Evaluating Methodological Quality in Pediatric C-Spine Injury Detection Studies

Reference: Tavender E, et al. Triage tools for detecting cervical spine injury in paediatric trauma patients. Cochrane Database Syst Rev. 2024 Date: May 29, 2024 Guest Skeptic: Dr. Caleb Ward is a pediatric emergency medicine attending and Associate Professor of Pediatrics and Emergency Medicine at Children’s National Hospital and The George Washington School of Medicine and Health Sciences in Washington, DC. His research focuses on the pre-hospital care of children by EMS. He is the principal investigator for EMSC State Partnership in Washington, DC and is also involved in various multi-center EMS studies with the Pediatric Emergency Care Applied Research Network (PECARN). Dr. Caleb Ward Case: A 4-year-old boy is brought to the emergency department (ED) by Emergency Medical Services (EMS) after falling from a tall tree. The fall was witnessed by his family. They tell you that he is going through a Spiderman phase and tries to climb everything. They saw him slip off the tree and landed in the grass below. He did not have any loss of consciousness. EMS placed him in a C-collar and brought him to you. On examination, you only notice a few scrapes, but he is cradling his left arm and complaining that it hurts. There looks to be an obvious deformity of his forearm. The family members ask you, “he seems uncomfortable in the neck collar, can we remove it? Do you think he injured his neck?” Background: Pediatric cervical spine injuries (CSI), though rare, can have devastating consequences if missed. Imaging studies include X-rays, CT scans, and MRIs. Typically, we see the use of X-ray or CT more often in the acute setting. The downside of these methods is exposing children to radiation. Clinical Decision Rules (CDRs) have been developed to help guide decision-making and minimize unnecessary tests and imaging while detecting significant injuries. Because there is no standardized process for identifying children with CSI after blunt trauma, practice varies based on clinician, institution, location, and available resources. We have covered some of these CDRs in pediatrics on the SGEM before: Ankle and Knee Injuries (SGEM#3, SGEM#5) Trauma (SGEM#127) Appendicitis (SGEM#155) Head Trauma (SGEM #412, SGEM #225) Febrile Infants (SGEM#171, SGEM #296) While there are CDRs for cervical spine injury in adults like the Canadian C-spine Rule and NEXUS criteria for C-spine imaging, we do not have a dedicated, accurate CDR for pediatric patients. Clinical Question: Which triage tools or Clinical Decision Rules (CDRs) are most effective for detecting cervical spine injuries in pediatric trauma patients?  Reference: Tavender E, et al. Triage tools for detecting cervical spine injury in paediatric trauma patients. Cochrane Database Syst Rev. 2024. Population: Children (aged 0 to <18 years) who underwent blunt trauma evaluation in emergency departments. (ED) Excluded: Patients with previous cervical spine surgery or congenital cervical spine anomalies Intervention: Application of various CDRs or sets of clinical criteria to evaluate the presence of cervical spine injuries following blunt trauma. Comparison: The CDRs were compared with each other and with reference standards like X-ray, CT, MRI, or clinical clearance/follow-up in low-risk children. Outcome: The primary outcome of interest was the diagnostic accuracy of the CDRs, specifically their sensitivity and specificity in detecting cervical spine injuries. Trial: Systematic review 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.” Quality Checklist for Systematic Review Diagnostic Studies: The diagnostic question is clinically relevant with an established criterion standard. Yes The search for studies was detailed and exhaustive. Yes The methodological quality of primary studies were assessed for common forms of diagnostic research bias. Yes The assessment of studies was reproducible. Yes There was low heterogeneity for estimates of sensitivity or specificity. Unsure The summary diagnostic accuracy is sufficiently precise to improve upon existing clinical decision-making models. No Were there any declared financial conflicts of interest? No financial conflicts of interest Results: The five included studies reported a total of 21,379 children, with age ranges from less than 3 years to under 18 years. Studies were conducted in various countries including the USA, UK, Australia, Canada, and Brazil. The median prevalence of CSI was 0.98% with IQR of 0.5% to 1.85% CSI was defined as fracture, dislocation, ligamentous injury or spinal cord injury involving