Physiotherapy 112 (2021) 121–134

Systematic review

The effect of spinal manipulative therapy on pain relief and

function in patients with chronic low back pain: an individual
participant data meta-analysis
Annemarie de Zoete a,∗ , Sidney M. Rubinstein a , Michiel R. de Boer g ,
Raymond Ostelo a , Martin Underwood b,c , Jill A. Hayden d , Laurien M. Buffart e ,
Maurits W. van Tulder a,f , on behalf of the International IPD-SMT group:,
G. Bronfort, N.E. Foster, C.G. Maher, J. Hartvigsen, P. Balthazard, F. Cecchi,
M.L. Ferreira, M.R. Gudavalli, M. Haas, B. Hidalgo, M.A. Hondras, C.Y. Hsieh,
K. Learman, P.W. McCarthy, T. Petersen, E. Rasmussen-Barr, E. Skillgate,
Y. Verma, . Vismara, B.F. Walker, T. Xia, N. Zaproudina
a Department of Health Sciences, Faculty of Science and Amsterdam Movement Science Research Institute, Vrije Universiteit,
Amsterdam, The Netherlands
b Warwick Clinical Trials Unit, Warwick Medical School, The University of Warwick, Coventry CV4 7AL, UK
c University Hospitals of Coventry and Warwickshire, Coventry CV2 2DX, UK
d Department of Community Health & Epidemiology, Dalhousie University, Halifax, Nova Scotia B3H 1V7, Canada
e Radboud UMC, Nijmegen, the Netherlands
f Department Physiotherapy & Occupational Therapy, Aarhus University Hospital, Aarhus, Denmark
g Department of General Practice and Elderly Care Medicine, UMCG, the Netherlands

Abstract
Background A 2019 review concluded that spinal manipulative therapy (SMT) results in similar benefit compared to other interventions for
chronic low back pain (LBP). Compared to traditional aggregate analyses individual participant data (IPD) meta-analyses allows for a more
precise estimate of the treatment effect.
Purpose To assess the effect of SMT on pain and function for chronic LBP in a IPD meta-analysis.
Data sources Electronic databases from 2000 until April 2016, and reference lists of eligible trials and related reviews.
Study selection Randomized controlled trials (RCT) examining the effect of SMT in adults with chronic LBP compared to any comparator.
Data extraction and data synthesis We contacted authors from eligible trials. Two review authors independently conducted the study
selection and risk of bias. We used GRADE to assess the quality of the evidence. A one-stage mixed model analysis was conducted. Negative
point estimates of the mean difference (MD) or standardized mean difference (SMD) favors SMT.

Abbreviations: IPD, individual participant data; RCT, randomized clinical trial; LBP, low back pain; SMT, spinal manipulative therapy; PRISMA-P,
Preferred Reporting Items of Systematic Reviews and Meta-Analyses Protocol; MD, mean difference; SMD, standardized mean difference; SD, standard
deviation; RR, relative risk; RMDQ, Roland Morris Disability Questionnaire.
∗ Corresponding author at: Department Health Sciences, Faculty of Science, Vrije Universiteit, De Boelelaan 1085, Room WN U-454, 1081 HV Amsterdam,

The Netherlands.
E-mail addresses: [email protected] (A. de Zoete), [email protected] (S.M. Rubinstein), [email protected] (M.R. de Boer), [email protected] (R. Ostelo),
[email protected] (M. Underwood), [email protected] (J.A. Hayden), [email protected] (L.M. Buffart), [email protected]
(M.W. van Tulder).

https://doi.org/10.1016/j.physio.2021.03.006
0031-9406/© 2021 The Author(s). Published by Elsevier Ltd on behalf of Chartered Society of Physiotherapy. This is an open access article under the CC BY
license (http://creativecommons.org/licenses/by/4.0/).
122 A. de Zoete et al. / Physiotherapy 112 (2021) 121–134

Results Of the 42 RCTs fulfilling the inclusion criteria, we obtained IPD from 21 (n = 4223). Most trials (s = 12, n = 2249) compared SMT to
recommended interventions. There is moderate quality evidence that SMT vs recommended interventions resulted in similar outcomes on pain
(MD −3.0, 95%CI: −6.9 to 0.9, 10 trials, 1922 participants) and functional status at one month (SMD: −0.2, 95% CI −0.4 to 0.0, 10 trials, 1939
participants). Effects at other follow-up measurements were similar. Results for other comparisons (SMT vs non-recommended interventions;
SMT as adjuvant therapy; mobilization vs manipulation) showed similar findings. SMT vs sham SMT analysis was not performed, because
we only had data from one study. Sensitivity analyses confirmed these findings.
Limitations Only 50% of the eligible trials were included.
Conclusions Sufficient evidence suggest that SMT provides similar outcomes to recommended interventions, for pain relief and improvement
of functional status. SMT would appear to be a good option for the treatment of chronic LBP.
Systematic Review Registration Number PROSPERO CRD42015025714
© 2021 The Author(s). Published by Elsevier Ltd on behalf of Chartered Society of Physiotherapy. This is an open access article under the
CC BY license (http://creativecommons.org/licenses/by/4.0/).

Keywords: Spinal manipulative therapy; Individual participant data; Low back pain

Introduction of life, recovery, return-to-work, medication use and treat-

ment satisfaction) at one, three, six and twelve months in
Low back pain (LBP) is the leading cause of pain and adults with chronic LBP.
disability worldwide, and has a major socioeconomic impact
[1]. Non-pharmacological approaches are the first choice of
treatment as the risk of adverse events is lower than with Methods
pharmacological approaches [2]. One non-pharmacological
approach includes spinal manipulation or mobilization, col- This study was conducted according to the Preferred
lectively known as spinal manipulative therapy (SMT). SMT Reporting Items of Systematic Reviews and Meta-
is used by a variety of heath care providers such as chiroprac- Analyses for IPD (PRISMA-IPD) guidelines [9] (Appendix
tors, osteopaths, manual therapists and physiotherapists. eTable 1). The protocol was registered with PROS-
Many systematic reviews and meta-analyses have ana- PERO (https://www.crd.york.ac.uk/prospero/display record.
lysed the effects of SMT and suggest that it is an effective php?RecordID=25714) and approved by the Scientific
intervention for the reduction of pain and improvement of Review Board of the Vrije Universiteit Amsterdam and by
function [3–5]. However, recommendations for SMT in inter- the Ethical Committee of the VU University Medical Centre
national guidelines for chronic LBP are not consistent [6–8]. Amsterdam (Projectnr. 2015.544).
Since each guideline development group is using the same A detailed description of our study design and procedures
evidence, this is likely to be a consequence of differences was published previously [10]. The methodology presented
in how groups approach appraisal and interpretation of the here gives a brief overview.
evidence.
One disadvantage of traditional meta-analyses, is that Data sources and searches
aggregate data are extracted at the study-level and the inves-
tigator is dependent upon how the data is analysed and Search methods for identification of new studies
presented. Individual participant data (IPD) meta-analysis We included RCTs published from the year 2000. We lim-
circumvents the issues of poor reporting and not correcting ited this inclusion, because it is difficult to trace authors
for baseline covariates, because the individual data are avail- of older trials, and there is a high probability that these
able, resulting in more precise and potentially, a more valid data would not be accessible. More importantly, more recent
estimate of the effect. studies of SMT for low-back pain are of better methodolog-
Our recent systematic review for SMT for chronic LBP [5] ical quality. Therefore, it is unlikely that this delineator will
reflects some of the potential limitations of traditional aggre- have introduced undesirable bias [11]. Studies in the 2011
gate meta-analysis. For example, the authors of included Cochrane Review which examined the effect of SMT for
studies used different definitions of LBP, included a few chronic LBP were included [12]. In addition, we updated the
subacute LBP patients, used different frequencies of treat- search in December 2016 following the same procedure used
ments, and different analytic techniques ranging from a t-test in the Cochrane review (Appendix eTable 10) [5,10]. This
to sophisticated regression models. In an IPD meta-analysis, was supplemented with reference checking of systematic
some of these problems can be resolved. reviews and meta-analyses, and personal communication. A
The specific objective of this IPD meta-analysis was to recent update of the search (May 2018) resulted in the identifi-
assess the effectiveness of SMT compared to any other cation of five new trials [5], all of which were small in size and
conservative therapy for primary outcomes (i.e. pain and considered to have a high risk of bias. A update search from
back-related function) and secondary outcomes (i.e. quality May 2018 until October 2020 identified five studies [13–17],
 A. de Zoete et al. / Physiotherapy 112 (2021) 121–134 123

