Write your review/opinion on the article below Diagnostic accuracy of upper limb neurodynamic tests for the assessment of peripheral neuropathic pain A syste

Contents lists available at ScienceDirect

Musculoskeletal Science and Practice

journal homepage: www.elsevier.com/locate/msksp

Systematic Review

Diagnostic accuracy of upper limb neurodynamic tests for the assessment of peripheral neuropathic pain: A systematic review Konstantinos Koulidisa, Yannis Veremisa, Christina Andersona, Nicola R. Heneghanb,∗

a School of Sport, Exercise & Rehabilitation Sciences, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK b Centre of Precision Rehabilitation for Spinal Pain, School of Sport, Exercise & Rehabilitation Sciences, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK

A R T I C L E I N F O

Keywords: Entrapment neuropathies Carpal tunnel syndrome Cervical radiculopathy Upper limb neurodynamics Validity

A B S T R A C T

Background: Upper limb neurodynamic tests (ULNTs) are used to identify a neuropathic pain component in patients' presenting with arm and/or neck pain. Clinical tests with established diagnostic accuracy are required to not only to inform clinical management but also minimise costs associated with expensive medical in- vestigations. Objective: To evaluate the role of ULNTs in assessment of peripheral neuropathic pain and to inform their value in clinical practice when assessing patients with arm and/or neck symptoms. Design: Systematic review was undertaken according to published guidelines, and reported in line with PRISMA- DTA. Method: Key databases were searched up to 21/11/2017. Inclusion criteria: Patient population experiencing arm and/or neck symptoms with suspected peripheral neuropathic involvement, studies that compared ULNT to a reference standard, any study design using primary diagnostic accuracy data. Two reviewers independently assessed risk of bias (ROB) using QUADAS-2. The overall quality of evidence was evaluated using GRADE. Results: Of eight included studies (n = 579), four were assessed as low ROB, although all had concerns regarding applicability. For carpal tunnel syndrome, ULNT1 sensitivity values ranged 0.4–0.93, specificity 0.13–0.93, positive likelihood ratio 0.86–3.67 and negative likelihood ratio 0.5–1.9. For cervical radiculopathy ULNT1 and the combined use of four ULNTs had sensitivity of 0.97 (95%CI 0.85–1.00) whereas the ULNT3 was the most specific (0.87, 95%CI 0.62–0.98). Positive likelihood ratio ranged 0.58–5.68 and negative likelihood ratio 0.12–1.62. Conclusion: Based on the available evidence ULNTs cannot be utilised as a stand-alone test for the diagnosis of CTS. Limited evidence suggests that ULNTs may be clinically relevant for the diagnosis of CR, but only as a “ruling out” strategy. However, the overall quality of the body of evidence after applying the GRADE approach was low to very low across studies. Further higher quality research is needed to establish firm conclusions.

1. Introduction

Peripheral neuropathic pain (PNP) is a term used to describe pain that results from a lesion or disease affecting the somatosensory ner- vous system (Finnerup et al., 2016). PNP can arise when a peripheral nerve trunk or a nerve root has been subject to injury, compression, inflammation or ischemia resulting in reduced physical capabilities of the nervous system (Nee and Butler, 2006). Symptoms and signs in neuropathies can be classified as positive (gain of function) or negative (loss of function). Positive symptoms include pain, paresthesia, dys- esthesia, hyperalgesia and allodynia and indicate abnormal excitability in the nervous system, whereas negative symptoms, such as

hypoesthesia or anesthesia and weakness reflect reduced impulse con- duction (Woolf, 2004).

The most common conditions affecting the peripheral nervous system are entrapment neuropathies (EN), with carpal tunnel syndrome (CTS), cubital tunnel syndrome and cervical radiculopathy (CR) being examples which contribute considerably to the socioeconomic burden of occupational related musculoskeletal complaints and the associated costs. Individually EN have been associated with severe pain, depres- sion and functional limitations (Fernández-de-las-Peñas et al., 2015). CTS is often observed in activities involving repetitive manual tasks, forceful wrist movements or with direct pressure on the wrist, estimated to affect 2–15% of workers (Atroshi et al., 1999) and costing more than

https://doi.org/10.1016/j.msksp.2019.01.001 Received 16 May 2018; Received in revised form 18 December 2018; Accepted 2 January 2019

∗ Corresponding author. E-mail addresses: [email protected] (K. Koulidis), [email protected] (Y. Veremis), [email protected] (C. Anderson),

[email protected] (N.R. Heneghan).

