Thoracic dysfunction in whiplash associated disorders: A systematic review

Background Research investigating Whiplash Associated Disorder (WAD) has largely focused on the cervical spine yet symptoms can be widespread. Thoracic spine pain prevalence is reported ~66%; perhaps unsurprising given the forceful stretch/eccentric loading of posterior structures of the spine, and the thoracic spine’s contribution to neck mobility/function. Approximately 50% WAD patients develop chronic pain and disability resulting in high levels of societal and healthcare costs. It is time to look beyond the cervical spine to fully understand anatomical dysfunction in WAD and provide new directions for clinical practice and research. Purpose To evaluate the scope and nature of dysfunction in the thoracic region in patients with WAD. Methods A systematic review and data synthesis was conducted according to a pre-defined, registered (PROSPERO, CRD42015026983) and published protocol. All forms of observational study were included. A sensitive topic-based search strategy was designed from inception to 1/06/16. Databases, grey literature and registers were searched using a study population terms and key words derived from scoping search. Two reviewers independently searched information sources, assessed studies for inclusion, extracted data and assessed risk of bias. A third reviewer checked for consistency and clarity. Extracted data included summary data: sample size and characteristics, outcomes, and timescales to reflect disorder state. Risk of bias was assessed using the Newcastle-Ottawa Scale. Data were tabulated to allow enabling a semi-qualitative comparison and grouped by outcome across studies. Strength of the overall body of evidence was assessed using a modified GRADE. Results Thirty eight studies (n>50,000) which were conducted across a range of countries were included. Few authors responded to requests for further data (5 of 9 contacted). Results were reported in the context of overall quality and were presented for measures of pain or dysfunction and presented, where possible, according to WAD severity and time point post injury. Key findings include: 1) high prevalence of thoracic pain (>60%); higher for those with more severe presentations and in the acute stage, 2) low prevalence of chest pain (<22%), 3) evidence of thoracic outlet syndrome, with some association to and involvement of the brachial plexus, 4) muscle dysfunction in the form of heightened activity of the sternocleidomastoid or delayed onset of action of the serratus anterior, 5) high prevalence of myofascial pain and trigger points in the scalene muscles, sternocleidomastoid and mid/lower fibres of trapezius muscle (48–65%), and 6) inconclusive evidence of altered thoracic posture or mobility. Conclusions Considerable evidence supports thoracic pain and dysfunction in patients with WAD, involving primarily nerves and muscles. Notwithstanding the low/very low level of evidence from this review, our findings do support a more extensive clinical evaluation of patients presenting with WAD. Additional high quality research is required to further characterise dysfunction across other structures in the thoracic region, including but not limited to the thoracic spine (mobility and posture) and thoracic muscles (stiffness, activation patterns). In turn this may inform the design of clinical trials targeting such dysfunction.


Introduction
The cumulative incidence of patients seeking healthcare post-whiplash from a road traffic accident has increased over the last 30 years to an annual incidence of between 3 and 6/1000 inhabitants in North America and Western Europe [1]. Following injury, individuals experience a range of clinical manifestations, described as Whiplash Associated Disorder (WAD), including neck pain, fatigue, nausea, low self-reported physical and mental health, cognitive impairments and pain in multiple sites [2]. The severity of presentation in WAD is categorised according to the Quebec Task Force Classification (QTF) where the presence of clinical signs and symptoms relate to the severity of the disorder [3].
Whilst research has identified risk factors for poor prognosis [4,5], and enhanced understanding of neurophysiological changes [6], it is not known why disability and pain persist beyond normal tissue healing times. With 40-60% patients progressing to experience chronic whiplash associated disorder (CWAD), estimated costs of~$4 billion (USA) and~€10 billion (Europe) associated with management and time off work [7,8], further research is needed to fully understand anatomical dysfunction in WAD and provide new directions for clinical practice and research. This includes the effects on anatomically related body regions, such as the thoracic spine. Potential ongoing dysfunction in the thoracic region may partially explain why there is inconclusive evidence for the effectiveness of physiotherapy management for WAD II, where interventions target a primary complaint of neck pain [9,10].
Although current research into WAD has focused on the primary complaint of neck pain [11], symptoms may also include stiffness [12,13] and pain in other regions including the jaw, head, upper and lower limbs, chest, abdomen and groin [14]. Moreover, data from a large cohort study (n = 6481) reported that 66% of individuals complained of thoracic mid-spine pain post whiplash injury, with 23% still experiencing symptoms one year later [14]. This is