three of which are small in size. The two large-sized studies Secondary outcomes included self-reported health-related
of which one examined SMT vs recommended therapies and quality of life, return-to-work, global improvement (i.e. per-
the other SMT as adjuvant therapy, reported similar results ceived recovery), treatment satisfaction and analgesic use.
to ours.
Data extraction and quality of assessment
Study selection
Risk of bias in individual studies
Type of studies and participants The 13 risk of bias criteria (scored as ‘low risk’, ‘high risk’
Inclusion criteria. Only randomized clinical trials (RCTs) or ‘unclear risk’) recommended by the Cochrane Back and
were included. Studies were included if they recruited adults Neck group were used (Appendix eTable 2) [19]. The risk
(≥18 years of age) with chronic (≥12 weeks duration) LBP. of bias was conducted by two independent reviewers (SMR,
LBP is defined as LBP not attributed to a specific pathol- AdeZ). To adjudicate disagreement, a third reviewer (RO)
ogy (e.g. infection fracture, tumour or radicular syndrome). was contacted.
Participants with diffuse leg pain due to a low-back condition Data of all participants was sought from the authors of the
were included as were participants from primary or secondary studies fulfilling the inclusion criteria. We extracted study
care. In those studies where a mixed population was included characteristics, patient characteristics, types of outcomes,
(e.g. subacute and chronic), where possible, we included only duration of follow-up, description of experimental and con-
those participants with >12 weeks of LBP. trol interventions.

Preparing data for analyses

Exclusion criteria. We excluded studies that: 1) used an We first compared the original data with the published data
inadequate randomization procedure (e.g. alternate alloca- to check for completeness and where necessary and possible,
tion, allocation based on birth date); 2) included participants attempted to resolve any discrepancies. All variables were
with LBP and other conditions such as pregnancy or post- harmonized in a data harmonization platform [10].
operative participants; 3) tested the immediate effect of a All outcomes were pooled following a decision rule
single treatment only; and 4) compared the effects of a multi- (Appendix eTable 5). All pain scores were converted to a
modal therapy including SMT to another therapy or any other 0–100 points pain scale. To allow pooling of different func-
study design whereby the contribution of SMT could not be tional status measures, we recoded the individual scores into
isolated. Z-scores for each separate time point using pooled stan-
dard deviations as denominator (Z score = xi−x SD ). Analysing
Types of interventions these Z-scores resulted in standardized mean differences
Experimental intervention. Studies of spinal manipulation (SMD’s). To ease interpretation of SMD’s, we converted
(i.e. high-velocity low-amplitude techniques) as well as these to a mean difference (MD) for the 24 point Roland
mobilization (i.e. low-velocity low-amplitude techniques) Morris Disability Questionnaire (RMDQ), by multiplying
were included. the SMD with the population standard deviation  (SD) of
 n
 (ni −1)∗S 2
the studies measuring RMDQ (SDpooled =  (ni −1)
Comparisons. We analysed the following comparisons: 1)
i=0
SMT vs recommended interventions including non-drug ni = sample size for each trial; S = standard deviation for each
treatment (e.g. exercise), and drug treatment (e.g. NSAIDs); trial).
2) SMT vs non-recommended interventions (e.g. diathermy), For quality of life, physical and mental component scales
3) sham ‘placebo’ SMT; 4) SMT + intervention vs inter- of SF12 and SF36 were combined.
vention alone; 5) high-velocity low-amplitude SMT vs Other secondary outcomes were all dichotomized
low-velocity low-amplitude SMT (i.e. manipulation vs mobi- (Appendix eTable 5). However, data were often insufficient
lization). (less than 3 trials) to perform any analyses for these outcomes.
We based the definition of ‘recommended’ and ‘non- Adverse events were not included in our protocol but we did
recommended’ interventions on recent international guide- examine these data.
lines for LBP from the USA [8], the UK [6], the Netherlands
[7] and COST B13 European guidelines [18]. We categorized Data synthesis and analysis
an intervention into ‘recommended’ or ‘non-recommended’
when this was consistently stated in at least two of these All analyses were based on the intention-to-treat (ITT)
guidelines. principle. Our primary analyses consisted of one-stage IPD
meta-analyses for the five main comparisons at one, three,
Types of outcome measures six and twelve months follow-up (see protocol) [10]. These
Primary outcomes were self-reported pain and back- chosen intervals are standard follow-up moments for treat-
specific functional status. ment in LBP. We did not examine the effect of SMT directly
124 A. de Zoete et al. / Physiotherapy 112 (2021) 121–134

post-intervention as there was a large variation in duration Results

and frequency of treatments among the studies. Furthermore,
many studies contained no follow-up data immediately fol- Identification of trials
lowing the end of treatment. Longitudinal analyses for all
time points simultaneously were not performed as the models In total, 43 RCTs met our inclusion criteria, of which
were deemed too computationally demanding. 21 (50%) provided data [24–44] (Fig. 1) representing 4223
Analyses were conducted using a random-effects analy- participants. In three trials, the results differed from the pub-
sis of covariance model adjusting for baseline outcome using lished results for the primary outcomes by more than five
REML (restricted maximum likelihood), where a separated percent (i.e. 5 points on a 0–100 point Visual analogue scale,
intercept and separate residual variance for each study is and 1.2 points for the 0–24 RMDQ), which we determined to
specified. Models extended with a separate baseline adjust- be a relevant difference. Of these, one trial provided only data
ment term per trial did not demonstrate convergence in most from participants who gave consent to share their data [40].
analyses and we omitted them from all analyses [20]. For another study, we received data for more participants and
The pooled treatment effect of SMT was estimated using longer follow-up than published [38] while for the third study,
an MD or SMD (for continuous outcomes) or as an odds ratio baseline data were very similar but our results of the analyses
(for dichotomous outcomes) including the 95% CI. Nega- deviated somewhat from the published results due to differ-
tive MD for pain and SMD for function favours SMT, while ent patient numbers and use of different statistical techniques;
positive MD for quality of life favours SMT. This was a small trial (n = 41) and therefore, these deviations
We did not assess the effects of imputing missing data were not likely to influence the results presented here [41].
on outcomes. We addressed the missing outcome data (see
results: characteristics of studies). Characteristics of studies

