Few objects are as visually synonymous with trauma care as the semi-rigid cervical collar. It appears in every training video, every simulation, every stock photograph of a resuscitation bay. It is applied by paramedics within minutes of arrival at scene, tolerated by patients for hours in emergency departments, and in some populations worn continuously for twelve weeks. It is also, on the balance of the published evidence, an intervention whose benefit has never been demonstrated and whose harms are increasingly well characterised.
That tension is not new. What has changed in 2026 is the weight of the material addressing it. In May, the National Institute for Health and Care Research published the full report of the Duration of External Neck Stabilisation (DENS) trial, the only randomised comparison of collar versus no collar in older or frail adults with odontoid fracture. In July, a comprehensive narrative review in Orthopedic Research and Reviews synthesised the guideline landscape and proposed a framework for selective use. In April, the Scandinavian Journal of Trauma, Resuscitation and Emergency Medicine published an essay arguing that the collar persists not because of evidence but in spite of it. Read together, these papers suggest that the question has quietly shifted from whether the collar works to why we are still applying it by default.
A device that arrived without evidence
The modern practice of spinal immobilisation traces to a 1971 position from the American Academy of Orthopaedic Surgeons recommending a cervical collar and long spine board for patients with a suspected spinal injury, selected on mechanism alone. That recommendation was consensus, not trial evidence. It was subsequently absorbed into Advanced Trauma Life Support and Prehospital Trauma Life Support teaching, and from there into ambulance service protocols worldwide, acquiring the authority of universality without ever acquiring an evidence base.
The underlying logic was reasonable enough. Traumatic spinal cord injury is devastating and, in principle, secondary injury during extrication and transport might be preventable. But the epidemiology has always sat awkwardly with universal application. Spinal injury occurs in somewhere between one and seven per cent of non-penetrating trauma patients, and the subset in whom mechanical instability is such that ordinary handling would cause neurological deterioration is smaller still. Applying a collar to every patient in order to protect that subset means accepting a very large denominator.
The first serious empirical challenge came in 1998, when Hauswald and colleagues compared neurological outcomes between a New Mexico cohort in which spinal injury patients were routinely immobilised and a Malaysian cohort in which they were not. They found no protective effect of immobilisation. Nearly three decades later, the situation has not fundamentally improved.
What the blunt trauma evidence actually shows
The most rigorous recent synthesis is a systematic review published in PLOS ONE in 2024 by Pandor and colleagues, which asked whether different immobilisation strategies - full immobilisation, movement minimisation, or none - affect outcomes after blunt trauma. Six comparative observational studies met inclusion criteria. Methodological quality was variable, with most carrying serious or critical risk of bias. The reviewers found no clear evidence that immobilisation prevented neurological deterioration, spinal injury or death compared with no immobilisation, while consistently identifying increased pain, discomfort and anatomical complications among immobilised patients.
A scoping review conducted for the International Liaison Committee on Resuscitation and published in 2025 reached a similar impasse from a different angle. Of sixty-six included studies, forty-seven were experimental, and three-quarters of those were performed on uninjured healthy volunteers measuring range of cervical motion. This is the central methodological problem of the field: the evidence base is dominated by biomechanical studies of intact spines, which can tell us how much a collar restricts movement but nothing about whether restricting movement changes outcomes in an injured spine. Patient-centred endpoints are largely absent, and neurological deterioration attributable to prehospital handling is so rare that it has never been reliably measured.
It is worth stating the epistemic position precisely, because it is frequently overstated in both directions. There is no high-quality evidence that collars improve outcomes in blunt trauma. That is not the same as evidence that they do not. For a rare, catastrophic and plausibly preventable outcome, the absence of a demonstrated benefit in six confounded observational studies is a weak basis for abolition. The honest reading is that we do not know, and that we have been running a universal intervention for fifty years without generating the data to find out.
Penetrating trauma: the clearest signal in the field
Where the evidence does point in one direction is penetrating injury. In 2010, Haut and colleagues analysed 45,284 penetrating trauma patients from the National Trauma Data Bank, of whom 4.3 per cent underwent prehospital spinal immobilisation. Unadjusted mortality was twice as high in immobilised patients (14.7 per cent versus 7.2 per cent) with an adjusted odds ratio of death of 2.06 (95% CI 1.35 to 3.13). Only thirty patients in the entire cohort, 0.01 per cent, had an incomplete spinal cord injury and underwent operative fixation. The authors calculated a number needed to treat of 1,032 to potentially benefit one patient, against a number needed to harm of 66 to potentially contribute to one death.
