Pediatric cervical spine injuries (CSI) are rare (1-2%) after blunt trauma. Decision rules to identify adults at low risk of cervical spine injury have been developed (NEXUS criteria (Hoffman, NEJM 2000, PMID: 10891516), Canadian C-spine rule (Stiell, JAMA 2001, PMID: 11597285)). A pediatric rule was developed as a subset of the Nexus study (Vicellio, Pediatrics 2001, PMID: 11483830). The pediatric NEXUS included only 30 patients less than 18 years of age with cervical spine injuries and few in those at the lower age range (2-9 years: 4 CSI, < 2 years: 0 CSI) While the sensitivity of the rule was 100%, the lower limit of the 95% confidence interval was 88% due to its small sample size.
The PECARN Network previously conducted a case-control study to derive a pediatric cervical spine clinical decision rule (Leonard, Ann Emerg Med. 2011, PMID: 21035905). The study identified 8 predictors of pediatric cervical spine injury. These included: predisposing conditions, diving, high risk motor vehicle collision, complaint of neck pain, focal neurologic deficit, altered mental status, substantial torso injury and torticollis. The rule had a sensitivity of 98% 95% CI (96, 99%) and specificity of 26% 95% CI (23, 29%) for cervical spine injury. The rule has not been validated.
In children less than 18 years of age who sustain blunt trauma, are clinical signs and symptoms accurate in identifying those at low risk of cervical spine injury in order to potentially forgo cervical spine imaging?
Observational: Prospective cohort
- < 18 years with blunt trauma
- Transported from the scene by EMS
- Present to the ED either directly via EMS or in transfer from another institution
- Underwent a trauma evaluation with or without cervical spine imaging
- Penetrating trauma
- Legal guardian with a significant language barrier
- Transferred from the study site for definitive care
- n=3, Level I Trauma Children’s Hospitals (U.S.), 3/2014-11/2016
Included factors with biologic or anatomic plausibility and good inter-rater reliability: Mechanism of injury/injury biomechanics variables and patient history, signs and symptoms variables. (See appendix in the complete study review)
- Cervical Spine Injury (Occiput → C7)
- Vertebral fracture
- Ligamentous injury (including ligaments attached to T1)
- Intraspinal hemorrhage
- Spinal cord injury: MRI or spinal cord injury without radiographic abnormality
ED Imaging Performed
Review of c-spine imaging reports and spine surgeon consultation notes if applicable
No ED Imaging Performed
Medical record review at 28 days for subsequent imaging. If no imaging obtained then phone follow-up at 21-28 days after initial ED visit
- Rule characteristics
- Potential reduction in XRAY utilization
- Cervical Spine Injury: 1.8% (74/4,091)
- Age < 8 years, 39.3% (1,608/4,091), CSI: 1.4% (23/1,608)
- Imaging obtained: 78.2%
|INDEPENDENT PREDICTORS OF CSI: REGRESSION ANALYSIS|
|PREDICTOR||PECARN MODEL1||DE NOVO MODEL1|
|Mechanism: High Risk MVC||1.58 (0.63, 3.97)|
|Mechanism: Diving||17.60 (5.60, 55.32)||9.16 (2.41, 34.83)|
|Mechanism: Axial Load||2.51 (1.22, 5.16)|
|History: Predisposing Condition||2.02 (0.27, 15.10)|
|History: Neck Pain2||1.65 (1.04, 2.62)||2.87 (1.50, 5.48)|
|History: Inability to Move Neck2||3.77 (2.00, 7.12)||3.51 (1.72, 7.17)|
|Exam: Altered Mental Status||5.67 (3.54, 9.09)||2.90 (1.37, 6.12)|
|Exam: Intubated||10.71 (4.43, 25.91)|
|Exam: Limited Neck Range of Motion||1.85 (0.88, 3.90)|
|Exam: Substantial Torso Injury||2.61 (1.24, 5.53)|
|Exam: Respiratory Distress||5.84 (1.56, 21.88)|
|Exam: Focal Neurologic Deficits||2.62 (1.04, 6.63)|
|1. Adjusted Odds Ratio (95% Confidence Interval), 2. “Neck pain” and “Inability to move neck” were assessed separately. These were combined as Torticollis in the derivation of the PECARN case-control study|
