• 
• 

Three Part Question

In [children who have sustained a mild or minor head injury with a GCS of 14-15] do [clinical findings] predict [intra-cranial injury on computerised tomography]?

Clinical Scenario

A 6-year-old girl presents to the emergency department with a teacher. She was missing at the end of the lunch break and was found dazed at the base of a climbing frame, and can’t remember what happened. Her examination is normal other than a scalp haematoma, with no evidence of an underlying skull fracture. You are aware of the adverse effects of radiation on the developing brain, but wonder if the history and clinical findings warrant performing a head CT despite the potential risks.

Search Strategy

The previous BestBet (Andrew Munro, Ian Maconochie 2008, search until October 2007) was used as a starting point and the same search strategy used from October 2007-November 2011.
Medline and Embase: 2007 – November 2011.
[(head AND trauma) OR exp brain injuries OR exp craniocerebral trauma OR exp head injuries, closed OR (head AND injur*)] AND [(exp adolescent OR exp child OR exp child, preschool OR exp child, hospitalized OR exp child, institutionalized, OR exp infant OR exp pediatrics) OR (pediatric* OR paediatric* OR adolescen*)] AND [(exp tomography OR exp tomography scanners, x-ray computed OR exp tomography, x-ray computed) OR (ct AND scan)] AND (exp prospective studies OR prospectiv*). Limited to [Abstracts and Humans and (Age Groups All Child 0 to 18 years)].

Search Outcome

The 2008 search found 347 papers, ten of which were considered relevant and of sufficient quality to address the question. In 2011, 235 additional papers were found which had been published since the previous update. Eight papers were considered relevant and were available in the English language. All eighteen are summarised in the table.
In addition, 2 good quality systematic reviews were identified, along with a comparison of clinical decision rules.

Relevant Paper(s)

