistically significant effects of the intervention, indicating a greater reduction in disability on the ODI in the flexion-distraction group as compared to those undergoing manipulation. However, the observed mean difference between these groups (5.081) was smaller than the MDC (12.72) and MCID (9.5) for the ODI in participants with chronic low back pain [76], suggesting the observed differences were not clinically meaningful. Independent of the known challenges with legacy patient-reported outcome measures [77] and the interpretation of the MCID of these measures [78], the observed effect size was small and likely not clinically meaningful.

There are challenges in comparing the results of this SR to previous reviews. One recent SR reported no immediate, consistent, or meaningful hypoalgesic effect of manual therapy on PPTs [24]. However, there were significant methodological differences from the current review. The SR by Jung et al. included underpowered pilot studies, 11 RCTs involved pain-free participants, and one included participants with rheumatoid arthritis [24]. Less than half of the included studies (10/22) included musculoskeletal diagnoses that manual therapy purports to treat. Including asymptomatic and participants with a non-musculoskeletal condition suggests this SR did not create a homogeneous clinical population [24].

Matching a patient’s dominant pain mechanism with a treatment that should improve the symptoms related to that mechanism may result in improved clinical outcomes [32]. The proposed purpose of manual therapy interventions involves decreasing pain and increasing the range of motion to facilitate exercise [79]. Unfortunately, the classification of pain mechanisms is based on expert opinion [80], and it is unknown if any specific pain mechanism is influenced explicitly by manual therapy. Manual therapy has been shown to produce local and remote hypoalgesia, suggesting a central effect of the intervention [14–16] and that QST may be related to pain outcomes following manual therapy [81]. Additionally, it has been reported that spinal manipulative therapy lessens central sensitization [82]. Unfortunately, based on high-quality, low-risk-of-bias evidence, it is still unknown if the proposed specific purpose of manual therapy addresses a specific mechanism or if the observed outcomes are related to shared mechanisms in a clinical population [83,84].

Future research should seek to create research with verified prospective intent that is moderate-to-high in methodological quality and has a moderate to low RoB. The research would need to answer these questions: are the changes in PPT after manual therapy clinically significant, and do they correlate with clinical outcomes, such as pain intensity and disability? Moreover, this investigation could not identify studies assessing the effects of manual therapy on PPT during the conditioned pain modulation test. This measurement will show if manual therapy can reactivate descending analgesic pathways that are commonly inhibited in patients with chronic pain. Also, there is a lack of studies assessing the temporal summation of pain. This test can be performed with a pressure algometer and is an indirect measure of central excitability. It is advisable that these tests, along with remote measurements of PPT, are performed in future studies to answer whether manual therapy has a predominant peripheral or centrally mediated effect.

Limitations

The primary limitation of this SR was the limited number of available RCTs with clear prospective, external, and internal validity that could be synthesized into strong clinical practice recommendations based on high confidence in the estimated effects. The limited number of studies also creates challenges related to statistical, clinical, and methodological heterogeneity.

Conclusion

This SR sought to identify if QST thresholds related to PPT meaningfully change in response to manual therapy. As a potential measure of centrally mediated pain changes, this outcome measure could provide clues regarding where we are in establishing the possible mechanisms of manual therapy or help to identify or refine research questions in this area. Unfortunately, there is a lack of homogeneous RCTs with high confidence in estimated effects that can be used to make strong clinical practice recommendations based on the GRADE criteria.

Biographies

Dr. Sean P. Riley is an Assistant Professor in the Doctor of Physical Therapy Program and a faculty member in the orthopaedic physical therapy residency at the University of Hartford. He is board certified in orthopaedics and a Fellow of the American Academy of Orthopaedic Manual Physical Therapists. Dr. Riley’s research interests include symptom modification approaches to evaluating and treating neuromusculoskeletal disorders, evidence-based practice, research methodology, and clinical reasoning.

Dr. Brian T. Swanson is an Associate Professor at the University of Hartford. He serves as Chair of the Department of Rehabilitation Sciences, Director of the Doctor of Physical Therapy Program, and co-director of the University of Hartford/HHCRN orthopedic physical therapy residency program. He is board certified in orthopaedics and a Fellow of the American Academy of Orthopaedic Manual Physical Therapists. Dr. Swanson’s research interests include validating tests and measures in orthopedic manual physical therapy, developing a further understanding of the mechanisms of manual physical therapy interventions, and evidence-based practice/research methodology.

Dr. Stephen M. Shaffer is a residency and fellowship-trained clinical specialist, educator, and scientist with nineteen years of experience in the physiotherapy profession. He has worked primarily in orthopaedic settings, is an Adjunct Professor at the University of Hartford, and is a Fellow of the American Academy of Orthopaedic Manual Physical Therapists and the Canadian Academy of Manipulative Physiotherapy. Dr. Shaffer has co-authored numerous peer-reviewed scientific papers and has presented at local, state, national, and international venues.

Dr. Daniel W. Flowers is an Assistant Professor in the Doctor of Physical Therapy and PhD in Rehabilitation Sciences Programs at LSU Health Shreveport. He also serves as program director of the orthopaedic residency. He is board-certified in orthopaedic physical therapy. His research interests include modifying the gait impairments of patients with knee osteoarthritis, post-traumatic rehabilitation, and educational outcomes of physical therapy students.

Margaret A. Hofbauer is a second-year physical therapy doctorate student at the University of Hartford. She is the class president and is actively involved within the program. She aids in the Scientific Inquiry class for the first year DPT students as a teaching assistant. Margaret completed her personal research investigating communication methods for individuals with Alzheimer’s Disease for the Honors Program in 2021. Her research interests include orthopaedics, neurologic degenerative disorders, and geriatrics.

Dr. Richard E. Liebano is an Associate Professor in the Department of Rehabilitation Sciences at the University of Hartford. He has published three books, 14 book chapters, and co-authored numerous peer-reviewed scientific papers. Dr. Liebano’s research interests include the mechanisms and effectiveness of biophysical agents and other physical therapy interventions to reduce pain and improve tissue healing.

Funding Statement

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

Disclosure statement

The primary author Sean P. Riley is a Deputy Editor at the Journal of Manual and Manipulative Therapy (JMMT). He also contributes to the Duke Center for Excellence in Manual and Manipulative Therapy. The authors’ otherwise report that there are no competing interests to declare.

Author contributions

The primary author, Sean Riley, coordinated the development of the research questions and protocol with his coauthors. Brian Swanson identified which search engines would be used and developed the search strategies for each with the assistance of a professional librarian. Margaret Hofbauer ensured that the full-text articles of interest were retrieved and uploaded into Covidence. Daniel Flowers and Stephen Shaffer screened the identified randomized clinical trials to determine if they met the inclusion and exclusion criteria. Richard Liebano was responsible for the interpretation of the data, recognizing gaps in knowledge, and identifying a potential path forward to answer the questions related to the mechanisms of manual therapy. All authors were involved in drafting and revising the protocol for important intellectual content and agreed to be accountable for the accuracy and integrity of the work.

Support

Sources

The resources for gathering and inputting the randomized clinical trials into Covidence were provided by the University of Hartford, College of Education, Nursing, and Health Care Professions.

Sponsor

The sponsors for this series of living systematic reviews are the Center of Excellence in Manual and Manipulative Therapy at Duke University and the University of Hartford.

Role of sponsor or funder

The funder, sponsor, and institutions were not involved in developing this systematic review.

SUPPLEMENTARY MATERIAL

Supplemental data for this article can be accessed online at https://doi.org/10.1080/10669817.2023.2247235