Information sources
The search employed sensitive topic-based strategies designed for each database from inception to 1/6/16. No language or geographical restrictions were included. Databases included, CINAHL, EMBASE, MEDLINE, ZETOC, Index to Chiropractic Literature ChiroAccess and Google Scholar. Selected Internet sites and Indexes including, Turning Research into Practice, PubMed, National Research Register and Cochrane Back Review Group were also searched. Hand searching of key journals included Spine and the European Spine Journal. Grey literature included British National Bibliography for Report Literature, Dissertation Abstracts, Index to Scientific and Technical Proceedings, National Technical Information Service and the System for Information on Grey Literature.

Search strategy
The search strategy included terms related to whiplash associated disorder and patient reported or performance based measures of thoracic dysfunction. Terms and keywords derived from the scoping search and experts [subject specific (NRH,AR) and methodological (NRH,AR)] included: 'whiplash', 'whiplash associated disorder', 'WAD', 'whiplash injury', 'motor vehicle accident OR collision', 'road traffic accident', 'cervical strain' and 'thoracic spine', 'dorsal spine', 'mid-spine', 'thoracic injuries', limiting to adults >19 years and diagnosis to achieve the best balance of sensitivity and specificity. An example a search from Medline is included S2 Table. Terms were adapted to reflect spelling differences and [14] unique searching features of individual databases. Reference lists of included papers were also searched.

Study selection
Two reviewers (NRH, RS) independently searched information sources and assessed identified studies for inclusion, facilitated by grading each eligibility criterion as eligible/not eligible/ might be eligible [27]. Full texts were reviewed and included when both reviewers agreed [23]. A third reviewer (IT) mediated in the event of disagreement [28].

Data collection process and items
Using a standardised form, the two reviewers extracted data independently [23]. A further reviewer (IT) independently examined data for accuracy and clarity. Authors were contacted for additional information or data where required.

Data items
Data were extracted from each study, including: study design, sample characteristics including age, gender, severity of WAD using the QTF Classification if reported, time point post injury and patient reported or performance based measures of thoracic dysfunction.

Risk of bias in individual studies
Risk of bias for each included study was independently assessed by the same initial reviewers. The third reviewer mediated in situations of disagreement. All tools and processes were piloted prior to use. Risk of bias was only assessed for cohort and case-control studies using the Newcastle-Ottawa Scale [29]; that includes eight items that are rated and categorised into three groups, namely selection, comparability and outcome.

Summary measures
Summary measures of patient reported or performance based measures of thoracic dysfunction are presented in the form of prevalence data and confidence intervals where provided, for thoracic pain, chest pain, thoracic outlet syndrome (TOS), myofascial pain and trigger points, dysfunction involving the brachial plexus, thoracic spine posture or mobility. Results are presented where possible, according to severity (QTF) and stage of WAD i.e. acute/sub-acute (< 3 months) or chronic (> 3 months).

Synthesis of results
In accordance with the protocol [22] meta-analyses would be performed where a sufficient number of studies share all of the stated characteristics; design, measure of dysfunction, severity based on QTF and stage post whiplash injury.

Quality of evidence across studies
Quality of evidence, including risk of bias across studies was evaluated using GRADE [30] for individual outcomes of interest. By their very nature, observational studies are considered 'low quality' although could be upgraded where a large dose response was evident, or the effects could not be accounted for by bias [30]. Likewise, findings could be downgraded to 'very low' where concerns were identified from the body of studies relating to precision, consistency, directness, precision or potentially other additional domains relating to strength of association e.g. magnitude of effect [30].

