Article Text

Download PDFPDF

Head home: a prospective cohort study of a nurse-led paediatric head injury clinical decision tool at a district general hospital
  1. Patrick Aldridge1,
  2. Heather Castle1,
  3. Clare Phillips1,
  4. Emma Russell1,
  5. Richard Guerrero-Luduena2,
  6. Raj Rout3
  1. 1 Paediatric Emergency Department, Frimley Park Hospital NHS Foundation Trust, Frimley, UK
  2. 2 School of Health Sciences, Centre for Implementation Science, University of Southampton, Southampton, Hampshire, UK
  3. 3 Global Medical Affairs, Sanofi Genzyme, Guildford, UK
  1. Correspondence to Dr Patrick Aldridge, Paediatric Emergency Department, Frimley Park Hospital NHS Foundation Trust, Camberley GU16 7UJ, UK; patrickjaldridge{at}


Objectives To assess if a nurse-led application of a paediatric head injury clinical decision tool would be safe compared with current practice.

Methods All paediatric (<17 years) patients with head injuries presenting to Frimley Park Emergency Department (ED), England from 1 May to 31 October 2018 were prospectively screened by a nurse using a mandated electronic 'Head Injury Discharge At Triage' questionnaire (HIDATq). We determined which patients underwent CT of brain and whether there was a clinically important intracranial injury or re-presentation to the ED. The negative predictive value of the screening tool was assessed. We determined what proportion of patients could have been sent home from triage using this tool.

Results Of the 1739 patients screened, 61 had CTs performed due to head injury (six abnormal) with a CT rate of 3.5% and 2% re-presentations. Of the entire cohort, 1052 screened negative. 1 CT occurred in this group showing no abnormalities. Of those screened negative, 349 (33%)/1052 had ‘no other injuries’ and 543 (52%)/1052 had ‘abrasions or lacerations’. HIDATq's negative predictive value for CT was 99.9% (95% CI 99.4% to 99.9%) and 100% (95% CI 99.0% to 100%) for intracranial injury. The positive predictive value of the tool was low. Five patients screened negative and re-presented within 72 hours but did not require CT imaging.

Conclusion A negative HIDATq appears safe in our ED. Potentially 20% (349/1739) of all patients with head injuries presenting to our department could be discharged by nurses at triage with adequate safety netting advice. This increases to 50% (543/1739), if patients with lacerations or abrasions were given advice and discharged at triage. A large multicentre study is required to validate the tool.

  • trauma, head
  • paediatric emergency med
  • head
  • paediatrics, paediatric injury

Statistics from

Key messages

What is already known on this subject?

  • >700 000 children per year attend hospital with a head injury in England and Wales.

  • Research indicates that <1% of all children with a head injury undergo neurosurgical intervention.

What this study adds?

  • In this prospective study at a single emergency department (ED) a screening tool applied by nurses had a negative predictive value for CT of 99.9%.

  • Between 20% and 50% of all paediatric patients with head injuries attending our ED may have been suitable for discharge soon after screening.

  • A large multicentre study is required to validate the tool.

It is estimated that over 700 000 children annually attend emergency departments (EDs) across England and Wales due to a head injury (National Institute for Health and Clinical Excellence (NICE) ‘Head Injury: assessment and early management’ CG176, 2014). It was recently shown1 that of 19 920 children attending EDs across Australia with a head injury, only 2.3% (n=457) had either a traumatic brain injury on computed tomography (TBI-CT) or a clinically important traumatic brain injury (ciTBI). The primary paper2 to this analysis1 showed that <1% of all children with a head injury undergo neurosurgical intervention. EDs therefore see numerous head injuries to identify the small percentage with a clinically significant brain injury. Recent discussion has centred on further risk stratification of children with minor head injuries.3

Minor head injuries at our clinical centre are regularly among the top three most common paediatric ED discharge diagnoses. In current practice all children are triaged by a trained paediatric ED nurse followed by an ED doctor or emergency nurse practitioner review. A standardised head injury proforma is used based on local guidelines compliant with current NICE guidance (‘Head Injury: assessment and early management’ CG176, 2014). This results in the closure/dressing of any wounds as well as discharge or CT/admission. Dependant on departmental demands and acuity, patients may have extended waiting times after triage. Some UK clinical centres have initiated ‘Nurse led’ discharge of paediatric patients with head injury without medical review. A literature review showed no published data on this pathway. Given the potentially significant reduction in waiting times for these patients and resultant ED occupancy, coupled with the need for robust evidence of its safety, we sought to analyse the impact of a nurse-led Head Injury Discharge At Triage (HIDAT) clinical decision tool (CDT) at our clinical centre.

