Chiropractic & Manual Therapies - Unravelling functional neurology… · 2019. 10. 10. ·...

SYSTEMATIC REVIEW Open Access

Unravelling functional neurology: doesspinal manipulation have an effect on thebrain? - a systematic literature reviewAnne-Laure Meyer1,2,3* , Michel-Ange Amorim1,2, Martin Schubert4, Petra Schweinhardt5 andCharlotte Leboeuf-Yde1,2,3,6

Abstract

Background: A recent hypothesis purports that spinal manipulation may cause changes at a brain level. FunctionalNeurology, a mainly chiropractic approach, promotes the use of spinal manipulation to improve ‘brain function’ asif it were a proven construct. No systematic review has been performed to investigate how well founded thishypothesis is.

Objective: To investigate whether spinal manipulation has an effect on ‘brain function’ that is associated with anyclinical benefits.

Method: In this systematic review, the literature was searched in PubMed, Embase, and PEDro (final search February2018). We included randomized or non-randomized controlled studies, in which spinal manipulation was performed toany region of the spine, applied on either symptomatic or asymptomatic humans, and compared to a sham or toanother type of control. The outcome measures had to be stated as direct or proxy markers of ‘brain function’. Articleswere reviewed blindly by at least two reviewers, using a quality checklist designed for the specific needs of the review.Studies were classified as of ‘acceptable’, ‘medium’, or ‘low’ methodological quality. Results were reported in relation to(i) control intervention (sham, ‘inactive control’, or ‘another physical stimulus’) and (ii) study subjects (healthy,symptomatic, or with spinal pain” subjects/spinal pain”), taking into account the quality. Only results obtained frombetween-group or between-intervention comparisons were considered in the final analysis.

Results: Eighteen of 1514 articles were included. Studies were generally of ‘low’ or ‘medium’ methodological quality,most comparing spinal manipulation to a control other than a sham. Thirteen out of the 18 studies could be includedin the final analysis. Transitory effects of different types of ‘brain function’ were reported in the three studies comparingspinal manipulation to sham (but of uncertain credibility), in “subclinical neck/spinal pain” subjects or in symptomaticsubjects. None of these three studies, of ‘medium’ or ‘acceptable’ quality, investigated whether the neurophysiologicaleffects reported were associated with clinical benefits. The remaining 10 studies, generally of ‘low’ or ‘medium’ quality,compared spinal manipulation to ‘inactive control’ or ‘another physical stimulus’ and similarly reported significantbetween-group differences but inconsistently.

Conclusion: The available evidence suggests that changes occur in ‘brain function’ in response to spinal manipulationbut are inconsistent across and - sometimes - within studies. The clinical relevance of these changes is unknown. It istherefore premature to promote the use of spinal manipulation as a treatment to improve ‘brain function’.

Keywords: Spinal manipulation, Brain, Functional Neurology, Chiropractic, Systematic review

© The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

* Correspondence: [email protected], Univ. Paris-Sud, Université Paris-Saclay, 91405 Orsay Cedex, France2CIAMS, Université d’Orléans, 45067 Orléans, FranceFull list of author information is available at the end of the article

Meyer et al. Chiropractic & Manual Therapies (2019) 27:60 https://doi.org/10.1186/s12998-019-0265-8

http://crossmark.crossref.org/dialog/?doi=10.1186/s12998-019-0265-8&domain=pdf

http://orcid.org/0000-0003-0840-4176

http://creativecommons.org/licenses/by/4.0/

http://creativecommons.org/publicdomain/zero/1.0/

mailto:[email protected]

Résumé

Introduction: Une hypothèse récente propose que la manipulation vertébrale causerait des changementsneurophysiologiques au niveau du cerveau. En Neurologie Fonctionnelle, approche principalement présente enchiropraxie, l’utilisation de la manipulation vertébrale est déjà promue comme capable d’améliorer le fonctionnementdu cerveau. A ce jour, aucune revue systématique de la littérature n’a été conduite afin de connaître l’étendue et laqualité de l’évidence scientifique susceptible de justifier cet usage de la manipulation vertébrale.

Objectif: Déterminer, à travers une revue systématique et critique de la littérature, si la manipulation vertébrale a uneffet (spécifique) sur le cerveau et, si oui, si celui-ci est associé à un effet clinique.

Méthode: Le moteur de recherche PubMed et deux bases de données, Embase et PEDro, ont fait l’objet d’une recherchebibliographique (actualisée en février 2018). Les critères d’inclusion étaient: essais contrôlés, randomisés ou non, danslesquels la manipulation vertébrale avait été comparée à un placébo ou à un autre type de contrôle, chez des sujetssymptomatiques ou asymptomatiques. La manipulation vertébrale pouvait avoir été effectuée au niveau de n’importequelle région de la colonne vertébrale et les critères de jugement utilisés devaient être indiqués comme capables demesurer, directement ou indirectement, une forme ‘d’activité cérébrale’. La qualité méthodologique des études a étéévaluée de manière indépendante par au moins deux chercheurs à l’aide d’une grille de qualité créée pour les besoins decette revue. Les études ont été classifiées comme étant de qualité méthodologique ‘acceptable’, ‘moyenne’, ou ‘faible’.Les résultats ont été rapportés de façon narrative, en fonction du type de contrôle utilisé (placébo, ‘inactif’, ou ‘autrestimulus physique’) et du type de sujets d’étude (sans problème de santé, symptomatiques, ou présentant des « douleursrachidiennes subcliniques »), en tenant compte de la qualité méthodologique. Seuls les résultats issus de comparaisonsinter-groupes ont été pris en compte dans notre analyse finale.

Résultats: Dix-huit articles parmi les 1514 titres obtenus ont été inclus. Les études étaient pour la plupart de qualitéméthodologique ‘faible’ ou ‘moyenne’ et avaient principalement comparé la manipulation vertébrale à une interventionautre que placébo. Les résultats rapportés dans 13 des 18 articles inclus ont finalement été pris en compte. Un effettransitoire sur différentes formes ‘d’activité cérébrale’ a été rapporté à l’issue de trois études dans lesquelles lamanipulation vertébrale avait été comparée à un placébo (de crédibilité cependant incertaine), chez des sujets présentantdes « douleurs rachidiennes subcliniques » (n = 2) ou souffrant de cervicalgies non spécifiques aiguës / subaiguës (n = 1).Une potentielle association avec un effet clinique n’a pas été étudiée dans ces trois études, de qualité méthodologique‘moyenne’ (n = 2) ou ‘acceptable’ (n = 1). Dans les 10 études restantes, la plupart de qualité méthodologique ‘faible’ ou‘moyenne’, la manipulation vertébrale avait été comparée à un contrôle ‘inactif’ ou à un ‘autre stimulus physique’. Desdifférences inter-groupes y ont également été rapportées, parfois de façon inconsistante.

Conclusion: La littérature scientifique suggère que des changements neurophysiologiques surviennent au niveau ducerveau en réponse à la manipulation vertébrale mais, de façon inconsistante. La pertinence clinique de ces changementsn’est pas connue. Ainsi, il est prématuré d’attribuer à la manipulation vertébrale des bénéfices cliniques via un effet sur lecerveau.

Mots clés: Manipulation vertébrale, Cerveau, Neurologie Fonctionnelle, Chiropraxie, Revue systématique

Meyer et al. Chiropractic & Manual Therapies (2019) 27:60 of 30

IntroductionSpinal manipulation (SM) is widely used by varioushealth practitioners, including physiotherapists, osteo-paths and chiropractors, to treat mainly musculoskel-etal conditions, but some also use it for a variety ofother health-related problems [1, 2]. While the litera-ture tends to support the benefit of SM as a usefultreatment in the musculoskeletal area, no clear evi-dence exists in relation to non-musculoskeletal condi-tions [3].This lack of evidence contrasts with claims of some

therapists, including (but not restricted to) those whopractice using the theoretical concepts of FunctionalNeurology (FN), a mainly chiropractic approach, foundedby a chiropractor, FR Carrick [4]. In addition to musculo-skeletal conditions, “functional neurologists” (i.e. FN prac-titioners) also provide treatment for complex disorderssuch as neurodevelopmental disorders, neurodegenerativedisorders, and post-traumatic stress disorders [5]. Alsobased on these FN concepts, some therapists claim to en-hance human performances (e.g. physical performances),including in asymptomatic individuals [5].In line with FN, a current hypothesis is that the clin-

ical benefits observed following SM would be, at leastpartially, due to neurophysiological changes within thebrain [6, 7]. Some practitioners already use this conceptclaiming it to be a fact that SM has a clinically relevanteffect on the brain, as shown through several sources ofinformation in a recent scoping review on FN [5]. Fur-thermore, for some proponents of this hypothesis, atleast for the “functional neurologists”, a multitude ofconditions results from dysfunction within the brain [5].It is stated that malfunctioning clusters of neurons, de-scribed as primarily located within the brain, could be thesingle cause of virtually any type of symptom and/or dis-order that a person may have. Examples extracted form aFN textbook are attention deficit and hyperactivity dis-order, depression, mechanical low back pain, and mi-graines [7]. In addition, these ‘malfunctions’ are stated tobe reversible through the use of stimulation of the nervoussystem, including by SM [5]. This would give SM the po-tential to be used for both musculoskeletal and non-mus-culoskeletal conditions. In fact, it is even stated that SM isone of the most easily available methods for manual prac-titioners to improve ‘brain function’ [7].Within the scientific literature, differently framed hy-

potheses exist in relation to the potential mechanismsinvolving the brain, which could explain clinical bene-fits following SM [6, 8, 9]. The one mechanism thatseems to prevail relates to the chiropractic concept of‘subluxation’, which has developed over time [10].Currently some authors purport that ‘subluxations’modify afferent inputs to the central nervous system [6,11]. These authors state, in addition to this, that the

‘subluxation’ is at the source of maladaptive neuralplastic changes, including in the cerebral cortex, whichin turn result in altered processing and integration ofsubsequent afferent inputs and, consequently, alteredmotor outputs [11]. As a consequence, SM is claimedto restore afferent inputs to the central nervous system(including to the brain) and result in appropriate motoroutputs from the central nervous system [11].Potential neurophysiological effects of SM on the

brain have been the focus of several recent experi-mental studies. As the brain is involved in a multi-tude of functions, its activities or alteration ofactivities after an intervention can be explored in sev-eral ways. Not surprisingly then, the studies in thisfield of research use various approaches and outcomemeasures to test the hypothesis that SM has an effecton ‘brain function’. For example, some studies investi-gated the potential effect of SM on brain areas in-volved in pain processing [9] and autonomicfunctions [8], whereas others reported on the poten-tial effect of SM on cortical somatosensory integrationof stimuli from the upper limb [6]. Therefore, in thepresent systematic review ‘brain function’ is used as ageneric expression referring to processes in which thebrain is involved.Because studies are quite heterogeneous, it is diffi-

cult to understand and interpret the evidence in thisarea. Nevertheless, this task is needed to understandif assertions of the ‘brain-mediated’ hypothesis propo-nents are substantiated by scientific evidence. A nar-rative review on the topic by Haavik and Murphy waspublished in 2012 [6]. In this review, they concludedthat some evidence supports a brain mechanism ofaction for SM but that it remains to be investigatedwhether this correlates to clinical benefits. They alsostated that such studies were underway [6]. Giventhat 7 years have passed since this narrative review,the aforementioned studies exploring potential associ-ations between clinical and ‘brain function’ changespost-SM were likely published, and thus may provideimportant updates to the state of the field. For thesereasons, we undertook a systematic critical review ofthe literature, which had as its overall aim to investi-gate whether SM has an effect on ‘brain function’ thatis associated with any clinical benefits in healthy and/or symptomatic subjects. The specific research ques-tions were:

In relation to sham controlled studies, i.e. ‘effect’studies:1 - Is there an effect of SM on ‘brain function’?2 - If there is an effect, for how long does it last?3 - If there is an effect, is it associated with any clinical

benefits?


In relation to other controlled studies (‘inactivecontrol’ or ‘another physical stimulus’), i.e. the‘differences in outcome’ studies:4 - Is there a difference in ‘brain function’ after SM vs.

an ‘inactive control’?5 - Is there a difference in ‘brain function’ after SM vs.

‘another physical stimulus’?

MethodsA systematic critical review of the literature was car-ried out to shed light on the research questionsabove. The review was registered in the PROSPEROinternational prospective register of systematic reviews(CRD42017074966). Some deviations from the originalprotocol were required in response to the materialavailable in the reviewed articles, which was unknownat the time of planning the review. These were: (i)the wording of the research questions was improved,(ii) the review was restricted to spinal manipulativetherapy (i.e. did not include extremities), and (iii) theresults were analyzed depending on three categoriesof study subjects instead of the two we planned.Regarding the latter, it was initially planned toanalyze the results depending on whether study sub-jects were (i) healthy or (ii) symptomatic. However, athird type of study subjects was identified, namely“subclinical neck/spinal pain” subjects. Further detailson these study subjects are given in a subsequent sec-tion (see Data analysis and synthesis).

Search for literatureA systematic literature search was conducted in threeelectronic databases: PubMed, Embase and PEDro in April2017 (updated between January and February 2018). Thesearch strategy was initially developed for PubMed (avail-able in Additional file 1) and then adapted to the twoother databases in collaboration with a health scienceresearch librarian. In short, the strategy was designed byassociating (i) terms related to SM, for example “manipu-lation, spinal”, “musculoskeletal manipulations”, or “high-velocity low-amplitude spinal manipulation”, (ii) termsrelated to brain or brain structures, for example “brain”,“cerebrum”, or “cerebellum”, and (iii) terms related to thedifferent ways of assessing ‘brain function’, for example“transcranial magnetic stimulation”, “electroencephalog-raphy”, or “positron-emission tomography”.

Eligibility criteriaThe eligible studies in this review had to include at leastone control group, with or without random allocation.Two- or several-arm trials were accepted as well ascrossover designs. These studies had to be conducted onhumans, with no restriction regarding their study

population such as age, sex, healthy or symptomatic sub-jects, or type of symptoms.The tested intervention had to consist of manually

performed, instrumentally assisted, or mechanicallyassisted SM. Studies with combined or concomitanttherapies were excluded, as it would not be possible toseparate results obtained from the SM and the othertherapies. However, if all the study groups of a reportwere subjected to the same combined or concomitanttherapies (e.g. pain medication), i.e. the only differencebetween the study groups being that one group was sub-jected to the tested intervention (e.g. pain medicationAND spinal manipulation) but not the other (i.e. painmedication only), the article could be included.The control group could be subjected to a sham pro-

cedure, an ‘inactive control’, or ‘another physical stimu-lus’ (other than SM). However, only studies using asham as comparator could be considered to investigatethe effect of SM (i.e. effect specific to SM) on the brainand could therefore be used to answer the search ques-tions 1 to 3 of the present review. The control was con-sidered as ‘another physical stimulus’ when it involved atleast a manual contact (e.g. passive movement of a spineregion, or joint preloading), or when it included otherforms of manual therapies (e.g. joint mobilization, thera-peutic touch). ‘Inactive control’ would consist of, for ex-ample, placing the study subject in side posture withoutmanual contact or just resting.Given our overall aim, the inclusion criteria were

not limited to specific outcome measures or to spe-cific measurement procedures. Studies were includedif their authors stated that the outcome measureswere used to assess ‘brain function’, meaning this wasnot necessarily expected that the outcome measureswere valid or markers exclusive of ‘brain function’.This lenient criterion was chosen for two reasons: (i)some outcome measures are known also to dependon segmental activity (e.g. the V-wave) [12] and (ii)the outcome measures used by “functional neurolo-gists” are not necessarily valid [13]. These potentialissues will be discussed later in the review.There was no restriction in relation to the date of pub-

lication of the studies but only articles in English orFrench were included.