Musculoskeletal Science and Practice 40 (2019) 21–33

2468-7812/ Crown Copyright © 2019 Published by Elsevier Ltd. All rights reserved.

T

2 billion dollars each year in the USA (work absenteeism, medical evaluation, treatment) (Saint-Lary et al., 2015). In the case of CR, the data regarding the prevalence and the epidemiology of the condition are very limited. The reported annual incident of CR is 83.2 per 100.000 persons (107.3 for men and 63.5 for women) with a peak in- cidence in the fifth and sixth decade for both genders (Radhakrishnan et al., 1994).

The diagnosis of EN is based on information received during the subjective (history taking) and physical examination, which is then confirmed via diagnostic imaging or electrophysiological studies. Clinical examination of EN encompasses a variety of tests (sensation, muscle strength and reflexes) assessing the integrity and ability of the nervous system to conduct afferent or efferent impulses (loss of func- tion) (Baselgia et al., 2017). In addition, a thorough examination in- cludes evaluation of increased mechanical sensitivity of the nervous system, since PNP can be present without or with minimal loss of nerve conduction (Schmid et al., 2009). Diagnostic imaging and electro- physiological studies are most commonly used to establish a diagnosis of EN (Wainner et al., 2003). For most clinicians, these methods are accessible but given the waiting time for patients and the high cost for the society it would be useful to establish accurate clinical examination tests for the diagnosis of EN.

Neurodynamic tests are used by musculoskeletal physiotherapists in order to identify changes of mechanosensitivity in the nervous system, thus assessing gain of function (Baselgia et al., 2017). Due to the in- terdependence of the mechanical, electrical and chemical properties of the nervous system, changes in one of these features may affect the others (Butler, 2008). Impairments in the surrounding musculoskeletal structures could apply mechanical or chemical stimuli to a nerve, re- sulting in venous congestion, impaired axoplasmic flow, inflammation and development of mechanosensitive abnormal impulse generating sites (Nee and Butler, 2006).

For disorders affecting the upper limbs four different neurodynamic tests have been proposed to assess mechanosensitivity of the brachial plexus, medial, radial and ulnar nerve (Elvey, 1980) (Table 1). Where symptoms are not related to central pain mechanisms (broader dis- tribution of symptoms due to central sensitization e.g. in case of per- sistent pain) a positive test response could be associated with neural or non-neural tissue sensitivity. A neurodynamic test is considered posi- tive if it can reproduce the patient's own symptoms and if those symptoms can be altered through structural differentiation (Butler, 2000). Schmid et al. (2009) assessed the reliability of ULNTs and found that those tests have moderate reliability. Wainner et al. (2003, 2005) reported substantial to almost perfect reliability for the interpretation of theULNT1 (median) and ULNT2b (radial).

Although used by clinicians the diagnostic accuracy of upper limb neurodynamic tests (ULNTs) has not yet been fully established and is important to optimise patient care. A recent systematic review has summarised the evidence on diagnostic performance of tests (including ULNTs) which are utilised for the identification of CR and concluded that when consistent with patient history, a combined result of four negative ULNTs (high sensitivity) and a negative Arm Squeeze test

could be used to rule out the disorder (Thoomes et al., 2017). Likewise an earlier systematic review, concluded that a positive Spurling's, traction/neck distraction, and Valsalva's test might be indicative of CR, while a negative ULNT1 might be used to rule it out (high sensitivity) (Rubinstein et al., 2007). Of the eight included studies in this systematic review only two had assessed the diagnostic accuracy of ULNTs. Finally in a previous clinical commentary the authors attempted to summarise the available evidence in regard to the diagnostic usefulness of neuro- dynamic tests (Nee et al., 2012). The authors, based on biomechanical and experimental studies, concluded that ULNTs can potentially dis- tinguish pain related to neural mechanosensitivity from pain arising from other tissues, and therefore could detect PNP. In the view of the growing body of evidence, a systematic review is required to evaluate the quality and synthesis the available current evidence of the diag- nostic accuracy of ULNTs and to inform clinical practice. The aim therefore of this study was to examine the intended role of ULNTs in assessment of PNP, by answering the following research question: What is the diagnostic accuracy of ULNTs when compared to diagnostic imaging or electrophysiologic studies, and how results from ULNTs can be interpreted when assessing patients with arm and/or neck symp- toms?

