Background and limitations of the current paradigm

Controversy continues to exist in the literature and in practice over how to ‘clear’.Footnote 1 the spine of the obtunded blunt trauma patient (OBTP) who is likely to remain unevaluable. Previous reviews and practice guidelines on this topic were produced in 2004, and adopted by the United Kingdom Intensive Care Society [1, 2]. Since 2004 there have been three meta–analyses on the subject [35] and an update from the Eastern Association of Trauma (EAST) [6]. The evidence base remains limited with the meta-analyses drawing on Individual cohort studies, which at best represent level 2b evidence [7].

Why is this still important?

Cervical spine injury complicates blunt poly-trauma in a significant minority of cases, with a typical incidence of 5% [1, 8, 9]. There is a delicate balance of risks and benefits in managing OBTPs. Certainly missed unstable spinal injuries are associated with potentially devastating neurological compromise. Balanced against this is the reality that the majority of patients have a stable spinal column and extensive imaging is expensive and delays patient mobility. Most patients are maintained with ‘spinal precautions’ during this time (e.g. immobile with log roll turns and a cervical collar) and this is associated with significant morbidity and on occasion mortality.

Therefore, to clear the spine of significant unstable injury while patients remain obtunded or unevaluable largely centres on imaging, possibly supported by clinical evaluation if and when patients become evaluable [10]. We will consider the imaging modalities below but the main options include:

  • Plain radiographs

  • Computerised tomography (CT), most commonly multidetector-row CT (MDCT)

  • Magnetic resonance imaging (MRI)

  • Dynamic studies, e.g. dynamic fluoroscopy (DF) or flexion/extension radiography (FE/R)

Clinicians are engaged in a ‘screening’ process, attempting to identify significant spinal instability while rejecting normal or minor spinal injuries, i.e. ‘clearing’ the spine. In common with all screening procedures, a number of concepts are important:

  • Prevalence of injury, i.e. pre-test probability. This is low, typically about 5%. Therefore if no screening were applied 95% of patients would suffer little consequence, an unclear proportion of the 5% could experience a complication and any intervention based upon the screening, especially if screening has high sensitivity and poor specificity, would subject many ‘normal’ patients to spinal precautions.

  • There remains controversy around what constitutes spinal stability following screening, and therefore the ‘result’ of the screening may be unclear. Furthermore, the performance of the individual interpreting the screening imaging is difficult to define beyond specialty and seniority, e.g. ‘a senior musculoskeletal radiologist’.

  • The sensitivity, specificity, negative and positive predictive values are not clearly defined for the various imaging modalities. As ever, there is a trade off, partly related to pre-test probability, between optimal sensitivity and specificity, and ultimately clinicians may not reach consensus on what they regard as optimal balances of these parameters [11].

  • The benefit or otherwise of available ‘therapies’ as a result of a positive test remain unclear. It is difficult to standardise management of certain injuries, e.g. prolonged collar or halo vest application vs. surgical stabilisation.

  • The complications of instituted therapies (e.g. cervical collars, radiation doses, transfers) are poorly described and this becomes important if the screening has a high sensitivity, low specificity and is applied to the ‘95% normal’ population.

  • The resource implications are significant, e.g. a ‘missed’ injury with associated care and legal implications must be balanced against liberal use of imaging and prolonged immobilisation.

  • It is likely that a rigorous scientific analysis is only part of the solution. Individual clinician and institutional ‘memory’ (e.g. Mrs X with the missed broken neck) tends to be long and deeply embedded. The significance of individual values to clinicians will therefore vary, e.g. a quoted miss rate of 5% may be acceptable to one clinician but not to another; therefore, different institutions will chose to implement the research differently.

  • Screening inevitably has false positives and negatives. Clinicians and patients must accept finite rates of these variables, while striving for the optimal values of them, e.g. when any patient is declared to have a ‘stable’ spine, mobilisation should be careful with maintained vigilance for missed injury.

  • This review does not consider subsequent determination of the significance of detected injuries, i.e. if screening detects an unstable injury, individualised assessment of detected injuries is required which may require any or all of clinical evaluation, plain radiographs, MDCT, MRI or dynamic studies.

Why undertake another review?

The meta-analyses on the subject draw on broadly equivalent data but have opposite conclusions underlining the importance of an iterative review of the subject.