ScreeningEligibility criteria were applied twice to the titles bythe first author, who also searched the reference listsof the included full texts for additional relevant stud-ies. Thereafter, the abstracts and then the relevant fulltexts were read independently by two authors (ALMand CLY) to determine if they could be included inthe review.


Extraction of informationThree types of specific checklists were developed for thisreview relating to: main descriptive features of includedarticles (Tables 1, 2 and 3), methodological quality as-sessment (Tables 4, 5, 6, 7 and 8), and report of results(Table 9, 10 and 11). Information of interest was ex-tracted from the Methods and Results sections only.

Descriptive informationMain descriptive features of the included articles were re-ported in three tables, one for each type of control, i.e.sham, ‘inactive’, and ‘another physical stimulus’ (see Ta-bles 1, 2 and 3). The descriptive data were extracted fromeach included article independently by ALM and CLY andwere later compared to minimize extraction errors.

Information related to methodological qualityA quality checklist was designed in order to evaluatemainly risk of bias of the type of studies included in thepresent review. This checklist was developed based onconcepts described in the CONSORT statements [30]and on usual concepts in relation to risk of bias, such asthose used by the Cochrane collaboration [31] and thescale proposed in the PEDro database [32]. The itemsused for the quality assessment and their rationale aredescribed in Additional file 2.When deemed necessary, three researchers (MAA, MS

and PS), with an expertise in at least one of the out-comes used in the included studies, provided commentsin relation to the methodology and technical aspects ofthe studies they assessed. These comments could beused to discuss the findings in relation to each researchobjectives. Some of these comments have been includedin this report (see Tables 4, 5, 6 and 7, col.10). Each ex-pert (MAA, MS and PS) dealt with articles withintheir own areas of expertise only. One of the authors(MAA), with special expertise on the types of statis-tical analyses used in experimental studies, reviewedall the statistical analyses. The articles were groupedby type of outcomes or families of outcomes in fivemethodological quality checklists of similar items (seeTables 4, 5, 6, 7 and 8). This was done in order tofacilitate the overview of the comments of the expertsin relation to each type of outcomes or families ofoutcomes. As we did not have access to experts onall the outcomes used in the included studies, com-ments related to methodology and technical aspectscould not always be provided (as mentioned in Tables5, 6 and 8, col.10).Each article was independently reviewed for each meth-

odological quality item by at least two of the authors(ALM and CLY or ALM, CLY and PS). Data were latercompared to minimize extraction errors. Discrepancies

were planned to be resolved by discussion between theauthors.

Information related to the resultsThe outcomes of the selected studies were reported inthree tables (see Tables 9, 10 and 11), one for each type ofcontrol (sham, ‘inactive’, and ‘another physical stimulus’).For each of these tables, results were reported grouped by(i) type of study subjects (healthy, symptomatic, and with“subclinical neck/spinal pain”), (ii) type of outcomes, and(iii) consecutively by year of publication.In accordance with the recommendations of Bland and

Altman (2011) [33], we planned to report only resultsthat reflected clearly differences between-groups (in trialsconsisting of at least two separate groups of study sub-jects). In crossover designs, differences should be testedbetween-types of interventions. This means that resultsof studies that did not perform and clearly report com-parisons between-groups or, in the case of crossoverstudies, failed to report between-types of interventionswould not be taken into consideration to answer our re-search questions. Therefore, if authors reported only sig-nificant within-groups or within-types of interventions,without taking into account the difference between read-ings of the two interventions, this would be ignored.However, our review revealed both unusual and con-

fusing statistical reporting. We therefore decided to takeinto consideration also some results in a ‘benefit of thedoubt’ approach, such as instances when none of the re-viewers was able to decide whether the authors had, infact, performed an appropriate between-groups or be-tween-types of intervention analysis.Only the primary outcomes of the included studies

were considered.

Classifying articles by their methodological qualityEach article was checked for each quality item, giving ei-ther half of a point or one point for each fulfilled item asdescribed in Additional file 2. The quality score was ar-bitrarily divided into ‘acceptable’ (68 to 100% of max-imum number of points), ‘medium’ (34 to 67% ofmaximum number of points) and ‘low’ (0 to 33% ofmaximum number of points), to indicate the quality ofthe methodological aspects mainly in relation to risk ofbias of studies.

Data analysis and synthesisThe various tables were used to report narratively themain findings in relation to our five research questions,taking into account the methodological quality of the in-dividual studies, so that we would have more confidencein the studies of better quality than those with additionalmethodological deficiencies.


For each type of control, the findings were reported bytype of study subjects: healthy, symptomatic, or “subclin-ical neck/spinal pain’” subjects. Following the definitiongiven by the authors of the respective studies, the “sub-clinical neck/spinal pain” subjects appeared to us as an in-dependent category of study subjects, neither healthy norin pain at the time of study. Although this latter definitionchanges somewhat from one publication to another, studysubjects were usually described as having a history of“mild intermittent spinal pain, ache or tension (subclinicalspinal pain), and evidence of dysfunction in the spinaland/or pelvic joints” [24]; spinal/pelvic dysfunction refer-ring to the chiropractic concept of ‘subluxation’ [14, 20,25, 27, 28]. In some of these studies, these study sub-jects were also defined as not yet having sought treat-ment for their complaint [14, 15, 24, 27].

ResultsFigure 1 shows a flow diagram of the study selectionprocess. Of the 1514 initially screened articles, 18 ful-filled our inclusion criteria and were included in the re-view. These were published in English between 2000 and2018. The majority of studies (n = 10), all on “subclinical

neck/spinal pain” subjects, were conducted by researchteams that included one specific author [14, 15, 20, 21,23–28].All articles reported an ethics approval from an ethics

committee or from a review board, with or without anidentification number of the application and approval.As for conflict of interest, 11 studies declared to havenone [8, 9, 14–16, 18, 21, 22, 24, 26, 28], whereas theissue of conflict was not mentioned at all in the others[17, 19, 20, 23, 25, 27, 29].

Description of studies (n = 18)Detailed descriptive information of each study is avail-able in Tables 1, 2 and 3 and briefly summarized below.The size of the study samples of the 18 included

studies ranged from 10 to 72. Ten were conducted on“subclinical neck/spinal pain” subjects (Tables 1, 2and 3), four on pain free healthy subjects (Tables 2and 3), and four on symptomatic subjects (Tables 1, 2and 3), including one on subjects with experimentallyinduced low-back myalgia [22]. In five of the tenstudies considered in the present review, as conductedon “subclinical neck/spinal pain” subjects, the study

Fig. 1 Description of the search for literature in a systematic review on the effect of spinal manipulation on ‘brain function’


Table

1Descriptio

nof

threestud

iesinclud

edin

asystem

aticreview

ontheeffect

ofspinalmanipulationon

‘brain

functio

n’,com

parin

gspinalmanipulationto

asham

interven

tion

1st

author

Yr Ref

Design

Type

ofstud

ysubjects

Num

berof

stud

ysubjects

(males/

females)

-Age

(rang

e)-M

ean

-Typeof

spinalmanipulation

-Typeof

control

-Sham

How

was

cerebralactivity

measured?

Whe

nwas

cerebralactivity

measured?

Clinicalou

tcom

esassessed

(measuremen

ttool

andtim

eof

assessmen

t)

Spars

2017

[9]

Rand

omized

controlled

trial

Symptom

aic:volunteers(unkno

wn

origin)with

mechanicaln

eckpain

<6

weeks

ofdu

ratio

n

24(4

/8)

manipulation

grou

p/(4

/8)

sham

grou

p

-?/?

-36

manipulation

grou

p/40

sham

grou

p

-HVLAmidthoracic(X1)

-‘No’control

-Sham:sim

ilarpo

sitio

ning

ofthesubjectand

investigator’shand

swhich

wereplaced

across

theskin

with

minim

alpressure

(to

mim

ictheHVLAproced

ure)

Bloo

doxygen

ationlevel-

depe

nden

tsign

al(in

respon

seto

noxiou

sstim

uli)

Before

After:immed

iately

Pain

intensity

(11-po

int

numericalpain

ratin

gscale)

(beforespinal

manipulationor

sham

proced

ures

andafter

thefinalfM

RI)

Lelic

2016

[14]

Crossover

controlled

trial(orde

rof

interven

tions

rand

omized

)

“Sub

clinicalne

ck/spinalp

ain”:

volunteers(orig

inun

know

n)with

recurren

tspinalache

,painor

stiffne

ssandeviden

ceof

spinaldysfun

ctionbu

twho

didno

tyetsoug

httreatm

entfor

thisandpain

freeat

thetim

eof

the

stud

y.

19(9

/10)

-? -26

-HVLA(whe

rene

eded

,in

anyspinelevelo

rsacroiliac

joints,n

bun

know

n_may

beat

severallevels)

-‘No’control

-Sham:p

assive

andactive

movem

entsof

thehe

ad,

spine,andbo

dy,sim

ilarto

whatwas

done

forHVLA

interven

tion,

with

out

loadingandthrust

SEPam

plitu

des:N30

peaks

Streng

thof

brainsources:

contralateral

somatosen

sory

cortex,

prefrontalcortex,

cing

ulatecortex,and

bilateralsecon

dary

somatosen

sory

cortex

Before

After:exact

timeun

know

nNon

e

Baarbé

2018

[15]

Rand

omized

controlled

trial

“Sub

clinicalne

ck/spinalp

ain”:

volunteers(unkno

wnorigin)with

recurren

tmild

neck

pain

andmuscle

tension,

butminim

alacutepain

onthe

dayof

testingandwho

neversoug

httreatm

entforthisne

ckcomplains.

27(6

/8)

interven

tion

grou

p/(5

/8)

sham

grou

p

-18–27

intervention

group/19–24

sham

grou

p-21(fo

rbo

thgrou

ps)

-HVLAcervical(X2to

4pe

rsubject)

-‘No’control

-Sham:n

eckge

ntlymoved

into

lateralflexion

and

rotatio

nin

asimilarmanne

rto

theactualne

ckmanipulation,

with

out

applying

theHVLAthrust

Cereb

ellarinhibitio

nBefore

After:exact

timeun

clear(said

tobe

immed

iatelyafter

motor

acqu

isition

task,i.e.

cerebe

llarinhibitio

nwas

re-

measuredabou

t20

min

after

spinalmanipulation)

Non

e

Articlesarepresen

tedby

(i)type

ofstud

ysubjects,i.e.sym

ptom

aticor

“sub

clinical

neck/spina

lpain”

subjects,and

(ii)consecutivelyby

year

ofpu

blication

fMRI

Functio

nalm

agne

ticresona

nceim

aging,

HVLAHighvelocity

low

amplitu

de,n

bNum

ber,SEPSo

matosen

sory

evok

edpo

tential


Table

2Descriptio

nof

eigh

tstud

iesinclud

edin

asystem

aticreview

ontheeffect


‘brain

functio

n’,com

parin


aninactive

control

1st

author

Yr Ref

Design

Type

ofstud

ysubjects

Num

berof

stud

ysubjects

(males/

females)

-Age

(rang

e)-M

ean

-Typeof

spinal

manipulation

-Typeof

control

How

was

cerebral

activity

measured?

Whe

nwas

cerebralactivity

measured?

Clinicalou

tcom

esassessed

(measuremen

ttool

andtim

eof

assessmen

t)

Kelly

2000

[16]

Rand

omized

controlledtrial

Health

y:volunteerchiro

practic

stud

entswith

eviden

ceof

uppe

rcervical“sub

luxatio

n”.

36(9

/9)

interven

tion

grou

p/(11

/7)

control

grou

p

-20-37

(bothgrou

ps)

-24(bothgrou

ps)

-Tog

gle(X1)

-Con

trol:2

min

ofresting

Men

tal

rotatio

nreactio

n-tim

etask

Before

After:exact

timeun

know

nNon

e

Dishm

an2002

[17]

Non

-rando

mized

controlledtrial

Health

y:he

althycollege

stud

ents,volun

teers

24(?/?)

(repartition

ineach

grou

pno

trepo

rted

)

-?/?

-25interven

tion

grou

p/27

control

grou

p

-HVLAL5-S1(X1)

-Con

trol:side

posturepo

sitio

ning

with

outlower

limb

flexion

,truncal

torque,o

rmanual

contact

MEP

amplitu

des

Before

After:

-immed

iately(20to

120s)

-5min

−10

min

Non

e

Dishm

an2008

a

[18]

Rand

omized

controlledtrial

Health

y:he

althychiro

practic

stud

ents,volun

teers

72(21/5)

interven

tion

grou

p/(15

/8)

control

1/(14/9)

control2

-?(3

grou

ps,saidto

bebe

tweentheir20s

and30s)

-?(3

grou

ps)

-HVLAL5-S1(X1)

-Con

trol

1:L5-S1

preloading

-Con

trol

2:side

posturepo

sitio

ning

MEP

amplitu

des

Before

(10MEP

recorded

durin

g100s)

After:immed

iately(10MEP

recorded

durin

g100s)

Non

e

Fryer

2012

[16]

Crossover

controlledtrial

(order

ofinterven

tions

rand

omized

)

Health

y:he

althyun

iversity

stud

ents,volun

teers

14(10/4)

-18-50

-23

-HVLAL5-S1(X2to

4)-Con

trol:b

ilateral

side

-posture

positio

ning

with

out

trun

caltorqu

e,or

manualcon

tact

MEP

latencies

and

amplitu

des

Silent

perio

ds

Before

After:exact

timeun

know

n(according

totheDiscussion

approxim

ately10

min

after)

Non

e

Ogu

ra2011

[19]

Crossover

controlledtrial

(order

ofinterven

tions

“cou

nterbalanced

”)

Symptom

atic:volun

teers,

recruitedat

thelocalu

niversity,

with

mechanicalcervicalp

ain

andshou

lder

stiffne

ss.