2. Design and methods

This systematic review was conducted according to a pre-defined protocol based on the Cochrane Handbook for Diagnostic Test Accuracy studies (Deeks et al., 2013) and the Center for Reviews and Dis- semination (CRD, 2009). In addition, the study is reported according to Preferred Reporting Items for a Systematic Review and Meta-analysis of Diagnostic Test Accuracy Studies (PRISMA-DTA) (McInnes et al., 2018). (Appendix 1)

2.1. Search strategy

Informed by subject (NH, KK, YV) and methodological experts (NH, CA) key bibliographic databases were searched independently by two reviewers (KK, YV). The search employed sensitive topic-based strate- gies designed for each database from inception to 21st November 2017. Databases of interest were: PEDro, MEDLINE (through PubMed), AMED, CINAHL, Cochrane Library, and EMBASE. The search strategy, informed by scoping search included MeSH terms and text words, as well as a combination of both for a comprehensive search. The fol- lowing keywords and combination of them were used: upper limb neurodynamic test, neural provocation test, upper limb tension test, diagnosis, peripheral neuropathic pain, peripheral entrapment neuro- pathy, radicular pain, cervical radiculopathy, brachial plexus, carpal tunnel syndrome, cubital tunnel syndrome, accuracy, specificity, sen- sitivity, validity.

The search was augmented using reference lists of included studies, as well as searching the grey literature. Box 1 details the MEDLINE search strategy.

Table 1 ULNT procedure.

Order of movements ULNT1 (median) ULNT2a (median) ULNT2b (radial) ULNT3 (ulnar)

1 Shoulder depression Shoulder depression Shoulder depression Shoulder depression 2 Shoulder abduction 110° Elbow extension Elbow extension Shoulder abduction 100° 3 Wrist and fingers extension Lateral rotation of the arm Medial rotation arm Lateral rotation arm 4 Forearm supination Wrist and finger extension Wrist and fingers flexion Forearm pronation 5 Shoulder lateral rotation Shoulder abduction 10° Shoulder abduction Elbow flexion 6 Elbow extension Contralateral lateral flexion of the

cervical spine Contralateral lateral flexion of the cervical spine

Wrist and fingers extension

7 Contralateral lateral flexion of the cervical spine

Contralateral lateral flexion of the cervical spine

K. Koulidis et al. Musculoskeletal Science and Practice 40 (2019) 21–33

22

2.2. Eligibility criteria

Eligibility criteria were established following the recommendations of The Cochrane Handbook for Diagnostic Test Accuracy studies (Leeflang et al., 2008) and informed using the SPIDER search concept (Cooke et al., 2012). Titles and abstract of the identified studies were screened by two independent reviewers (KK, YV) for eligibility using pre-specified inclusion criteria.

Inclusion criteria (based on SPIDER) included that the sample (S) comprised populations aged > 18 years with arm and/or neck symp- toms with suspected peripheral neuropathic involvement (signs and symptoms suggesting excitability in the nervous system such as pain, paresthesia, dysesthesia, spasm or reduced impulse conduction such as hypoesthesia or anesthesia and weakness) (Nee and Butler, 2006); the phenomenon of interest (PI) was the diagnostic accuracy of ULNTs; investigated using a diagnostic accuracy study design (D); with com- parison of the index test (ULNTs) to a reference standard, such as, electrophysiologic examination (electromyography and nerve conduc- tion studies) or advanced imaging (e.g. Magnetic Resonance Imaging (MRI), CT, myelography) (E). Although not perfect, these tests are considered to be the most accurate diagnostic tests available (Wainner et al., 2003; Jablecki et al., 1993, 2002; Kuijper et al., 2009).

Exclusion criteria: case series, case reports, surgical or cadaveric studies; publications for which full text not available.

2.3. Quality assessment

Two reviewers (KK, YV) independently conducted the risk of bias (ROB) assessment using the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) – tool, a development of the original tool (Whiting et al., 2011). It consists of four key domains: patient selection, index test, reference standard and, flow and timing. All key areas are assessed for ROB, whereas the first three are also assessed in terms of applicability to the review question. Each domain is judged as “high risk”, “low risk” or “unclear risk” based on signaling questions aiming to assist judgment (Whiting et al., 2011). Overall, a study can be judged

as having “low risk of bias” if every domain has been ranked as “low risk”. Assessment of applicability is based on the first three domains and whether they are in line with the review question. The study is judged as having “no concerns” regarding applicability if these domains are in line with the review question and “with concerns” if deviates from the review objective. The QUADAS-2 has been used in recent systematic reviews (Grodahl et al., 2016; Hegedus et al., 2012) and is re- commended by the Cochrane Collaboration and the U.K National In- stitute for Health and Clinical Excellence (Reitsma et al., 2009).