Study selection
A total of 38 studies met the eligibility criteria, including 19 cohort studies, 16 case control studies and 3 single case studies/reports. The process of selection is detailed in Fig 1, with the list of excluded studies and reasons provided S3 Table. Study characteristics Study characteristics are detailed in Table 1. A range of measures of thoracic dysfunction were identified, including thoracic pain, chest pain, involvement of the brachial plexus, thoracic outlet syndrome, changes in muscle activation etc. An overview of the types of dysfunction is provided in Table 2 and, where possible, is presented according to severity and stage following injury.

Risk of bias within studies
Agreement of risk of bias following discussion was excellent with studies ranging considerably in quality from 1/9 to 7/9. Key methodological flaws included poor definition of cases, representativeness of cases, lack of non-exposed cohort, lack of blinding, controlling for risk factors beyond age and gender and clarity of follow up time point in cohort studies. There were many instances where information was missing and email requests for additional data or clarification was unforthcoming, resulting in ratings being 'unclear'. See Table 3 for risk of bias for cohort  studies and Table 4 for case control studies.

Synthesis of results
Synthesis of findings from cohort and case control studies across outcomes of dysfunction are provided in Tables 5-8. Sub-grouping according to stages and defined populations are included where reported. A summary of overall quality is provided based on GRADE following appraisal of risk of bias, consistency, precision, directness and effect size. Clinical heterogeneity across samples with respect to severity of presentation, time post injury and outcomes precluded meta-analysis being performed. Few studies stated a primary aim which accurately reflected the aims and objectives of this review.
Thoracic spine pain. Despite the very low quality of included studies, there is evidence of thoracic spine pain in a sub-acute WAD population (n = 11,576) [11,14,17,34,39,43,52,60], with prevalence ranging between 21%-66%,. Findings were inconsistent in chronic WAD, with prevalence ranging 0-94% [17,50]. Study quality, differing time points post injury, differing measurement approaches and higher degrees of WAD severity could partly account for the inconsistency. For example, chronic WAD, studies with less severe presentations (minor or WADII) [32,50] demonstrated lower prevalence rates (0-15.3%) compared to studies investigating WADII/III where prevalence was 54% [58]. Just one study reporting thoracic pain  specifically associated with central sensitisation (allodynia) where pain prevalence was 71% in individuals with CWADI/II [33]. Chest pain. Prevalence of chest pain in acute/sub-acute was reported to range 0.0-19% [11,14,17,52], although when considered as part of a widespread pain presentation ranged 9-19% [11,14]. In one study where they only considered individuals with numerical rating scale (NRS 0-10) 4+ chest pain prevalence was lower at 8% [52]. In CWAD prevalence rates for chest pain ranged 6-22% [17,53] although this reflects a, broad timescale with one study reporting results 6 month post injury [17] to one exploring prevalence at any time point following injury [53]. Drawing definitive conclusions on prevalence of chest pain is difficult given variation in approaches used to record pain (pain drawings, VAS, etc.), time points post injury, sample heterogeneity and the overall methodological low quality of research. This may also be a related to the focus of this review, being to those with mild to moderate presentations of WAD where studies including WADIV or fractures were excluded.
Thoracic posture. Evidence of thoracic postural dysfunction is inconclusive given conflicting findings from a small number of studies of low methodological quality evidence where postural assessment was not a primary focus [44,61]. Future studies should consider the use of a gold standard measure for postural evaluation [63].