12(12/0)

-21–40

-28

-Instrumen

tally

assisted

manipulation

(locatio

nandnb

ofspinallevelsadjusted

unknow

n)-Con

trol:20min

ofresting

Region

alcerebral

metabolic

rate

(rate

ofglucose

consum

ption)

Nobe

fore

measuremen

tAfter:b

etween35

to55

min

post-interven

tionor

resting

-StressRespon

seScale

(immed

iatelyafterinterven

tions)

-Europ

eanOrganizationfor

Research

andTreatm

entof

CancerQualityof

Life

Questionn

aire-Core30

(immed

iatelyafterinterven

tions)

-Painintensity

(visualanalogu

escale)

(beforeandim

med

iately

afterspinalmanipulation,

not

before-after-20

min

ofresting)

Inam

i2017

[8]

Crossover

controlledtrial

(order

ofinterven

tions

rand

omized

)

Symptom

atic:volun

teers

(unclear

origin

–prob

ablythe

sameas

Ogu

raet

al.2011)

with

mechanicalcervicalp

ainand

shou

lder

stiffne

ss.

21(21/0)

-? -26

-Instrumen

tally

assisted

manipulation(whe

rene

eded

,anywhe

reat

thespine,sacroiliac

jointsand/or

scapulae,m

eanof

8pe

rsubject)

-Con

trol:20min

ofresting

Region

alcerebral

metabolic

rate

(rate

ofglucose

consum

ption)

Nobe

fore

measuremen

tAfter:b

etween35

min

to1.05

h.po

st-in

terven

tionor

resting

Pain

intensity

(visualanalogu

escale)

(beforeandim

med

iately

afterspinalmanipulation,

and

before

andafter20

min

ofresting,

onlyfor9/21

subjects)


Table

2Descriptio

nof

eigh

tstud

iesinclud

edin

asystem

aticreview

ontheeffect


‘brain

functio

n’,com

parin


aninactive

control(Co

ntinued)

1st

author

Yr Ref

Design

Type

ofstud

ysubjects

Num

berof

stud

ysubjects

(males/

females)

-Age

(rang

e)-M

ean

-Typeof

spinal

manipulation

-Typeof

control

How

was

cerebral

activity

measured?

Whe

nwas

cerebralactivity

measured?

Clinicalou

tcom

esassessed

(measuremen

ttool

andtim

eof

assessmen

t)

Haavik-

Taylor

2007aa

[20]

Crossover

controlledtrial

(order

ofinterven

tions

rand

omized

)

“Sub

clinicalne

ck/spinalp

ain”:

volunteers(unkno

wnorigin)

with

ahistoryof

recurringne

ckpain

orstiffne

ssandwith

eviden

ceof

cervicalspinal

dysfun

ction,

pain

freeat

the

timeof

thestud

y.

13(5

/8)

-22-45

-31

-HVLAcervical(X2to

3pe

rsubject)

-Con

trol

1:passivehe

admovem

entwith

out

loadingandthrust

-Con

trol

2:no

thing

MEP

amplitu

des

CSP

duratio

ns

Before

After:

-with

in0–10

min

-with

in10–20min

-with

in20–30min

Non

e

Haavik-

Taylor

2010b

[21]

Crossover

controlledtrial

(order

ofinterven

tions

rand

omized

)

“Sub

clinicalne

ck/spinalp

ain”:

stud

entandun

iversity

staff

mem

bers,volun

teers,with

reoccurringne

ckprob

lemsand

eviden

ceof

cervicalspine

dysfun

ction,

pain

freeat

the

timeof

thestud

y.

11(4

/7)

-22-40

-29

-HVLAcervical(nb

unknow

n,may

beat

severallevels)+20

min

oftyping

task

-Con

trol:20min

oftyping

task

only

SEPMU/M

+Upe

akratio

s:-P14-N18

complex

-ParietalN

20(N20-P25

complex)

-FrontalN30

(P22-N30

complex)

Before

After:

exacttim

eun

clear(saidto

beim

med

iatelyafterHVLA+20

min

oftyping

task

orafter20

min

typing

task

only,b

utalso

said

tobe

with

in25

min

post

interven

tions,i.e.p

ossiblywith

in45

min

afterspinalmanipulation)

Non

e

Articlesarepresen

tedby

(i)type

ofstud

ysubjects,i.e.h

ealth

y,symptom

aticor

“sub

clinical

neck/spina

lpain”

subjects,(ii)

type

ofou

tcom

esor

family

ofou

tcom

es,and

(iii)consecutivelyby

year

ofpu

blication

CSPCortical

silent

perio

d,HVLAHigh-velocity

low-amplitu

de,M

EPMotor

evok

edpo

tential,nb

Num

ber,SEPSo

matosen

sory

evok

edpo

tential

a Article

presen

tedin

Tables

2an

d3


Table

3Descriptio

nof

nine

stud

iesinclud

edin

asystem

aticreview

ontheeffect


‘brain

functio

n’,com

parin


anothe

rph

ysical

stim

ulus

1stauthor

Yr Ref

Design

Type

ofstud

ysubjects

Num

berof

stud

ysubjects(m

ales/

females)

-Age

(rang

e)-M

ean

-Typeof

spinalmanipulation

-Typeof

control

How

was

cerebral

activity

measured?

Whe

nwas

cerebral

activity

measured?

Clinicalou

tcom

esassessed

(measuremen

ttool

andtim

eof

assessmen

t)

Dishm

an2008

a

[18]

Rand

omized

controlled

trial

Health

y:he

althychiro

practic

stud

ents,

volunteers.

72(21/5)

interven

tion

grou

p/(15/8)

control1

/(14/9)

control2

-?(3

grou

ps,

said

tobe

betw

eentheir

20sand30s)

-?(3

grou

ps)

-HVLAL5-S1(X1)

-Con

trol

1:L5-S1preloading

-Con

trol

2:side

posture

positio

ning

MEP

amplitu

des

Before

(10

MEP

recorded

durin

g100s)

After:

immed

iately

(10MEP

recorded

durin

g100s)

Non

e

Gay

2014

[22]

Rand

omized

controlled

trial

Symptom

atic:volun

teersfro

maprevious

clinicaltrial,recruitedat

thelocalu

niversity,

hospitaland

surrou

ndingcommun

ity,w

hocompleted

anexercise-in

jury

protocol

toindu

cemyalgiain

thelow

back.

24 (1/5)

manipulation

grou

p/(1

/7)

mob

ilizatio

ngrou

p/(5

/5)

therapeutic

touch

grou

p(7

/17)

-?/?

/?(re

quiredto

bebe

tween18

and

44)

−21

manipulation

grou

p/21

mob

ilizatio

ngrou

p/23

therapeutic

touchgrou

p

-HVLA(X1,prob

ablyin

the

lumbarspine)

-Con

trol

1:gradeIIIlumbar

spinalmob

ilizatio

n-Con

trol

2:therapeutic

touch

(ligh

tpressure

with

acontact

tothesacroiliacjoints)

Functio

nal

conn

ectivity

Before

After:

immed

iately

Pain

intensity

(101-

pointnu

merical

ratin

gscale)

(beforeand

immed

iatelyafter

ineach

grou

p)

Haavik-Taylor

2007aa

[20]

Crossover

controlled

trial(orde

rof

interven

tions

rand

omized

)

“Sub

clinicalne

ck/spinalp

ain”:volun

teers

(unkno

wnorigin)with

ahistoryof

recurring

neck

pain

orstiffne

ssandwith

eviden

ceof

cervicalspinaldysfun

ction,

pain

freeat

the

timeof

thestud

y.

13(5

/8)

−22-45

−31

-HVLAcervical(X2to

3pe

rsubject)

-Con

trol

1:passivehe

admovem

entwith

outloading

andthrust

-Con

trol

2:no

thing

MEP

amplitu

des

CSP

duratio

nsBefore

After:

-with

in0–10

min

-with

in10–

20min

-with

in20–

30min

Non

e

Haavik-Taylor

2008

[23]

Crossover

controlled

trial(orde

rof

interven

tions

rand

omized

)

“Sub

clinicalne

ck/spinalp

ain”:adu

lts(unkno

wnorigin)with

ahistoryof

reoccurringne

ckpain

orstiffne

ssandwith

eviden

ceof

cervicalspinaldysfun

ction,

pain

freeat

thetim

eof

thestud

y.

12(7

/5)

−19-45

−27

-HVLAcervical(nbun

know

n,may

beat

severallevels)

-Con

trol:p

assive

head

movem

entwith

outloading

andthrust

MEP

amplitu

des

CSP

duratio

nsSICI

SICF

Before

After:exact

time

unknow

n

Non

e

Haavik2016

[24]

Crossover

controlled

trial

“Sub

clinicalne

ck/spinalp

ain”:volun

teers

(unkno

wnorigin)with

ahistoryof

spinal

symptom

sandwith

eviden

ceof

spinal

and/or

pelvicdysfun

ctionbu

twho

didno

tyetsoug

httreatm

entforthisandpain

free

atthetim

eof

thestud

y.

12(?)

-? −28

-HVLAcervical(nbun

know

n,may

beat

severallevels)

-Con

trol:p

assive

head

movem

entwith

outloading

andthrust

MEP

amplitu

des

Slop

eof

the

steepe

stpartof

the

curve(k)

Stim

ulus

intensity

requ

iredto

obtain

arespon

sethat

is50%

ofthemax

(S50)

Before

After:exact

time

unknow

n

Non

e


Table

3Descriptio

nof

nine

stud

iesinclud

edin

asystem

aticreview

ontheeffect


‘brain

functio

n’,com

parin


anothe

rph

ysical

stim

ulus

(Con

tinued)

1stauthor

Yr Ref

Design

Type

ofstud

ysubjects

Num

berof

stud

ysubjects(m

ales/

females)

-Age

(rang

e)-M

ean

-Typeof

spinalmanipulation

-Typeof

control

How

was

cerebral

activity

measured?

Whe

nwas

cerebral

activity

measured?

Clinicalou

tcom

esassessed

(measuremen

ttool

andtim

eof

assessmen

t)

Haavik-Taylor

2007b

[25]

Twogrou

ps“pseud

o-rand

omized

”trial

“Sub

clinicalne

ck/spinalp

ain”:volun

teers

(orig

inun

know

n)with

reoccurringne

ckprob

lemsandeviden

ceof

cervicalspine

dysfun

ction,

pain

freeat

thetim

eof

the

stud

y.

24(7

/5)

interven

tion

grou

p/(4

/8)

controlg

roup

−20-53

interven

tion

grou

p/21–35

controlg

roup

−30

interven

tion

grou

p/27

controlg

roup

-HVLAcervical(X2to

3pe

rsubject)

-Con

trol:p

assive

head

movem

entwith

outloading

andthrust

SEPlatenciesand

amplitu

des:P14–18

complex,N

20(P14-

N20

andN20-P27)

andN30

(P22-N30)

peaks

Before

After:

-with

in0–10

min

-with

in10–

20min

-with

in20–

30min

Non

e

Haavik-Taylor

2010a

[26]

Crossover

controlled

trial(orde

rof

interven

tions

rand

omized

)

“Sub

clinicalne

ck/spinalp

ain”:volun

teers

(orig

inun

know

n)with

reoccurringne

ckprob

lemsandeviden

ceof

cervicalspine

dysfun

ction,

pain

freeat

thetim

eof

the

stud

y.

13 (5/8)

−18-40

−28

-HVLAcervical(nbun

know

n,may

beat

severallevels)

-Con

trol:p

assive

head

movem

entwith

outloading

andthrust

SEPMU/M

+Upe

akratio

s:P14-N18

complex,N

20-P25

complex,and

P22-

N30

complex

ratio

s

Before

After:

with

in25

min

Non

e

Niazi2015

[27]

Crossover

controlled

trial

(order

ofinterven

tions

rand

omized

)

“Sub

clinicalne

ck/spinalp

ain”:volun

teers

(orig

inun

know

n)with

recurring,

interm

itten

tlow-grade

spinalpain,ache,or

tension,

with

eviden

ceof

spine

dysfun

ction,

butwhich

didno

tsoug

httreatm

entforthisprob

lem

andarepain

freeat

thetim

eof

thestud

y.

10 (10/0)

-?(re

quiredto

bebe

tween18

and40)

−28

-HVLA(whe

rene

eded

,inany

spinelevelo

rsacroiliacjoints,

nbun

know

n_may

beat

severallevels)

-Con

trol:p

assive

andactive

movem

entsof

thesubject’s

head,spine

,and

body

into

themanipulationsetup

positio

ns,w

ithou

tloading

andthrust

V-waveam

plitu

deBefore

After:

exacttim

eun

know

n

Non

e

Christiansen

2018

[28]

Crossover

controlled

trial(orde

rof

interven

tions

rand

omized

)

“Sub

clinicalne

ck/spinalp

ain”:elite

Taekwon

doathletes,from

theAuckland

area,w

ith“sub

clinicalspinalpain”and

eviden

ceor

spinedysfun

ction,

pain

freeat

thetim

eof

thestud

y.

12(6

/6)

-?(re

quiredto

bebe

tween17

and50)

−25

-HVLA(whe

rene

eded

,inany

spinelevelo

rsacroiliacjoints,

nbun

know

n_may

beat

severallevels)

-Con

trol:p

assive

andactive

movem

entsof

thesubject’s

head

andspineinto

the

manipulationsetuppo

sitio

ns,

with

outloadingandthrust

V-waveam

plitu

deBefore

After:

-immed

iately

−30

min

−60

mins

Non

e

Articlesarepresen

tedby

(i)type

ofstud

ysubjects,i.e.h

ealth

y,symptom

aticor

“sub

clinical

neck/spina

lpain”

subjects,(ii)

type

ofou

tcom

esor

family

ofou

tcom

es,and


year

ofpu

blication

CSPCortical

silent

perio

d,HVLAHigh-velocity

low-amplitu

de,M

EPMotor

evok

edpo

tential,nb

Num

ber,SEPSo

matosen

sory

evok

edpo

tential,SICF

Shortinterval

intra-cortical

inhibitio

n,SICI

Shortinterval

intra-cortical

facilitation,

SMSp

inal

man

ipulation

a Article

presen

tedin

Tables

2an

d3


subjects were not explicitly described as such [20, 21, 23,25, 26]. However, the description provided by their au-thors clearly referred to the definition of “subclinicalneck/spinal pain” subjects [24].All the included studies were controlled trials, includ-

ing two to three experimental groups, most of them witha random allocation (n = 14) and mostly conducted usinga crossover design (n = 10). Only one study reported thedates and duration of data collection [9].Most of the included articles investigated high-velocity

low-amplitude SM, whereas three investigated instru-mentally or mechanically assisted techniques. The areawhere SM was provided varied across studies to includeall areas of the spine, whereas one study did not indicatewhere [19]. In most of them SM was provided ‘wheredeemed necessary’.Most studies used as a control group some passive

type of procedure, considered in the present systematicreview as ‘another physical stimulus’, or used a com-pletely ‘inactive control’, whereas three attempted to usesome types of sham comparators [9, 14, 15]. One studycompared SM to two other manual therapies, i.e. spinalmobilization and therapeutic touch of the lumbosacralarea [22].The outcomes of all these studies were either de-

scribed as reflecting some type of ‘brain function(s)’ or assuggesting some type of ‘brain function(s)’, meaning thatsome outcomes could also reflect, for example, neuro-physiological changes at a segmental level (e.g. V-wave,motor evoked potentials, or cortical silent periods) [12,34]. The outcome measures and measurement tools usedin the selected studies are briefly described in Add-itional file 3. In two studies, outcomes were assessedonly after intervention or control, presumably for ethicalreasons [8, 19]. As can be seen in Tables 1, 2 and 3, inall other studies outcomes were assessed before and afterintervention at various time points. However, six studiesdid not specify the time of re-assessment at all and fourdid not report it clearly.The four studies conducted on symptomatic subjects

described in their Methods section that they alsoassessed clinical outcomes, mainly pain intensity (seeTables 1, 2 and 3) [8, 9, 19, 22]. However, only two ofthem had as one of their research objectives to investi-gate whether a relation exists between potential neuro-physiological changes and pain intensity changes afterintervention, and none of these two assessed this againsta sham intervention [19, 22].