Of these 21 RCT’s, 12 evaluated the effect of SMT

Subgroup and sensitivity analyses vs recommended interventions of which eight were com-
Subgroup analyses were pre-specified in our protocol [10] pared to exercise therapy [26,27,29,30,32–34,38,40–43], one
and conducted for the following variables: 1) type of clini- evaluated the effects of SMT vs sham SMT [44], five eval-
cian (i.e. chiropractor vs other); 2) ‘multi-modal’ SMT (i.e. uated the effect of SMT vs non-recommended interventions
SMT delivered alone as opposed to in conjunction with other [24,31,33,37,39], five evaluated the effect of SMT as an adju-
modalities have limiting or no effect); 3) country where the vant therapy [25,33,35,36,43] and three evaluated the effect
study was conducted (USA vs other); 4) only chronic LBP of manipulation vs mobilization [28,32,39] (Appendix eTable
participants (some trials included participants with subacute 3).
LBP), and 5) only trials with exercise therapy as a compara- Sample sizes ranged from 21 to 1334 (median = 192;
tor. We conducted sensitivity analyses for studies: 1) with IQR = 45–271). However, some trials included multiple arms,
low risk of bias on random sequence generation and alloca- and some included non-chronic LBP patients; Therefore, the
tion concealment, 2) with overall low risk of bias (defined as sample size for a given comparison should be considered
fulfilling six or more of the criteria items); 3) with a follow- potentially smaller. The included trials varied with respect
up period of eight weeks (data from eight weeks follow-up to the recruitment method, type of SMT technique, number
analysed with the three months instead of one month and 4) and duration of treatments and type of practitioner (Appendix
where we were able to reproduce published results. eTable 3).
Furthermore, sensitivity analyses were performed by cal- Of the 4223 participants, 2249 were randomized to the
culating functional status scores ourselves instead of using SMT group and 1974 to the comparison group. Table 1
the received overall score. Also, we examined the different presented the patient characteristics at baseline for SMT vs
functional status measures (e.g. RMDQ) and pain scales (e.g. recommended interventions. Data for the other comparisons
average pain, pain intensity), separately. are tabulated in eTable 7 (Appendix).
Lastly, to examine whether the RCTs included in this Missing data for primary outcomes ranged from 11% at
IPD meta-analysis were a representative sample of all known one month to 21% at 12 months. The UK BEAM trial pro-
RCTs published since 2000, we conducted a two-stage sensi- vided the largest dataset (n = 1334) and as a result, contributed
tivity analysis wherein we examined the effect sizes of RCTs most to the missing outcome data (50% of the total amount).
included in this IPD meta-analysis vs those which were eligi- The UK BEAM authors did not find a difference across ran-
ble for inclusion, but for which no IPD was available (using domized groups between responders and non-responders and
published aggregate data) [5]. drop-out appeared to be unrelated to the treatment [43].
Assessment of clinical relevance was defined as a small,
medium or large effect and based on the recommendations Risk of bias
of the Cochrane Back and Neck group [21,22]. The overall
quality of the evidence for each outcome was evaluated using Approximately three quarters of the studies (n = 15)
GRADE [23] adapted for IPD (see Appendix eTable 6). reported an adequate random sequence generation
 A. de Zoete et al. / Physiotherapy 112 (2021) 121–134 125

Fig. 1. Flow diagram of study inclusion.

126 A. de Zoete et al. / Physiotherapy 112 (2021) 121–134

Table 1
Patient characteristics at baseline for groups receiving SMT vs groups receiving recommended interventions (s = 12; n = 2475).
Demographic data SMT Recommended interventions
Age, mean (SD) years (s = 11, n = 2409) 47 (14) 47 (14)
Sex, n (%) female (s = 11, n = 2412) 667 (57) 684 (55)
Body mass index, mean (SD) (s = 8, n = 1434) 27 (5) 27 (5)
Ethnicity, n (%) white (s = 5, n = 861) 409 (91) 388 (88)
Lifestyle factors
Physical activity, (%) (s = 6, n = 824)
Low (1 or less than once a week) 115 (32) 166 (36)
Medium (2–3× a week) 146 (40) 166 (36)
High (more than 3× a week) 100 (28) 131 (28)
Smoking, n (%) non-smokers (s = 6, n = 1173) 451 (80) 453 (75)
Alcohol use, n (%) * *
Socio-demographics
Marital status, n (%) married; living with a partner (s = 6, n = 1173) 397 (69) 404 (68)
Level of education, n (%) low/middle (s = 7, n = 1672) 600 (68) 534 (68)
Income, n (%) * *
Employment status, n (%) at work (s = 9, n = 2126) 818 (78) 770 (72)
Nature and severity of LBP
Duration of LBP, n (%) less than 12 months (s = 7, n = 1252) 121 (20) 149 (23)
Leg pain, n (%) (s = 5, n = 1038) 320 (59) 281 (57)
Previous LBP treatment received, n (%) (s = 5, n = 930) 258 (28) 218 (23)
Previous physiotherapy for low back pain received, n (%) (s = 5, n = 771) 64 (8) 72 (9)
Previous SMT for low back pain received, n (%) (s = 6, n = 988) 209 (21) 111 (11)
Used medication for low back, n (%) (s = 6, n = 1018) 200 (20) 269 (26)
Non-specific, n (%) * *
Comorbidities * *
Type of treatment * *
Psychosocial factors SMT Control
Depression, n (%) (s = 5, n = 1297) 43 (6) 75 (13)
Treatment preference/expectations * *
Primary outcomes
Pain
Combined pain score at baseline, mean (SD), (s = 12, n = 2441) 49.5 (22.3) 49.8 (21.6)
Combined pain score at one month, mean (SD), (s = 10, n = 1948) 34.2 (23.0) 35.8 (23.9)
Combined pain score at three months, mean (SD), (s = 9, n = 1673) 27.92 (23.0) 32.1 (24.3)
Combined pain score at six months, mean (SD), (s = 8, n = 1321) 27.35 (23.1) 32.3 (23.9)
Combined pain score at twelve months, mean (SD), (s = 10, n = 1816) 31.80 (26.8) 33.3 (25.4)
Functional status
RMDQ sum score at baseline, mean (SD), (s = 9, n = 2174) 9.0 (5.0) 10.1 (5.4)
RMDQ sum score at one month, mean (SD), (s = 8, n = 1760) 5.6 (5.0) 6.7 (5.4)
RMDQ sum score at three months, mean (SD), (s = 8, n = 1648) 4.8 (5.1) 5.5 (5.3)
RMDQ sum score at six months, mean (SD), (s = 8, n = 1348) 5.0 (5.4) 6.3 (6.0)
RMDQ sum score at twelve months, mean (SD), (s = 7, n = 1575) 5.4 (5.7) 6.2 (5.92)
Secondary outcomes
SF36 Physical Component Scale of SF36 at baseline, mean (SD), (s = 5, n = 1362) 40.7 (7.2) 41.1 (7.6)
SF36 Physical Component Scale of SF36 at one month, mean (SD), (s = 3, n = 865) 44.1 (7.9) 45.7 (8.1)
SF36 Physical Component Scale of SF36 at three months weeks, mean (SD), (s = 4, n = 1154) 46.7 (8.2) 46.9 (8.5)
SF36 Physical Component Scale of SF36 at six months, mean (SD), (s = 5, n = 839) 47.3 (7.8) 47.9 (7.7)
SF36 Physical Component Scale of SF36 at twelve months, mean (SD), (s = 5, n = 1249) 46.4 (8.6) 46.8 (8.8)
SF36 Mental Component Scale of SF36 at baseline, mean (SD), (s = 5, n = 1362) 43.8 (9.1) 45.1 (9.6)
SF36 Mental Component Scale of SF36 at one month, mean (SD), (s = 3, n = 865) 45.5 (8.8) 46.9 (8.8)
SF36 Mental Component Scale of SF36 at three months, mean (SD), (s = 4, n = 1154) 46.9 (9.0) 47.3 (9.5)
SF36 Mental Component Scale of SF36 at six months, mean (SD), (s = 5, n = 839) 46.9 (9.1) 48.0 (8.9)
SF36 Mental Component Scale of SF36 at twelve months, mean (SD), (s = 5, n = 1249) 45.7 (8.9) 46.2 (10.0)
Medication use at baseline, n (% medication use) (s = 3, n = 668) 145 (22) 216 (32)
Medication use at one month, n (% medication use) (s = 3, n = 646) 84 (13) 146 (23)
Medication use at three months, n (% medication use) (s = 3, n = 626) 78 (13) 132 (21)
Medication use at six months, n (% medication use) s = 3, n = 593) 67 (11) 143 (24)
Medication use at twelve months weeks, n (% medication use) (s = 3, n = 582) 82 (14) 141 (24)
SD = standard deviation; s = number of studies; n = number of participants; * combining categories was not meaningful or no data available.
 A. de Zoete et al. / Physiotherapy 112 (2021) 121–134 127

and allocation concealment (Appendix eTable 4) 4) SMT + intervention vs intervention alone