Author, date and country	Patient group	Study type (level of evidence)	Outcomes	Key results	Study Weaknesses
Dietrich et al 1993 USA	n=322 <20y Head trauma All CT scanned	Single centre prospective cohort study.	Intracranial injury (ICI) if Glasgow Coma Score (GCS) = 15	11/195 (5%) All had one of: LOC, amnesia for the event, vomiting, headache, seizures, altered mental status, focal neurological deficit Loss of consciousness (LOC), amnesia, GCS<15, and neurologic deficit were significant for ICI.	All patients had CT scans but no clear criteria for scan. No data on neurosurgical (NS) interventions in minor Head Injury (HI).
Quayle et al. 1997 USA	n=321 <18y Blunt, non-trivial HI	Single centre prospective cohort study.	ICI if normal neurology (normal mental status, no focal findings)	16/266 (6%) All had headache or vomiting. 5 had NS intervention. Independent predictors of ICI = altered mental status, focal deficit, signs of a base of skull (BOS) fracture, seizure. Those <1 year may have ICI with few signs/symptoms.	Trivial HI excluded. Only 27 ICIs. 89 missed eligible children had lower admission rate.
Wang et al. 2000 USA	n=157 <15y Blunt HI Paramedic-assessed GCS 13-14	Multi-centre prospective cohort study	ICI if normal neurology (normal mental status, no focal findings)	30/157 (19%). 5 (3%) had haematoma evacuation.	52 eligible children not included. No data on other potential predictors of ICI e.g. focal neurology. Small number of ICI.
			Rate of SF or ICI if GCS=13 vs. rate of same if GCS=14	No significant difference
			Correlation of ICI incidence with changing GCS	60% of patients with ICI had improving GCS in transit or ED.
			Rate of LOC in ICI	No LOC in 67% with ICI.
Greenes and Schutzman 2001 USA	n=422 <2y Asymptomatic with no sign depressed/BOS fracture	Single centre prospective cohort study. Derivation of score based clinical decision rule (CDR) to predict SF (score based on age, haematoma size/location).	ICI rate	13/422 (3%), 1 haematoma evacuation. All had SF.	CDR to predict SF only. Reliance on skull x-ray to stratify ICI risk. Subjective measure of haematoma size.
			ICI with scalp haematoma	12/46 with scalp haematoma had ICI.
			Missed ICI	No missed ICI at 2 week follow-up (F/U) on discharged patients.
			CDR performance in predicting SF	Sensitivity (sens) 0.98, specificity (spec) 0.49.
			Impact of CDR implementation	CT rate 35%. No ICI missed.
Murgio et al. 2001 Argentina, Brazil, France, Hong Kong & Spain	n=4690 <16y with HI n=3710 GCS 14-15	Multinational prospective observational study	CT abnormality (fracture, ICI).	GCS 14-15: 55/3,710 (1.5% of total, 19% of those scanned). Propensity for abnormal scan if 3-9y. All GCS: 236/2690 (5.6% of total, 35% of those scanned).	Not restricted to minor trauma. CT indications unclear. No differentiation between ICI & fracture.
Palchak et al. 2003 USA	n=2043 <18y Blunt, non-trivial HI	Single centre prospective observational cohort study. Derivation of CDR. Individual decision trees given for specific outcomes	CDR performance: ICI on CT or requiring intervention	Sens 99%, Spec 25.8%, Negative Predictive Value (NPV) 99.7%, Positive Predictive Value (PPV) 10%	Not restricted to minor HI. Discretionary CT (not all cases of LOC/amnesia). Record of LOC unreliable in many cases. Younger children unable to describe headache, LOC or amnesia. Incomplete F/U (but no missed ICI found).
			Performance of 4-point tree: Intervention (Neurosurgical (NS) procedure, antiepileptic drugs >1 week, persistent neurologic deficits, or hospitalization ≥ 2 nights)	Sens 100%, Spec 42.7%, NPV 100%, PPV 8.6%
			Performance of 4-point tree: ICI on CT if GCS 14-15	Sens 94.9%, Spec 49.5%, NPV 99.6%, PPV 6.5%
			Performance of 3-point tree: NS procedure	Sens 100%, Spec 64.3%, NPV 100%, PPV 3.9%
			Performance of 2-point tree: ICI on CT <2y	Sens 100%, Spec 33.5%, NPV 100%, PPV 11.2%
Palchak et al. 2004 USA	n=2043 <18y Blunt, non-trivial HI See (Palchak et al. 2003)	Single centre prospective observational cohort study.	ICI or intervention (see Palchak et al 2003 above)	0% in those with isolated LOC (no vomiting, headache, fracture, seizure, mental alteration, neurology, scalp haematoma)	See (Palchak et al. 2003)
			ICI or intervention	0% in those with isolated LOC and/or amnesia
Boran et al. 2006 Turkey	n=421 <17y Blunt HI, GCS 15, no focal neurology. All had XR and CT.	Single centre prospective cohort study	ICI	37/421 (8.8%). 16 required neurosurgery. No significant association with headache, vomiting or scalp laceration.	Some SFs (BOS and depressed) classed as ICI.
			ICI if linear fracture on plain film	11/38 (29%)
			ICI if LOC	14/23 (61%)
			ICI if Seizure	5/6 (83%)
			New ICI at 24h.	Those with linear fracture had 2nd CT at 24h. 1 child with no ICI on initial CT had extradural haematoma at 24h.
Dunning et al. 2006 UK	n=22772 <16y with any head injury.	Multi-centre prospective cohort study. Derivation of CHALICE CDR.	CDR performance: ICI or depressed fracture if GCS 13-15	Sens 97.6%, Spec 87.3%, PPV 5.44%, NPV 99.9%, CT rate 13.3%.	Selective CT (based on Royal College of Surgeons guidelines). No direct patient follow-up – selected information gathered from participating hospitals, local neurosurgical units and Office for National Statistics. No validation.
