Background The National Institute for Health and Clinical Excellence (NICE) published guidelines containing clear criteria for CT scanning of children with head injury in 2007. The aim of this study was to quantify the effects of adherence to these guidelines on the number of head scans requested.
Method A retrospective case note review was carried out of all patients under the age of 16 years presenting to the emergency department with head injury in 2007. The number of CT head scans actually performed was recorded, and the number that would have been requested using the hospital guidelines and the 2007 NICE guidelines was calculated.
Results 25 (6.7%) of the 394 patients included in our study had head CT scans. 47 (12.7%) children would have been scanned had the hospital guidelines been rigidly followed and 74 (19.7%) children would have had head CT scans if the 2007 NICE guidelines had been adhered to.
Conclusion Considerably fewer children with head injury had CT scans in 2007 than would have been indicated by the hospital guidelines or 2007 NICE guidelines.
Statistics from Altmetric.com
Head injury is a common presentation to emergency departments in the UK, resulting in approximately 1 million attendances annually.1 Almost half of these patients are aged 15 years or under. The majority of these presentations are of minor head injury, defined in the UK as head injury with a Glasgow Coma Scale of 13–15.
The challenge of managing paediatric minor head injury lies in identifying which children have the very rare intracranial injuries which will require neurosurgical intervention. CT scanning is required to establish a definitive diagnosis but this process is more problematic in children than adults. CT scanning involves significant doses of ionising radiation to vulnerable developing tissues, and general anaesthesia or sedation may occasionally be required for technical reasons.2
Prior to 2007, we used a consensus based local departmental guideline (table 1). The National Institute for Health and Clinical Excellence (NICE) guidelines on the management of head injury in infants, children and adults were updated in September 2007 (table 2).3 The guidelines list a number of criteria as indications for CT scanning. It is obvious that adherence to the revised guidelines would result in many more scans being requested in children. The scale of this effect was unclear.
We planned to quantify the effect of compliance with the NICE 2007 guidelines on the number of CT head scans requested in children presenting with minor head injury. This would be compared with the rate of scanning indicated under the existing local hospital guidelines. The local guidelines were based on the original NICE guidelines for management of head injury published in June 2003 but modified to take the availability of local services into account.4
We carried out a single centre, retrospective, cross sectional study to determine the effect of adherence to the 2007 NICE guideline (CG 56) on the number of CT head scans performed on children with minor head injury.
We conducted a case note review of all children under the age of 16 years who presented to the emergency department at Addenbrooke's Hospital between 1 January 2007 and 31 December 2007 with a minor head injury. Addenbrooke's Hospital emergency department sees approximately 90 000 patients a year, of which 18 000 are children, and is a regional referral centre for neurosurgery and paediatric intensive care.
We included children aged 15 years or under who had a clear history of trauma to the head. Patients were excluded if the history suggested that the trauma was not a primary event (eg, primary seizure). The cases were selected by a search for all patients presenting to the emergency department in the year 2007, before their 16th birthday, with ‘minor head injury’ or ‘head injury with intracranial abnormality’ coded by the treating clinician. These are the only codes used for head injured patients in our emergency department.
The clinical variables that warranted CT scanning of the head were identified from the NICE 2007 guidelines and the hospital guidelines (see tables 1 and 2). The case notes were then reviewed to establish which of the variables were present in each case and whether or not a scan was carried out. We compared the number of CT scans actually performed with the number of CT head scans that would have been carried out if the hospital guidelines had been followed, and also the number of scans that would have been performed if the 2007 NICE guidelines had been followed. We analysed these numbers using the χ2 test for categorical data in STATA V.7. We generated absolute percentage increases in the number of CT scans that would have been requested by using CI Analysis (BMJ Publishing Group, London, UK) We also examined the data to identify which clinical variables lead to increased CT scanning. We did not seek ethics approval as we considered this to be an audit. We used the hospital computer system to identify any children that re-attended with complications related to their head injury.
A total of 426 case notes were examined; 31 of the 426 cases were excluded because the presentation was not a primary head injury or had not waited to be seen. One case was excluded because the case note could not be found. The electronic patient registration system details indicated that this patient was receiving care at the hospital for an unrelated condition.