[24–27,29–32,34,37–40,42,43]. Fifteen trials provided
an adequate overview of withdrawals or drop-outs and Primary outcomes
were able to keep these to a minimum for the subsequent Pain. There is moderate quality evidence that
follow-up measurements [24–29,31,33,34,36–40,44]. SMT + intervention has a similar benefit compared to
intervention alone at one, three and twelve months and
low quality evidence that SMT has a similar benefit to
Effect of SMT on primary and secondary outcomes: one
intervention alone at six months (largest difference at one
stage meta-analysis
month) (Table 2).
Functional status. There is moderate quality evidence that
Negative point estimates of the mean difference (MD) or
SMT + intervention has similar benefit compared to inter-
standardized mean difference (SMD) favours SMT.
vention alone at one, three and twelve months and low
1) SMT vs recommended Interventions quality evidence that SMT + intervention has similar bene-
fit compared to the intervention alone at six months (largest
Pain and function improved by the end of treatment and difference at three months) (Table 2).
this improvement was sustained up to twelve months after Secondary outcomes
randomization for all groups (Appendix eFigs. 3 and 4). Quality of life
Primary outcomes There is moderate quality evidence that
Pain. There is moderate quality evidence that SMT has SMT + intervention has similar benefit compared to
similar benefit to recommended interventions at all time the intervention alone at one, three and twelve months
points (largest difference at three months; Table 2). and low quality evidence that SMT + intervention has
Functional status. There is moderate quality evidence that similar benefit to the intervention alone at 6 months (largest
SMT has similar benefit to recommended interventions at all difference at twelve months) (Table 3).
time points (largest difference at one month; Table 2).
A subgroup analysis for SMT vs exercise showed similar 5) Manipulation vs mobilization
results (see Appendix eTable 8).
Pain. There is moderate quality evidence that manipula-
Secondary outcomes
tion has a similar benefit compared to mobilization at one
There is moderate quality evidence that SMT results in
month (Table 2).
a medium reduction in medication use compared to recom-
Functional status. There is moderate quality evidence that
mended interventions at two of the four time points (largest
manipulation has a similar benefit compared to mobilization
difference at six months. For all other secondary analyses,
at one month (Table 2).
there is low to high quality evidence that SMT has a similar
There are no data for the other time points and secondary
benefit to recommended interventions (Table 3).
outcomes.
2) SMT vs non-recommended interventions
Subgroup and sensitivity analyses
Primary outcomes
Pain. There is moderate quality evidence that SMT has The results from all one-stage sensitivity analyses suggest
similar benefit compared to non-recommended interventions similar results for pain and functional status at all time points
at one and six months (largest difference at six months). There (Appendix eTable 8).
are insufficient data for the three and twelve months analyses We found no differences in pain and functional sta-
(Table 2). tus between RCTs included and eleven eligible RCTs not
Functional status: There is moderate quality evidence included in the IPD repository (Table 4 and Appendix eTable
that SMT has similar benefit compared to non-recommended 9). The results of the two-stage analysis were comparable
interventions at one, three, and six months (largest differ- with the one-stage analysis. Sensitivity analysis, including
ence at six months). There are insufficient data for the twelve studies published since 2016, did not change our results.
months analysis (Table 2).
Secondary outcomes
Quality of life Discussion
There is low quality evidence that SMT has a similar ben-
efit to non-recommended interventions at one and six months Our results suggest there is moderate quality evidence
(largest difference at six months). There are insufficient data that SMT has similar effects as recommended treatments
for the three and twelve months analyses (Table 3). for pain reduction and improved functional status at short-
3) SMT vs Sham SMT , intermediate- and long-term follow-up. Additionally, there
is moderate evidence that SMT has similar effects for pain
The analysis for this comparison was not performed, relief and improvement in function when compared to non-
because we only had data from one study [44]. recommended therapies and when examined as an adjuvant
128 A. de Zoete et al. / Physiotherapy 112 (2021) 121–134

Table 2
Main treatment effects and GRADE summary of findings for all comparisons for the primary outcomes. Regression coefficients (␤) and 95% confidence
intervals (CI) of the intervention effects of random-effect models adjusted for baseline using REML (one stage analysis) are presented.
Comparison 1: SMT vs recommended therapies

Time measurement Difference in effect (␤) (95% CI) # studies N Quality of the evidence (and Comments
reason for downgrading)
Outcome: paina
1 month MD −3.0, 95% CI −6.9 to 0.9 10 1922 Moderate (inconsistency)
3 months MD −6.6, 95% CI −13.0 to −0.2 9 1647 Moderate (inconsistency)
6 months MD −5.6, 95% CI −9.6 to −1.5 8 1321 Moderate (inconsistency)
12 months MD −2.5, 95% CI −7.1 to 2.1 10 1791 Moderate (inconsistency)
Outcome: functional status SMD converted to a MD on
the 24 point RMDQ scale
1 month SMD −0.2, 95% CI −0.4 to 0.0 10 1939 Moderate (inconsistency) −0.8
3 months SMD −0.1, 95% CI −0.4 to 0.1 11 1892 Moderate (inconsistency) −0.6
6 months SMD −0.2, 95% CI −0.3 to 0.0 9 1490 Moderate (inconsistency) −0.8
12 months SMD −0.1, 95% CI −0.3 to 0.1 10 1826 Moderate (inconsistency) −0.5

Comparison 2: SMT vs non-recommended therapies

Outcome: paina
1 month MD −6.6 95% CI −10. 8 to −2.3 5 755 Moderate (inconsistency)
3 months Not enough data
6 months MD −8.3, 95% CI −20.5 to 3.8 3 419 Moderate (imprecision)
12 months Not enough data
Outcome: functional status SMD converted to a MD on
the 24 point RMDQ scale
1 month SMD −0.3, 95% CI −0.6 to 0.0 5 835 Moderate (inconsistency) −0.6
3 months SMD −0.0, 95% CI −0.8 to 0.7 2 375 Moderate (inconsistency) −0.9c
6 months SMD −0.3, 95% CI −0.6 to −0.0 3 414 Moderate (imprecision) −0.9
12 months Not enough data

Comparison 3: SMT vs sham SMT

No results only data of one study

Comparison 4: SMT + intervention vs intervention alone

Outcome: paina
1 month MD −7.44, 95% CI −12.7 to −2.1 5 762 Moderate (inconsistency)
3 months MD −5.2, 95% CI −11.0 to 0.7 2 619 Moderate (inconsistency)
6 months MD −1.4, 95% CI −6.7 to 3.8 2 222 Low (inconsistency, imprecision)
12 months MD −2.2, 95% CI −5.9 to 1.4 2 603 Moderate (inconsistency)
Outcome: functional statusb
1 month MD −0.6, 95% CI −2.3 to 1.1 a 4 746 Moderate (inconsistency)
3 months MD −1.3, 95% CI −2.6 to −0.1a 3 681 Moderate (inconsistency)
6 months MD −1.0, 95% CI −2.1 to 0.1a 2 218 Low (imprecision, inconsistency)
12 months MD −0.9, 95% CI −1.6 to −0.2a 2 626 Moderate (inconsistency)