Imaging technology has dramatically advanced since 2004; however, worldwide clinical practice remains varied.

There is nothing in the Cochrane database, BMJ clinical evidence, the trip database or any other evidence based medicine resource on the subject.

The majority of literature is from the U.S.A. and due complex social, financial, medico-legal, and political agendas is not always directly translatable to the rest of the world.

Methods

Ovid was used to search Medline and Embase. Google Scholar, BestBETs, the trip database, BMJ clinical evidence and the Cochrane library were searched. See Figs 1 and 2. 129 articles were retrieved in the writing of this review. The bibliographies of all relevant studies were then reviewed. Hand searches of reference lists were used to identify additional references. Direct communication with authors was sought where necessary.

Fig. 1
figure 1

Flow diagram showing MeSH terms and search combinations. The numbers refer to the number of papers found for the given MeSH term combination. The boxes in red show lists where titles and abstracts were read. Remaining papers and editorials were found from the reference lists of the studies

Fig. 2
figure 2

Flow diagram showing MeSH terms and search combinations. The numbers refer to the number of papers found for the given MeSH term combination. The boxes in red show lists where titles and abstracts were read. Remaining papers and editorials were found from the reference lists of the studies

We focused on papers published since our previous guidelines and review (2004) and included articles considering spine screening protocols and secondary outcome measures during prolonged immobilisation of the spine in OBTPs.

Screening protocols

Certain assumptions remain valid or have emerged since our last review [1]:

  1. 1.

    Clinical evaluation in multiply injured or obtunded patients is unreliable and screening imaging of the spine is mandated.

  2. 2.

    Plain radiography misses injuries of the spinal column in both the alert and the obtunded patient. There is overwhelming evidence for the superiority of CT over plain radiography to detect spinal injury. CT also has time and logistical advantages [6, 1224].

  3. 3.

    Dynamic fluoroscopy (DF)/Flexion extension (FE/R) radiography to detect instability is hard to perform, time consuming, inaccurate and potentially harmful. There has been little advance in this area, and our previous work suggested at least 177 patients would need to undergo DF to detect one further injury beyond plain radiographs and CT [6, 2528].

  4. 4.

    MRI is the gold standard if there is a positive neurological examination (referable to the spinal cord) at presentation or if it becomes clear at any stage that there is an abnormal neurological examination [1, 6, 20, 29]. Furthermore, MRI is a highly sensitive test for soft tissue spinal injury, traded against specificity (relative to CT).

Summary of the available imaging options

Plain radiography

The sensitivity of plain radiographs to detect significantFootnote 2 spinal injuries, including the three view combination of anteroposterior, lateral and open mouth ‘odontoid’ views ranges widely from 31 to 94% [12, 14, 15, 17, 18, 23, 24, 30]. Usefully Holmes quoted a pooled sensitivity of 52% [22]. Specificity is typically better and has approached 98% in some studies [23]. Most authors agree that there is little place for routine plain radiographs in the management of the OBTP, whenever alternative imaging modalities, in particular MDCT, are available.

Computerised tomography (CT)

The technology behind CT has progressed considerably since 2004, with most centres using helical or multidetector-row CT rather than ‘directed CT’ (see abbreviations). The ability of CT to detect spinal column injury is excellent with sensitivity typically approaching 100% [1]. CT is especially good at detecting bony injuries [15, 2123, 3033] with specificities approaching 100%. Harris et al. [34] quote a negative predictive value of 99.7% for the ability of CT to detect clinically significant injury. It is emphasised that purely ligamentous cervical spine injuries are exceptionally rare, representing between 0.1 and 0.7% of blunt trauma victims, with most published studies at the lower end of that range [1, 34, 35]. CT is more time and cost effective than plain radiography [3638].

It has been and remains a widely held belief that only MRI is able to demonstrate isolated ligamentous injury [39]. However, whilst MRI is undoubtedly the most sensitive modality CT can demonstrate ligamentous injury in a high proportion of casesFootnote 3 in some series approaching 100% [1, 4143]. This is particularly true of modern CT scanners [44]. While CT is a commonly employed modality in managing poly-trauma and patients can typically be scanned ‘from head to pelvis’ in one investigation, CT does expose patients to significant radiation.