Da Dalt et al. 2006 Italy	n=3806 <16y Blunt HI Analysed in 5 groups	Multi-centre prospective cohort study. CDR derivation.	ICI if A: GCS 15, no neurology, no clinical BOS or non-frontal fracture, asymptomatic	0/2748	Discretionary CT. Possible underestimated ICI rate. Difficult to separate those with minor HI.
			ICI if B: A + <30s LOC, impact seizure, non-prolonged vomiting or headache	0/550
			ICI if C: A + >30s LOC, amnesia, drowsiness or prolonged headache	3/233
			ICI if D: A + clinical non-frontal fracture.	6/201
			ICI if E: abnormal GCS / neuro exam or clinical BOS fracture.	13/66
			Accuracy of rule (A+B vs C+D+E) to predict ICI	Sens 100% Spec 87.3%, PPV 4.4%, NPV 100%
Oman et al. 2006 USA	n=1666 <19y Blunt HI who had CT	Multi-centre prospective cohort study. Derivation of NEXUS-II CDR, accuracy in paediatric subset.	CDR performance: Clinically important ICI (neurosurgical intervention [craniotomy, intracranial pressure {ICP} monitoring, or mechanical ventilation] or likely significant long-term neurologic impairment)	Sens 98.6%. NPV 99.1%. Spec 15.1%. 2 missed, one due to clinician error, neither required intervention.	CDR derived from mixed adult and child data. Only applied to those who had discretionary CT. Small group <3y. Not validated.
			Performance of CDR: Clinically important ICI in children <3y	Sens 100%, NPV 100%, spec 5.3%.
Atabaki et al. 2008 USA, Canada	n=1000 <21y Closed non-trivial HI, GCS 13-15 who had CT.	Multi-centre prospective observational study. Derivation of CDR.	Performance of CDR: ICI on CT	Sens 95.4%, Spec 48.9%, NPV 99.3%. 46.3% of sample would have avoided CT. Only 6 required NS intervention.	Discretionary CT. Small number of patients requiring intervention.
Turedi et al. 2008 Turkey	n=240; 120 <16y, 120>16y. GCS 13-15, LOC<15min, Post-Traumatic Amnesia (PTA)<1h	Single centre prospective cohort study. High risk group (using criteria identified in meta-analysis (Dunning et al. 2004)) compared with low risk.	ICI on CT	Low-risk group: 7 ICI (4 <7y; 3 >6y), none needing intervention. High-risk group: 25 ICI (13 <7y; 12 >6y), undisclosed number intervention. 2nd scan in high-risk at 16-24h: 2 new ICI, no intervention.	Small sample. Paediatric data not analysed separately from adults.
Kuppermann et al. 2009 North America	n=42412 <18y GCS 14-15, non-trivial HI. Derivation and validation populations: 8502 and 2216 aged <2y, and 25 283 and 6411 aged ≥2y <18y	Multicentre prospective cohort study. Derivation and validation CDR.	CDR performance accuracy <2y in terms of clinically important Traumatic Brain Injury (ciTBI) (death, neurosurgery, intubation >24 h, admission ≥2 nights)	Derivation: NPV 99.9%, sens 98.6%, spec 53.7% Validation: NPV 100%, sens 100% , spec 53.7%	Discretionary CT. Selected population Internal validation only.
			CDR performance accuracy ≥2y: cTBI	Derivation: NPV 99.95%, sens 96.7%, spec 58.5% Validation: NPV 99.95%, sens 96.8%, spec 59.8%
Osmond et al. 2010 Canada	n=3866 <16y Blunt minor HI	Multi-centre prospective cohort study. Derivation of CATCH CDR.	CDR performance: High risk criteria in terms of NS intervention	Sens 100%, Spec 70.2%, CT rate 30.2%.	Discretionary CT. Relatively few <2y. No validation
			CDR performance: High or medium risk criteria in terms of TBI on CT	Sens 98.1%, spec 50.1%, CT rate 51.9% for specific population included in derivation study.
Bin, Schutzman, and Greenes 2010 USA	n=417 <2y Blunt HI 203 with CT or X-ray. 214 with no imaging not included in validation set.	Single centre prospective observational study. Validation of previously derived CDR (Greenes and Schutzman 2001)	CDR performance accuracy: Score ≥4 & SF	All patients: sens 0.9, spec 0.78 Asymptomatic: sens 0.93, spec 0.68	Almost 30% eligible patients not enrolled and this group much less likely to get skull XR. Discretionary imaging - CT or XR not evaluated separately. Qualitative estimate of haematoma size. 78% F/U (phone call at 1 & 4w) – possible missed ICI.
			CDR performance accuracy: Score ≥4 & ICI	All patients: sens 0.62, spec 0.64. Asymptomatic: sens 0.56, spec 0.59
			CDR performance accuracy: Score ≥3 & ICI	All patients: sens 0.93, spec 0.42. Asymptomatic: sens 1, spec 0.33
Nigrovic et al. 2011 North America	n=40113 <18y GCS 14-15, non-trivial	Secondary analysis of PECARN subset – see (Kuppermann et al. 2009) above.	Impact of observation on rates of CT, clinically significant TBI, emergent NS intervention.	Significantly lower CT rate (31.1% vs 35%) if observed, no significant difference in ciTBI (0.75% vs 0.8%) or NS intervention rates (0.11% vs 0.14%).	No analysis of length of stay, length of observation or of significance of any delay to NS intervention.
			When was observation used?	Intermediate risk more likely to be observed than high or low risk.
Fabbri et al. 2011 Italy	n=2391 <11y Blunt HI	Retrospective validation of CDRs (Oman et al. 2006; Da Dalt et al. 2006) using single centre prospective cohort data.	Performance of NEXUS-II CDR in terms of any post-traumatic lesion on CT	Sens 88.9%, spec 59.4, NPV 99.9.	Discretionary CT. Not restricted to minor injury. Small no. of ICI (18) and NS intervention (2)
			Performance of Da Dalt CDR in terms of any post-traumatic lesion on CT	Sens 100%, spec 76.1, NPV 100