2.4. Data extraction

Diagnostic accuracy data and study characteristics were extracted by one reviewer (KK) using a pre-designed data extraction sheet which covered five areas. The data were audited by a second reviewer (YV) for accuracy. The following data were extracted: authors and publication details, studies' methods (aim of study, study design, method of re- cruitment, eligibility criteria, and ethical approval), participant details, diagnostic test data (sensitivity, specificity, predictive values, like- lihood ratios and other). Finally, the fifth section was 2 × 2 con- tingency tables for the diagnostic tests.

2.5. Summary measures

Sensitivity, specificity, likelihood ratios (LR) and predictive values (PV) were the outcomes for which data were sought. True positive, false positive, true negative and false negative values were summarised. In cases where only incomplete or raw data were presented, a 2 × 2 contingency table was used to re-estimate these values. Sensitivity and specificity were graded as low (< 0.50), low/moderate (0.51–0.64), moderate (0.65–0.74), moderate/high (0.75–0.84) and high (> 0.85) in line with previous systematic reviews of diagnostic accuracy studies (Grodahl et al., 2016; Schneiders et al., 2012). Clinical interpretation of likelihood ratios was based on Jaeschke et al. (1994a,b) as follows: conclusive evidence (LR+ > 10 and LR- < 0.1), strong diagnostic evi- dence (LR+ 5 to 10 and LR- 0.1 to 0.2), weak diagnostic evidence (LR

Box 1 MEDLINE search strategy

1. peripheral neuropathic pain.mp or exp Neuralgia/ 2. radicular pain.mp or exp Hereditary Sensory and Autonomic Neuropathies/ 3. peripheral entrapment neuropathy.mp 4. cervical radiculopathy.mp or exp Radiculopathy/ 5. carpal tunnel syndrome.mp or exp Carpal tunnel syndrome/ 6. cubital tunnel syndrome.mp or exp Cubital tunnel syndrome/ 7. brachial plexus neuropathies.mp or exp Brachial plexus neuropathies/ 8. exp Nerve compression syndromes/ 9. 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8

10. upper limb neurodynamic test.mp 11. upper limb tension test.mp 12. neural provocation test.mp 13. exp Diagnosis/ 14. exp Pain measurements/ 15. exp Neurologic examination/ 16. exp Physical examination/ 17. 10 or 11 or 12 or 13 or 14 or 15 or 16 18. diagnostic accuracy.mp 19. sensitivity and specificity.mp or exp Sensitivity and specificity/ 20. validity.mp 21. exp Reproducibility of results/ 22. exp Predictive value of tests/ 23. 18 or 19 or 20 or 21 or 22 24. 9 and 17 and 23

K. Koulidis et al. Musculoskeletal Science and Practice 40 (2019) 21–33

23

+2 to 5 and LR- 0.2 to 0.5) and negligible evidence (LR+ 1 to 2 and LR- 0.5 to 1).

2.6. Data analysis

Homogeneity among studies was explored to evaluate if the studies were suitable for combining in a meta-analysis. Areas of exploration were: study designs, patient population, comparable reference tests and diagnostic data, no differences in diagnostic thresholds (Burgess et al., 2011). In addition, quality assessment of the included studies was conducted, since studies with high ROB often over-estimate the per- formance of a test (Lijmer et al., 2002). Given the heterogeneity of the included studies a narrative synthesis was undertaken.

2.7. Quality of evidence across studies

Quality of evidence, including risk of bias across studies was eval- uated using GRADE (Schünemann et al., 2008) for individual tests. Quality of overall body of evidence is influenced by amongst other factors, study design, patient populations, precision, consistency, di- rectness and as such each outcome was evaluated by both reviewers independently (Schünemann et al., 2008).

3. Results

3.1. Study identification

The searches identified 1802 studies with screening of title and abstract resulting in 15 studies that were retrieved for full-text eva- luation and 8 studies (n = 579) meeting the eligibility requirements for inclusion. (Fig. 1). There was 100% of agreement between the re- viewers on selecting studies.