Confounding by indication is an obvious concern: patients who look sicker may be more likely to receive additional prehospital interventions. But the direction and magnitude of the effect, corroborated by a meta-analysis within the 2018 Eastern Association for the Surgery of Trauma practice management guideline showing a relative risk of mortality of 2.4 (95% CI 1.07 to 5.41), was sufficient for EAST to recommend against routine spinal immobilisation in adult penetrating trauma. The mechanism is not mysterious. In a patient exsanguinating from a stab wound to the chest, a collar obstructs airway access, delays transfer and treats a threat that is essentially hypothetical while the actual threat proceeds unimpeded.
The other side
Whatever one concludes about benefit, the collar is not a neutral intervention, and the harms have been quantified with more precision than the benefits. A 2022 systematic review and meta-analysis by Brannigan and colleagues in Global Spine Journal examined adverse events from prolonged hard collar immobilisation and identified pressure ulceration, dysphagia and raised intracranial pressure as the dominant complications, with a pooled pressure ulcer prevalence of approximately seven per cent.
Each of these matters in a specific population. Raised intracranial pressure, mediated through impaired jugular venous outflow, is precisely the wrong physiological insult in the patient with traumatic brain injury - the patient in whom a collar is most likely to be left in place because the examination is unreliable. Dysphagia and respiratory compromise fall disproportionately on older adults; a retrospective cohort of geriatric patients with traumatic brain injury and cervical spine injury found collar placement associated with higher rates of dysphagia and respiratory failure. Semi-rigid collars reduce mouth opening and worsen intubating conditions, which is why emergency airway management routinely requires removal of the anterior portion with manual in-line stabilisation - an admission, in effect, that the device and the priority are in conflict. In frail patients, prolonged collar use contributes to immobility, delirium, poor sleep and reduced participation in rehabilitation.
DENS: a UK trial
The DENS trial addressed a narrower but clinically substantial question. Odontoid fractures are among the commonest cervical injuries in older adults, typically following a fall from standing. The Trauma Audit and Research Network identifies at least 1,700 each year in England and Wales, some 85 per cent in people over 65. One-year mortality is 20 to 50 per cent, comparable to or higher than hip fracture. In the United Kingdom, 85 to 90 per cent are managed non-operatively, most in a hard collar worn twenty-four hours a day for six to twelve weeks. Yet bony fusion occurs in only 20 to 80 per cent of these patients, stable fibrous non-union in 60 to 80 per cent, and fibrous non-union is an acceptable outcome with comparable pain, function and mortality.
The trial randomised older or frail adults with a new odontoid fracture to early collar removal or standard management in a hard collar for twelve weeks, with EQ-5D-5L quality of life at twelve weeks as the primary outcome. The target was 887 participants. It recruited 138 (15.6 per cent of target) before being terminated for slow recruitment. Mean age was 82.3 years; 72.5 per cent of fractures were type II.
On intention-to-treat analysis, adjusted mean EQ-5D-5L was 0.650 in the early removal group and 0.667 in the hard collar group, an adjusted mean difference of −0.017 (95% CI −0.103 to 0.068, p = 0.836). Neck Disability Index scores at twelve weeks showed no difference. Subgroup analyses by fracture type, frailty and age were all non-significant. There were numerically more deaths in the early removal arm (10 versus 5 at six months; odds ratio 1.92, 95% CI 0.61 to 6.04, p = 0.266), none attributed to the fracture or its management, and more serious adverse events (13.3 versus 5.4 per cent), largely unrelated to allocation. One episode of fracture displacement in the early removal arm was classed as possibly related.
The most revealing findings were not the outcome data. Mean collar wear was 55.5 days in the standard arm against 0.2 days in the intervention arm, but 15.6 per cent of collar-allocated patients crossed over, and by twenty-eight days after randomisation 80 of the recruited patients (61.5 per cent) were not wearing a collar despite equal allocation. Of 1,185 patients screened, 102 were judged unsuitable for management without a collar by the assessing spinal surgeon, on subjective criteria, because no evidence-based guidance exists to inform that judgement. The embedded qualitative study found that healthcare professionals had lost equipoise in both directions: some were uncomfortable randomising patients to no collar, while others were distressed by the collar itself and reluctant to randomise patients into it. Ward staff continued to tell patients they risked paralysis if the collar came off, which made subsequent recruitment conversations untenable. The authors concluded that further randomised evaluation is unlikely to be feasible because equipoise no longer exists.
Two qualitative findings deserve emphasis. Patients in the collar arm reported distinct harms their counterparts did not: having food cut up or being fed, difficulty descending stairs because they could not look down, problems with washing and grooming, and, strikingly, difficulty wearing hearing aids, with consequent isolation. In the health economic analysis, collar patients reported significantly greater need for family assistance at every time point. Against this, some patients described the collar as making them feel safe.