Thoracic mobility and chest mobility. There is a notable gap in the evidence exploring thoracic and chest mobility in WAD, with just two studies (n = 54) of very low quality suggesting a trend for reduced chest mobility (p>0.05) and flexion hypomobility at the cervico-thoracic junction in women. In terms of thoracic mobility, evidence from one study (n = 7) [61] suggests a slight increase in thoracic flexion-extension, although with such a small sample and lack of inclusion of a valid approach to quantify thoracic mobility, meaningful conclusions cannot drawn.
Thoracic outlet syndrome in CWAD. Five studies found, a prevalence of 31-74+% [31,35,36,47,51] of thoracic outlet syndrome in CWAD. Although this suggests a relatively high prevalence, the methodological quality of included studies was generally poor (<4/9), resulting in a very low rating of quality according to GRADE.
Brachial plexus. Acute/sub-acute WAD Thoracic dysfunction in WAD Notwithstanding that evidence is drawn principally from one research group, there is evidence of thoracic dysfunction in relation to the brachial plexus provocation test (BPPT) in WADI-III. From the research by Sterling et al [12,57,59], there is evidence of a negative association between self-report symptom severity (VAS) and range of elbow extension during the BPPT in chronic WAD. Moreover, the extent of this dysfunction is in turn related to the degree of pain and disability, with those participants presenting with higher levels of selfreported pain and disability (NDI) having greater levels of dysfunction during BPPT.
Chronic WAD For participants with CWADII and/or III, evidence indicates dysfunction detected during the BPPT, with pain levels and restriction in elbow extension almost double those found in asymptomatic controls [38,55]. Furthermore, those with higher levels of self-reported pain and disability (NDI>30) continued to have evidence of dysfunction 6 months after the injury which was not seen in the mild group who were no different to the recovered group at 2 months [57]. Although quality of individual studies varied (4-6/9), the overall body of evidence for dysfunction of the brachial plexus remains very low overall, primarily due to risk of bias of the included studies.
Muscle activation. There are limited and very variable findings of thoracic muscle dysfunction (activation) in WAD with studies investigating a relatively small number of muscles: sternocleidomastoid [12,49,56]; serratus anterior [45]; middle and lower fibres of trapezius [45,54]. Although it is difficult to derive meaningful conclusions with respect to serratus anterior and trapezius, there is evidence supporting changes in sternocleidomastoid muscle activation with heightened levels of activation during a task of cranio-cervical flexion; this increase in sternocleidomastoid activity however, was not seen during neck rotation [49]. There appears to be a positive relationship between sternocleidomastoid activation and higher levels of pain severity, with participants with CWADII/III and moderate to severe levels of disability demonstrating increased levels of sternocleidomastoid activation of between 27-47% [12,56]. Thoracic dysfunction in WAD  Individuals with mild, moderate/severe presentations all share comparative levels of sternocleidomastoid activation in the acute [56] and chronic phases [12]. Myofascial pain and trigger points. Myofascial trigger points are highly prevalent in WAD with estimates ranging between 48-64% [16,37,40]. Muscles that have been investigated include the middle/lower trapezius [16], scaleneus medius [40,41] and sternocleidomastoid [37,40,41], all with similar prevalence levels. Findings suggest that latent trigger points are more prevalent in sub-acute WADII [41] whereas in chronic WAD there is a higher prevalence of active trigger points [37]. It should however be noted that the sample of chronic WAD comprised both WADII and III, so perhaps severity could partly explain the differences seen. Likewise, age may account for some of the differences given the sub-acute sample was 28.7 years [41] compared 41.6 years in the chronic WAD group [37].

