Reduction of paediatric head CT utilisation at a rural general hospital emergency department

Abstract

Background

Blunt head injury is a common pediatric injury and often evaluated in general emergency departments. It estimated that 50% of children will undergo a head computed tomography (CT), often unnecessarily exposing the child to ionizing radiation. Pediatric academic centers have shown quality improvement (QI) measures can reduce head CT rates within their emergency departments. We aimed to reduce head CT utilization at a rural community emergency department.

Methods

Children presenting with a complaint of blunt head injury and were evaluated with or without a head CT. Head CT rate was the primary outcome. We developed a series of interventions and presented these to the general emergency department over the duration of the study. The pre and intervention data was analysed with control charts.

Results

The preintervention and intervention groups consisted of 576 children: 237 patients with a median age of 8.0 years and 339 patients with a median age of 9.00 years (p=0.54), respectively. The preintervention HCT rate was 41.8% (95% CI 35.6% to 48.1%) and the postintervention rate was 27.7% (95% CI 23.3% to 32.7%), a decrease of 14.1% (95% CI 6.2% to 21.9%, p=0.0004). During the intervention period, there was a decrease in HCT rate of one per month (OR 0.96, 95% CI 0.92 to 1.00, p=0.07). The initial series of interventions demonstrated an incremental decrease in HCT rates corresponding with a special cause variation.

Conclusion

The series of interventions dispersed over the intervention period was an effective methodology and successfully reduced HCT utilisation among children with blunt head injury at a rural community emergency department.

INTRODUCTION

Evaluation for head injury is the most frequent emergency department (ED) presentation leading to utilisation of head CT (HCT) in children, and accounts for 50% or more of all paediatric trauma CTs.^1–5 ED rates of HCT utilisation for paediatric head injury vary widely, both by institution, with rates as high as 69%,^3–8 and by ED provider.^8,9 While the rate of paediatric HCT utilisation appears to be decreasing in recent years,^4,10–13 the prevalence of intracranial traumatic brain injury (TBI) continues to be less than 10%, and the need for neurosurgical intervention, particularly for majority of children who present with a Glasgow Coma Scale (GCS) score of 14–15, is only 1%.¹⁴

In 2009, a large multicentre research network, the Pediatric Emergency Care Applied Research Network (PECARN), published the results from a prospective study intended to identify those children presenting with minor TBI (GCS score of 14–15) who have a clinically important traumatic brain injury (ciTBI).¹⁴ These clinical prediction rules performed well, with very high sensitivities and negative predictive values for paediatric ciTBI, and have been widely adopted since publication.¹⁴ Several paediatric EDs have independently validated the PECARN head injury guidelines (HIG).^15–17

To date, implementation of the PECARN HIG to reduce utilisation of HCT and its associated cancer risk has predominantly been applied to children evaluated in paediatric EDs at children’s hospitals.^7,9–11,18 However, 89% of paediatric ED visits occur in general (non-children’s) hospitals, where the vast majority of paediatric HCTs are performed.^3,19,20 General EDs, as opposed to paediatric EDs, have been found to have significantly higher HCT utilisation rates,^3,6,8,21 at higher radiation levels,^22–24 and with more associated cervical thoracic or abdominal CT scans.¹ If paediatric trauma patients are then transferred for further care, they frequently have duplicate CT scans performed at the referral children’s or trauma hospital, further increasing their radiation exposure and cancer risk.^1,23,25–27

We sought to use quality improvement (QI) methodology and data analysis to implement the PECARN HCT decision rules and effectively reduce paediatric HCT rates in a rural community hospital general ED that treats both adults and children, without inhouse paediatric resources.

METHODS

Study setting

University emergency medicine faculty collaborated with a rural general community hospital (GCH) ED for a paediatric HCT reduction QI project. The GCH is located 50 min northeast of a major metropolitan area and staffed by 15 emergency medicine physicians (12 emergency medicine and 3 family practice) and 4 physician assistants. It is a state-designated level III trauma centre with an annual ED census of approximately 30 000 adult and 2000 paediatric (age less than 18 years) patients. The GCH does not have inhouse paediatric hospitalist coverage or other paediatric resources to support the ED. The GCH is among seven other hospitals in the healthcare system, which includes a university children’s hospital, all with a common electronic health record (EHR). The researchers were paediatric emergency medicine (PEM) physicians from a university children’s hospital located in an urban metropolitan city.