Data extraction (n = 18)The data extraction process was relatively problem freewith only few exceptions. These concerned some articlesin which the statistics and/or results sections were un-clear [9, 14, 21–23, 26, 27]. This was resolved through

multiple discussions. In addition, experts’ opinions weresought in these areas and the experts (MAA, MS andPS) also reported difficulty to interpret some of the stud-ies [9, 14, 21, 23, 26].

Data synthesis: methodological quality of the studies (n =18)The level of methodological quality was generally ‘low’(n = 7) or ‘medium’ (n = 8), except for three articles thatwere considered to be of ‘acceptable’ quality (see Table12 for a summary). The most frequently encounteredmethodological weaknesses were: (i) the success of theblinding of the subjects was uncertain or unsuccessful(the three ‘effect’ studies), (ii) no clear reporting whetherthe study was conducted on naïve subjects (most of the‘differences in outcome’ studies), (iii) no reportingwhether the assessor was blinded to treatment group(most studies), and (iv) no reporting whether the personwho analyzed the data was blinded to treatment group(most studies). In addition, the experts (MAA, MS andPS) sometimes commented on unusual procedures (fordetailed information see Tables 4, 5, 6, 7 and 8 col.10).In two of the studies it was clear that the authors did

not report having performed a between-group analysis[20, 25]. Thus their results were not taken into accountfor our five research questions, and were therefore notreported in Tables 10 and 11. Another study did appearto compare the outcomes of SM on two different handmuscles rather than to compare the effect of SM to acontrol intervention (see Table 5 col.10) [23]. This art-icle was therefore not reported in Table 11.Also, three studies did not report results in relation to all

the statistical between-group comparisons that they stated intheir respective Methods sections that they would do [14,22, 27]. For these studies, only the results from between-group comparisons, if present, were therefore reported.A total of 13 studies were finally used to answer our

five research questions. One of these studies appears intwo of the three results tables (Tables 10 and 11) [18].

Data synthesis: answers to research questions (n = 13)Sham controlled studies (Table 9), i.e. ‘effect’ studies (n = 3)Only three studies used a sham comparator and weretherefore considered as potentially able to provide an-swers to the research questions 1 to 3 [9, 14, 15]. How-ever, in two of these the credibility of the sham isunclear [9, 15], and in the third, the sham was recog-nized as such by most of the study subjects [14]. Twowere considered of ‘medium’ methodological quality[14, 15] and one of ‘acceptable’ methodological quality[9] (see Table 12). These studies, reporting on symp-tomatic subjects or on “subclinical neck/spinal pain”subjects, investigated the potential effect of SM on‘brain function’ by using three different outcome


Table

4Qualityitemsandscoreof

onestud

yusingareactio

n-tim

etask

includ

edin

asystem

aticreview

ontheeffect


‘brain

functio

n’

1stAutho

rYr

ofpu

blication

Ref

-Were

stud

ysubjectsin

sham

controlled

stud

ies

repo

rted

tobe

blind?

(Yes

/No/

Unclear)

-Ifyes/

unclear,

was

the

blinding

tested

for

success?

(Yes

/No)

-Ifyes,was

it successful?

(Yes

/No)

-Werestud

ysubjects

instud

ieswith

controlg

roup

repo

rted

tobe

naive?

(Yes

/No/Unclear)

-Was

theorigin

ofthe

subjectsrepo

rted

(Yes

/No)

-Ifyes,do

esitallow

toexclud

eany

interest?(Yes

/No/

Unclear)

Were

stud

ysubjects

repo

rted

tohave

been

rand

omly

allocated

tostud

ygrou

ps?

(Yes

/No

/Unclear)

Werestud

ygrou

pscomparable

inrelatio

nto

symptom

swhe

nstud

ying

symptom

atic

subjects

(durationand

pain

intensity)(NA

whe

ncross-

over

stud

yde

sign

)?(Yes

/No)

Werethe

interven

tion

and

control(s)

well

describ

ed(atleast

whe

reand

how)?(Yes

/No)

Was

the

assessor

repo

rted

tobe

blind

togrou

pallocatio

n?(Yes

/No)

Were

losses

and

exclusions

ofstud

ysubjects

repo

rted

orob

viou

sin

result

section

(includ

ing

intables

or graphs)?

(Yes

/No/

Unclear)

Was

the

person

who

statistically

analyzed

thedata

repo

rted

tobe

blind

togrou

pallocatio

n?(Yes

/No)

Com

men

tsby

thetechnical

expe

rts(i)

onthestatistical

analysis,and

(ii)in

relatio

nto

themetho

dology

and/or

technicalaspects

Qualityscore(risk

ofbias,alsoinclud

ingan

externalvalidity

criteria)andclassification

Kelly

2000

[29]

-Yes

(but

inrelatio

nto

theou

tcom

e)-Yes

-No

1: -The

authorsused

aStud

ent

tteststo

compare

means

insteadof

usingamixed

-mod

elANOVA

,followed

bypo

st-hoc

testsifne

eded

.-The

authorsdidno

tstud

yho

wRT

(forcorrectansw

ers)varied

with

angle,which

isthemain

analysiscond

uctedin

the

literatureon

such

data.

Therefore,with

outsuch

a(usuallylinear)tren

danalysis

itisno

tpo

ssibleto

unde

rstand

iftheoverallm

eaneffect

observed

bytheauthorsisdu

eto

achange

inslop

e(re

flecting

achange

inprocessing

speed)

orin

intercep

t(re

flectinga

change

instim

ulus

encoding

).3: -Between-grou

pdifference

pre-po

stsign

ificant

onlywith

one-side

dt-test.

-The

betw

een-grou

pdifference

pre-po

stisno

trepo

rted

forthesimple

RTtask

butitseem

sthat

acontrib

ution

ofthesimpleRT

totheRT

ofthe

complex

task

cann

otbe

exclud

ed.

-Unclear

whe

ther

errorswerealso

coun

ted.

3.5/6(58%

)med

ium

NA

=Unclear

0.5pt

Yes1pt

NA(health

ysubjects)

-Yes

0.5pt

-Yes

0.5pt

No0pt

Yes1pt

No0pt

NANot

applicab

le,R

TRreactiontim

e


Table

5Qualityitemsandscores

ofsevenstud

iesusingtranscranialmagne

ticindu

cedou

tcom

emeasuresinclud

edin

asystem

aticreview

ontheeffect

ofspinalmanipulation

on‘brain

functio

n’1stAutho

rYr

ofpu

blication

Ref

-Werestud

ysubjectsin

sham

controlledstud

iesrepo

rted

tobe

blind?

(Yes

/No/Unclear)

-Ifyes/un

clear,was

theblinding

tested

forsuccess?

(Yes

/No)

-Ifyes,was

itsuccessful?

(Yes

/No)

-Werestud

ysubjectsin

stud

ieswith

controlg

roup

repo

rted

tobe

naive?

(Yes

/No/Unclear)

-Was

theorigin

ofthe

subjectsrepo

rted

(Yes

/No)

-Ifyes,do

esitallow

toexclud

eanyinterest?

(Yes

/No/Unclear)

Werestud

ysubjects

repo

rted

tohave

been

rand

omly

allocatedto

stud

ygrou

ps?(Yes

/No/

Unclear)

Werestud

ygrou

pscomparablein

relatio

nto

symptom

swhe

nstud

ying

symptom

atic

subjects(duration

andpain

intensity)(NA

whe

ncross-over

stud

yde

sign

)?(Yes

/No)

Werethe

interven

tion

and

control(s)

well

describ

ed(atleast

whe

reand

how)?(Yes

/No)

Was

the

assessor

repo

rted

tobe

blind

togrou

pallocatio

n?(Yes

/No)

Were

losses

and

exclusions

ofstud

ysubjects

repo

rted

orob

viou

sin

result

section

(includ

ing

intables

or graphs)?

(Yes

/No/

Unclear)

Was

the

person

who

statistically

analyzed

thedata

repo

rted

tobe

blind

togrou

pallocatio

n?(Yes

/No)

Com

men

tsby

thetechnicalexperts(i)

onthestatisticalanalysis,and

(ii)in

relatio

nto

themetho

dology

and/or

technicalaspects

Qualityscore(risk

ofbias,

also

includ

ingan

external

validity

criteria)and

classification

Dishm

an2002

[17]

-No

-Yes

-No

2: -MEP

metho

dology

does

notcorrespo

ndto

standard:n

omotor

threshold,

noforcecontrol,

andlack

ofrand

omintervalsbe

tweenstim

ulus

-The

coilpo

sitio

ning

seem

sno

tapprop

riate

tolower

legMEPs.

2.5/6(42%

)med

ium

NA

=No0pt

Unclear

("cou

nterbalanced

")0.5pt

NA(health

ysubjects)

-Yes

0.5pt

-Yes

0.5pt

No0pt

Yes1pt

No0pt

Haavik-Taylor

2007a

[20]

-No

-No

-NA

1: -The

authorsstated

having

used

plan

ned

compa

rison

sinsteadof

postho

canalysis

inorde

rto

minim

izeType

1error.How

ever,

plan

nedcomparison

sdo

notminim

ize

Type

1error.

-The

ymen

tionrunn

ingaon

e-way

repe

ated

measures

ANOVA

with

thefactor

“interven

tion”.

How

ever,the

degreesof

freed

omof

theF

fortheresult

clearly

show

that

authors

treated“in

terven

tion”

asa

betw

een-subjectsfactor,

which

isno

tcorrect.

2/6(33%

)low

NA

=No0pt

Yes1pt

NA(SCPsubjects

/cross-over)

-Yes

0.5pt

-Yes

0.5pt

No0pt

No0pt

No0pt

Dishm

an2008

[18]

-No

-Yes

-No

2: -MEP

metho

dology

isno

tstandard:lack

ofprecisemotor

threshold,

andlack

ofrand

omintervalsbe

tweenstim

ulus.

-Fig.1Cindicatesan

inhibitio

nin

thetim

eintervalpriorto

SM,w

hich

may

berespon

sibleforsign

ificant

differences

and

relativeincrease

ofam

plitu

deafterSM

.

2/6(33%

)low

NA

=No0pt

Yes1pt

NA(health

ysubjects)

-Yes

0.5pt

-Yes

0.5pt

No0pt

No0pt

No0pt

Haavik-Taylor

2008

[23]

-No

-No

-NA

1: -The

authorsmen

tionrunn

ing2-way

ANOVA

sforrepe

ated

measureswith

thefactors“m

uscle”

and“in

terven

tion”

wereappliedto

compare

theeffects

ofSM

onthetw

odifferent

uppe

rlim

bmuscles.H

owever,the

degreesof

freed

omof

theFfortheresults

clearly

show

that

authorstreatedthetw

ofactorsbe

tween-

subjects,w

hich

isno

tcorrect.

-The

yusettestsinsteadof

post-hoc

test

fortestingpairw

isecomparison

ssubseq

uent

totheANOVA

.

2/6(33%

)low

NA

=No0pt

Yes1pt

NA(SCPsubjects

/cross-over)

-Yes

0.5pt

-Yes

0.5pt

No0pt

No0pt

No0pt


Table


ofsevenstud

iesusingtranscranialmagne

ticindu

cedou

tcom

emeasuresinclud

edin

asystem

aticreview

ontheeffect

ofspinalmanipulation

on‘brain

functio

n’(Con

tinued)

1stAutho

rYr

ofpu

blication

Ref

-Werestud

ysubjectsin

sham

controlledstud

iesrepo

rted

tobe

blind?

(Yes

/No/Unclear)

-Ifyes/un

clear,was

theblinding

tested

forsuccess?

(Yes

/No)

-Ifyes,was

itsuccessful?

(Yes

/No)

-Werestud

ysubjectsin

stud

ieswith

controlg

roup

repo

rted

tobe

naive?

(Yes

/No/Unclear)

-Was

theorigin

ofthe

subjectsrepo

rted

(Yes

/No)

-Ifyes,do

esitallow

toexclud

eanyinterest?

(Yes

/No/Unclear)

Werestud

ysubjects

repo

rted

tohave

been

rand

omly

allocatedto

stud

ygrou

ps?(Yes

/No/

Unclear)

Werestud

ygrou

pscomparablein

relatio

nto

symptom

swhe

nstud

ying

symptom

atic

subjects(duration

andpain

intensity)(NA

whe

ncross-over

stud

yde

sign

)?(Yes

/No)

Werethe

interven

tion

and

control(s)

well

describ

ed(atleast

whe

reand

how)?(Yes

/No)

Was

the

assessor

repo

rted

tobe

blind

togrou

pallocatio

n?(Yes

/No)

Were

losses

and

exclusions

ofstud

ysubjects

repo

rted

orob

viou

sin

result

section

(includ

ing

intables

or graphs)?

(Yes

/No/

Unclear)

Was

the

person

who

statistically

analyzed

thedata

repo

rted

tobe

blind

togrou

pallocatio

n?(Yes

/No)

Com

men

tsby



(ii)in

relatio

nto

themetho

dology

and/or

technicalaspects

Qualityscore(risk

ofbias,

also

includ

ingan

external

validity

criteria)and

classification

2: Theconclusion

sarefarfe

tche

das

assumptions

andde

ductionaremade

which

cann

otno

tbe

backed

bytheresults.

Fryer

2012

[16]

-No

-Yes

-No

2:Thecoilpo

sitio

ning

seem

sno

tapprop

riate

tolower

legMEPs.

2.5/6(42%

)med

ium

NA

=No0pt

Yes1pt

NA(health

ysubjects)

-Yes

0.5pt

-Yes

0.5pt

No0pt

Unclear

0.5pt

No0pt

Haavik

2016

[24]

-Unclear

(mostsubjectswere

“noviceto

chiro

practic”)

-No

-NA

2: -The

recruitm

entcurves

lack

measure

ofvariance.

-Feedb

ackfro

mbackgrou

ndEM

Gislacking,

which

isaconcep

tualconcernandcould

explainob

served

increasedin

amplitu

des.