We therefore included 394 children in our study; 221 (56%) were boys Median age was 4.2 years (range 7 days to 15 years and 11 months). We performed 25 CT head scans (6.7%, 95% CI 4.2 to 9.2) on these children. Seven of these scans showed an intracranial abnormality. Applying either guideline rigidly would have led to a statistically significant increase in the number of children requiring CT scanning.
If the departmental 2003 guideline had been rigorously applied, 47 children (12.7%, 95% CI 8.7 to 15.1) would have been scanned. This would have been a 5.5% (95% CI 1.5% to 9.6%) increase in the proportion of children who were actually scanned (χ2 (1 df)=79.8, p<0.001).
If the NICE 2007 guideline had been applied, 74 children (19.7%, 95% CI 14.9% to 22.6%) would have been scanned. This would have been a 12.4% (95% CI 7.9% to 17.0%) increase in the proportion of children scanned (χ2 (1 df)=104.3, p<0.001).
The number of children presenting with the clinical criteria outlined in the local hospital guidelines is shown in table 1. The main contributing variable to increased scanning with both guidelines was vomiting, either more than three times (2003 guidelines) or three or more times (2007 guidelines). More than one of the criteria were sometimes present in the same child. Young children were no more likely not to be scanned, despite fulfilling criteria, than older children.
We admitted 18 (4.6%) children to our short stay observation ward; 11 of these had been scanned. Thirteen children re-attended after discharge within 1 month, 10 with presentations clearly unrelated to the initial head injury. Box 1 shows the clinical features of these children who returned with head injury related problems.
Clinical characteristics of children re-attending after assessment for a head injury with a head injury related problem
Case 1: An 8-month-old child fell backwards from a height of 1 m. The child did not fulfil the NICE 2007 criteria but did fulfil local guidelines because of the dangerous mechanism of injury. The child was discharged without CT scanning and represented 3 weeks later with drowsiness. A CT scan showed bilateral subdural haematomas which were tapped by the neurosurgeons. The child was discharged home with no functional deficit.
Case 2: A 6-month-old child presented after a fall. The child did not fulfil any criteria for CT scanning and was sent home with written advice. The child returned with subsequent vomiting. CT scanning showed a parietal skull fracture and a small subdural haematoma. The child was managed conservatively, admitted for 1 week, and was discharged from outpatient care 3 months later with no apparent deficits.
Case 3: An 8-year-old child fell onto his face while playing football, briefly losing consciousness. He did not fulfil any scanning criteria. He returned the same day with lethargy and headache. He was observed on an observation ward overnight without scanning and discharged home. He has continued to access care at our hospital for unrelated conditions and is apparently well.
Applying the 2007 NICE guidelines would have led to a nearly threefold increase in CT scanning in the subject population, a statistically significant increase. The main criterion contributing to this increase is frequency of vomiting. It appears unlikely that the 2007 guidelines will be adopted as we have not yet achieved compliance with 2003 guidance. The number of patients fulfilling criteria for scanning under the NICE 2007 guidelines also increased compared with the hospital guidelines. Most of this increase can be attributed to two criteria. Firstly, the number of episodes of vomiting warranting a CT scan was reduced from more than three in the hospital guidelines to three or more in the NICE 2007 guidelines. Secondly, the length of amnesia mandating a scan was reduced from 30 min to 5 min. In keeping with previous studies in adults, we found that NICE guidelines increased the proportion of patients requiring CT scanning.5
We have also demonstrated that neither of these guidelines can detect all intracranial injuries while increasing the number of CT scans that we perform (see box 1). We accept that no decision rule is ever going to provide ‘no risk’ and that these guidelines help identify children at low or very low risk of intracranial haematoma.
Children who presented post-head injury with either just repeated vomiting or a dangerous mechanism of injury alone were often not scanned, despite the inclusion of both of these criteria in the hospital guidelines. Those who presented with the more ‘serious’ criteria such as seizure or focal deficit were far more likely to be scanned.