Comparison 5: Manipulation vs mobilization

Outcome: pain
1 month MD −1.5, 95% CI −6.8 to 3.9 3 321 Moderate (limitations)
Not enough data for other time points
Outcome: Functional status SMD converted to a MD on
the 24 point RMDQ scale
1 month SMD 0.0, 95% CI −0.0 to 0.1 3 356 Moderate (limitations) −0.6
Not enough data for other time points
Negative difference in effect indicates higher estimated decrease in pain or improvements in function for SMT group compared to the control.
MD = mean difference of combined pain score on a 0–100 scale.
SMD = standardized mean difference of combined functional status score.
a Pain measured on a 0–100 point scale.
b All studies in the SMT + intervention vs intervention alone measured Roland Morris Disability questionnaire, therefore we use a mean difference.
c Based on one small study.
 A. de Zoete et al. / Physiotherapy 112 (2021) 121–134 129

Table 3
Main treatment effects and GRADE summary of findings for all secondary outcomes.
Comparison 1: SMT vs recommended therapies

Time measurement Difference in effect (95% CI) # studies N Quality of the evidence (and
reason for downgrading)
Outcome: Quality of life: Physical Component Scale of SF36 and SF12 combined
1 month MD −0.6, 95% CI −1.4 to 0.1 4 844 High
3 months MD −0.2, 95% CI −1.0 to 0.7 3 967 High
6 months MD −0.3, 95% CI −1.5 to 0.91 4 688 High
12 months MD 0.1, 95% CI −0.8 to 1.0 4 1055 High
Outcome: Mental Component Scale of SF36 and SF12 combined
1 month MD 0.4, 95% CI −0.4 to 1.2 4 844 High
3 months MD 0.8, 95% CI −0.0 to 1.6 3 967 High
6 months MD −0.1, 95% CI −1.4 to 1.2 4 688 High
12 months MD 0.5, 95% CI −0.9to 2.0 4 1055 High
Outcome: Recoverya : Yes vs No
1 month OR 1.3, 95% CI 0.9 to 1.9 2 499 Moderate (inconsistency)
3 months OR 1.2, 95% CI 0.8 to 1.8 3 538 Moderate (inconsistency)
6 months OR 1.1, 95% CI 0.8 to 1.6 3 651 Moderate (inconsistency)
12 months OR 0.8, 95% CI 0.5 to 1.2 2 445 Moderate (inconsistency)
Outcome: Medication Usea : Yes vs No
1 month OR 0.7, 95% CI 0.5 to 1.0 3 646 Moderate (inconsistency)
3 months OR 0.7, 95% CI 0.4 to 1.2 3 626 Moderate (inconsistency)
6 months OR 0.5, 95% CI 0.3 to 0.9 3 593 Moderate (inconsistency)
12 months OR 0.7, 95% CI 0.3 to 1.3 3 582 Moderate (inconsistency)
Outcome: Return to worka : Yes vs No
1 month Not enough data
3 months OR 1.0, 95% CI 0.5 to 1.9 3 190 Low (inconsistency,
imprecision)
6 months OR 0.6, 95% CI 0.3 to 1.3 3 189 Low (inconsistency,
imprecision)
12 months OR 1.3, 95% CI 0.6 to 2.7 3 180 Low (inconsistency,
imprecision)
Outcome: Satisfactiona : Yes vs No
1 month OR 0. 8, 95% CI 0.4 to 1.6) 2 319 Low (inconsistency,
imprecision)
3 months OR 6.6, 95% CI 1.5 to 29.9 2 429 Low (inconsistency,
imprecision)
6 months Not enough data
12 months Not enough data

Comparison 2: SMT vs non-recommended therapies

Outcome: Quality of life: Physical Component Scale of SF36 and SF12 combined
1 month MD 0.4, 95% CI −0.8 to 1.6 3 345 Low (inconsistency, imprecision)
3 months Not enough data
6 months MD −0.1, 95% CI −2.5 to 2.2 2 202 Low (inconsistency, imprecision)
12 months Not enough data
Outcome: Quality of life: Mental Component Scale of SF36 and SF12 combined
1 month MD 1.0, 95% CI −3.1; 5.2 3 345 Low (inconsistency, imprecision)
3 months Not enough data
6 months MD 1.7, 95% CI −0.6; 4.0 2 202 Low (inconsistency, imprecision)
12 months Not enough data
Other outcomes not enough data

Comparison 3: SMT vs sham SMT

No results only data of one study

Comparison 4: SMT + intervention vs intervention alone

Outcome: Quality of life: Physical Component Scale of SF36 and SF12 combined
1 month MD 0.1, 95% CI −1.1 to 1.5 3 708 Moderate (inconsistency)
3 months MD 0.3, 95% CI −0.7 to 1.3 2 619 Moderate (inconsistency)
130 A. de Zoete et al. / Physiotherapy 112 (2021) 121–134

Table 3 (Continued)
Comparison 1: SMT vs recommended therapies

Time measurement Difference in effect (95% CI) # studies N Quality of the evidence (and
reason for downgrading)
6 months MD −1.4, 95% CI −2.9 to 0.1 2 221 Low (inconsistency,
imprecision)
12 months MD −2.2, 95% CI −5.9 to 1.4 2 603 Moderate (inconsistency)
Outcome: Quality of life: Mental
Component Scale of SF36 and
SF12 combined
1 month MD −0.2, 95% CI −1.6 to 1.2 3 708 Moderate (inconsistency)
3 months MD 1.9, 95% CI −0.7 to 4.5 2 619 Moderate (inconsistency)
6 months MD 1.8, 95% CI −0.1 to 3.7 2 221 Low (inconsistency,
imprecision)
12 months MD 1.0, 95% CI −0.3 to 2.2 2 605 Moderate (inconsistency)
Other outcomes not enough data

Comparison 5: Manipulation vs mobilization

All outcomes not enough data
Positive difference in effect indicates higher increased quality of health for SMT group compared to the control.
MD = mean difference.
OR = odds ratio.
a Recovery was classified as ‘recovered’ if the participant scored more than 50% improvement or were (much) better or had no symptoms. Medication use

was classified for those using taking any using medication for LBP, while not taking any medication was classified as no medication use. Return to work was
classified as participants had returned to work or if there were no sick days recorded. Satisfaction was classified as ‘satisfied with care’ if participants were
(completely) satisfied or had scores >75%.

therapy. We have no results for the SMT vs sham compari- tions into recommended or non-recommended interventions
son, because we could only include one study. Finally, there was not always straightforward (e.g. myofascial therapy), and
is moderate quality evidence that manipulation has similar therefore, open for interpretation. While a sensitivity analy-
effects as mobilisation. sis could have helped to resolve this issue, the data were not
Our results are consistent with the recently published sufficiently robust to make this possible. Lastly, the catego-
aggregate data review [5] and with other recently published rization of an intervention as ‘non-recommended’ does not
systematic reviews [4,45,46]. imply that these interventions do not have an effect or are
It is somewhat difficult to interpret these findings, dangerous or ill-advised. While trials whereby patients are
particularly when SMT demonstrates similar effects to ‘blinded’ (i.e. sham) would help to resolve this issue; in our
recommended and non-recommended therapies or when estimation, no single study was adequately able to do so. An
examined as an adjuvant therapy. This appears confusing important difference of our IPD analysis compared to tradi-
and requires explanation. Firstly, most studies we identified tional aggregate meta-analyses is that we could adjust for the
examined the effect of SMT vs recommended therapies. In covariates, baseline pain and functional status, and were not
general, these studies were larger, had more data on follow- dependent upon how these data were reported in the original
up time-points and were of better methodological quality (i.e. publications. This has increased precision of our estimates
low risk of bias) than the studies in the other comparisons. compared to aggregate data meta analyses, but did not lead
Meaning, these findings were more robust and therefore, we to a different conclusion for the main effects.
have more confidence in their effect estimate. Even though It will be difficult to justify the required financial and
for all these comparisons, there is generally moderate qual- participant resources for further trials comparing SMT vs
ity evidence according to GRADE. While there are general current recommended therapies, as this is unlikely to change
guidelines for applying GRADE, there is no consensus. For our overall conclusions. Others have previously made the
example, we used a general rule-of-thumb when evaluating same observation with regard to trials of exercise treatment
‘imprecision’ in accordance with what might be considered for low back pain. A 2019 IPD meta-analysis of exercise
an ‘optimal information size. Applying a more stringent opti- therapy for low back pain has also produced precise esti-
mal information size would result in lower quality evidence mates for effectiveness [47]. Therefore, future studies should
for SMT vs non-recommended therapies or SMT as an adju- focus on cost-effectiveness, optimal dosage, delivery route
vant therapy, but not when applying this criterion to SMT vs to minimize side-effects, specificity of the location treated
recommended therapies (because the latter analyses included and maximize the non-specific effects of care, instead of
more than 1000 subjects). Secondly, categorizing interven- reproducing the same type of trials.
 A. de Zoete et al. / Physiotherapy 112 (2021) 121–134 131