MRI

Most authorities state that MRI is more sensitive than CT for detecting soft tissue and ligamentous injuries that could contribute to an occult instability of the cervical spine [3, 4, 4548]. Unfortunately MRI is less sensitive for detecting bony injury and injuries of the posterior spinal column and the significance of many detected injuries (specificity) remains controversial. Routinely performing MRI also carries significant resource implications, requires transfer and introduces the patient to the ferromagnetic environment.

Traumatologists have agreed on the use of MRI in patients with a suspected myelopathy and cord injury, and CT is well established to be inferior in this regard [29, 49]. The false positive rate for MRI has been quoted to be between 25 and 40% [50]. Oversensitivity for presumably clinically insignificant injuries has limited its widespread use as a screening modality [4].

“Postmortem data suggests that MRI demonstrates lesions within the ligaments with high sensitivity, but that many may not reflect instability. Established criteria for distinguishing significant from inconsequential apparent abnormalities on MRI do not exist. The range of ‘normal’ anatomical variations has only become apparent as MRI has become established, and the distinctions between ‘lesion’ and ‘variation’ are blurred. Without proven guidelines, many physicians use through-and-through tears of ligaments to indicate definite mechanical failure, with lesser evidence of injury, such as simple high signal on T2-weighted images, being considered ambiguous or suspicious. These less specific findings tend to be incorporated with clinical findings, evidence of subluxation and other imaging findings, mechanism of injury, and likelihood of successful compliance with conservative treatment.” [51, 52].

Dynamic fluoroscopy (DF)/Flexion extension (FE/R) radiography

The ability to visualise the relevant anatomy is poor, as low as 4% [25]. They have a low ability to detect injury [26, 28]. It is highly resource and labour intensive [26].

We will consider the available studies in more depth below but essentially plain radiographs and dynamic studies have largely been made redundant and the main debate within the literature centres on whether patients can be cleared following CT alone, or whether CT and MRI is mandated.

Summary of literature since 2004

The first meta-analysis (see Table 1): Muchow et al. [3] examined studies that prospectively or retrospectively looked at blunt trauma patients who were entered into a cervical spine clearance protocol that included MRI. They included five papers which either pre-date 2004 or are included in subsequent meta-analyses. With regard to resolving which imaging combination should be used to clear the spine of the adult OBTP this meta-analysis contains only 12 patients appropriate for consideration.

Table 1 Studies included in the Muchow meta-analysis [3]

The Schuster et al. [53] paper features in the Muchow et al. [4] meta-analysis and the second meta-analysis by Schoenfeld. It looked at all blunt trauma patients (2,854) who when first examined were moving all four limbs. They didn’t restrict to unevaluable patients, but of the unevaluable patients, only 12 had MRI scans despite a normal CT scan. These were all normal. This was not a true comparison of the ability of CT versus MRI to detect clinically significant injuries. However, interestingly, they conclude that, as all the MRI scans were normal, CT only is required to clear the CS of the blunt trauma victim.

The second meta-analysis (see Table 2): Schoenfeld et al. [4] included 11 studies from 2000 to 2008 consisting of 1,550 patients (1,295 of which were OBTPs) (Table 2). “Investigations included in this meta-analysis were any prospective or retrospective studies in which patients had an MRI for the purposes of cervical spine clearance after a negative CT scan.” [4]. None of the 11 studies were prospective, randomized, controlled trials comparing MRI and CT to CT alone. As yet (2011) we await such a study.

Table 2 Studies included in the Schoenfeld meta-analysis from 2010

The results need to be interpreted with caution. Of the 12 patients who required operative intervention after an initially negative CT and subsequent positive MRI scan, nine of them were from the Sarani et al. [45] study and none of them were obtunded. One of the patients had surgery for a chance fracture [54] at T7, leaving only two patients from the Menaker study [43, 55]. Table 1 in the paper by Menaker documents the findings of these two patients. These two patients [48], who significantly, represent the only OBTPs in the literature (when this meta-analysis was written) required operative intervention after a negative CT scan. With regard to the first patient Como claims that the injuries described would be extremely unusual without evidence of other injuries and in the second remarks that it is unclear why they needed an operation at all. The authors conclude that the 1% (12 of 1,550) is unacceptable and that any protocol should include an MRI scan to ‘clear’ the CS, in the OBTP.