The case for not abandoning the collar
An honest account must set out the counter-argument properly. DENS recruited under a sixth of its target and is straightforwardly underpowered; its confidence interval around the primary outcome comfortably includes differences that would be clinically meaningful. High crossover diluted the contrast between arms. The excess of deaths and serious adverse events in the early removal group was not statistically significant but was not zero either, and a trial this size could not have detected a modest safety signal. Crucially, the population studied was pre-selected by spinal surgeons who excluded patients they judged unsafe for collar-free management, so the results do not license removing collars from every odontoid fracture.
In the acute setting the argument is different again. The blunt trauma evidence is observational, heavily confounded, and addresses an outcome rare enough that no realistic trial will resolve it. The collar also performs a communication function that is easy to dismiss and hard to replace: it flags to every subsequent clinician, porter and radiographer that a question about the spine remains unanswered. DENS clinicians raised precisely this in reverse - several were reluctant to randomise to early removal because collar patients received weekly district nurse visits for pad changes and skin checks, and removing the collar removed the contact. And in the obtunded patient after negative high-quality CT, genuine uncertainty persists about whether adjunct MRI is needed, with recent meta-analytic data showing MRI does detect additional injuries even if treatment-changing findings are uncommon.
From "collar or no collar" to "which patient, which phase"
The July 2026 review in Orthopedic Research and Reviews reframes the question usefully. Rather than asking whether collars work, it separates four sequential decisions: whether to apply, whether to clear, whether to continue, and whether to de-escalate. Each rests on different evidence and different populations, and conflating them is the source of much of the confusion.
Application should be selective, reserved for blunt trauma patients with unreliable examination, neurological deficit, midline tenderness, deformity or high-risk mechanism, and generally avoided in isolated penetrating trauma. Clearance in the awake, evaluable patient should proceed through validated decision rules - NEXUS or the Canadian C-Spine Rule - with high-quality CT where imaging is indicated; persistent neck pain after a negative CT in a neurologically intact patient does not mandate prolonged rigid immobilisation. In the obtunded adult, a negative high-quality CT may support removal but should not be treated as automatic authorisation, with CT adequacy, neurological findings and radio-clinical concordance all feeding the decision. Once a fracture is confirmed, the collar stops being a diagnostic holding measure and becomes a treatment, to be justified on stability, frailty, tolerance and treatment goals rather than on habit or a fixed twelve-week ritual. Danish evidence- and consensus-based guidance has already issued a weak recommendation against routine rigid collar use in ABCDE-stable adult trauma patients.
Why the evidence has not been enough
If the case for selective use is this well established, the persistence of routine collaring requires explanation. An essay published in April 2026 in the Scandinavian Journal of Trauma, Resuscitation and Emergency Medicine argues that the answer lies outside the evidence. Fear of neck injury, it suggests, has roots that predate clinical practice, and threat associations attached to survival-relevant anatomy are acquired quickly and extinguished slowly. On this account the collar functions partly as an anxiety-management device for clinician and patient alike, and reform will require engaging with the historical, linguistic and cognitive determinants of that fear rather than simply publishing better trials. The DENS qualitative data support the argument: what obstructed recruitment was not disagreement about evidence but a ward culture in which patients were told they would be paralysed if the collar came off.
What this means for the ATLS candidate
The eleventh edition of ATLS, launched in July 2025, is the version you will be examined on, and it has moved in the direction this literature indicates without going as far as the literature might permit. The primary survey is now xABCDE, with control of exsanguinating haemorrhage formally preceding airway. Head and spine are consolidated under Disability. The approach to spine motion restriction has been revised, and the language throughout favours selective, proportionate motion restriction over universal immobilisation - note that the manual speaks of restriction, not immobilisation, and the distinction is examinable.
Expect questions to probe four areas. First, penetrating trauma: routine spinal motion restriction is not indicated in isolated penetrating injury without neurological deficit, and a question that offers "apply a cervical collar" as an option in a haemodynamically unstable stabbing is testing whether you will prioritise a hypothetical threat over an actual one. Second, the harms: raised intracranial pressure, pressure ulceration, dysphagia, aspiration risk and impaired airway access are all fair game, particularly the interaction between collar and traumatic brain injury. Third, clearance: know NEXUS and the Canadian C-Spine Rule cold, including which patients each was validated in, and know that removing the anterior portion of the collar with manual in-line stabilisation is the correct answer for airway management in a patient with a potential cervical injury. Fourth, the older patient: the expanded chapter on trauma in the older adult, combined with the geriatric odontoid literature, makes frailty-aware decision-making a plausible target.
One strategic caution. ATLS examinations assess the manual, not the most recent publication. Where the two diverge, answer the manual and save the argument for the ward round. But candidates who understand why the guidance has moved - that the collar was adopted on consensus, that its benefits have never been demonstrated, that its harms have been quantified, and that a British randomised trial could not be completed because clinicians had already changed their minds - will reason more reliably through the scenarios where the algorithm runs out.