Summary of evidence
This is the first methodologically rigorous systematic review investigating thoracic dysfunction in whiplash associated disorders. From a comprehensive search, 38 studies were included and evaluated as part of the review. Many studies were at risk of bias, primarily due to poor reporting with most studies published prior to the introduction of the STROBE reporting guidelines for observational studies. Notwithstanding the low quality of the evidence, there is unequivocal evidence of thoracic dysfunction in WAD. Pain. Although there is a high prevalence of acute and chronic pain experienced in the thoracic spine region following injury, distinguishing the interplay of pain mechanisms is however challenging. Whilst we know peripheral and central sensitisation begin immediately following injury [64], it is plausible that damage to thoracic musculoskeletal tissues contributes to the relatively high prevalence of pain reported in the acute/sub-acute stages from peripheral nociceptor stimuli [11,14,17,43,52]. Relatively few studies in this review reported perceived pain levels e.g. VAS, making it difficult to consider an association between injury severity and tissue damage. Although reviewed in detail by Van Oosterwijck et al, [64] the current review found just one study reporting thoracic pain specifically associated with central sensitisation, with allodynia reported in 71% of individuals with CWADI/II [33], arguably contributing to the lack of consistency with findings for chronic WAD where differing pain mechanisms may co-exist. Likewise distinguishing local from referred pain is challenging where injury in the cervical spine may refer pain caudally to the upper and mid thoracic region [65]. Unlike the thoracic spine, reported chest pain prevalence was relatively low, perhaps more closely associated with severe injuries, including fractures, which were excluded from this review. Future studies should, in addition to using the QTF Classification, include self-reported pain severity for each anatomical region.
Thoracic posture and mobility. With limited very low quality evidence and relative to the cervical spine, there is a paucity of research investigating thoracic posture and mobility in WAD [44,45,48,61]. This may be explained with priority being given to areas with most severe pain, with the cervical spine and associated tissues being most vulnerable to stress and damage compared to the relatively stable and stiff thoracic spine [21], arguably enhanced with the mandatory use of seatbelts offering additional stability. However, considering the effect of a forceful injury on posterior structures, and neck stiffness being a hallmark of chronic WAD further research is required, specifically to investigate thoracic posture, mobility and muscle stiffness in WAD, all of which may offer new directions for research into management of WAD.
Thoracic outlet syndrome. Although this review suggests a relatively high prevalence of thoracic outlet syndrome, the quality of evidence is very low and derived from relatively older studies, including two studies from the same group of authors [35,36]. Whilst the mechanism of injury and resultant strain on the scalene triangle in WAD would, in theory, place the thoracic outlet at risk of injury, this does needs to be investigated further. Likewise as a 'syndrome', this does not provide primary evidence of a primary structural dysfunction; it merely provides evidence of a dysfunction which, in turn, could be neurogenic, myogenic, vasculogenic in nature. Perhaps, in the absence of evidence of thoracic outlet syndrome in acute WAD, this condition is secondary to the consequences of whiplash, altered posture, changes in muscle behaviour etc. [66]. Future studies should use robust observational study designs and include valid assessment techniques to diagnose thoracic outlet syndrome.
Brachial plexus. There is considerable evidence of brachial plexus dysfunction in both acute/sub-acute and chronic WAD from research investigating the brachial plexus provocation test. Although coming from a relatively small group of researchers, the evidence supports further investigation. Future research could usefully consider the relationship of brachial plexus dysfunction to other musculoskeletal changes in the cervical and thoracic spine following a whiplash injury, but also approaches to managing this; a notable gap within the whiplash management evidence [9,10,66]. Although inclusion of the brachial plexus in this review may at first appear tenuous, with contributions from the level of T1 and its relationship to the thoracic outlet, its inclusion provides good evidence to further explore this anatomically and functionally challenging cervico-thoracic-supraclavicular region; a transitional zone between the stable/stiff thoracic spine and relatively mobile cervical spine.
Muscle activation. There is unequivocal evidence of altered function of muscles following whiplash injury [66], however this evidence is largely limited to cervical muscles, with this review identifying just three muscles with insertions to the thoracic region, sternocleidomastoid [12,49,56]; serratus anterior [45]; middle and lower fibres of trapezius [45,54]. Although it is difficult to derive meaningful conclusions with so few muscles investigated and the quality of the evidence, the observed positive relationship between sternocleidomastoid activation and pain severity, and evidence across all stages post injury [12,49,56] supports the need for further research into altered activation of cervico-thoracic and thoracic muscles, ideally involving functional spinal movements; something that is now feasible with advances in technology, including high density EMG.