Overall there were 182 positive MRI scans after a normal CT, 84 (46%) patients had prolonged collar application and 12 (7%) patients required surgery. It is not clear if the patients who were treated with prolonged collar application actually had significant injuries that required this treatment and it is debateable if those that had surgery required it. However, if one considers studies including OBTPs the results become: 102 positive MRI following normal CT, 61 (60%) with prolonged collar application, and 2 (2%) patients required surgery.

The EAST review from 2009 (lead author J Como) included 52 studies. Of the studies included in the Muchow paper only three were used in the EAST guidelines: [53, 56, 57].

Of the 52 studies included in the guideline the following three papers are relevant to the issue of what to do with the OBTP that are not included in the above two meta-analyses:

Ghanta et al. [58] retrospectively reviewed 51 obtunded patients who had undergone both CT and MRI; 10 of 46 patients (22%) with a normal CT CS had an abnormal MRI CS. Of these, seven were felt to be potentially unstable. Even so, it is unclear how significant these injuries are. The authors concluded that the 2000 EAST guidelines might not be sensitive enough in the obtunded patient [6].

Sanchez et al. [59] validated their own clearance protocol; they cleared the CS (provided that the patient was moving all four extremities when initially examined) on the basis of a normal CT. They only used MRI where there was neurological abnormality. From a population of 2,854 they claim to have missed one injury (0.03%) using their protocol (in a patient with syringomyelia). They concluded that CT is adequate for a clearance protocol.

EAST concludes that for the OBTP: “The risk/benefit ratio of obtaining MRI in addition to CT is not clear, and its use must be individualized in each institution (level 3).”

Additional (more recent) relevant papers not included above: (See Table 3)

The following two studies by Menaker et al. [47] and Brown et al. [44], both published in the same journal 2 months apart, re-emphasise the contrasting opinion in the literature. Menaker et al. [47] presented a retrospective cohort study to ascertain if newer CT technology (40-detector row CT) was sufficient for ruling out CS injury in OBTPs. Ninety-six were OBTPs (see Table 3). Of these; 15 (15.6%) had an abnormal MRI scan, seven (7.3%) were managed with a hard cervical collar and one (1%) patient was managed operatively. They found that 8.7% of patients in the 2010 study vs. 8.3% of patients in the 2008 study had their management changed on the basis of the MRI. Only one patient required surgery (1%) suffering a cord compression without contusion but the authors conclude that MRI must be part of any spinal evaluation algorithm because it alters management in approximately 8% of patients [47].

Table 3 Relevant studies published post the Schoenfeld meta-analysis [4]

Conversely, Brown et al. [44]: This study was the first to directly compare two different CT scanners with MRI (106 patients). They compared the missed injury rate between a four-slice (43 patients) and a 64-slice MDCT scanner (63 patients). CT missed three injuries that were picked up by MRI (all were in the four-slice group), only one of which required operative fixation. They conclude that newer CT scanners do not appear to miss clinically significant injuries and may allow clearance of the CS in OBTPs without MRI.

Steigelman [41]: cleared the spine after a negative MRI in the OBTP group. With a large cohort of >14,000 patients (Not all OBTPs) 120 in the OBTP group; they concluded that the use of MRI in patients with normal results on CS MDCT (four-row detector) does not appear to alter treatment [41].

Como et al. [43]: Como’s group prospectively evaluated their revised protocol with 197 patients. Their view of MRI for the OBTP is clear: “Use of MRI in OBTPs is costly, time-consuming, and potentially dangerous. Our study evaluated the safety of a protocol to discontinue the cervical collar in OBTPs based on CT scan alone” [43]. They all had to be moving all four extremities on admission and have a negative CT prior to collar removal; they were then (importantly) followed up for any developing signs of injury.