2/6(33%

)low

NA

=No0pt

No0pt

NA(SCPsubjects

/cross-over)

-Yes

0.5pt

-Yes

0.5pt

No0pt

Yes1pt

No0pt

Baarbé

2018

[15]

-Yes

-No

-NA

Non

ein

relatio

nto

statistics

Noexpertwas

availablein

relationto

thetechnical

aspectsof

thisou

tcom

emeasure

3.5/6(58%

)med

ium

=Unclear

0.5pt

NA

Yes1pt

NA(SCPsubjects)

-Yes

0.5pt

-Yes

0.5pt

No0pt

Yes1pt

No0pt

EMGElectrom

yograp

hy,M

EPMotor-evo

kedpo

tential,NANot

applicab

le,SCP

“sub

clinical

neck/spina

lpain”,SM

Spinal

man

ipulation


Table


offour

stud

iesusingou

tcom

emeasuresin

relatio

nto

somatosen

sory-evokedpo

tentialsinclud

edin

asystem

aticreview

ontheeffect

ofspinal

manipulationon

‘brain

functio

n’1stAutho

rYr

ofpu

blication

Ref

-Werestud

ysubjectsin

sham

controlledstud

ies

repo

rted

tobe

blind?

(Yes

/No/

Unclear)

-Ifyes/un

clear,

was

theblinding

tested

for

success?

(Yes

/No)

-Ifyes,was

itsuccessful?(Yes

/No)

-Werestud

ysubjectsin

stud

ieswith

controlg

roup

repo

rted

tobe

naive?

(Yes

/No/

Unclear)

-Was

theorigin

ofthesubjects

repo

rted

(Yes

/No)

-Ifyes,do

esit

allow

toexclud

eanyinterest?(Yes

/No/Unclear)

Werestud

ysubjectsrepo

rted

tohave

been

rand

omly

allocatedto

stud

ygrou

ps?

(Yes

/No/Unclear)

Werestud

ygrou

pscomparable

inrelatio

nto

symptom

swhe

nstud

ying

symptom

atic

subjects

(durationand

pain

intensity)(NA

whe

ncross-

over

stud

yde

sign

)?(Yes

/No)

Werethe

interven

tion

and

control(s)

well

describ

ed(atleast

whe

reand

how)?(Yes

/No)

Was

theassessor

repo

rted

tobe

blindto

grou

pallocatio

n?(Yes

/No)

Were

losses

and

exclusions

ofstud

ysubjects

repo

rted

orob

viou

sin

result

section

(includ

ing

intables

or graphs)?

(Yes

/No/

Unclear)

Was

the

person

who

statistically

analyzed

thedata

repo

rted

tobe

blind

togrou

pallocatio

n?(Yes

/No)

Com

men

tsby


onthestatistical

analysis,and

(ii)in

relatio

nto

themetho

dology

and/or

technicalaspects

Qualityscore(risk

ofbias,alsoinclud

ing

anexternalvalidity

criteria)and

classification

Haavik-Taylor

2007b

[25]

-No

-No

-NA

1: -Norepo

rtof

thetestingof

theno

rmality

ofthedata

distrib

ution.

-Tominim

izeType

1error,po

stho

ctestswou

ldbe

approp

riate

(insteadof

planne

dcomparison

s).

-Nobe

tweengrou

pcomparison

was

perfo

rmed

.Noexpertwas

availablein

relationto

thetechnicalaspectsof

thisou

tcom

emeasure

2/6(33%

)low

NA

=No0pt

Unclear

("pseud

orando

mized

")0.5p

t

NA

(SCPsubjects)

-Yes

0.5p

t-Yes

0.5p

tNo(but

data

were

code

dby

aninde

pend

ent

person

toredu

ceanybias

durin

ganalysis)0.5p

t

No0pt

No0pt

Haavik-Taylor

2010a

[26]

-No

-No

-NA

1:Bo

thparametric

andno

nparam

etric

results

onthesame

data

arerepo

rted

.Usually,eith

erdata

areno

rmallydistrib

uted

andparametric

testscanbe

used

ordata

areno

tno

rmally

distrib

uted

andno

n-parametric

testsmustbe

used

.Noexpertwas

availablein

relationto


thisou

tcom

emeasure

2.5/6(42%

)med

ium

NA

=No0pt

Yes1p

tNA(SCP

subjects/

cross-over)

-Yes

0.5pt

-Yes

0.5pt

No(id

emHaavik-Taylor

2007a)

0.5p

t

No0pt

No0pt

Haavik-Taylor

2010b

[21]

-No

-Yes

-Unclear

(“stude

ntsand

staffp

opulation

attheUniversity

ofAuckland”)

1:Seecommen

tsin

relatio

nto

Haavik-Taylor

2010a

Noexpertwas

availablein

relationto


thisou

tcom

emeasure

2.5/6(42%

)med

ium

NA

=No0pt

Yes1pt

NA(SCP

subjects/

cross-over)

-Yes

0.5pt

-Yes

0.5pt

No(id

emHaavik-Taylor

2007b)

0.5pt

No0pt

No0pt

Lelic

2016

[14]

-Unclear

(saidto

benaïve)

-Yes

-No(sham

interven

tionwas

discovered

assuch

bymostof

thesubjects)

1:Unu

sualrepo

rtingof

statistics:no

repo

rtof

which

werethe

expe

rimen

talfactorsandho

wthey

weretreated(but

prob

ablypre/po

stwas

treatedwith

in-sub

jectsand

interven

tions

asbe

tween-subjects),andof

thede

tailedresults

fortheFtestsof

theANOVA

.Noexpertwas

availablein

relationto


thisou

tcom

emeasure

2.5/6(42%

)med

ium

=No0pt

NA

Yes1pt

NA(SCP

subjects/

cross-over)

-Yes

0.5pt

-No0pt

No0pt

Yes1pt

No0pt

NANot

applicab

le,SCP

“sub

clinical

neck/spina

lpain”


Table


offour

stud

iesusingne

uroimagingou

tcom

emeasuresinclud

edin

asystem

aticreview

ontheeffect


‘brain

functio

n’

1stAutho

rYr

ofpu

blication

Ref

-Were

stud

ysubjectsin

sham

controlled

stud

ies

repo

rted

tobe

blind?

(Yes

/No/

Unclear)

-Ifyes/

unclear,

was

the

blinding

tested

for

success?

(Yes

/No)

-Ifyes,was

it successful?

(Yes

/No)

-Werestud

ysubjectsin

stud

ieswith

control

grou

prepo

rted

tobe

naive?

(Yes

/No/

Unclear)

-Was

theorigin

ofthe

subjectsrepo

rted

(Yes

/No)

-Ifyes,do

esitallow

toexclud

eanyinterest?(Yes

/No/Unclear)

Werestud

ysubjects

repo

rted

tohave

been

rand

omly

allocatedto

stud

ygrou

ps?(Yes

/No/

Unclear)

Werestud

ygrou

pscomparable

inrelatio

nto

symptom

swhe

nstud

ying

symptom

atic

subjects

(durationand

pain

intensity)(NA

whe

ncrossover

stud

yde

sign

)?(Yes

/No)

Werethe

interven

tion

and

control(s)

well

describ

ed(atleast

whe

reand

how)?(Yes

/No)

Was

the

assessor

repo

rted

tobe

blind

togrou

pallocatio

n?(Yes

/No)

Were

losses

and

exclusions

ofstud

ysubjects

repo

rted

orob

viou

sin

result

section

(includ

ing

intables

or graphs)?

(Yes

/No/

Unclear)

Was

the

person

who

statistically

analyzed

thedata

repo

rted

tobe

blind

togrou

pallocatio

n?(Yes

/No)

Com

men

tsby



(ii)in

relatio

nto

themetho

dology

and/or

technicalaspects

Qualityscore(risk

ofbias,also

includ

ingan

externalvalidity

criteria)and

classification

Ogu

ra2011

[19]

-No

-Yes

-Unclear

(recruitedon

the

campu

sof

Toho

kuUniversity)

1:Theextent

thethresholdforthevoxel

clustersize

was

defined

as“10to

50voxels

minim

um”.Thepu

rposeof

thisvarying

thresholdisun

clear.

3:Lenien

tstatisticalthreshold:

Z=3,extent

threshold;

10voxels.

2/6(33%

)low

NA

=No0pt

Unclear

(“cou

nterbalanced

”)0.5pt

NA(cross-

over)

-No0pt

-Yes

0.5pt

No0pt

Yes1pt

No0pt

Inam

i2017

[8]

-No

-No

-NA

1:Theph

rasing

“(e.g.,10

voxelsminim

um)”

sugg

estsagain(see

thecommen

tin

relatio

nto

Ogu

ra2011)that

thisthresholdwas

not

fixed

.2/6(33%

)low

NA

=No0pt

Yes1pt

NA(cross-

over)

-Yes

0.5pt

-Yes

0.5pt

No0pt

No0pt

No0pt

Gay

2014

[22]

-No

-No

-Unclear

(recruitedfro

mthecampu

sof

the

University

ofFloridaand

UFHealth

Hospitaland

thelocalcom

mun

ity)

1: -Autho

rs“corrected

forthenu

mbe

rof

separate

RM-ANOVA

scond

uctedacross

the

120RO

I-to-RO

Ipairsby

usingapvalueless

than

.01as

sign

ificant.”(p.618).Thisthreshold

(p=0.01)correctio

nformultip

lecomparison

sisno

tconservativeen

ough

.-The

rewas

neith

erbe

tween-grou

psstatistical

testat

“pre”,no

rat

“post”.

3:Lenien

tstatisticalthreshold:

p=0.01

with

120comparison

s.

5/7(71%

)acceptable

NA

=No0pt

Yes1pt

Yes1pt

Yes0.5pt

-Yes

0.5pt

Yes1pt

Yes1pt

No0pt

Sparks

2017

[9]

-Yes

-No

-NA

1:Theauthorsused

analph

a=0.01

threshold

forthefM

RIanalysis.Itisno

tconservative

enou

ghin

myop

inion(asdiscussedby

Eklund

etal.2015,andLieb

erman

&


Table


offour

stud

iesusingne

uroimagingou

tcom

emeasuresinclud

edin

asystem

aticreview

ontheeffect


‘brain

functio

n’(Con

tinued)

1stAutho

rYr

ofpu

blication

Ref

-Were

stud

ysubjectsin

sham

controlled

stud

ies

repo

rted

tobe

blind?

(Yes

/No/

Unclear)

-Ifyes/

unclear,

was

the

blinding

tested

for

success?

(Yes

/No)

-Ifyes,was

it successful?

(Yes

/No)

-Werestud

ysubjectsin

stud

ieswith

control

grou

prepo

rted

tobe

naive?

(Yes

/No/

Unclear)

-Was

theorigin

ofthe

subjectsrepo

rted

(Yes

/No)

-Ifyes,do

esitallow

toexclud

eanyinterest?(Yes

/No/Unclear)

Werestud

ysubjects

repo

rted

tohave

been

rand

omly

allocatedto

stud

ygrou

ps?(Yes

/No/

Unclear)

Werestud

ygrou

pscomparable

inrelatio

nto

symptom

swhe

nstud

ying

symptom

atic

subjects

(durationand

pain

intensity)(NA

whe

ncrossover

stud

yde

sign

)?(Yes

/No)

Werethe

interven

tion

and

control(s)

well

describ

ed(atleast

whe

reand

how)?(Yes

/No)

Was

the

assessor

repo

rted

tobe

blind

togrou

pallocatio

n?(Yes

/No)

Were

losses

and

exclusions

ofstud

ysubjects

repo

rted

orob

viou

sin

result

section

(includ

ing

intables

or graphs)?

(Yes

/No/

Unclear)

Was

the

person

who

statistically

analyzed

thedata

repo

rted

tobe

blind

togrou

pallocatio

n?(Yes

/No)

Com

men

tsby



(ii)in

relatio

nto

themetho

dology

and/or

technicalaspects

Qualityscore(risk

ofbias,also

includ

ingan

externalvalidity

criteria)and

classification

Cun

ning

ham

2009).

3: -Unclear

whe

ther

statisticalthresholdapplied

across

thewho

lebrainor

justfortheregion

ofinterest.

-Itisun

clearho

wtheregion

ofinterestwas

defined

5.5/7(79%

)acceptable

=Unclear

0.5pt

NA

Yes1pt

Yes1pt

-Yes

0.5pt

-Yes

0.5pt

Yes1pt

Yes1pt

No0pt

fMRI

functio

nalm

agne

ticresona

nceim

aging,

NANot

applicab

le


Table


oftw

ostud

iesusingV-waveas

outcom

emeasuresinclud

edin

asystem

aticreview

ontheeffect


‘brain

functio

n’

1stAutho

rYr

ofpu

blication

Ref

-Were

stud

ysubjectsin

sham

controlled

stud

ies

repo

rted

tobe

blind?

(Yes

/No/

Unclear)

-Ifyes/

unclear,

was

the

blinding

tested

for

success?

(Yes

/No)

-Ifyes,was

it successful?

(Yes

/No)

-Were

stud

ysubjects

instud

ies

with

control

grou

prepo

rted

tobe

naive?

(Yes

/No

/Unclear)

-Was

the

origin

ofthe

subjects

repo

rted

(Yes

/No)

-Ifyes,

does

itallow

toexclud

eany

interest?

(Yes

/No

/Unclear)

Were

stud

ysubjects

repo

rted

tohave

been

rand

omly

allocated

tostud

ygrou

ps?

(Yes

/No

/Unclear)

Werestud

ygrou

pscomparable

inrelatio

nto

symptom

swhe

nstud

ying

symptom

atic

subjects

(durationand

pain

intensity)(NA

whe

ncross-

over

stud

yde

sign

)?(Yes

/No)

Werethe

interven

tion

and

control(s)

well

describ

ed(atleast

whe

reand

how)?(Yes

/No)

Was

the

assessor

repo

rted

tobe

blind

togrou

pallocatio

n?(Yes

/No)

Were

losses

and

exclusions

ofstud

ysubjects

repo

rted

orob

viou

sin

result

section

(includ

ing

intables

or graphs)?

(Yes

/No/

Unclear)

Was

the

person

who

statistically

analyzed

thedata

repo

rted

tobe

blind

togrou

pallocatio

n?(Yes

/No)

Com

men

tsby


onthe

statisticalanalysis,and

(ii)in

relatio

nto

the

metho

dology

and/or

technicalaspects

Qualityscore(risk

ofbias,alsoinclud

ingan

externalvalidity

criteria)

andclassification

Niazi

2015

[27]

-No

-No

-NA

Non

ein

relatio

nto

statistics

Noexpertwas

availablein

relationto

the

technicalaspectsof

thisoutcom

emeasure

2.5/6(42%

)med

ium

NA

=No0pt

Yes1pt

NA(SCP

subjects)

-Yes

0.5pt

-No0pt

No0pt

Yes1pt

No0pt

Christiansen

2018

[28]

-No

-Yes

-Yes

Non

ein

relatio

nto

statistics

Noexpertwas

availablein

relationto

the

technicalaspectsof

thisoutcom

emeasure

5/6(83%

)acceptable

NA

=Unclear

0.5pt

Yes1pt

NA(SCP

subjects)

-Yes

0.5pt

-No0pt

Yes1pt

Yes1pt

Yes1pt

NANot

applicab

le,SCP

“sub

clinical

neck/spina

lpain”


Table

9Results

from

threestud

iesinclud

edin

asystem

aticreview

ontheeffect


‘brain

functio

n’,com

parin


asham

interven

tion

1st

Autho

rYear

Ref

Type

ofstud

ysubjects

Outcomevariable

Was

astatisticallysign

ificant

differencebe

tweengrou

psob

served

?Was

therearelatio

nship

betw

eenbrainchange

sandanyclinical

outcom

e?