Table 4
Representativeness of the pooled effects of studies providing data for the IPD study and those not providing data. Two stage analysis; SMT vs recommended
therapies.
Representativeness Number of studies Difference in Test of Prediction Interval
effect (CI 95%) heterogeneity
Outcome I2 P-value
Pain Mean difference
Combined pain score one month
All eligible studies 16 −2.4 (−4.9; 0.1) 66% <0.001 −2.4 (−11.6; 6.9)
Studies providing data 10 −2.5 (−5.9; 0.9) 75% <0.001 −2.7 (−13.8;8.8)
Studies not providing data 6 −2.2 (−6.1; 1.7) 55% 0.02 −2.218 (−12.9; 8.5)
Combined pain score three months
All eligible studies 14 −3.1 (−7.0; 0.8) 80% <0.001 −3.1 (−18.6; 12.4)
Studies providing data 9 −5.9 (−11.4; −0.5) 86% <0.001 −5.9 (−25.0; 13.0)
Studies not providing data 5 1.1 (−2.7; 4.9) 29% 0.196 1.1 (−7.4; 9.6)
Combined pain score six months
All eligible studies 13 −4.1 (−6.7; −1.6) 66% <0.001 −4.1 (−13.5; 5.2)
Studies providing data 8 −4.8 (−8.9; −0.6) 72% 0.001 −4.8 (−17.8;8.3)
Studies not providing data 5 −3.5 (−7.0; −0.1) 61% 0.009 −3.5 (−13.8; −6.4)
Combined pain score twelve months
All eligible studies 12 −1.8 (−4.8; 1.3) 71% <0.001 −1.8 (−12.5; 9.0)
Studies providing data 10 −2.1 (−6.5; 2.2) 79% <0.001 −2.1 (−16.8; 12.5)
Studies not providing data 2 −0.9 (−3.5; 1.8) 0% 0.6 −0.9 (−6.7; 5.0)
Functional status Difference in Z-score
Combined Functional Status one month
All eligible studies 13 −0.2 (−0.3; −0.0) 49% 0.014 −0.2 (−0.6; 0.3)
Studies providing data 9 −0.3 (−0.5; −0.0) 65% 0.003 −0.3 (−1.0; 0.4)
Studies not providing data 4 −0.1 (−0.2; 0.0) 0% 0.739 −0.1 (−0.2; 0.1)
Combined Functional Status three months
All eligible studies 15 −0.1 (−0.2; 0.1) 75% <0.001 −0.1 (−0.8; 0.7)
Studies providing data 11 −0.2 (−0.4; 0.0) 79% <0.001 −0.2 (−0.9; 0.6)
Studies not providing data 4 0.2 (−0.1; 0.4) 45% 0.089 0.2 (−0.5; 0.8)
Combined Functional Status six months
All eligible studies 13 −0.1 (−0.2; 0.0) 56% 0.002 −0.1 (−0.6; 0.4)
Studies providing data 9 −0.2 (−0.4; −0.0) 69% 0.001 −0.2 (−0.9; 0.5)
Studies not providing data 4 0.0 (−0.1; 0.1) 0% 0.778 0.0 (−0.1; 0.2)
Combined Functional Status twelve months
All eligible studies 13 −0.2 (−0.3; 0.1) 72% <0.001 −0.2 (−0.7; 0.4)
Studies providing data 10 −0.2 (−0.4; 0.1) 79% <0.001 −0.2 (−0.9; 0.6)
Studies not providing data 3 −0.1 (−0.3; 0.1) 43% 0.132 −0.1 (−0.7; 0.5)
CI = confidence interval; I2 = I2 statistic, which is the percentage of total variance that can be explained by heterogeneity, and 25% is considered low, 50%
moderate, and 75% high heterogeneity.

Strengths and limitations The most important limitation is potential selection bias.
We included only 50% of the eligible trials, which is com-
The most important strength is the sample size and the parable to other IPD studies [47,48]. In a two-stage analysis
diversity of studies, meaning these results are likely to be we examined the effect sizes of those that were eligible but
broadly generalizable to clinical practice. On the other hand, did not provide data. Results suggest only small differences
this diversity in studies led to a difference in methodological between included studies and those for which we were unable
quality and types of outcomes and covariates across trials to source the original data, indicating that the RCTs included
and introduced statistical heterogeneity, which can intro- are likely to be a representative sample of all published stud-
duce difficulties in interpreting the data. We investigated this ies. Also, the range of studies based upon publication date and
diversity with sensitivity analyses and two-stage analyses for methodological quality of the studies we included is compa-
primary outcomes at all follow-up measurements, but could rable with the non-included studies in the recently published
not explain the statistical heterogeneity. Our understanding of review [5]. Therefore, this facilitates an effective comparison
the effects of SMT would improve if we had an understanding of interventions across trials.
of the aetiology of LBP and how SMT works.
132 A. de Zoete et al. / Physiotherapy 112 (2021) 121–134