Third meta-analysis

Panczykowski et al. [5] recently published a meta-analysis from the Tomycz et al. group [42] and considered studies comparing CT with other imaging modalities. Of the 17 studies that were included 12 of them have been discussed in detail above [17, 21, 23, 29, 30, 34, 41, 42, 48, 53, 54, 60]. Sekula et al. did not look exclusively at obtunded patients (it is not clear how many were obtunded) and they did not compare patients with a negative MDCT with MRI [61]. Spiteri et al. [62] retrospectively analysed a clearance protocol from 1994 to 2004, which did not compare CT to MRI. The remaining three were published prior to 2004 and have been considered elsewhere [1]. The following papers [21, 23, 30, 34] have all been considered above comparing plain radiographs to CT. The remaining eight studies [17, 29, 41, 42, 48, 53, 54, 60], which can be used to answer the question: ‘Which screening combination should be used? MDCT alone or MDCT followed by MRI for the OBTP?’ were included in the meta-analysis by Schoenfeld [4] apart from Steigelman et al. [41] due to publication timing.

The paper essentially uses the same data as the meta-analysis by Schoenfeld et al. but concludes the opposite, stating that CT alone is sufficient to detect unstable cervical spine injuries.

Complications of ‘spinal precautions’: prolonged immobilisation and cervical collars (See Table 4)

The numerous serious complications of cervical collars and their ability to stabilise the neck have been reviewed elsewhere. The risks of prolonged immobilisation, beyond 48–72 h, are poorly appreciated and exceed those of a serious missed cervical spine injury [1, 2]. Whilst most clinicians are cognisant and have experience of the many complications of ‘spinal precautions’ and immobilisation with collar complications, there is paradoxically sparse literature available on this topic. The Cochrane group found little evidence that spinal immobilisation improves outcomes following unstable spinal injury. [63].

Table 4 Complications of prolonged immobilisation and spinal precautions

Stelfox et al. demonstrated that clearing the CS based on MDCT was associated with less delirium and less ventilator associated pneumonia (VAP), both of which have been associated with increased mortality in critically ill patients [64, 65]. To illustrate the complexities in decision-making VAP may have an attributable mortality (6%), in some series approaching that of the incidence of unstable spine injury itself (5%). They also proved that morbidity increased with increasing duration of collar application in line with previous research [66]. Hence, strategies with high sensitivity but low specificity will inevitably result in liberal collar application and prolonged immobilisation, the risks of which must be balanced against the risks of missed injuries [1, 67].

Dunham et al. [68] reviewed the risks of prolonged cervical collars and MRI scanning. They concluded that secondary brain injury was more likely than CS instability in OBTPs, and therefore advocated individualised risk assessments.

Clearing the thoracolumbar spine

Full evaluation of the entire spine (i.e. cervical and thoracolumbar) should be considered after identification of a fracture, because there is an estimated 16% incidence of non-contiguous spine fractures [69]. Isolated ligamentous injury is rare in the TL spine without a fracture due to the centripetal location and supporting musculature. There is therefore less controversy around screening protocols for the T/L spine.

Thoracolumbar (TL) spinal injuries occur in 2–3% of all blunt trauma victims, increasing to 10% in the OBTP population [15]. CT is superior to plain imaging for the detection of significant injury to the TL spine [15, 16, 7072]. Berry et al. [73] quote 100% sensitivity and 97% specificity for CT to detect TL fracture compared to 73% sensitivity and 100% specificity for plain radiographs. Furthermore, radiographs are inherently compromised in the lateral projection due to the shoulders and hence the cervicothoracic junction region tends to be evaluated and cleared on the anteroposterior view alone.

Where a CT of the chest/abdomen/pelvis (CT-CAP) has already been performed on admission then the use of reformats is superior to plain imaging [73]. In the rare circumstances where a CT-CAP has not been deemed necessary then we would argue that a CT of the entire spine should be performed. Some authors have advocated a ‘pan- CT’ approach to the ‘poly-trauma victim’ in recent years, although this has been called into question [74]. When Tilou et al. [75] attempted to reduce their scan rate they concluded that they would have missed 17% of injuries. Therefore, we would strongly advocate including routine CT to screen the thoracolumbar spine in obtunded blunt trauma victims, and not performing plain radiographs as previously suggested [1].