Timeof

assessmen

tQuality

classification

Sparks

2017

[9]

Symptom

atic

(mechanicaln

eck

pain

<of

6weeks

ofdu

ratio

n)

Bloo

doxygen

ation-level

depe

nden

tsign

al(in

respon

seto

noxiou

sstim

uli)

Yes(p<.05)

Statisticallysign

ificant

increase

ofactivationin

the

insularandsensorim

otor

cortices

post-SM

comparedto

control;

andin

theanterio

randpo

steriorcing

ulate,supp

lemen

tary

motor

area,and

precen

tralgyruspo

st-con

trol

comparedto

SM

Pain

intensity

assessed

butno

relatio

nship

tested

Immed

iatelyafter

Accep

table

Lelic

2016

[14]

“Sub

clinicalne

ck/

spinalpain”

N30

somatosen

sory

evoked

potentialp

eak

amplitu

des

Yes(significantpo

st-in

terven

tiondifferencebetween-grou

psrepo

rted

butwith

outinclusionof

thecorrespo

ndingp-value

andmen

tionof

thestatisticalthresholdforsign

ificance)

Statisticallysign

ificant

decrease

post-SM

(p=.02)

butno

statisticallysign

ificant

change

spo

st-con

trol

(p=.4)

Noclinicalou

tcom

einclud

edNot

repo

rted

Med

ium

Baarbé

2018

[15]

Cereb

ellarinhibitio

nYes(p<.001)

Statisticallysign

ificant

redu

cepo

st-SM

comparedto

control

Noclinicalou

tcom

einclud

edUnclear

(according

toFig.

1im

med

iatelyafterthemotor

acqu

isition

task,i.e.abo

ut20

min

afterinterven

tion)

Med

ium

Results

arerepo

rted

(i)grou

pedby

type

ofstud

ysubjects

(sym

ptom

aticor

with

“sub

clinical

neck/spina

lpain”),an

d(ii)consecutivelyby

year

ofpu

blication

SMSp

inal

man

ipulation


Table

10Results

from

sevenstud

iesinclud

edin

asystem

aticreview

ontheeffect


‘brain

functio

n’,com

parin


aninactive

control

1st

Autho

rYear

Ref

Type

ofstud

ysubjects

Outcome

Was

astatisticallysign

ificant

differencebe

tweengrou

psob

served

?Timeof

assessmen

tQuality

classification

Kelly

2000

[29]

Health

yReactio

n-tim

eto

amen

tal

rotatio

ntask

Yes(p<.05)

Statisticallysign

ificant

decrease

post-SM

comparedto

control

Unkno

wn

Med

ium

Dishm

an2002

[17]

MEP

amplitu

des

Yes(p<.05)

Statisticallysign

ificant

increase

from

20to

120s.po

st-SM

comparedto

control

-Immed

iatelyafter(each20sdu

ring120s

afterSM

orcontrol)

−5min

−10

min

Med

ium

Dishm

an2008

[18]

MEP

amplitu

des

Yes(p<.05)

Statisticallysign

ificant

increase

at10

s.po

st-SM

comparedto

control

Immed

iatelyafter(each10sdu

ring100s

afterSM

orcontrol)

Low

Fryer

2012

[16]

MEP

amplitu

des

Yes(p=.04)

Statisticallysign

ificant

decrease

post-SM

comparedto

control

Unkno

wn(app

roximately10

min

afteron

einterven

tionor

theothe

r)Med

ium

MEP

latencies

No

CSP

duratio

nsNo

Ogu

ra2011

[19]

Symptom

atic

(mechanicaln

eckpain

andshou

lder

stiffne

ss)

Region

alcerebral

metabolicrate

Yes(p<.001)

Statisticallysign

ificant

increase

post-SM

comparedto

controlintheinferio

rprefrontalcortex,anteriorcing

ulatecortex,and

middletempo

ralg

yrus;and

statisticallysign

ificant

decrease

post-SM

comparedto

controlinthecerebe

llar

verm

isandvisualassociationcortex

Betw

een35

to55

min

Low

Inam

i2017

[8]

Region

alcerebral

metabolicrate

Yes(p<.05)

Statisticallysign

ificant

increase

post-SM

comparedto

controlintheBroca’sarea,

anterio

rcing

ulatecortex,som

atosen

sory

associationcortex,W

ernike’sarea,visual

associationcortex,cereb

ellarverm

is,and

visualcortex;and

statisticallysign

ificant

decrease

post-SM

comparedto

controlintheinferio

rparietallob

ule,fro

ntalpo

le,

inferio

rfro

ntalgyrus,parstriang

ularis,p

remotor

area/sup

plem

entary

motor

area,

prim

arymotor

cortex,frontaleyefield,d

orsolateralp

refro

ntalcortex,ang

ular

gyrus,

fusiform

gyrus,inferio

rtempo

ralg

yrus,and

tempo

ralp

ole.

Betw

een35

to65

min

Low

Haavik

2010b

[21]

“Sub

clinicalne

ck/spinal

pain”

P14-N18

SEP

peak

ratio

No

Unclear

(saidto

bewith

in25

min

post-SM

orcontrol,po

ssibly45

min

afteron

einterven

tionor

theothe

r)

Med

ium

N20-P25

SEP

peak

ratio

No

P22-N30

SEP

peak

ratio

Yes(p=.00005)

Statisticallysign

ificant

decrease

post-SM

comparedto

control

Results

arerepo

rted

(i)grou

pedby

type

ofstud

ysubjects

(health

y,symptom

atic,o

rwith

“sub

clinical

neck/spina

lpain”),(ii)grou

pedby

type

ofou

tcom

es,and


year

ofpu

blication

CSPCortical

silent

perio

d,MEP

Motor

evok

edpo

tential,SEPSo

matosen

sory

evok

edpo

tential,SM

Spinal

man

ipulation


Table

11Results

from

four

stud

iesinclud

edin

asystem

aticreview

ontheeffect


‘brain

functio

n’,com

parin


anothe

rph

ysical

stim

ulus

1stAutho

rYear

Ref

Type

ofstud

ysubjects

Outcome

Was

astatisticallysign

ificant

differencebe

tweengrou

psob

served

?Timeof

assessmen

tQuality

classification

Dishm

an2008

[18]

Health

yMEP

amplitu

des

Yes(p<0.05)

Statisticallysign

ificant

increase

at10

s.po

st-SM

comparedto

control

Immed

iatelyafter(each10

s.du

ring100s.after

SMor

control)

Low

Haavik

2010a

[26]

“Sub

clinicalne

ck/

spinalpain”

P14-N18

SEPpe

akratio

No

Unclear

(saidto

bewith

in25

min

post-SM

orcontrol)

Med

ium

N20-P25

SEPpe

akratio

No

P22-N30

SEPpe

akratio

Yes(p=.003)

Statisticallysign

ificant

decrease

post-SM

comparedto

control

Haavik

2016

[24]

MEP

amplitu

des

Yes(p=.01)

Statisticallysign

ificant

increase

post-SM

comparedto

control

Not-rep

orted

Low

k(slope

ofthesteepe

stpartof

thecurve)

No

S 50(stim

ulus

intensity

toob

tain

arespon

se50%

ofthemaxim

um)

No

Christiansen

2018

[28]

V-waveam

plitu

des

Yes(p<0.01–0.03)

Statisticallysign

ificant

increase

ateach

timepo

intpo

st-SM

comparedto

control

-Immed

iatelyafter

−30

min

after

−60

min

after

Accep

table

Results

arerepo

rted

(i)grou

pedby

type

ofstud

ysubjects

(health

y,symptom

atic,o

rwith

“sub

clinical

neck/spina

lpain”),(ii)grou

pedby

type

ofou

tcom

es,and


year

ofpu

blication

MEP

Motor

evok

edpo

tential,SEPSo

matosen

sory

evok

edpo

tential,SM

Spinal

man

ipulation


measures, which did not allow us to compare their re-spective results (Tables 9, 10 and 11).

Summary of finding in relation to the research questions 1–3In summary, and in relation to our three first researchquestions, three studies reported a transient (immedi-ately to about 20 min post-intervention) effect on ‘brainfunction’ of varied types after SM vs. a sham comparatorin symptomatic subjects and in “subclinical neck/spinalpain” subjects. However, in these studies SM was com-pared to sham procedures with unclear credibility, ordiscovered as such by the study subjects. Also, the ex-perimental findings were untested in relation to clinicalbenefits. Detailed results are reported in the nextsection.

1 - Is there an effect of SM on ‘brain function’? (n = 3)Symptomatic subjects (n = 1)One study of ‘acceptable’ methodological quality [9],

conducted on subjects suffering from mechanical neckpain, reported an effect on activation in response to nox-ious stimuli, as measured by fMRI using the blood oxy-genation level-dependent (BOLD) signal, after SM vs. asham comparator. As shown previously, mechanicalnoxious stimulation resulted in increased activation inseveral brain areas associated with pain processing. Agroup comparison was reported, although it was unclearwhether it was performed on the pre- post-interventiondifferences, as it should. This indicated increased

activation in the SM group relative to the sham group inthe insular cortex, supramarginal gyrus and superiorparietal lobe (presumably in sensory association/integra-tion areas). On the other hand, there was increased acti-vation in the sham group relative to the SM group inthe cingulate cortex, the supplementary motor area, andthe middle temporal gyrus.“Subclinical neck/spinal pain” subject (n = 2)Two studies on “subclinical neck/spinal pain” sub-

jects [14, 15], both of ‘medium’ methodological qual-ity, reported an effect of SM vs. a sham comparator.One of these studies found a statistically significantdecrease of N30 somatosensory evoked potential peakamplitudes post-SM compared to a sham group [14].The other one reported a statistically significant de-crease of cerebellar inhibition following SM comparedto a sham intervention [15].

2 - If there is an effect, for how long does it last? (n = 3)Symptomatic subjects (n = 1)The effect reported by Sparks et al. 2017 [9] (in a

study of ‘acceptable’ methodological quality) on symp-tomatic subjects was immediate, with no effect investi-gated beyond this time point.“Subclinical neck/spinal pain” subject (n = 2)The effect reported by Baarbé et al. 2018 [15] (in a

study of ‘medium’ methodological quality) on “subclin-ical neck pain” subjects was measured after interventiononly once, at about 20 min post-intervention. Anotherpotentially relevant study (Lelic et al. 2016) [14], also of

Table 12 Summary of quality scores and quality classification for 18 articles included in a systematic review on the effect of spinalmanipulation on ‘brain function’

Type of study First author / Year [ref] Scorea (risk of bias and external validity) Quality classification

Sham studies Sparks, 2017 [9] 5.5/7 (79%) acceptable

Baarbéé, 2018 [15] 3.5/6 (58%) medium

Lelic, 2016 [14] 2.5/6 (42%) medium

Comparison studies Christiansen, 2018 [28] 5/6 (83%) acceptable

Gay, 2014 [22] 5/7 (71%) acceptable

Kelly, 2000 [29] 3.5/6 (58%) medium

Dishman, 2002 [17]Haavik-Taylor, 2010a [26]Haavik-Taylor, 2010b [21]Fryer, 2012 [16]Niazi, 2015 [27]

2.5/6 (42%) medium

Haavik-Taylor, 2007a [20]Haavik-Taylor, 2007b [25]Haavik-Taylor, 2008 [28]Dishman, 2008 [23]Ogura, 2011 [19]HaaviK, 2016 [24]Inami, 2017 [8]

2/6 (33%) low

aThe quality score for each study could range from 0 to 6 or 7, depending on their respective study design and the type of study subjects included. Each qualityscore was then converted on percentage to allow comparisons. Quality classification: ‘low’: 0–33%; ‘medium’: 34–67%; ‘acceptable’: 68%-100


‘medium’ methodological quality, did not report the timeof assessment after interventions.

3- If there is an effect, is it associated with any clinicalbenefits? (n = 3)Symptomatic subjects (n = 1)The study by Sparks et al. 2017 [9] on subjects suf-

fering from mechanical neck pain (a study consideredto be of ‘acceptable’ methodological quality), inaddition to assessing brain activation in response to anoxious stimulus by means of fMRI, assessed pain in-tensity pre- post-interventions. However, they did notinvestigate whether there was an association betweenpain intensity changes and cerebral activity changes,making it impossible to answer this third researchquestion. It is worth noting that the authors investi-gated whether there was a correlation between sub-jective ratings of the noxious stimulus intensity andchange in activation in the insular cortex, but nosuch relationship was found.“Subclinical neck/spinal pain” subjects (n = 2)None of the two studies on “subclinical neck/spinal

pain” subjects [14, 15], both of ‘medium’ methodologicalquality, included clinical outcomes.

Other types of controls, specifically ‘inactive control’(Table 10) or ‘another physical stimulus (Table 11),i.e the (‘differences in outcome’ studies (n = 10)

4 - Is there a difference in ‘brain function’ after SM vs. an‘inactive control’? (n = 7)Seven studies could be used for the fourth researchquestion [8, 16–19, 21, 29], four considered to be of‘medium’ methodological quality and three of ‘low’methodological quality (see Table 12). These studies,reporting on three different types of subjects, investi-gated the potential changes on ‘brain function’ in re-sponse to SM by using varied outcome measures and/orexperimental protocols, which makes comparisons be-tween studies difficult.

Summary of findings in relation to the fourth re-search questionIn these studies, of ‘low’ or ‘medium’ methodological

quality, there were, in general, statistically significant dif-ferences in outcome between SM and the controls butnot necessarily in the same direction. When brain areaswere compared, differences were found, but again withsome conflicting results. Detailed results are reported inthe next section.

Healthy subjects (n = 4)In healthy subjects, an immediate and transient in-

crease of motor-evoked potential amplitudes after

lumbar SM was reported in two studies [17, 18],whereas one reported a decrease of motor-evokedpotential amplitude after lumbar SM (approximately10 min after intervention) [16]. The third reported nostatistically significant findings for motor-evokedlatencies and cortical silent period durations [16].Two of these studies were considered of ‘medium’methodological quality [16, 17] and one of ‘low’methodological quality [18].A fourth study [29], of ‘medium’ quality, reported a

statistically significant greater decrease of reaction-timeto a mental reaction task post-SM vs. post-resting with-out reporting the time of reassessment.

Symptomatic subjects (n = 2)The two studies conducted on symptomatic subjects

were from the same research team and both of ‘low’methodological quality [8, 19]; the first that waspublished being considered by its authors as a “proofof concept” study [19], which apparently led to theirsecond experiment [8]. Both reported a statisticallysignificant increase of regional cerebral metabolic rate(glucose uptake) in some brain areas and a statisti-cally significant decrease of glucose uptake in otherbrain areas, sometimes with conflicting results (seeTable 10).