the cervical region. The reference standard was X-ray, CT, MRI. In low-risk children who did not undergo imaging, follow-up (clinical evaluation in the ED) served as the reference standard. They also wanted information about additional follow-up after the ED in this population to avoid misclassification. A total of 7 CDRs were included: PECARN retrospective criteria Canadian C-spine Rule NEXUS Leonard de novo National Institute for Health and Care Excellence (NICE) guideline CG56 National Institute for Health and Care Excellence (NICE) guideline CG176 PEDSPINE Key Results: Pediatric cervical spine injuries are rare and the existing CDRs have high sensitivity but low specificity. There is insufficient evidence to recommend one CDR over others for detecting CSIs in children. 1. Criterion Standard for CSI: One thing to flag up front is that the criterion standard used in the review and all of the papers included is just any CSI (fracture, dislocation, ligamentous injury or spinal cord injury - whether or not detectable on conventional imaging). This is not like head trauma CDRs where we talk about clinically significant intracranial injuries, rather than focusing on picking up a nondisplaced linear skull fracture. We are skeptical that missing a tiny spinous process fracture or ligamentous strain is the same as missing a complete transection of a ligament with associated subluxation and cord impingement. 2. Sensitivity and Specificity: Sensitivity and specificity are often reported statistical terms. But what does this mean for our patients and practice when it comes to clinical decision rules and CSI? Most decision rules had high sensitivity which means that they did accurately identify children who had CSI. However, the specificity was not great. This means that the CDRs cannot accurately identify kids who do not have CSI. While this may be good in that we are not missing children with CSI, a CDR with low specificity can end up INCREASING the proportion of children who get imaging which potentially means that a large proportion of children are getting unnecessary radiation from imaging. It is also important to note that even though the point estimates of sensitivities were high, the 95% confidence intervals for many of these studies were quite wide. The bottom range of these confidence intervals ranged from 0.48-0.88, reflecting the lack of true positive cases. 3. Bias: Biases are things that may systematically steer us away from the “truth.” They did a thorough exploration of bias in four areas: 1) Patient selection 2) Index test 3) Reference standard 4) Flow and timing. Under patient selection, two studies were considered at high risk of bias because they were retrospective studies looking at medical records and trauma registries. Regarding the index test, two studies applied the index test retrospectively and were rated as high risk of bias. The other three had an unclear risk of bias because there was no information about the masking of imaging results before the interpretation of the index test. The authors mention that the development of a CDR is a three-step process: derivation, validation, and impact analysis. Two studies that derived a new CDR were high risk because of no validation. The reference standard varied quite a bit amongst the studies. Two studies had some kind of radiographic imaging in all children. One study had imaging and clinical clearance in ED. Two studies had either imaging or clinical clearance in ED with follow-up phone calls. One study had a low risk of bias. Three studies had an unclear risk of bias. One study had a high risk of bias (that is because there was no follow-up on discharged children). For flow and timing (the authors had this mean “Did all patients receive the same reference standard.”), one study had a high risk of bias. 4. Clinical Judgement: Whenever we read about clinical decision rules or tools now, we have to wonder, “How do any of these CDRs compare to clinician judgement?” Some CDRs have criteria that one would deem almost kind of obvious. For example, if someone has a focal peripheral neurologic deficit and neck pain, we don’t need a CDR to tell us that they probably require imaging. Is it possible that our own clinical gestalt is better than any tool/rule out there? Our colleague, Justin Morgenstern, of First10EM has a thought-provoking post titled, “Clinical Decision Rules are Ruining Medicine.” We  would love to see more studies on CDRs compare them with clinician judgement. Some of those studies have been done, and we must stay humble. Leonard’s team showed that for her cohort of kids in the 2009 study,