 

Limitations of the research

• The trials are almost exclusively retrospective cohort series of institutional protocols

• There seems a persistent reluctance to balance the costs and impacts of missed injuries against those of subjecting a broadly ‘normal’ population to prolonged immobilisation, transfers and imaging [43]

• CT technology has advanced and continues to advance since 2004, see abbreviations [20, 44]. At present only Brown et al. [44] have studied this. The advancement in scan technology is vast and is beyond the scope of this review. It has been reviewed elsewhere [20]. The ability of helical scanners to gather submillimeter section data has become more rapid with larger numbers of detection rays, 256-slice scanners now becoming commonplace. These systems cover 128 mm of anatomy with 0.5 mm slices in a single rotation of the gantry

•  The evidence for the timing of MRI is controversial. EAST state that MRI should be performed within 72 h. More recently this had been questioned and needs further research. Menaker et al. claim that the notion that MRI has low sensitivity after 48 h is ‘based on poorly documented anecdotes, poor image quality and no evidence that the delay between injury and imaging was responsible for the false negative MRI’ [47].

•  MRI detects injury with high sensitivity and significant rates of false positive studies, perhaps up to 40% [3, 39, 50, 79, 80].

•  The literature is now including increasing numbers of reviews and meta-analyses while the methodological quality of the included studies is poor. Accordingly many, analyses are reaching diametrically opposed conclusions based on the same original data

 What constitutes stability of the spine?

•  Absolute stability or instability of the spine is relatively easy to determine, however the gradations between these are less easy to quantify [7678].

•  Hogan et al. highlighted that many studies lacked outcome data about the stability of injuries of the patients that were kept in cervical collars for prolonged periods. There is some agreement in the literature regarding ligamentous columns and what constitutes instability. “The question of greater import is whether MDCT can depict unstable ligament injuries, that is, injuries involving two adjacent ligament support columns as defined by Denis [40].” [29] The limitations of the widely applied Denis model include it being based around thoracolumbar spine stability and predominantly based upon plain radiographs i.e. it was never validated with CT or MRI.

•  There is generally very poor information regarding the details of ligamentous injuries, i.e. the structures that were injured aside from the work by Steigelman et al. [41]. Many authors would argue that newer CT technology could detect these injuries [41, 43, 44, 53].

Resolving the controversy between clearing the spine on imaging: CT or CT and MRI?

As discussed, the on-going controversy largely revolves around the need for routine MRI to supplement CT, or whether CT alone is adequate to clear the cervical spine. One of the key obstacles in resolving the CT or CT + MRI controversy is that the performance, in particular, sensitivity for significant injuries, of CT alone has been so variously reported. The ‘miss rate’ for unstable injuries following a ‘normal CT’ (variously described as needing prolonged collar application, halo vest or surgical stabilisation) varies from almost 0% [59], <1% [60], 2.5% [68] to approximately 4.3% as we have described above, among whom approximately 0.29% would require surgical stabilisation. This must be balanced however against the false positive rate with MRI, perhaps up to 25–40% [50], which if applied routinely would require all OBTPs to undergo prolonged stabilisation (with associated complications), even though 95% have no actual injury. Furthermore, difficulties in standardised interpretation of imaging of any modality make exact rates of sensitivity and specificity difficult to define.

There are therefore two valid interpretations of the available literature and little prospect of resolving their disparate conclusions; namely it appears both modern CT and combined CT + MRI are acceptable ways of screening for spinal instability. The decision as to which modality to use will depend on the features outlined above when considering ‘screening’ but also:

  • Previous performance by individuals and the institution. It is essential centres audit their data because it is likely this data will be more informative than research in determining what the likely screening performance (sensitivity, specificity, etc.) is for a patient in any given institution.

  • It is likely that a strategy that revolves around CT alone will result in faster liberation from spinal precautions and associated complications and have a lower rate of false positive tests.

  • Conversely it is likely that a routine strategy of combined CT and MRI is likely to have the best sensitivity for injuries, including ligamentous injuries. This will be at the expense of an increased rate of false positive tests and prolonged application of spinal precautions, amongst a population where the actual injury rate approximates 5%.

  • The decision around which strategy to adopt is partly scientific but also influenced by previous events (e.g. missed injuries or complications of spinal immobilisation) at any given institution.

  • It is likely clinicians will continue to struggle to reach consensus on what constitutes an ‘acceptable’ rate of missed injuries or unnecessary spinal precautions. It is likely, therefore, that a dichotomous approach remains for the foreseeable future (Table 5).

    Table 5 Advantages and disadvantages of CT alone vs. CT followed by MRI

What recommendations for practice can be made?

Given the low level of higher quality evidence, considered in more detail above, we would make the following recommendations for practice based upon the GRADE system [81], the recommendations being ‘strong’ or ‘conditional’. (See Appendix).