|Italics = Statistically Significant, Bold = Not Statistically Significant|
|PECARN RULE1||DE NOVO RULE|
|Sensitivity||90.54% (83.87, 97.21%)||91.88% (85.7, 98.11%)|
|Specificity||45.58% (44.04, 47.12%)||50.26% (48.72, 51.81%)|
|Predictive Value (+) Test||2.97% (2.27, 3.68%)||3.29% (2.52. 4.06%)|
|Predictive Value (-) Test||99.62% (99.34, 99.90%)||99.71% (99.47, 99.94%)|
|Likelihood Ratio (+) Test||1.66 (1.54, 1.80)||1.85 (1.71, 1.99)|
|Likelihood Ratio (-) Test||0.21 (0.10, 0.42)||0.16 (0.07, 0.35)|
|1 Any 9 factors in the PECARN rule including the 3 that were not statistically significant|
Potential Reduction in Imaging
Imaging would potentially be reduced from a baseline imaging rate of 78.2% by:
- 22.1% (78.2% – 55.1% = 22.1%) for the PECARN
- 26.6% (78.2% – 51.6% = 26.6%) for the De Novo rule
- Extensive list of candidate factors with biologic or anatomic plausibility and good inter-rater reliability included in the derivation process. However, good inter-rater reliability was not defined and kappa statistics were not provided.
- Prospective validation of the previous derivation of the PECARN c-spine rule that was a Case-Control study allows for calculation of predictive values and likelihood ratios.
- > 2.5 times the number of patients with CSI compared to the Pediatric Nexus study
- For transfer patients, data was collected prior to imaging interpretation by the radiologist at the study institution but clinicians may have been aware of imaging results from the transferring institution. 42% (31/74) of those with cervical spine injury were transfers. However, when a subgroup analysis that excluded transfer patients was conducted, the test characteristics for both models remained similar.
- The inclusion of transferred patients likely improves the generalizability of the study’s results.
- The proportion of patients with each of the significant predictors was not presented. The authors did note that predisposing conditions occurred in < 1% of patients
- Only 74 patients with CSI included (n=23 < 8 years of age)
- The De Novo rule is a level IV rule requiring further validation before it can be applied clinically
- The PECARN rule can is a level IV rule (a re-derivation with different predictors) requiring further validation before it can be applied clinically
- The PECARN rule had a lower sensitivity and wider confidence intervals (90.54%, 95% CI (83.87, 97.21%) then the original case-control derivation (Sensitivity (98% 95% CI (96, 99%)).
- Rule characteristics for c-spine injury requiring neurosurgical intervention were not presented (as in the original PECARN derivation)
“In this prospective cohort of children with blunt trauma, we confirmed that there are risk factors with good test accuracy in identifying cervical spine injury. We also demonstrated that incorporating these risk factors into a clinical prediction rule has the potential to substantially reduce cervical spine imaging during trauma evaluation of children. A future, adequately powered prospective observational study aimed at using these risk factors to construct a definitive pediatric cervical spine injury prediction rule is warranted.”
This is a pilot study in one of the PECARN network nodes that will be further investigated in the larger PECARN network. The study demonstrated the use of the rule could potentially decrease imaging usage by 20-25% at the expense of rarely missing patients with cervical spine injury (8-10% of those with CSI were not identified by the rules). However, none of the missed patients required a surgical intervention.
Potential Impact To Current Practice
The study’s results should not be applied clinically at present. I eagerly await the follow-up study in the entire PECARN network to validate the two models.