3.2. Study description

Table 2 summarises the specific characteristics of all eight studies. Three studies investigated the diagnostic accuracy of ULNTs in in- dividuals with suspected CR (Wainner et al., 2003; Apelby-Albrecht et al., 2013; Ghasemi et al., 2013). Two of the studies used electro- physiologic procedures as the reference standard (Wainner et al., 2003; Ghasemi et al., 2013). One study used MRI, clinical examination and history as a reference standard (Apelby-Albrecht et al., 2013). Five studies investigated the diagnostic accuracy of ULNTs in individuals with suspected CTS with nerve conduction studies as the reference standard (Wainner et al., 2005; Vanti et al., 2011; 2012; Bueno-Gracia et al., 2016; Trillos et al., 2018).

3.3. Risk of bias assessment

Agreement of risk of bias following discussion was excellent (100%). Four studies were assessed as “low risk of bias” (ROB) (Wainner et al., 2003, 2005; Vanti et al., 2012; Trillos et al., 2018), but all of them had concerns with regards to applicability (Table 3). Patient selection procedures and poor reporting of flow and timing were the main areas of ROB. Only two studies were assessed as no concerns for applicability (Fig. 2) (Apelby-Albrecht et al., 2013; Bueno-Gracia et al., 2016). Interpretation of the index test was the main reason for concern regarding applicability since it was not in agreement with our review question. In our study an ULNT is considered positive only when it reproduces the patient's clinical symptoms and those symptoms are modified with structural differentiation (Nee et al., 2012; Butler, 2000; Coppieters et al., 2002).

3.4. Synthesis of results

The main limitations for performing a meta-analysis were the

heterogeneity in terms of the reference standard utilised, as well as in the interpretation of the index test and the methodological quality of the included studies. Since a meta-analysis was not possible, diagnostic accuracy data (sensitivity, specificity, predictive values and likelihood ratios) are presented using a narrative approach. The overall body of the evidence in terms of ROB, inconsistency, indirectness, imprecision, and the presence of potential reported bias after applying the GRADE approach was low to very low across studies and across outcomes. Diagnostic accuracy for all clinical indicators is summarised in Tables 4 and 5 and outcome of GRADE evaluation in Tables 6 and 7.

3.5. Diagnostic accuracy of upper limb neurodynamic tests

3.5.1. Carpal tunnel syndrome Five studies examined the diagnostic accuracy of ULNTs in patients

with suspected CTS (Wainner et al., 2005; Vanti et al., 2011; 2012; Bueno-Gracia et al., 2016; Trillos et al., 2018). From these studies two were at ROB (Vanti et al., 2011; Bueno-Gracia et al., 2016) and four had concerns regarding applicability (Wainner et al., 2005; Vanti et al., 2011; 2012; Trillos et al., 2018). Those at ROB had limitations related to patient selection and flow and timing. The study of Vanti et al. (2011) was at ROB because the number of patients enrolled in the study was different from the number of patients that were included in the analysis (Whiting et al., 2011), whereas in the study by Bueno-Gracia et al. (2016) the authors provided limited information in regards to the methods used for the enrollment of the sample (consecutive or random sample). The studies that had concerns regarding applicability used a definition for a positive ULNT that differs from that being used in this review.

Three studies assessed the validity of ULNT1 (median) considering the test positive in the presence of only one of the following criteria: 1) reproduction of patient's symptoms; 2) side to side differences (> 10°) in elbow extension; 3) contralateral neck side-flexion increased symp- toms or ipsilateral side-flexion decreased symptoms (Wainner et al., 2005; Vanti et al., 2011; Trillos et al., 2018). Sensitivity was moderate/ high 0.75 (95%CI 0.58–0.92) (Wainner et al., 2005) to high 0.91 (95%CI 0.74–0.98) (Vanti et al., 2011) and 0.93 (95%CI 0.88–0.96) (Trillos et al., 2011). Specificity was low in all 3 studies: 0.13 (95%CI 0.04–0.22) (Wainner et al., 2005), 0.15 (95%CI 0.05–0.36) (Vanti et al., 2011) and 0.06 (95%CI 0.0–0.33) (Trillos et al., 2018). In the study by Vanti et al. (2011) the authors conducted a second analysis in which “reproduction of patient's symptoms” changed to “reproduction of symptoms in the first, second or third digit”, but again only one of the three criteria was required for a positive ULNT1. The second analysis revealed low to moderate sensitivity (0.54, 95%CI 0.35–0.72) and moderate specificity (0.70, 95%CI 0.48–0.85). Overall, none of the in- terpretations of ULNT1 was capable of ruling in or ruling out a diag- nosis of CTS because LRs were between 0.5 and 2.0.