Myofascial pain and trigger points. This review has identified that myofascial trigger points are highly prevalent in WAD [16,37,40], although again from very low quality evidence and limited to a small number of muscles; middle/lower trapezius [16], scaleneus medius [40,41] and sternocleidomastoid [37,40,41]. Nonetheless all muscles have similar prevalence levels of trigger points, with a higher prevalence of latent trigger points in sub-acute WAD [41] and active trigger points in chronic WAD [37]. With evidence of trigger points across all muscles and across the stages and severity of presentations, research is now required to explore other muscles in the thoracic region and better understand the development of pain, and persistent pain seen in chronic WAD, perhaps with longitudinal studies. Notwithstanding the quality, evidence was found of myofascial-entheseal dysfunction [16], a relatively new clinical entity and arguably similar to insertional tendinopathies. With rapid advances in our understanding of injury induced tendinopathies in the lower limb, this does offer new insights to possible muscle pathologies which may contribute to persistent pain and disability seen in WAD.
This review provides unequivocal evidence of thoracic dysfunction in WAD, albeit from evidence of low/very low quality. The findings do support a more extensive clinical evaluation of patients following a whiplash injury and the need for more methodologically robust observational studies to further characterise thoracic dysfunction in WAD across stages of the condition and levels of severity. Knowledge and understanding of thoracic dysfunction, where anatomical and biomechanical relationships with the cervical spine exist, offers novel directions for research into management of this disabling condition.
Research into WAD management has been, and continues to be primarily focused on managing cervical spine dysfunction, with interventions such as manual therapy, exercise etc. targeting the cervical spine [66], and targeting the psychological impact of a whiplash injury [66]. With inconclusive evidence of the therapeutic value of the above [67] it is perhaps time to consider new directions for research.
Whilst exercise as part of multimodal packages of care is recommended in the management of acute and chronic WAD [66,68], the range of approaches available are considerable e.g. graded functional exercise, postural exercises, and strengthening and motor control exercises etc. [66]. The authors are not however aware of any recommendations or research specifically supporting the inclusion of thoracic spine exercises, although these could reasonably be incorporated within functional exercise programmes. With at best short term modest improvements in current exercise interventions [9,10] and some evidence of some therapeutic value of thoracic spine manipulation in WAD [69] further research to investigate interventions targeting the thoracic spine and related dysfunction is justified.
It has not until recent years that the thoracic spine, coined the 'Cinderella' of the spine [21], has started to receive more research interest. This has been attributed partly to the relatively lower prevalence of symptoms than the cervical and lumbar spine more generally, but also due to the lack of affordable, non-invasive and valid measurement tools to evaluate motion in this anatomically complex and relatively stiff spinal region [21]. With the development of new measurement approaches for thoracic posture and mobility [63,70,71] we now have tools to support research of thoracic posture and mobility in the thoracic spine region in painful neck disorders.

Strengths and limitations of this review
This review is rigorous and original, with a design and focus on the thoracic spine region using a pre-defined rigorous and published protocol with subject and methodological experts contributing to the evaluation. The key limitation of the review is lack of high quality evidence and compounding this was that few authors responded to requests for additional information or data.

Conclusions
This first and rigorous systematic review found considerable evidence of thoracic pain and dysfunction in patients at all stages following whiplash injury. Notwithstanding the low/very low level of evidence, our findings do support a more extensive clinical evaluation of patients presenting with WAD. Key findings include 1) a high prevalence of thoracic spine pain, with the highest levels of pain immediately following injury and in more severe presentations 2) evidence of muscle dysfunction (delayed onset or heightened levels of activity) in a limited number of muscles 3) evidence of thoracic outlet syndrome and brachial plexus involvement 4) inconclusive/limited evidence of postural changes and effect on thoracic spinal mobility. Additional high quality research is required to further characterise dysfunction across other structures in the thoracic region, including but not limited to the thoracic spine (mobility and posture) and thoracic muscles (stiffness, activation patterns). In turn this may inform the design of clinical trials targeting such dysfunction.
Supporting information S1