 

 1. Document movement of limbs at first presentation; if intubated ensure robust handover from pre-hospital team (strong).

 2. Routine plain radiographs and dynamic flexion/extension views are out (strong).

 3. MRI if there is a positive neurological examination referable to the spinal cord (myelopathy) (strong).

 4. CT the entire (cervical, and thoracolumbar) spine in OBTPs using a modern MDCT (strong).

 5. Get images reported by a senior radiologist who is skilled in musculoskeletal/neuro-radiology (strong).

 6. Some centres may still feel an MRI is mandated after a normal CT and the consequences of this must be carefully considered (conditional).

 7. It remains acceptable to remove spinal precautions after CT and combined CT and MRI but there are clear implications to both approaches (conditional).

 8. Considering that only approximately 5% of OBTP have an actual injury whatever strategy is used, spinal precautions and the cervical collar should be removed as soon as is feasibly possible and mobilisation should be achieved (strong).

 9. Be vigilant for developing neurological signs despite a spine having been ‘cleared’ (strong)

Suggestions for future research and audit

As has been suggested above, it is as important for individual institutions to undergo rigorous audit rather than solely awaiting a definitive research study. The complexity of processes involved in screening for spinal injuries from scanner capability, image acquisition to interpretation and correlation of clinical features means that all centres must be able to demonstrate the typical performances of screening in their practice. Suggested end points for audit would be related to the recommendations for practice above including:

  • Accurate rates of sensitivity, specificity, and predicative values for any imaging undertaken for screening for spinal injuries.

  • Time to clearing the spine and rates of prolonged collar or halo vest application or surgical stabilisation.

  • The multidisciplinary team reviews all missed injuries.

  • Complications related to screening, e.g. during transfer, pressure sores, rates of ventilator associated pneumonia, outcomes from traumatic brain injury and intracranial pressure records.

Despite the challenges there is precedent in conducting high quality research of this nature in screening for spinal injuries [8, 82]. Many of the conclusions drawn from recent meta-analyses on this subject and older research called for good quality prospective trials of MDCT versus MRI in their respective abilities to detect unstable spinal injuries in the OBTP population [6, 29, 79]. We believe the study that would be most informative would be a prospective, multiple centre comparison of clearing the cervical spine in OBTPs by comparing MDCT versus combined MDCT and MRI. This study would need to be powered to detect a difference in detection rates at the lower end of the quoted ranges (i.e. approaching 0%) and given pre-existing clinician practice, prejudice or lack of equipoise this could be run most feasibly as a cluster randomised model. In addition to accurately defining the true performances of these imaging modalities such a large cohort could also include secondary tiers of research [83, 84], e.g. comparing different brands of cervical collars and pressure sore rates.

In the absence of such work comparative audit and centres publishing their series and cohorts, can do much to inform likely best practice. It is hoped that industry can become involved more energetically when one considers the return from installing and running modern scanning machines.

This review is principally limited by the quality of the studies, notably there have been no prospective randomised controlled trials comparing CT with MRI for the OBTP, and therefore our primary conclusion is that MDCT and combined MDCT and MRI are both acceptable approaches. The paucity of follow up data and indications for prolonged collar application along with a lack of information about decisions that occurred to keep a cervical collar on limit our ability to interpret the relevance of some of these studies.

Conclusions

From the data that we have presented in Tables 2 and 3 for all studies that compared CT with MRI for the OBTP group, where the patient had an MRI after a ‘normal’ CT scan. We have worked out a ‘worst case’ operative fixing rate of 0.29% and a prolonged collar application rate of 4.3% with a positive MRI scan rate of 7.5%. Many workers suggest modern CT can identify all significant injuries in their practice. The key issue clinicians need to consider is what is the screening performance (sensitivity, specificity and predicative values) of the imaging undertaken in their institution, and what rates do they conceptually find acceptable to work with. This must be considered in the context of a population where 95% do not have an actual injury, and failure to remove spinal precautions produces secondary morbidity and mortality that could potentially rival the rates of complications due to missed injuries. In common with all screening processes false positive and negative results will occur and have consequences. All clinicians must be vigilant to missed injuries and attempt to clear the spine of instability as soon as possible following injury.