Our aim was to assess whether application of HIDAT would be safe compared to current practice in our clinical centre. ‘Safe’ is defined as no child screening negative (using HIDAT) undergoing CT imaging.


Study setting

Frimley Park is a district general hospital in England with a co-located adult and paediatric ED. The paediatric ED is staffed with trained paediatric nurses, is separate from the adult unit and staffed 24 hours/day. Paediatric ED has approximately 28 000 attendances aged <17 years old per annum and functions as a local trauma unit. Frimley Park does not currently operate a ‘single front door policy’ meaning many paediatric general practitioner (GP) referrals do not attend ED and instead directly attend our paediatric assessment unit (PAU). PAU does not accept children with injuries and they are streamed to ED. As noted all head injuries are assessed following a local guideline compliant with current national/NICE guidance.


Inclusion criteria was any child <17 years old who attended Frimley Park Hospital ED in the 6-month period from 1 May to 31 October 2018 with an attending complaint of head or facial injury.

Derivation of study tool/CDT

A number of clinical tools and algorithms have been generated4–6 to identify children with a clinically relevant brain injury. These all attempt to risk stratify children into low risk of a ciTBI and those who require CT and/or observation. The Paediatric Emergency Care Applied Research Network is arguably the most validated and sensitive/specific.2 The referenced tools and algorithms2 4–6 were reviewed by the authors along with the current NICE guidelines (2014). An ‘ultra low risk’ HIDAT questionnaire (HIDATq) was developed (figure 1) and applied prospectively to all children attending our ED within the inclusion criteria.

Figure 1

HIDATq—answers: yes or no. GCS, Glasgow Coma Score; BP, blood pressure; HIDATq, Head Injury Discharge At Triage questionnaire.

A hybrid paper/electronic medical records system currently exists at our centre. Triage is carried out by trained paediatric ED nurses with triage and safeguarding competencies. All triage nurses can close wounds up to 5 cm. Triage notes are fully electronic. HIDATq was made into a mandatory electronic questionnaire at triage for all children presenting with head or facial injury as their chief complaint.

A retrospective pilot study was conducted using only the triage notes for all ED presenting paediatric head injuries in 2017, with a TBI-CT or ciTBI (n=13). HIDATq was applied independently by two paediatric triage nurses. The nurses were only aware of what was written in the triage notes (including observations) and were blinded to outcome. No patient was discharged at triage and no interobserver variability was noted.

Data collection

All children with ‘head injury’ or ‘facial injury’ as the attending complaint were electronically mandated to have HIDATq screening at triage (figure 1). Triage staff were asked to complete this as accurately as possible as part of the medical records and were aware it would be used to analyse the efficacy/safety of HIDAT. Data were also collected on the presence/absence of other injuries (figure 1, Question 14).

For questions with an unknown answer (eg, loss of consciousness or amnesia in a non-verbal child), staff were advised to put what they felt was the most appropriate/pragmatic answer. Given the large number of triage nurses using HIDATq it was felt that any errors or issues with data entry would become clear over the study period. Interobserver variation during the study was not measured as triage is carried out by a single nurse.

Study procedures

All patients with ‘head injury’ or ‘facial injury’ complaints were sourced from the ED electronic medical record system (Symphony by EMIS Health). The discharge diagnoses of these patients were screened for a head injury diagnosis—‘minor, moderate or severe head injury’. All patient diagnosis not fitting these had manual screening to include or exclude them from analysis (figure 2). Data from these presentations (including HIDATq, diagnosis, time to triage, length of stay and admission/discharge/transfer) were downloaded from the patient record.

Figure 2

Patient flow through HIDATq study. n=1740 initially, n=1739 eligible. Screened negative=HIDATq −ve (n=1052), screened positive HIDATq +ve (n=687). Imaging, admissions, discharges and reattenders <72 hours as indicated. HIDATq, Head Injury Discharge At Triage questionnaire.

