Triage tools for detecting cervical spine injury in paediatric trauma patients

Background: Paediatric cervical spine injury (CSI) after blunt trauma is rare but can have severe consequences. Clinical decision rules (CDRs) have been developed to guide clinical decision-making, minimise unnecessary tests and associated risks, whilst detecting all significant CSIs. Several validated CDRs are used to guide imaging decision-making in adults following blunt trauma and clinical criteria have been proposed as possible paediatric-specific CDRs. Little information is known about their accuracy.

Objectives: To assess and compare the diagnostic accuracy of CDRs or sets of clinical criteria, alone or in comparison with each other, for the evaluation of CSI following blunt trauma in children.

Search methods: For this update, we searched CENTRAL, MEDLINE, Embase, and six other databases from 1 January 2015 to 13 December 2022. As we expanded the index test eligibility for this review update, we searched the excluded studies from the previous version of the review for eligibility. We contacted field experts to identify ongoing studies and studies potentially missed by the search. There were no language restrictions.

Selection criteria: We included cross-sectional or cohort designs (retrospective and prospective) and randomised controlled trials that compared the diagnostic accuracy of any CDR or clinical criteria compared with a reference standard for the evaluation of paediatric CSI following blunt trauma. We included studies evaluating one CDR or comparing two or more CDRs (directly and indirectly). We considered X-ray, computed tomography (CT) or magnetic resonance imaging (MRI) of the cervical spine, and clinical clearance/follow-up as adequate reference standards.

Data collection and analysis: Two review authors independently screened titles and abstracts for relevance, and carried out eligibility, data extraction and quality assessment. A third review author arbitrated. We extracted data on study design, participant characteristics, inclusion/exclusion criteria, index test, target condition, reference standard and data (diagnostic two-by-two tables) and calculated and plotted sensitivity and specificity on forest plots for visual examination of variation in test accuracy. We assessed methodological quality using the Quality Assessment of Diagnostic Accuracy Studies Version 2 tool. We graded the certainty of the evidence using the GRADE approach.

Main results: We included five studies with 21,379 enrolled participants, published between 2001 and 2021. Prevalence of CSI ranged from 0.5% to 1.85%. Seven CDRs were evaluated. Three studies reported on direct comparisons of CDRs. One study (973 participants) directly compared the accuracy of three index tests with the sensitivities of NEXUS, Canadian C-Spine Rule and the PECARN retrospective criteria being 1.00 (95% confidence interval (CI) 0.48 to 1.00), 1.00 (95% CI 0.48 to 1.00) and 1.00 (95% CI 0.48 to 1.00), respectively. The specificities were 0.56 (95% CI 0.53 to 0.59), 0.52 (95% CI 0.49 to 0.55) and 0.32 (95% CI 0.29 to 0.35), respectively (moderate-certainty evidence). One study (4091 participants) compared the accuracy of the PECARN retrospective criteria with the Leonard de novo model; the sensitivities were 0.91 (95% CI 0.81 to 0.96) and 0.92 (95% CI 0.83 to 0.97), respectively. The specificities were 0.46 (95% CI 0.44 to 0.47) and 0.50 (95% CI 0.49 to 0.52) (moderate- and low-certainty evidence, respectively). One study (270 participants) compared the accuracy of two NICE (National Institute for Health and Care Excellence) head injury guidelines; the sensitivity of the CG56 guideline was 1.00 (95% CI 0.48 to 1.00) compared to 1.00 (95% CI 0.48 to 1.00) with the CG176 guideline. The specificities were 0.46 (95% CI 0.40 to 0.52) and 0.07 (95% CI 0.04 to 0.11), respectively (very low-certainty evidence). Two additional studies were indirect comparison studies. One study (3065 participants) tested the accuracy of the NEXUS criteria; the sensitivity was 1.00 (95% CI 0.88 to 1.00) and specificity was 0.20 (95% CI 0.18 to 0.21) (low-certainty evidence). One retrospective study (12,537 participants) evaluated the PEDSPINE criteria and found a sensitivity of 0.93 (95% CI 0.78 to 0.99) and specificity of 0.70 (95% CI 0.69 to 0.72) (very low-certainty evidence). We did not pool data within the broader CDR categories or investigate heterogeneity due to the small quantity of data and the clinical heterogeneity of studies. Two studies were at high risk of bias. We identified two studies that are awaiting classification pending further information and two ongoing studies.

Authors' conclusions: There is insufficient evidence to determine the diagnostic test accuracy of CDRs to detect CSIs in children following blunt trauma, particularly for children under eight years of age. Although most studies had a high sensitivity, this was often achieved at the expense of low specificity and should be interpreted with caution due to a small number of CSIs and wide CIs. Well-designed, large studies are required to evaluate the accuracy of CDRs for the cervical spine clearance in children following blunt trauma, ideally in direct comparison with each other.