“Subclinical neck/spinal pain” subjects (n = 1)In a study on “subclinical neck pain subjects” of

‘medium’ methodological quality [21], the authorsreported a statistically significant decrease of the P22-N30 somatosensory potential peak ratio post-SM buta statistically significant increase of this ratio post-control intervention. They found no statisticallysignificant between-group differences for the othersomatosensory evoked potential peak ratios theyinvestigated.

5 - Is there a difference in ‘brain function’ after SM vs.‘another physical stimulus’? (n = 4)Four articles were able to be used in relation to ourfifth research question [18, 24, 26, 28]. Two were of‘low’, one of ‘medium’, and one of ‘acceptable’methodological quality (see Table 12). Again, as theywere conducted on different types of study subjectsand/or most often used different outcomes measures,the possibility of making comparisons between studieswas limited.

Summary of findings in relation to the fifth researchquestionSome statistically significant differences in outcome

between SM and the controls were reported but resultswere mixed, in studies of ‘low’ to ‘acceptable’


methodological quality. Detailed results are reported in thenext section.

Healthy subjects (n = 1)One study of ‘low’ methodological quality [18], con-

ducted on healthy chiropractic students, reported at 10 spost-intervention statistically significant greater motor-evoked potential amplitudes in the SM group vs. a pre-loading control group.

“Subclinical neck/spinal pain” subjects (n = 3)One study on “subclinical neck/spinal pain” subjects

[26], of ‘medium’ methodological quality, found a statis-tically significant decrease of the P22-N30 somatosen-sory potential peak ratio post-SM vs. post-control [26].There were no statistically significant between-group dif-ferences for the other somatosensory evoked potentialpeak ratios investigated.Haavik et al. 2016 [24], in a study of ‘low’ methodo-

logical quality, reported a statistically significant increasein motor-evoked potential amplitudes in the SM groupcompared to the control group. They did not find anystatistically significant between-group differences for twoother variables they studied.Christiansen et al. 2018 [28], in a study of ‘acceptable’

methodological quality, conducted on elite taekwondoathletes with “subclinical spinal pain”, found a statisti-cally significant greater V-wave amplitude post-SM vs.post-control at each time point of assessment (immedi-ately, 30, and 60 min after).

DiscussionSummary of findings and their interpretationThis systematic review consists of 18 relevant articles.Once classical risk of bias aspects, necessary in this typeof experimental design, had been taken into account,most of these articles were considered of ‘low’ or‘medium’ methodological quality. In addition, their stat-istical methods and results sections were often difficultto interpret because of unclear and/or unusual descrip-tions. For methodological reasons, the results of only 13of these 18 articles were considered for final analysis.These 13 articles reported on (i) whether SM has an

effect on ‘brain function’ compared to a sham interven-tion, and (ii) whether SM alters ‘brain function’ in a dif-ferent way compared to an ‘inactive control’ or ‘anotherphysical stimulus’, and this on different type of studysubjects. Based on the studies using a sham interventionas comparator to SM, it seems that SM does have an ef-fect on ‘brain function’. As a result, we also studied ourtwo other main objectives, namely (i) how long this effectwould last, and (ii) whether this effect was associatedwith clinical benefits.

The three studies using a sham intervention as com-parator, two of ‘medium’ and one of ‘acceptable’ methodo-logical quality, provided some evidence to support thehypothesis that SM has supra-segmental neurophysio-logical effects. It was thus shown that SM seems to havethe potential to transiently alter (i) somatosensory integra-tion of afferent inputs from the upper limb [14], (ii) cere-bellar inhibition [15], both on “subclinical neck/spinalpain” subjects and (iii) activation of several brain areas as-sociated with pain processing on acute or subacute mech-anical neck pain subjects [9]. However, none reportedwhether such effects were lasting and none investigatedwhether these were associated with any clinical benefits.Also, according to the authors of these three studies, whatthese effects mean for the brain remains to be established.The 10 studies not using a sham intervention as

comparator, most of which were of ‘low’ or ‘medium’methodological quality, also reported statistically sig-nificant between-group differences but not necessarilyin the same direction and also not systematically foreach of the outcomes they studied. Most of thesestudies reported on healthy or on “subclinical neck/spinal pain” subjects.In summary, based on both the ‘effect’ studies and

the ‘differences in outcome’ studies, it seems thatsomething does indeed happen within the brain in re-sponse to SM. However, what this means remainselusive both in the brain and at a clinical level, andthe researchers provide only hypotheses rather thaninterpretations. In addition, the reported findings haveto be interpreted with caution given the general levelof methodological quality (‘low’ or ‘medium’) of theincluded studies.Showing that SM is reflected by brain activity does not

necessarily mean that something ‘positive’ and clinicallyrelevant happens in response to SM. The significance ofany putative effect on ‘brain function’ must thus be putinto perspective by comparing it to effects in response toother types of (comparable) physical stimuli or othertypes of treatment. The question is therefore: Are thefindings in relation to ‘brain function’ specific to SM?For various methodological reasons, none of the stud-ies could clearly answer this question. Furthermore,in order to claim brain involvement in the effects ofSM it should be expected that changes in ‘brain func-tion’ following SM can be shown related to the de-sired clinical effects of SM. However, no informationrelated to any clinical significance of such findingswas unearthed in this review.

Methodological considerations of our own reviewThree databases were searched and only one authorapplied eligibility criteria to the titles. Thus, it is pos-sible that not all relevant articles on the topic were


found. Nevertheless, the additional search of referencelists produced only three additional titles. All the fol-lowing steps of the screening process and of the dataextraction were made independently by at least twoof the reviewers.Most of the articles we obtained reported on experi-

mental studies, in which no clinical outcomes were in-cluded. Studies of this type are not strictly comparableto ordinary clinical studies using the randomized clinicaltrial design, for which well-established critical appraisaltools exist. The quality checklist used in the present re-view was therefore not standard. For example, someusual risk of bias items, such as allocation concealmentwere not assessed, as they were judged less relevant fornon-clinical randomized controlled trials. Regarding al-location concealment, we assumed that it would be diffi-cult to predict which study subjects would react howregarding the outcome variables used in the includedstudies. Nevertheless, most of the items we selected con-sisted of accepted items to evaluate risk of bias [31, 32].These items related to selection, performance, detection,attrition, and analysis risk of bias. Additional methodo-logical concerns specific to the different studies, voicedby the experts, were summarized in a separate columnof the quality checklists for the readers who would be in-terested in more information. These comments can beused as a basis for discussion on how to proceed withfuture studies of this type.Because several of the statistical analyses and/or report-

ing were unclear and/or unusual, we finally resorted to a‘benefit-of-the-doubt’ approach. Thus, after many discus-sions and attempts at interpreting some confusing reports,we deviated from our previous criterion, which was toinclude in the data synthesis only results of the studiesthat reported clearly differences between-groups or be-tween-types of interventions. However, such exceptionswere noted explicitly in the Results section, either in thetext [9] or in Table 9 [14].Studies were included also when the outcome variables

they tested were not necessarily a reflection of ‘brain func-tion’ only, i.e. some would depend on both segmental andsupra-segmental changes (e.g. motor evoked potential am-plitudes, V-waves, and cortical silent period duration) [12,34]. This means that results obtained via these outcomevariables must be interpreted with caution; a fact that isoften admitted by the authors of the reports. On the otherhand, being unrestrictive allowed us to cover the literatureon the topic more exhaustively.Many different outcome variables are used in research

to measure brain activity, and this was also the case forthe articles we included. Their heterogeneity in relation to(i) study subjects (healthy, symptomatic, and “subclinicalneck/spinal pain”), (ii) outcome variables (16 different out-come variables for 18 articles), (iii) experimental protocols

for each single variable, and (iv) generally rather lowmethodological quality, makes comparison of studies diffi-cult and one or several meta-analyses impossible.

Methodological considerations of the included studiesThe methodological quality was quite low in relation towell accepted risk of bias items. Admittedly, these types ofstudies require a lot of knowledge on technical aspects butthis must not remove focus from the fundamental meth-odological requirements of research, namely the necessityto collect and interpret data in an objective manner.For example, the studies considered to be of ‘low’ and

‘medium’ methodological quality often failed to reportusing either a credible sham comparator or having beenconducted on naïve subjects. As suggested by Fryer andPearce [16], the blinding or naivety of the study subjectswhen ‘objective’ outcomes are used could potentially beconsidered not as important in purely experimentalstudies. ‘Objective’ here means that study subjects can-not usually willfully or inadvertently influence outcome.However, the placebo effect implies complex neurophysi-ologic responses involving the brain [35]. In our opinion,this makes the use of a sham comparator and the evalu-ation of its success relevant also for the ‘objective’ out-come measurements used in the included studies.According to this review, the credibility of the sham

comparator used in the three ‘effect’ studies must beconsidered uncertain for two and was recognized as suchby the participants in the third. Thus it cannot be ruledout that the effect of SM on ‘brain function’ was the re-sult of contextual factors, rather than truly caused by theSM, as discussed by Rossettini et al. 2018 [36]. This wasacknowledged in one of these reports [9], where the au-thors noted that changes in cerebral activation in re-sponse to noxious stimuli post-SM may reflect subjects’expectations.In relation to the ‘difference in outcome’ studies, the

origin of the study subjects was reported in only a fewcases. Thus we do not know if they had any precon-ceived ideas/expectations with respect to the study out-comes. This problem could be compounded if severalstudies were conducted on the same study subjects.Another example was that the blinding of the assessor

and of the person who analyzed the data was generallypoorly reported. Although it is fair to recognize that thisreporting may be unusual in some fields of research (e.g.neuroimaging studies), some authors were transparentin relation to this point, which should encourage otherresearchers also to do the same.In addition, comments were provided by our experts,

suggesting that several of the experimental protocols ofthese reports lacked some of the standards, specific tosuch studies. Several comments were also provided fromthe experts on the statistical analyses, indicating that this


was an area of concern, as the statistical analysis is atthe heart of the validity of any statistically significantfindings.Neuroimaging studies, which produce ‘visual’ answers,

are perhaps easier to interpret for people without spe-cific knowledge in neurological testing. Nevertheless,they present a challenge for formal analysis. For example,quantification of data is difficult. There are many analytictechniques available for these types of studies and there isa lack of consensus with respect to the most appropriatestatistical thresholds to be used [37]. Therefore, this typeof study needs to be replicated by other independent re-search teams. Obviously, this is required for any type ofresearch, particularly when one specific research teamdominates the area or when there are potential or realconflicts of interest.

Conceptual concernsIn relation to all the studiesThe rationale for investigating whether SM acts throughmodulation of ‘brain function’ was generally not clear in theincluded studies. Nevertheless, most studies proposed thatchanges observed at the brain level would result (at least par-tially) from a ‘bottom-up’mechanism, due to altered afferentinputs in response to SM [8, 9, 14, 15, 17, 18, 20, 21, 23–28].However, in addition to ‘bottom-up’ effects, SM mightchange brain activity through ‘top-down’ effects, i.e. throughcontextual factors. This means that in absence of trulyblinded subjects one cannot exclude a ‘top-down’ effect. Inaddition, a blinded assessor would be required. These aretwo methodological aspects often lacking in the reports wescrutinized, and therefore ‘top-down’ effects cannot be ruledout to explain some findings.

In relation to the studies using “subclinical neck/spinalpain” subjectsArticles, in which Haavik was one of the authors, included“subclinical neck/spinal pain” subjects in their studies [14,15, 20, 21, 23–28]. However, the definitions of “subclinicalneck/spinal pain” were not consistent in the various studies,so this concept remains unclear. In fact, it is uncertainwhether these “subclinical neck/spinal pain” study subjectsare clearly different from ‘ordinary’ healthy subjects interms of neurophysiological parameters, such as somato-sensory evoked potentials and motor evoked potentials [15,38]. Most authors of these articles proposed that the effectsor changes they measured in the SM groups reflect im-provement of ‘brain function’ [14, 15, 20, 21, 23–28]. This,obviously, raises the following question: If these subjectsare not different from healthy subjects, what, exactly, wouldbe improved?Additionally, these studies rest on the assumption

that it is possible to detect ‘subluxations’ in people

with “subclinical neck/spinal pain”; a concept that re-mains hypothetical. Overall, it is our considered opin-ion that some clarifications are needed regarding this“subclinical neck/spinal pain” with ‘subluxations’ con-cept to ensure appropriate interpretation of the re-sults of these studies.

Gap between scientific level of evidence and itsimplementation in clinical practiceFN practitioners use SM as a treatment of ‘brain lesions’[5] despite a lack of evidence of its clinical effect, asunearthed in this review. One example of how this con-cept is taught within FN is the seminar in which PFreud, a chiropractor, proposes to show how to ‘adjustthe brain’ [39]. Furthermore, this is stated to be basedon the latest scientific knowledge on the topic (as shownin Additional file 4).

Gap between scientific level of evidence and itsmodification for popularizationBased on this systematic review, we conclude that thereis presently no evidence indicating that SM has a benefi-cial effect on ‘brain function’ or that the diverse findingsidentified in this review would be in any way indicatorsof a health benefit in general. It is therefore prematurefor authors of this type of research to promote clinicalbenefits. Nevertheless, this can be observed in at leastsome commercial videos [40] and in an ‘information’book for patients subtitled “A quest to understandChiropractic from the inside out” [41]. This informationprovided to laymen regarding published research on thetopic claims clinical effects on ‘brain function’ followingSM, contrary to the more careful interpretations in re-search articles produced by the same researcher [14, 15,20, 21, 23–28].In an example of this overreach, Haavik wrote in her

book: “Having your spine checked regularly, to ensure yourbrain is accurately aware of what is going on in and aroundyour body, should be just as common as exercising everyday and brushing your teeth. Everyone should have accessto chiropractic care right from birth through to the daythey pass away. I believe a lot of suffering could be pre-vented if this was the case.” [41].Another example is extracted from one of Haavik’s

commercial videos [42], which is mainly based on one ofher own studies included in the present review [14]. Thisvideo starts with the following message: “Chiropracticcare really does change brain function!”. After havinggiven a lay interpretation of this study [14], it concludes:“Have you seen your chiropractor lately? You may wantto have your brain’s conductor fine-tuned too”. It wouldbe easy to interpret this as a suggestion that the brain isunable to do its job properly but that a chiropractor can


improve the situation. However, the section “Study Con-siderations” of that article [14] obviously does not sup-port such claims and the authors of that article pointout that it is not clear how long the changes observed inthe brain last. They also state that it is not known if theobserved changes are, at all, beneficial. There is thereforea gap between the guarded discussion in the peer-reviewed study and the more positive message of thecommercial material [42].

RecommendationsRecommendations for future researchFurther research should be undertaken in this area andwe recommend attention to the following:

– The clinical relevance of any brain changes shouldbe investigated using symptomatic study subjects.Also, clinical outcomes should be included and thecorrelation between these and brain changes shouldbe tested to establish if there is some type ofbenefits.