References
1. Brennan PM, Adamczyk M, Closs M, et al. Duration of external neck stabilisation in the management of older and frail patients with a new odontoid fracture: the DENS RCT. Health Technology Assessment 2026. doi:10.3310/GJPB0728. https://www.ncbi.nlm.nih.gov/books/NBK622230/
2. Woodfield J, Edlmann E, Black PL, et al. Duration of External Neck Stabilisation (DENS) following odontoid fracture in older or frail adults: protocol for a randomised controlled trial of collar versus no collar. BMJ Open 2022;12(7):e057753. doi:10.1136/bmjopen-2021-057753
3. Besharaty S, Sheikhalishahi S, Jafarinasab H. Cervical Collar in Adult Trauma: Evidence, Controversies, and Clinical Decision-Making. Orthopedic Research and Reviews 2026;18:1–20. doi:10.2147/ORR.S621602
4. Nutbeam T. Beyond the evidence: the cervical collar as cultural artefact. Scandinavian Journal of Trauma, Resuscitation and Emergency Medicine 2026;34:74. doi:10.1186/s13049-026-01600-w
5. Pandor A, Essat M, Sutton A, et al. Cervical spine immobilisation following blunt trauma in pre-hospital and emergency care: a systematic review. PLOS ONE 2024;19(4):e0302127. doi:10.1371/journal.pone.0302127
6. Haut ER, Kalish BT, Efron DT, et al. Spine immobilization in penetrating trauma: more harm than good? Journal of Trauma 2010;68(1):115–120. doi:10.1097/TA.0b013e3181c9ee58
7. Velopulos CG, Shihab HM, Lottenberg L, et al. Prehospital spine immobilization/spinal motion restriction in penetrating trauma: a practice management guideline from the Eastern Association for the Surgery of Trauma (EAST). Journal of Trauma and Acute Care Surgery 2018;84(5):736–744. doi:10.1097/TA.0000000000001764
8. Brannigan JFM, Dohle E, Critchley GR, et al. Adverse events relating to prolonged hard collar immobilisation: a systematic review and meta-analysis. Global Spine Journal 2022;12(8):1968–1978. doi:10.1177/21925682221087194
9. Jarvis S, Sater A, Gordon J, et al. Cervical collars and dysphagia among geriatric TBIs and cervical spine injuries: a retrospective cohort study. Journal for Healthcare Quality 2023;45(3):160–168. doi:10.1097/JHQ.0000000000000379
10. Maschmann C, Jeppesen E, Rubin MA, Barfod C. New clinical guidelines on the spinal stabilisation of adult trauma patients - consensus and evidence based. Scandinavian Journal of Trauma, Resuscitation and Emergency Medicine 2019;27(1):77. doi:10.1186/s13049-019-0655-x
11. Hauswald M, Ong G, Tandberg D, Omar Z. Out-of-hospital spinal immobilization: its effect on neurologic injury. Academic Emergency Medicine 1998;5(3):214–219. doi:10.1111/j.1553-2712.1998.tb02615.x
12. Hoffman JR, Mower WR, Wolfson AB, Todd KH, Zucker MI. Validity of a set of clinical criteria to rule out injury to the cervical spine in patients with blunt trauma (NEXUS). New England Journal of Medicine 2000;343(2):94–99. doi:10.1056/NEJM200007133430203
13. Stiell IG, Wells GA, Vandemheen KL, et al. The Canadian C-Spine Rule for radiography in alert and stable trauma patients. JAMA 2001;286(15):1841–1848. doi:10.1001/jama.286.15.1841
14. Patel MB, Humble SS, Cullinane DC, et al. Cervical spine collar clearance in the obtunded adult blunt trauma patient: a systematic review and practice management guideline from EAST. Journal of Trauma and Acute Care Surgery 2015;78(2):430–441. doi:10.1097/TA.0000000000000503
15. Huybregts JGJ, Polak SB, Jacobs WC, et al. Surgical versus conservative treatment for odontoid fractures in older people: an international prospective comparative study. Age and Ageing 2024;53(8):afae189. doi:10.1093/ageing/afae189
16. American College of Surgeons. ATLS 11: Advanced Trauma Life Support, 11th edition. Chicago: American College of Surgeons; 2025. https://www.facs.org/quality-programs/trauma/education/advanced-trauma-life-support/atls-11/
17. Milne WK, Leeper R. SGEM Xtra: This One Goes to 11 - ATLS 11th Edition. The Skeptics’ Guide to Emergency Medicine, 28 March 2026. https://thesgem.com/2026/03/sgem-xtra-this-one-goes-to-11-atls-11th-edition/