Also, our review differs slightly from our protocol with

regard to the classification of the comparator. In our proto- Key messages
col, we classified therapies into effective and non-effective,
whereas in this review we classified them into recommended • Randomised controlled trials of varying methodolog-
and non-recommended therapies. It was thought this would ical quality and size have examined the benefits and
best help translation of findings to clinical practice. This did harms of spinal manipulative therapy (SMT) for the
not affect the reported result, but was more a wording differ- treatment of chronic low back pain. These trials have
ence. Finally, longitudinal analyses would have provided us been summarised in systematic reviews with varying
with more information on the individual pattern of changes results. SMT is not currently recommended as a first
over time. In the future, these can be run, when programs are line treatment for chronic low back pain and its effects
able to process these large amounts of data. are uncertain.
• SMT is a good option for the treatment of chronic low
back pain.
Implications for clinicians • Future trials of SMT for low back pain should include
SMT is similarly effective as recommended and non- an economic evaluation and a better description of the
recommended interventions and when added an adjuvant qualitative and quantitative components of SMT (e.g.
therapy, in reducing pain and improving function in patients context of the visit, patient beliefs, and preferences
with chronic LBP. For patients with chronic LBP, SMT is a and factors that are likely to influence treatment).
treatment option. SMT can be delivered as a standalone ther- Future initiatives should also focus on standardizing
apy, although it is typically offered within the constructs of the manner in which inclusion and exclusion criteria,
a broader treatment package, together with exercise therapy outcomes and moderators are defined, measured and
or combined with usual care, as is recommended in recent reported. This will facilitate an effective comparison
national guidelines for low back pain [6–8]. This is impor- of interventions across trials.
tant because SMT is by nature a passive treatment. Therefore,
to prevent inappropriate behaviour and to empower patients
SMR reports grants from European Chiropractor’s Union,
to take control of their condition it is vital that practitioners
grants from Netherlands Chiropractic Association, grants
impart evidence-based messages about passive interventions
from European Centre for Chiropractic Research Excellence,
such as SMT. The choice of treatment should be the result of
grants from Belgian Chiropractic Association, during the
a shared decision-making process, taking patient preferences
conduct of the study; and SMR works in clinical practice
and clinicians experience and skills into account. No more
as a chiropractor and treats patients with chronic low-back
research is needed to support these recommendations. Further
pain.
similar research is unlikely to change these conclusions.
MU was Chair of the NICE accreditation advisory com-
Adverse events were often not recorded and when
mittee until March 2017 for which he received a fee. He
recorded, were measured differently across trials. Conse-
is chief investigator or co-investigator on multiple previous
quently, we were not able to pool these data. These data
and current research grants from the UK National Institute
did not provide more information than the adverse events
for Health Research, Arthritis Research UK and is a co-
described in our systematic review of aggregate data [5].
investigator on grants funded by the Australian NHMRC.
He is a NIHR Senior Investigator. He has received travel
expenses for speaking at conferences from the professional
Conclusion/clinical implication organisations hosting the conferences. He is a director and
shareholder of Clinvivo Ltd. that provides electronic data col-
Sufficient evidence suggest that SMT provides similar lection for health services research. He is part of an academic
outcomes to recommended therapies for pain relief and partnership with Serco Ltd. related to return to work initia-
improvement of functional status. SMT would appear to be tives. He is a co-investigator on a study receiving support in
a good option for the treatment of chronic LBP. kind from Orthospace Ltd. He is an editor of the NIHR jour-
Ethical approval: The study protocol was approved by the nal series, and a member of the NIHR Journal editors group,
Review Board of the coordinating institution (EMGO Insti- for which he receives a fee. He has published multiple papers
tute VU University Amsterdam). The protocol has also been on LBP some of which are referenced in this paper. He was
approved by the Ethical Committee (Ref. No. 2015.554) of lead author on one study included in the IPD meta-analysis.
the VU University. The authors declare they have no competing interests.
Funding: This systematic review was funded by a grant from
European Chiropractic Union Research Fund Contract No Acknowledgements
A.14.03.
Conflict of interest: ADZ works in clinical practice as a chi- We thank Martin Roosenberg for the advice in writing
ropractor and treats patients with chronic low-back pain. the article and correcting the English. We thank the Euro-
 A. de Zoete et al. / Physiotherapy 112 (2021) 121–134 133

pean Chiropractic Union for receiving the 2nd price ECCRE in the treatment of nonspecific chronic low back pain: a randomized
Research Award at the ECU congress 2018 in Budapest, controlled trial. Pain Med 2019;20(12):2571–87.
[14] de Oliveira Meirelles F, de Oliveira Muniz Cunha JC, da Silva
Hungary.
EB. Osteopathic manipulation treatment versus therapeutic exercises
in patients with chronic nonspecific low back pain: a randomized,
controlled and double-blind study. J Back Musculoskelet Rehabil
Appendix A. Supplementary data 2020;33(3):367–77.
[15] Ford JJ, Slater SL, Richards MC, Surkitt LD, Chan AYP, Tay-
lor NF, et al. Individualised manual therapy plus guideline-based
Supplementary material related to this article can be advice vs advice alone for people with clinical features of lumbar
found, in the online version, at doi:https://doi.org/10.1016/j. zygapophyseal joint pain: a randomised controlled trial. Physiotherapy
physio.2021.03.006. 2019;105(1):53–64.
[16] Goertz CM, Long CR, Vining RD, Pohlman KA, Walter J, Coulter I.
Effect of usual medical care plus chiropractic care vs usual medical
care alone on pain and disability among US service members with low
References back pain: a comparative effectiveness clinical trial. JAMA Netw Open
2018;1(1):e180105.
[1] GBD Disease Injury Incidence Prevalence Collaborators. Global, [17] Schulz C, Evans R, Maiers M, Schulz K, Leininger B, Bronfort
regional, and national incidence, prevalence, and years lived with dis- G. Spinal manipulative therapy and exercise for older adults with
ability for 354 diseases and injuries for 195 countries and territories, chronic low back pain: a randomized clinical trial. Chiropr Man Therap
1990–2017: a systematic analysis for the Global Burden of Disease 2019;27:21.
Study 2017. Lancet 2018;392(10159):1789–858. [18] Airaksinen O, Brox JI, Cedraschi C, Hildebrandt J, Klaber-Moffett
[2] Foster NE, Anema JR, Cherkin D, Chou R, Cohen SP, Gross DP, et al. J, Kovacs F, et al. Chapter 4. European guidelines for the manage-
Prevention and treatment of low back pain: evidence, challenges, and ment of chronic nonspecific low back pain. Eur Spine J 2006;15(Suppl
promising directions. Lancet 2018;391(10137):2368–83. 2):S192–300.
[3] Coulter ID, Crawford C, Hurwitz EL, Vernon H, Khorsan R, Sut- [19] Furlan AD, Malmivaara A, Chou R, Maher CG, Deyo RA, Schoene
torp Booth M, et al. Manipulation and mobilization for treating M, et al. 2015 updated method guideline for systematic reviews
chronic low back pain: a systematic review and meta-analysis. Spine J in the cochrane back and neck group. Spine (Phila Pa 1976)
2018;18(5):866–79. 2015;40(21):1660–73.
[4] Ruddock JK, Sallis H, Ness A, Perry RE. Spinal manipulation vs sham [20] Riley RD, Lambert PC, Abo-Zaid G. Meta-analysis of individual par-
manipulation for nonspecific low back pain: a systematic review and ticipant data: rationale, conduct, and reporting. BMJ 2010;340:c221.
meta-analysis. J Chiropr Med 2016;15(3):165–83. [21] Furlan AD, Pennick V, Bombardier C, van Tulder M, Group EBCBR.
[5] Rubinstein SM, de Zoete A, van Middelkoop M, Assendelft WJJ, de 2009 updated method guidelines for systematic reviews in the Cochrane
Boer MR, van Tulder MW. Benefits and harms of spinal manipulative Back Review Group. Spine (Phila Pa 1976) 2009;34(18):1929–41.
therapy for the treatment of chronic low back pain: systematic review [22] Cohen J. Statistical power analysis for the behavioral sciences. 2nd edn
and meta-analysis of randomised controlled trials. BMJ 2019;364:l689. Hillsdale: Lawrence Earlbaum Associates; 1988.
[6] NICE guideline [NG59]. Low back pain and sciatica in over [23] Higgins JPT, Green S. Cochrane handbook for systematic reviews of
16s: assessment and management; 2016 https://www.nice.org.uk/ interventions. The Cochrane Collaboration Version 510; 2011 [updated
guidance/NG59/chapter/Recommendations#non-invasive-treatments- March 2011]. Available from wwwhandbookcochraneorg.
for-low-back-pain-and-sciatica. [24] Balthazard P, de Goumoens P, Rivier G, Demeulenaere P, Ballabeni
[7] Bons SCS, Borg MAJP, Van den Donk M, Koes BW, Kuijpers P, Deriaz O. Manual therapy followed by specific active exercises
T, Ostelo RWJG, et al. NHG guideline for aspecific low- versus a placebo followed by specific active exercises on the improve-
back pain. https://www.nhg.org/standaarden/samenvatting/aspecifieke- ment of functional disability in patients with chronic non specific low
lagerugpijn#idp23613872.2017. back pain: a randomized controlled trial. BMC Musculoskelet Disord
[8] Qaseem A, Wilt TJ, McLean RM, Forciea MA, Clinical Guidelines 2012;13:162.
Committee of the American College of P. Noninvasive treatments [25] Bronfort G, Hondras MA, Schulz CA, Evans RL, Long CR, Grimm
for acute, subacute, and chronic low back pain: a clinical practice R. Spinal manipulation and home exercise with advice for subacute
guideline from the American College of Physicians. Ann Intern Med and chronic back-related leg pain: a trial with adaptive allocation. Ann
2017;166(7):514–30. Intern Med 2014;161(6):381–91.
[9] Stewart LA, Clarke M, Rovers M, Riley RD, Simmonds M, Stewart [26] Bronfort G, Maiers MJ, Evans RL, Schulz CA, Bracha Y, Svendsen
G, et al. Preferred reporting items for systematic review and meta- KH, et al. Supervised exercise, spinal manipulation, and home exer-
analyses of individual participant data: the PRISMA-IPD statement. cise for chronic low back pain: a randomized clinical trial. Spine J
JAMA 2015;313(16):1657–65. 2011;11(7):585–98.
[10] de Zoete A, de Boer MR, van Tulder MW, Rubinstein SM, Underwood [27] Cecchi F, Molino-Lova R, Chiti M, Pasquini G, Paperini A, Conti
M, Hayden JA, et al. Rational and design of an individual participant AA, et al. Spinal manipulation compared with back school and with
data meta-analysis of spinal manipulative therapy for chronic low back individually delivered physiotherapy for the treatment of chronic low
pain-a protocol. Syst Rev 2017;6(1):21. back pain: a randomized trial with one-year follow-up. Clin Rehabil
[11] Rubinstein SM, van Eekelen R, Oosterhuis T, de Boer MR, Ostelo 2010;24(1):26–36.
RW, van Tulder MW. The risk of bias and sample size of trials of [28] Cook C, Learman K, Showalter C, Kabbaz V, O’Halloran B. Early use
spinal manipulative therapy for low back and neck pain: analysis and of thrust manipulation versus non-thrust manipulation: a randomized
recommendations. J Manipulative Physiol Ther 2014;37(8):523–41. clinical trial. Man Ther 2013;18(3):191–8.
[12] Rubinstein SM, van Middelkoop M, Assendelft WJ, de Boer MR, van [29] Ferreira ML, Ferreira PH, Latimer J, Herbert RD, Hodges PW, Jennings
Tulder MW. Spinal manipulative therapy for chronic low-back pain. MD, et al. Comparison of general exercise, motor control exercise and
Cochrane Database Syst Rev 2011;(2):CD008112. spinal manipulative therapy for chronic low back pain: a randomized
[13] Grande-Alonso M, Suso-Marti L, Cuenca-Martinez F, Pardo-Montero trial. Pain 2007;131(1–2):31–7.
J, Gil-Martinez A, La Touche R. Physiotherapy based on a biobehav- [30] Gudavalli MR, Cambron JA, McGregor M, Jedlicka J, Keenum M,
ioral approach with or without orthopedic manual physical therapy Ghanayem AJ, et al. A randomized clinical trial and subgroup analysis
134 A. de Zoete et al. / Physiotherapy 112 (2021) 121–134