A record of all CT scans of the brain performed on children<17 years over the study period, who had attended via ED, was obtained. The notes/CT requests were reviewed to include all ED CTs for trauma by any means (eg, suspected non-accidental injury, fall, road traffic accident). These were cross-referenced with the initial study population to ensure all had undergone HIDATq and were not missed from analysis. Any patient who underwent CT without triggering HIDATq had their notes reviewed and the triage-presenting complaint category recorded.

Re-presentations within 72 hours were sourced electronically via the electronic record system. Patients within this group had their notes reviewed for HIDATq from both presentations and whether CT or admission occurred.

Measures and definitions

Negative screening in HIDATq (HIDATq –ve) corresponds to all triage questions being answered negatively (figure 1, Questions 1–11), as well as the child being alert with normal observations (figure 1, Questions 12–13).

Positive screening (HIDATq +ve) corresponds to any triage question being positive (figure 1, Questions 1–11) or the child having abnormal observations, abnormal neurology or there was nursing concern (figure 1, Questions 12–13).

The definitions for TBI-CT and ciTBI (figure 3) are derived from large and often-cited paediatric head injury studies.4–6 The definition of TBI-CT was expanded beyond these4–6 to explicitly include orbital fractures as the study definitions were ambiguous. Though an orbital fracture is not explicitly a brain injury, the authors felt it to be a significant injury and should not be potentially missed.


The primary outcome measure was the number of CT brain studies requested on those eligible for discharge following the HIDAT CDT.

The authors’ consensus was that the tool should not miss any TBI-CT or ciTBI and thus CT of brain was decided on as the primary endpoint, rather than admissions, as not all patients who undergo imaging are necessarily admitted and vice versa.

The secondary outcome measures were the number of admissions or transfers of paediatric patients with head injury, number of TBI-CT or ciTBI and the number of re-presentations to ED within 72 hours.

Data analysis

A power calculation was not possible due to changes in local attendance/discharge classifications during the study period, 2017–2018, meaning the incidence could not be accurately determined retrospectively. The study duration was set at 6 months of ED patient attendances as the preceding year (2017) had n=13 ciTBI or TBI-CT over the 12-month period and so over 6 months we would expect at least six abnormal CTs due to head injury. As noted, HIDATq applied retrospectively did not miss any (n=13) of these cases either. Within the limitations of our study, we decided this would give a fair reflection on the effectiveness of the CDT.

A statistician ran all statistical testing. A statistical computer software program (Python) was used and all standard assumptions when calculating sensitivity and specificity were applied. Qualitative data are presented as the number of observations and percentage. Standard 2×2 tables were used to measure sensitivity, specificity, negative predictive value and positive predictive value.7

Sensitivity is the proportion of patients who required a CT or had an important intracranial injury (ICI) following positive screening with HIDATq. Specificity is the proportion of patients who did not require CT or did not have an ICI following a negative screening using HIDATq.

Negative predictive value is the proportion of patients screened negative with HIDATq who did not require CT or had ICI (the probability that a patient who screened negative using HIDATq is a true negative—ie, does not require CT or did not have ICI).

Positive predictive value is the proportion of patients with a positive HIDATq who actually required a CT or had ICI (the probability that a positive HIDATq accurately indicates that the patient requires a CT or has ICI).


There were 13 223 paediatric ED attendances over the 6-month study period with 11% (n=1415) triaged ‘Immediate resuscitation’ or ‘Very urgent’, 16% (n=2168) ‘Urgent’, 55% (n=7236) ‘Standard’ and 18% (n=2404) ‘Non urgent’.

The initial study population of 1739 from 13 223 patients was selected into analysis as shown in figure 2. The median age was 6 years with an IQR of 2–11 years.

From the study group (n=1739) 61 CTs occurred with 6 (10%) showing an abnormality (table 1). Eleven further CT scans occurred in patients for trauma/presumed trauma but did not undergo HIDATq screening because the triage category was not head or facial injury. One of these scans was abnormal. The other six abnormal scans were among those screened positive. There was one intracranial haemorrhage. The most common injury was an orbital fracture. Those patients intubated or transferred did not have any intracranial injuries and had good outcomes (Drowning and Polytrauma).

Table 1

Imaging, abnormalities and outcomes

Primary outcome

A total 1052 (61%)/1739 screened negative (table 1). One of these patients had a CT scan (no abnormalities). A total of 687 patients screened positive and 60 underwent CT.