Comment(s)

Cranial computed tomography (CT) of children with minor HI needs to balance the benefits of the resultant data with the risks associated with the test. Minor HI can be associated with intracranial injury requiring neurosurgery, and cranial CT may be more cost effective than inpatient observation (Oman et al. 2006; Da Dalt et al. 2006). However, the long-term clinical significance of minor intracranial changes on CT following acute HI is unknown, and ionising radiation in the developing brain carries an increased lifetime risk of neoplastic change. Recent studies have concentrated on identifying clinically important traumatic brain injury (ciTBI) rather than any traumatic change on CT. However, there is no widely agreed definition of what is a ciTBI. Clinical decision rules (CDRs) use clinical variables to identify children who are either high or low risk for ciTBI, and may guide a clinician’s decision regarding a CT scan. The number of CDRs for paediatric HI has increased markedly since the last BestBet and a systematic review from the same time period (Maguire et al., Pediatrics 2009;124:e145–e154). However, a further systematic review (Pickering et al., Arch Dis Child 2011;96:414-421), and a comparison of the CATCH, CHALICE and PECARN CDRs (Lyttle et al., Emerg Med J 2012;29:785-794) have identified that the CDR derivation studies vary widely in their inclusion/exclusion criteria, outcomes, and predictor variables. For example, when considering CHALICE, CATCH and PECARN, GCS is a variable in all three, but whereas PECARN and CATCH use GCS<15, CHALICE uses GCS<14. PECARN includes a ‘history of vomiting’, whereas CHALICE specifies 3 discrete episodes of vomiting. For brevity, we have not included a full description of the CDRs in the table above, though references are supplied below. Economic analysis (Pandor et al., Health Technol Assess. 2011;15(27):1-202) shows that a selective CT policy is the most cost-effective and that admission is only cost-effective for those with an abnormal CT head. CDRs would therefore seem to be a cost-effective tool in managing paediatric HI. However, the use of some tools would likely increase CT rates if applied verbatim to the UK population, and there is insufficient data validating their use in different populations. Further studies validating and comparing the performance accuracy and cost-effectiveness of CDRs directly in new populations are required. Further research may focus on identifying TBI associated with adverse long-term neurological outcome. Current research investigating the role of biomarkers (Castellani et al., Acta Paediatrica 2009;98:1607-1612) suggests they may be of use in identifying patients who do not need a CT and future CDRs may incorporate such markers. However, there is insufficient evidence for their use in the management of paediatric HI at present.

Clinical Bottom Line

Clinical findings in the form of a clinical decision rule are useful in helping to predict which children have intra-cranial injury on CT scan after minor head injury. A number of such clinical rules exist. Each has differing advantages, and different rules may be more effective in different healthcare systems. Current UK guidance is in the form of the NICE guideline, based on the CHALICE CDR. In the absence of comparative validation studies of clinical decision rules, there appears to be insufficient evidence to support deviating from current practice in the UK. The NICE guidance is in the early stages of revision.