to compare flexion-distraction with active exercise for chronic low back [40] Zaproudina N, Hietikko T, Hanninen OO, Airaksinen O. Effective-
pain. Eur Spine J 2006;15(7):1070–82. ness of traditional bone setting in treating chronic low back pain: a
[31] Haas M, Vavrek D, Peterson D, Polissar N, Neradilek MB. Dose- randomised pilot trial. Complement Ther Med 2009;17(1):23–8.
response and efficacy of spinal manipulation for care of chronic low [41] Rasmussen-Barr E, Nilsson-Wikmar L, Arvidsson I. Stabilizing train-
back pain: a randomized controlled trial. Spine J 2014;14(7):1106–16. ing compared with manual treatment in sub-acute and chronic low-back
[32] Hondras MA, Long CR, Cao Y, Rowell RM, Meeker WC. A ran- pain. Man Ther 2003;8(4):233–41.
domized controlled trial comparing 2 types of spinal manipulation [42] Petersen T, Larsen K, Nordsteen J, Olsen S, Fournier G, Jacobsen S. The
and minimal conservative medical care for adults 55 years and older McKenzie method compared with manipulation when used adjunctive
with subacute or chronic low back pain. J Manipulative Physiol Ther to information and advice in low back pain patients presenting with
2009;32(5):330–43. centralization or peripheralization: a randomized controlled trial. Spine
[33] Hsieh CY, Adams AH, Tobis J, Hong CZ, Danielson C, Platt K, (Phila Pa 1976) 2011;36(24):1999–2010.
et al. Effectiveness of four conservative treatments for subacute [43] UK Beam Trial Team. United Kingdom back pain exercise and
low back pain: a randomized clinical trial. Spine (Phila Pa 1976) manipulation (UK BEAM) randomised trial: effectiveness of physical
2002;27(11):1142–8. treatments for back pain in primary care. BMJ 2004;329(7479):1377.
[34] Skillgate E, Vingard E, Alfredsson L. Naprapathic manual therapy or [44] Hidalgo B, Pitance L, Hall T, Detrembleur C, Nielens H. Short-term
evidence-based care for back and neck pain: a randomized, controlled effects of Mulligan mobilization with movement on pain, disability,
trial. Clin J Pain 2007;23(5):431–9. and kinematic spinal movements in patients with nonspecific low back
[35] Verma Y, Goyal M, Narkeesj D. Pain, range of motion and back strength pain: a randomized placebo-controlled trial. J Manipulative Physiol
in chronic low back pain before and after lumbar mobilisation. J Phys- Ther 2015;38(6):365–74.
iother Res 2013;(3):48–57. [45] Chou R, Deyo R, Friedly J, Skelly A, Hashimoto R, Weimer M, et al.,
[36] Vismara L, Cimolin V, Menegoni F, Zaina F, Galli M, Negrini S, et al. Rockville (MD) Noninvasive treatments for low back pain. AHRQ
Osteopathic manipulative treatment in obese patients with chronic low comparative effectiveness reviews; 2016.
back pain: a pilot study. Man Ther 2012;17(5):451–5. [46] Franke H, Franke JD, Fryer G. Osteopathic manipulative treatment for
[37] Walker BF, Hebert JJ, Stomski NJ, Losco B, French SD. Short-term nonspecific low back pain: a systematic review and meta-analysis. BMC
usual chiropractic care for spinal pain: a randomized controlled trial. Musculoskelet Disord 2014;15:286.
Spine (Phila Pa 1976) 2013;38(24):2071–8. [47] Hayden JA, Wilson MN, Stewart S, Cartwright JL, Smith AO, Riley
[38] Wilkey A, Gregory M, Byfield D, McCarthy PW. A comparison RD, et al. Exercise treatment effect modifiers in persistent low back
between chiropractic management and pain clinic management for pain: an individual participant data meta-analysis of 3514 participants
chronic low-back pain in a national health service outpatient clinic. from 27 randomised controlled trials. Br J Sports Med 2019;0:1–16.
J Altern Complement Med 2008;14(5):465–73. [48] Buffart Lm, Kalter J, Sweegers Mg, Courneya Ks, Newton Ru, Aaron-
[39] Xia T, Long CR, Gudavalli MR, Wilder DG, Vining RD, Rowell RM, son Nk, et al. Effects and moderators of exercise on quality of life and
et al. Similar effects of thrust and nonthrust spinal manipulation found physical function in patients with cancer: an individual patient data
in adults with subacute and chronic low back pain: a controlled trial meta-analysis of 34 RCTs. Cancer Treat Rev 2017;52:91–104.
with adaptive allocation. Spine (Phila Pa 1976) 2016;41(12):E702–9.

Available online at www.sciencedirect.com

ScienceDirect