Table 2 shows the test characteristics for the HIDATq tool. For determining the need for CT, sensitivity was 98.4% (95% CI 90.0%–99.9%) and negative predictive value was 99.9% (99.4%–99.9%). For detecting ICI, negative predictive value was 100% (99.0%–100%).

Table 2

HIDATq screening sensitivity/specificity table

Secondary outcomes

In the HIDATq negative group, 349 (33%)/1052 had no other injuries and 543 (52%) presented only with abrasions or lacerations. There were four admissions from the entire HIDATq −ve group and all were discharged within 24 hours of attendance. TBI-CT,ciTBI, admissions and transfers are noted in table 1.

In total, 20% (349/1739) of all patients with head injury screened negative with no other injuries and 31% screened negative with abrasions or lacerations.

There were 35 (2%) reattenders (figure 2) to ED within 72 hours of discharge from the study population. From this group 86% (30/35) screened positive on initial presentation and all screened positive on re-presentation. Two CTs were performed among the reattenders on re-presentation with no abnormalities detected. The 5/35 who screened negative on initial presentation all re-presented with vomiting within 24 hours of discharge.

Five of the reattenders were admitted and all were discharged within 24 hours. In this group, four patients screened positive on both presentations and one patient screened negative initially with their second presentation being positive. This patient was admitted with hypoglycaemia secondary to vomiting but a CT was not performed.


The sensitivity and specificity of HIDATq (table 2) performs favourably in comparison to other screening tools/CDTs4–6 with a negative predictive value of 100%. The positive predictive value is low but this CDT was not designed for this purpose.

A total of 349 screened negative and had ‘no other injuries’, with no CTs were performed. In this group of patients with head injury it suggests that all (349/1739, 20%) could have been safely discharged at triage with appropriate discharge and safety netting advice.

In the 543 screened negative with ‘abrasions or lacerations’, the number for potential discharge at triage increases to 51% (892/1739), if their wounds could be cleaned and/or closed at triage. This is possible at our centre as triage nurses are competent to close wounds but this may not be practical at others. From this group, one CT (no abnormalities) occurred contrary to local guidelines. This was a 2-year old who presented with an alleged fall from standing and one episode of vomiting. The doctor involved in the child’s care was not a regular ED staff member.

Variance is well known to occur in healthcare.8–10 Within this small study we have demonstrated CT imaging and admissions occurred despite screening negative. The implementation of HIDAT could potentially prevent or reduce this variance. However, the decision to admit and/or CT scan a patient is multifaceted. Babl et al 11 showed that clinician accuracy can be better than any of the CDT’s currently in the literature and this variance may not necessarily be reduced by HIDAT.

The potential benefit of the tool is that patients would not occupy ED for prolonged periods as they are discharged quickly and could reduce ED crowding. It could also reduce patient waiting times to see a doctor as they would not need to see the patients who screened negative. However, the tool could have a negative impact on other aspects of ED patient care (eg, triage time) as resources/nursing time are diverted to these patients when closing/cleaning wounds or giving head injury discharge advice. Modelling and/or real-time monitoring would be required, if implemented.

Kuppermann et al 4 in their derivation study did not include children with ‘trivial head injuries’ as defined by ‘ground-level falls or walking or running into stationary objects, and no signs or symptoms of head trauma other than scalp abrasions and lacerations’.4 We believe this group of ‘trivial head injuries’ forms a proportion of the 20%–51% of patients who screened negative (table 3) and have ‘no other injuries’ or ‘abrasions or lacerations’. Despite being defined as ‘trivial’ the general public (or even other health professionals) may not define them as such and patients attend ED. Dedicated studies into this subgroup of paediatric head injuries appear warranted as there are none currently in the literature and as a group have been actively excluded from the most validated CDT to date.2 4 A sustained public health education programme may also have an impact on these attendances.

Table 3

Negative HIDATq screening and outcomes (n=1052)

The median age attending with a head injury was 6 years. This is similar to other studies.12 Those <1 year could not screen negative due to a local policy of all children <1 year requiring formal examination by a doctor to ensure there are no safeguarding concerns for the child (n=194).