References

1. Dietrich, A M, M J Bowman, M E Ginn-Pease, E Kosnik, and D R King. Pediatric head injuries: can clinical factors reliably predict an abnormality on computed tomography? Annals of Emergency Medicine 1993; 22 (10) (October): 1535-1540
2. Quayle, Kimberly S., David M. Jaffe, Nathan Kuppermann, Bruce A. Kaufman, Benjamin C. P. Lee, T. S. Park, and William H. McAlister. Diagnostic Testing for Acute Head Injury in Children: When Are Head Computed Tomography and Skull Radiographs Indicated? Pediatrics 1997; 99 (5) (May 1): e11.
3. Wang, M Y, P Griffith, J Sterling, J G McComb, and M L Levy. A prospective population-based study of pediatric trauma patients with mild alterations in consciousness (Glasgow Coma Scale score of 13-14) Neurosurgery 2000; 46 (5) (May): 1093-1099
4. Greenes, D S, and S A Schutzman. Clinical significance of scalp abnormalities in asymptomatic head-injured infants. Pediatric Emergency Care 2001; 17 (2) (April): 88-92
5. Murgio, A, P D Patrick, F A Andrade, S Boetto, K M Leung, and M A Muñoz Sanchez. International study of emergency department care for pediatric traumatic brain injury and the role of CT scanning. Child’s Nervous System: ChNS: Official Journal of the International Society for Pediatric Neurosurgery 2001; 17 (4-5) (April): 257-262
6. Palchak, Michael J., James F. Holmes, Cheryl W. Vance, Rebecca E. Gelber, Bobbie A. Schauer, Mathew J. Harrison, Jason Willis-Shore, Sandra L. Wootton-Gorges, Robert W. Derlet, and Nathan Kuppermann. A decision rule for identifying children at low risk for brain injuries after blunt head trauma. Annals of emergency medicine 2003; 42 (4) (October 1): 492-506.
7. Palchak, Michael J, James F Holmes, Cheryl W Vance, Rebecca E Gelber, Bobbie A Schauer, Mathew J Harrison, Jason Willis-Shore, Sandra L Wootton-Gorges, Robert W Derlet, and Nathan Kuppermann. Does an isolated history of loss of consciousness or amnesia predict brain injuries in children after blunt head trauma? Pediatrics 2004; 113 (6) (June): e507-513.
8. Boran, Burak O, Perran Boran, Nehir Barut, Cem Akgun, Erhan Celikoglu, and Mustafa Bozbuga. Evaluation of mild head injury in a pediatric population. Pediatric Neurosurgery 2006; 42 (4): 203-207.
9. Dunning, J, J Patrick Daly, J-P Lomas, F Lecky, J Batchelor, and K Mackway-Jones. Derivation of the children’s head injury algorithm for the prediction of important clinical events decision rule for head injury in children. Archives of Disease in Childhood 2006; 91 (11) (November 1): 885 -891.
10. Da Dalt, Liviana, Alberto G Marchi, Lorenzo Laudizi, Giovanni Crichiutti, Gianni Messi, Lucia Pavanello, Francesca Valent, and Fabio Barbone. Predictors of intracranial injuries in children after blunt head trauma. European Journal of Pediatrics 2006; 165 (3) (March): 142-148.
11. Oman, Jennifer A., Richelle J. Cooper, James F. Holmes, Peter Viccellio, Andrew Nyce, Steven E. Ross, Jerome R. Hoffman, William R. Mower, and for the NEXUS II Investigators. Performance of a Decision Rule to Predict Need for Computed Tomography Among Children With Blunt Head Trauma. Pediatrics 2006; 117 (2) (February): e238 -e246.
12. Atabaki, Shireen M., Ian G. Stiell, Jeffrey J. Bazarian, Karin E. Sadow, Tien T. Vu, Mary A. Camarca, Scott Berns, and James M. Chamberlain. A Clinical Decision Rule for Cranial Computed Tomography in Minor Pediatric Head Trauma. Archives of Pediatrics & Adolescent Medicine 2008; 162 (5) (May 1): 439-445.
13. Turedi, S., A. Hasanbasoglu, A. Gunduz, and M. Yandi. Clinical Decision Instruments for CT Scan in Minor Head Trauma. Journal of Emergency Medicine 2008; 34 (3): 253-259.
14. Kuppermann, Nathan, James F Holmes, Peter S Dayan, John D Hoyle, Shireen M Atabaki, Richard Holubkov, Frances M Nadel, et al. Identification of children at very low risk of clinically-important brain injuries after head trauma: a prospective cohort study. The Lancet 2009; 374 (9696) (October 3): 1160-1170.
15. Osmond, Martin H., Terry P. Klassen, George A. Wells, Rhonda Correll, Anna Jarvis, Gary Joubert, Benoit Bailey, et al. CATCH: a clinical decision rule for the use of computed tomography in children with minor head injury. Canadian Medical Association Journal 2010; 182 (4) (March 9): 341 -348.
16. Bin, S.S., S.A. Schutzman, and D.S. Greenes. Validation of a clinical score to predict skull fracture in head-injured infants. Pediatric Emergency Care 2010; 26 (9): 633-639.
17. Nigrovic, Lise E., Jeff E. Schunk, Adele Foerster, Arthur Cooper, Michelle Miskin, Shireen M. Atabaki, John Hoyle, et al. The Effect of Observation on Cranial Computed Tomography Utilization for Children After Blunt Head Trauma. Pediatrics 2011; 127 (6) (June 1): 1067 -1073.
18. Fabbri, A., F. Servadei, G. Marchesini, A. Raggi, and A. Vandelli. Analysis of different decision aids for clinical use in pediatric head injury in an emergency department of a general hospital. Journal of Trauma - Injury, Infection and Critical Care 2011; 70 (5): E79-E83.