– To study the specific effect of SM, proper shamprocedures must be adopted and checked forsuccess after the intervention to control for anyunspecific effects, including placebo responses.

– Appropriate methodology in relation to randomizedcontrolled trials, with appropriate attention to thepotential sources of bias (e.g. blinding of studysubjects, assessor, and statistician) should berespected.

– In relation to the technical procedures, standardprotocols should be employed to ensurereproducibility of the outcome measures.

– Appropriate statistical methods and thresholdsshould be used.

– Any conflict of interest should be reported.– Results should be replicated by independent

research teams before their clinical acceptance.

Recommendations for the chiropractic professionPresently, the chiropractic profession might wish to con-sider the potential consequences of encouraging under-graduate- and postgraduate courses on chiropracticapproaches relating to the treatment of the brain via thespine. Obviously, this is also true for other health careproviders who may be tempted to practice followingsuch concepts in relation to SM. Moreover, statementsaimed at the public relating to this phenomenon shouldbe discussed as findings of unknown clinical effect.

ConclusionAccording to the results of the present systematic review,it is premature to promote the use of SM as a treatmentto improve ‘brain function’ from a clinical point of view.

Further, studies on this topic should (i) include symp-tomatic subjects, (ii) relate to the clinical significance byusing outcomes relevant to test associations with neuro-physiological changes, and (iii) take into considerationbasic methodological and technical requirements per-taining to this type of randomized controlled trials.

Additional files

Additional file 1: Search strategy developed for PubMed for asystematic critical review of the literature on the effect of spinalmanipulation on ‘brain function’ (DOCX 126 kb)

Additional file 2: Items selected for the quality checklists and theirrationale in relation to a systematic critical review on the effect of spinalmanipulation on ‘brain function’ (DOCX 22 kb)

Additional file 3: The various ways used in the scientific literature fortesting the hypothesis that spinal manipulation would have an effect on‘brain function’ (DOCX 83 kb)

Additional file 4: Commercial announcement of a chiropractic seminarentitled “Adjusting the Brain” (PDF 1462 kb)

AbbreviationsBOLD: Blood oxygenation level-dependent; fMRI: Functional magneticresonance imaging; FN: Functional Neurology; SM: Spinal manipulation

AcknowledgmentsThe authors would like to acknowledge Catherine Tellaa, librarian at theBibliothèque interuniversitaire de Santé Paris Descartes, for the assistance givenfor the search of literature. We are also grateful to Søren O’Neill DC, PhD, foreditorial assistance.

DeclarationsThe present systematic critical review was registered in PROSPERO, with thereference CRD42017074966.

Authors’ contributionsALM and CLY planned the review. All the authors contributed to thereviewing of the included articles. MAA, MS and PS provided comments onmethodological and technical aspects of articles depending on theirrespective areas of expertise (in Tables 4, 5, 6 and 7, col.10, each numberrefers to one of these co-authors, 1 for MAA, 2 for MS, and 3 for PS). ALMand CLY, with the assistance of MAA, MS and PS, interpreted the findings.ALM wrote the first draft. CLY, MAA, MS and PS provided comments for sub-sequent drafts. All the authors revised and approved the final manuscript.

Authors’ informationALM is a chiropractor and this work was done as part of her PhD program atthe University of Paris-Sud. MAA is a Professor in Sport sciences and physicalactivity at Université Paris-Sud where he is the Head of the CIAMS laboratory.He has a background in cognitive neuroscience and motor control. MS is aMD and PD (University lecturer) with a background in clinical and experi-mental neurophysiology and neurorehabilitation, specifically of spinal corddisorders. PS is an MD and a PhD with a background in neuroimaging(mostly functional and structural MRI, to a smaller extent, positron emissiontomography). CLY is a chiropractor and a Professor in Clinical Biomechanicsat the University of Southern Denmark. She has a background in epidemi-ology and systematic critical reviews and was the main supervisor on thisPhD project.

FundingApart from the authors being funded from their institutions, there were noexternal grants for this project.

Availability of data and materialsThe scientific articles scrutinized during the current critical review areavailable from the corresponding author on reasonable request.


https://doi.org/10.1186/s12998-019-0265-8

https://doi.org/10.1186/s12998-019-0265-8

https://doi.org/10.1186/s12998-019-0265-8

https://doi.org/10.1186/s12998-019-0265-8

Ethics approval and consent to participateNot applicable.

Consent for publicationNot applicable.

Competing interestsAuthors declare there are no conflicts of interest. CLY is a senior editorialadviser to and PS is a member of the editorial board of the journalChiropractic & Manual Therapies but played no part in the peer review of thesubmission.

Author details1CIAMS, Univ. Paris-Sud, Université Paris-Saclay, 91405 Orsay Cedex, France.2CIAMS, Université d’Orléans, 45067 Orléans, France. 3Institut FrancoEuropéen de Chiropraxie, 24 Bd Paul Vaillant Couturier, 94200 Ivry sur Seine,France. 4Spinal Cord Injury Center, University Hospital Balgrist, Zürich,Switzerland. 5Integrative Spinal Research Group, Department of ChiropracticMedicine, University Hospital Balgrist and University of Zürich, Zürich,Switzerland. 6Institute for Regional Health Research, University of SouthernDenmark, Odense, Denmark.

Received: 23 February 2019 Accepted: 2 July 2019

References1. Adams J, Lauche R, Peng W, Steel A, Moore C, Amorin-Woods LG, Sibbritt D.

A workforce survey of Australian chiropractic: the profile and practicefeatures of a nationally representative sample of 2,005 chiropractors. BMCComplement Altern Med. 2017;17(1):14.

2. Beliveau PJH, Wong JJ, Sutton DA, Simon NB, Bussieres AE, Mior SA, FrenchSD. The chiropractic profession: a scoping review of utilization rates, reasonsfor seeking care, patient profiles, and care provided. Chiropr Man Therap.2017;25:35.

3. Clar C, Tsertsvadze A, Court R, Hundt GL, Clarke A, Sutcliffe P. Clinicaleffectiveness of manual therapy for the management of musculoskeletaland non-musculoskeletal conditions: systematic review and update of UKevidence report. Chiropr Man Therap. 2014;22(1):12.

4. Beck RW. Dedication. In: Functional Neurology for Practitioners of ManualMedicine edn: Churchill Livingston Elsevier; 2011.

5. Meyer AL, Meyer A, Etherington S, Leboeuf-Yde C. Unravelling functionalneurology: a scoping review of theories and clinical applications in acontext of chiropractic manual therapy. Chiropr Man Therap. 2017;25:19.

6. Haavik H, Murphy B. The role of spinal manipulation in addressingdisordered sensorimotor integration and altered motor control. JElectromyogr Kinesiol. 2012;22(5):768–76.

7. Beck RW. Approaches to treatment. In: Functional neurology for practitionersof manual medicine. edn: Churchill Livingstone: Elsevier; 2011. p. 343–79.

8. Inami A, Ogura T, Watanuki S, Masud MM, Shibuya K, Miyake M, Matsuda R,Hiraoka K, Itoh M, Fuhr AW, et al. Glucose metabolic changes in the brainand muscles of patients with nonspecific neck pain treated by spinalmanipulation therapy: a [(18)F]FDG PET study. Evid Based ComplementAlternat Med. 2017;2017:4345703.

9. Sparks CL, Liu WC, Cleland JA, Kelly JP, Dyer SJ, Szetela KM, Elliott JM.Functional magnetic resonance imaging of cerebral hemodynamicresponses to pain following thoracic thrust manipulation in individuals withneck pain: a randomized trial. J Manip Physiol Ther. 2017;40(9):625–34.

10. Vernon H. Historical overview and update on subluxation theories. J ChiroprHumanit. 2010;17(1):22–32.

11. Haavik-Taylor H, Holt K, Murphy B. Exploring the neuromodulatory effects of thevertebral subluxation and chiropractic care. Chiropr J Aust. 2010;40(1):37–44.

12. McNeil CJ, Butler JE, Taylor JL, Gandevia SC. Testing the excitability ofhuman motoneurons. Front Hum Neurosci. 2013;7:152.

13. Meyer AL, Leboeuf-Yde C. Unravelling functional neurology: a critical reviewof clinical research articles on the effect or benefit of the functionalneurology approach. Chiropr Man Therap. 2018;26:30.

14. Lelic D, Niazi IK, Holt K, Jochumsen M, Dremstrup K, Yielder P, Murphy B,Drewes AM, Haavik H. Manipulation of dysfunctional spinal joints affectssensorimotor integration in the prefrontal cortex: a brain source localizationstudy. Neural Plast. 2016;2016:3704964.

15. Baarbe JK, Yielder P, Haavik H, Holmes MWR, Murphy BA. Subclinical recurrentneck pain and its treatment impacts motor training-induced plasticity of thecerebellum and motor cortex. PLoS One. 2018;13(2):e0193413.

16. Fryer G, Pearce AJ. The effect of lumbosacral manipulation on corticospinaland spinal reflex excitability on asymptomatic participants. J Manip PhysiolTher. 2012;35(2):86–93.

17. Dishman JD, Ball KA, Burke J. First prize: central motor excitability changesafter spinal manipulation: a transcranial magnetic stimulation study. J ManipPhysiol Ther. 2002;25(1):1–9.

18. Dishman JD, Greco DS, Burke JR. Motor-evoked potentials recorded fromlumbar erector spinae muscles: a study of corticospinal excitability changesassociated with spinal manipulation. J Manip Physiol Ther. 2008;31(4):258–70.

19. Ogura T, Tashiro M, Masud M, Watanuki S, Shibuya K, Yamaguchi K, Itoh M,Fukuda H, Yanai K. Cerebral metabolic changes in men after chiropracticspinal manipulation for neck pain. Altern Ther Health Med. 2011;17(6):12–7.

20. Haavik-Taylor H, Murphy B. Transient modulation of intracortical inhibitionfollowing spinal manipulation. Chiropr J Aust. 2007;37:106–16.

21. Haavik Taylor H, Murphy B. The effects of spinal manipulation on centralintegration of dual somatosensory input observed after motor training: acrossover study. J Manip Physiol Ther. 2010;33(4):261–72.

22. Gay CW, Robinson ME, George SZ, Perlstein WM, Bishop MD. Immediatechanges after manual therapy in resting-state functional connectivity asmeasured by functional magnetic resonance imaging in participants withinduced low back pain. J Manip Physiol Ther. 2014;37(9):614–27.

23. Taylor HH, Murphy B. Altered sensorimotor integration with cervical spinemanipulation. J Manip Physiol Ther. 2008;31(2):115–26.

24. Haavik H, Niazi IK, Jochumsen M, Sherwin D, Flavel S, Turker KS. Impact ofspinal manipulation on cortical drive to upper and lower limb muscles.Brain Sci. 2016;7(1):E2.

25. Haavik-Taylor H, Murphy B. Cervical spine manipulation alters sensorimotorintegration: a somatosensory evoked potential study. Clin Neurophysiol.2007;118(2):391–402.

26. Taylor HH, Murphy B. Altered central integration of dual somatosensory inputafter cervical spine manipulation. J Manip Physiol Ther. 2010;33(3):178–88.

27. Niazi IK, Turker KS, Flavel S, Kinget M, Duehr J, Haavik H. Changes in H-reflexand V-waves following spinal manipulation. Exp Brain Res. 2015;233(4):1165–73.

28. Christiansen TL, Niazi IK, Holt K, Nedergaard RW, Duehr J, Allen K, Marshall P,Turker KS, Hartvigsen J, Haavik H. The effects of a single session of spinalmanipulation on strength and cortical drive in athletes. Eur J Appl Physiol.2018;118(4):737–49.

29. Kelly DD, Murphy BA, Backhouse DP. Use of a mental rotation reaction-timeparadigm to measure the effects of upper cervical adjustments on corticalprocessing: a pilot study. J Manip Physiol Ther. 2000;23(4):246–51.

30. Moher D, Hopewell S, Schulz KF, Montori V, Gotzsche PC, Devereaux PJ, ElbourneD, Egger M, Altman DG. CONSORT 2010 explanation and elaboration: updatedguidelines for reporting parallel group randomised trials. BMJ. 2010;340:c869.

31. Cochrane Handbook for Systematic Reviews of Interventions http://handbook-5-1.cochrane.org/. Accessed 18 Feb 2019.

32. PEDRo scale [https://www.pedro.org.au/wp-content/uploads/PEDro_scale.pdf]. Accessed 18 Feb 2019.

33. Bland JM, Altman DG. Comparisons against baseline within randomisedgroups are often used and can be highly misleading. Trials. 2011;12:264.

34. Rossini PM, Burke D, Chen R, Cohen LG, Daskalakis Z, Di Iorio R, Di LazzaroV, Ferreri F, Fitzgerald PB, George MS, et al. Non-invasive electrical andmagnetic stimulation of the brain, spinal cord, roots and peripheral nerves:basic principles and procedures for routine clinical and research application.An updated report from an I.F.C.N. committee. Clin Neurophysiol. 2015;126(6):1071–107.

35. Wager TD, Atlas LY. The neuroscience of placebo effects: connectingcontext, learning and health. Nat Rev Neurosci. 2015;16(7):403–18.

36. Rossettini G, Carlino E, Testa M. Clinical relevance of contextual factors astriggers of placebo and nocebo effects in musculoskeletal pain. BMCMusculoskelet Disord. 2018;19(1):27.

37. Eklund A, Nichols TE, Knutsson H. Cluster failure: why fMRI inferences forspatial extent have inflated false-positive rates. Proc Natl Acad Sci U S A.2016;113(28):7900–5.

38. Andrew D, Yielder P, Haavik H, Murphy B. The effects of subclinical neckpain on sensorimotor integration following a complex motor pursuit task.Exp Brain Res. 2018;236(1):1–11.

39. Adjusting the brain. Accueil [https://www.adjustingthebrainfrancais.com/].Accessed 30 Jun 2018.


http://handbook-5-1.cochrane.org/

http://handbook-5-1.cochrane.org/

https://www.pedro.org.au/wp-content/uploads/PEDro_scale.pdf

https://www.pedro.org.au/wp-content/uploads/PEDro_scale.pdf

https://www.adjustingthebrainfrancais.com/

40. Heidi Haavik. Shop: Videos [https://www.heidihaavik.com/collections/videos]. Accessed 18 Feb 2019.

41. Haavik H. The reality check. A quest to understand chiropractic from theinside out. : Haavik research - heidihaavik.com; 2014.

42. Chiropractic Research: Prefrontal Cortex [https://www.youtube.com/watch?v=9c09gXL0pSM]. Accessed 18 Feb 2019.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.


https://www.heidihaavik.com/collections/videos

https://www.heidihaavik.com/collections/videos

https://www.youtube.com/watch?v=9c09gXL0pSM

https://www.youtube.com/watch?v=9c09gXL0pSM

Chiropractic & Manual Therapies - Unravelling functional neurology… · 2019. 10. 10. ·...

Documents

Transcript of Chiropractic & Manual Therapies - Unravelling functional neurology… · 2019. 10. 10. ·...