Only 3.5% of all children screened had a CT. This is markedly lower than rates reported in the literature of 10%–58%.12–14 This could be attributed to a locally higher threshold for CT imaging of children and a lower threshold to ‘admit and observe’. Of those who represented <72 hours, only 6% had a CT and 14% were admitted. At other institutions these patients may have undergone CT. The low CT rate could also be due to the large numbers of patients with minor head injury presenting and who may not present to other EDs.

Eleven CTs did not trigger HIDATq (table 1) but clinicians felt a head injury was related to the presenting complaint or the patient’s condition. For example a patient with ‘Polytrauma’ was involved in a high-speed vehicle crash with multiple injuries and presented with a Glasgow Coma Score <14 (abnormal CT). An infant with ‘Seizure’ required intubation and had a CT scan to rule out intracranial pathology (normal CT).


The number of abnormal CTs for head injuries was low with only 6 (10%) showing any abnormality. This may reflect the type of patients presenting to our ED which is not a major trauma centre. However, HIDATq shows concordance with the recently published NEXUS II criteria.12 Coupled with the high negative predictive value for CT or TBI-CT, HIDATq may still be applicable.

Not all children with a head or facial injury were captured by HIDATq. This may have led to an underestimate of the number of head injuries within our study. However, those children ‘missed’ would not be the children potentially discharged using the CDT and are at less risk of an inappropriate discharge.

Sensitivity, specificity and the secondary endpoints of ICI have been calculated assuming all those screened negative and who did not undergo CT did not have an ICI. Other than those returning to the ED, patients were not able to be followed up. Monitoring of our local patient incident system has not highlighted any missed ICIs and the authors had to assume any missed cases would be highlighted as a clinical incident. Future studies will require more detailed follow-up.

It was noted that 85 children had ‘dangerous mechanism’ using HIDATq (table 4). Review showed 15 (18%)/85 had a ‘dangerous mechanism’ as per figure 3 and 70 (82%)/85 did not. This highlights an area of potential bias. Staff were aware the data entered would be used to validate the CDT so may have been over cautious. ‘Dangerous mechanism’ also has a different definition locally (eg, fall >2 m) for when urgently calling the trauma team and may have contributed to this result. All 70 patients triggered HIDATq on other variables (vomiting n=60 and loss of consciousness n=30) and would not have been potentially discharged at triage. Routine checking of other patient answers did not occur.

Table 4

Summary of all patients (n=1739) screened using HIDATq. results given as n=positive answers

The number of patients in our study are low compared with those referenced2 4–6 and is a major limitation. A larger study involving multiple centres is required to validate HIDAT for more widespread application.

Reattendance <72 hours was felt to be an appropriate time period for a significant brain injury (TBI-CT or ciTBI) to represent. It has been shown2 not all children with TBI-CT or ciTBI re-present within 72 hours, though neither child (n=2) in the referenced study2 underwent neurosurgical intervention. A longer time period is required in future studies.

Interobserver variation using HIDATq could not be measured. Questions within HIDATq, for example Question 6 ‘Focal neurology?’ could be subject to different interpretation. From our small study this does not appear to have led to an adverse outcome locally but is a limitation for generalisability. Whether patients with ambiguous or unknown answers should be considered separately or excluded from analysis is for future studies to consider.

Re-presentations were sourced via an electronic algorithm and not manually screened. Some re-presentations may have been missed from analysis. Patients may have re-presented to their GP or other local EDs. Monitoring of our safety reporting system has not shown any known incidents.


A negative HIDATq on paediatric patients presenting to our ED appears to be safe and would not have missed any significant brain injuries found on CT scan. Between 20% and 50% of all patients with head injuries presenting to our department could potentially be discharged by nurses after triage with adequate safety netting advice as ‘ultra low risk’ for significant intracranial pathology. This could reduce paediatric ED crowding and waiting times at our centre but a large multicentre study is required to validate the tool.



  • Handling editor Mary Dawood

  • Contributors PA is the overall guarantor who planned, conducted and reported the study. HC and RR planned and reported the study. ER and CP reported the study. RGL reported the study, including statistical analysis.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting or dissemination plans of this research.

  • Patient consent for publication Not required.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement Data are available upon reasonable request. Deidentified participant data available from main author, Data available for 2 years from publication. Reuse permitted in correspondence with main author. No additional information available.

Request Permissions

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.