The reluctance to scan children presenting with relatively ‘soft’ criteria may be due to clinicians' concerns about exposing a developing brain to radiation.4
We cannot be certain what the reasons are for our comparatively low scanning rates. This is not due to a lack of availability of scanning as there is easy access to CT imaging, with a resident radiology registrar. There is a children's observation unit, which is used for extended observation of children. This may explain the low CT rates as there are alternatives to CT scanning. Our results, however, suggest that this is not the case, as most children who were admitted to the observation unit underwent CT scanning. We have noticed, anecdotally, that parents are often reluctant to subject their child to a CT head scan.
Previous studies have been carried out in this area in an attempt to discover those clinical factors that most closely predict those children who will require neurosurgical intervention. A meta-analysis of 16 studies concluded that intracerebral haemorrhage in children correlated most closely with the presence of skull fracture, focal neurology, loss of consciousness and Glasgow Coma Scale abnormality.6 This meta-analysis also found that headache and vomiting were not predictive of significant injury, and that post-traumatic seizures were of variable significance.
The CATCH study was a prospective, multicentre, cohort study that included 3781 children.7 It demonstrated that the high risk factors for predicting neurosurgical intervention were Glasgow Coma Scale <15 at 2 h, open skull fracture, a worsening headache and irritability on examination. Medium risk factors included large scalp haematoma, signs of basal skull fracture and a dangerous mechanism of injury. Vomiting was found not to be an important predictor.
The CHALICE study included 22 772 children presenting with minor head injury to 10 different hospitals.8 It derived a rule for CT scanning of paediatric head injury which the 2007 NICE guidelines are based on. The high risk criteria they identified are identical to those outlined in table 1. The authors of this study calculated that their decision rule had 98% sensitivity and 87% specificity for identifying those with significant pathology. The application of these rules led to a scan rate of 14% of all children presenting. Our department however, had a scan rate of 3.7%. Applying the NICE 2007 guidelines to our patient population would have led to an increased scan rate of 19%, as would be expected from the CHALICE study. Our results should not be seen as a validation study of the CHALICE paper. Our primary aim was to quantify the increase in CT scanning.
A more recent, very large, observational study from the USA derived a decision rule which is likely to advocate increased scanning rates in the UK although it may reduce scanning rates in the USA.9 In this study, 31% of children with head injuries underwent CT scanning. We cannot comment on the variables in this decision rule in our sample as we did not measure them.
There are some important limitations to this study. It was conducted in a single centre and it is a matter of judgement whether our findings can be applied to other departments. It is certainly plausible that our low scanning rates reflects national practice. We struggled to provide a clear definition of ‘minor head injury’ as this is a spectrum of trauma severity and there is no clear cut-off distinguishing ‘head injury’ from ‘bump to the head’. Our case definition was pragmatic and based on clinician coding. Another limitation is that the case notes were reviewed retrospectively and there were significant missing data around clinical variables. For instance, amnesia is almost impossible to assess in pre-verbal children and difficult to assess in young children. With the cases that were missed, we cannot be completely certain if the cases did not fulfil scanning criteria as we collected the data from case note review, which can be an unreliable way to collect clinical data.
Our method of follow-up is weak as we only examined the hospital computer system. We did not contact patients or carers. However, our hospital provides the only neurosurgical unit and paediatric intensive care unit for the Eastern Region. It is possible that the patients we saw subsequently deteriorated and were admitted elsewhere. This is, however, unlikely because of our geographical context.
The NICE guidelines are based on the best available evidence. The CHALICE study has not been validated. A validation study of the CHALICE and PECARN rules is necessary. Future qualitative work should also aim to understand why clinicians do not follow these guidelines.
Applying the 2007 NICE criteria for CT scanning of children would have led to a significant increase in the number of children with minor head injury who underwent CT scanning. Previous hospital guidelines, which advocated less scanning, were not followed.
This rise in scanning rate may not have led to any increase in detection of intracranial injury requiring intervention in our study population. The criteria that contribute most significantly to increased scanning are frequency of vomiting and duration of amnesia.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.