Two studies examined the diagnostic accuracy of ULNT1 using a different interpretation for a positive test. In these studies the test was considered positive if it was able to reproduce patient's symptoms and these symptoms were altered with structural differentiation (Vanti et al., 2012; Bueno-Gracia et al., 2016). Sensitivity ranged from low 0.05 (95%CI 0.02–0.19) (Vanti et al., 2012) to low/moderate 0.58 (95%CI 0.45–0.71) (Bueno-Gracia et al., 2016). Specificity ranged from moderate/high 0.84 (95%CI 0.72–0.96) (Bueno-Gracia et al., 2016) to high 0.93 (95%CI 0.82–0.98) (Vanti et al., 2012). Bueno-Gracia and colleagues (2016) suggested that the ULNT1 may be clinically useful to determine patients with CTS due to high +LR (3.67). However the high number of false negatives results challenges this notion (Table 4).

3.5.2. Cervical radiculopathy Three studies investigated the concordance of ULNT1 with a re-

ference standard in patients with suspected CR (Wainner et al., 2003; Apelby-Albrecht et al., 2013; Ghasemi et al., 2013). The reference standard in two of these studies was NCS and needle electromyography

K. Koulidis et al. Musculoskeletal Science and Practice 40 (2019) 21–33

24

(Wainner et al., 2003; Ghasemi et al., 2013), whereas in the third study the authors used the combination of patient history, clinical examina- tion and MRI findings as the reference standard (Apelby-Albrecht et al., 2013). In two of these studies ULNT1 showed moderate to high (0.83, 95%CI 0.66–0.93) and high sensitivity (0.97, 95%CI 0.90–1.0) (Apelby- Albrecht et al., 2013; Wainner et al., 2003) whereas in the third study the sensitivity was low 0.35 for chronic CR and low/moderate 0.6 for acute CR (Ghasemi et al., 2013). Specificity ranged from low 0.22 (95%CI 0.12–0.33) (Wainner et al., 2003) and 0.4 (Ghasemi et al., 2013) to moderate/high 0.75 (95%CI 0.48–0.93) (Apelby-Albrecht et al., 2013). Moreover, in the study of Wainner et al. (2003) the ULNT1 demonstrated negative likelihood ratio (LR) of 0.12, meaning that a negative ULNT1 could rule out CR. This study had low ROB, but had

concerns regarding applicability related to the different interpretation of the index test from the authors compared with the review question (Whiting et al., 2011). In addition, due to wide 95% CI the results of this study should be interpreted cautiously. Wide CIs reduce the strength of evidence by influencing the precision of the pooled esti- mates.

The validity of ULNT2b (radial) was assessed by two studies (Wainner et al., 2003; Apelby-Albrecht et al., 2013). Sensitivity was moderate in both studies: 0.66 (95%CI 0.48–0.81) (Apelby-Albrecht et al., 2013) and 0.72 (95%CI 0.52–0.93) (Wainner et al., 2003). Spe- cificity ranged from low 0.33 (95%CI 0.21–0.45) (Wainner et al., 2003) to moderate/high 0.75 (95%CI 0.48–0.93) (Apelby-Albrecht et al., 2013).

Fig. 1. PRISMA flow diagram for systematic reviews.

K. Koulidis et al. Musculoskeletal Science and Practice 40 (2019) 21–33

25

Ta bl

e 2

Ch ar

ac te

ri st

ic s

of in

cl ud

ed st

ud ie

s.

A ut

ho r.

(y ea

r) ,

co un

tr y

Ty pe

of st

ud y

Pa th

ol og

Collepals.com Plagiarism Free Papers

Are you looking for custom essay writing service or even dissertation writing services? Just request for our write my paper service, and we'll match you with the best essay writer in your subject! With an exceptional team of professional academic experts in a wide range of subjects, we can guarantee you an unrivaled quality of custom-written papers.

Get ZERO PLAGIARISM, HUMAN WRITTEN ESSAYS

Why Hire Collepals.com writers to do your paper?

Quality- We are experienced and have access to ample research materials.

We write plagiarism Free Content

Confidential- We never share or sell your personal information to third parties.

Support-Chat with us today! We are always waiting to answer all your questions.