Planning the intervention

Developing a collaborative relationship with the community ED

Prior to the initiation of this project, an emergency medicine researcher met with the GCH ED medical director and presented a strategy to change ED culture on the management of paediatric head injuries, with agreement on the following principles. First, all emergency medicine research interactions should occur on-site and only during existing staff meetings, with an allotted time of 30 min. Second, all educational materials and outcome data would be provided to the medical director prior to dissemination to ED staff. Third, the medical director would use the data, particularly data regarding individual provider HCT rates, at his discretion. Fourth, the medical director, as the site champion, could at his preference provide additional education to ED staff, without the knowledge of or assistance from the research team.

Introduction of HIG: preceding intervention of the QI study

In April 2012, before the QI start date on 1 January 2013, one of the authors introduced, summarised and distributed a paediatric HIG based on the PECARN paediatric head injury study¹⁴ at a system-wide trauma conference, with the GCH ED medical director in attendance. The conference was designed to educate and provide standardisation throughout the multihospital system. The GCH ED medical director discussed the HIG with the ED group in May 2012. In communication with the GCH medical director, no other interventions were used to encourage providers to reduce HCT use among children. This ‘passive diffusion’ of the HIG will be identified as the ‘preceding intervention’ in order to clarify and document real-time events with the study results.

Changing culture

We employed a variety of formats to change ED culture and created several categories. The educational format consisted of didactic talks on the PECARN study,¹⁴ the HIG and ionising radiation risks to children, and question and answer session (Q and A). Data feedback provided overall and individual HCT rates. The system change pertained to posting HIG by computers and creating EHR templates and ED staff handouts.

Timeline

Table 1 outlines the timeline of events and interventions. In May 2012, the ED medical director discussed the HIG with the ED staff, as mentioned above, approximately 7 months before the QI project start date on 31 January 2013. HIG was created and developed in March 2010 as a guideline for PEM providers at the university children’s hospital. The QI project began in June 2012, and the research team started developing a key driver diagram, educational handouts and a series of interventions. In December 2012, an emergency medicine researcher met with the medical director to develop a mutual plan to decrease the ED HCT rate. As a proponent of reducing unnecessary radiation exposure among children, the medical director also volunteered to become the site champion. Key dates are shown in figure 1.

Table 1.

Timeline of the QI study

Description	Key dates and intervention months correspond to that in figure 1.
Preintervention (historical control data) period.	1 January 2012–31 December 2012
Data collection for preintervention and intervention begins.	1 January 2012
Preceding intervention	April 2012
►   Introduction of the HIG at a trauma system conference in April 2012. The GCH ED medical director discussed the HIG to providers in May 2012.
Creation of the QI project with the research team.	June 2012
EM researcher meeting with the GCH ED director.	December 2012
►   Outlined methods for culture change.
►   ED director agrees to be site champion.
Intervention period (also refer to figure 1).	1 January 2013–30 June 2014
Site visit 1	January 2013
Educational
►   Presented PECARN article.
►   Discussed the HIG.
►  Discussed radiation risks in children.
►  Q and A session.
System change
►  HIG posted at GCH ED workstations.
Site visit 2	March 2013
Educational
►  Presented PECARN article.
►  Discussed the HIG.
►  Discussed radiation risks in children.
►  Q and A session.
System change
►  HIG posted at GCH ED workstations.
►  Presented DOT phrases.
►  Incorporated DOT phrases into EHR.
–  Low risk.
–  Intermediate risk.
–  High risk.
–  Discharge instructions.
►  Created ED staff handouts.
Site visit 3	June 2013
Educational
►  Discussed the HIG.
►  Q and A session.
Site visit 4	August 2013
Data feedback
►  Presented preintervention and intervention data (from January 2013 until May 2014).
Educational
►  Q and A session.
Site visit 5	November 2013
Data feedback
►  Presented preintervention and intervention data (from January 2013 until May 2014).
Educational
►  Q and A session provided the medical director with individual provider HCT rates.
Site visit 6	March 2014
Data feedback
►  Presented funnel chart demonstrating individual provider HCT rates.
Educational
►  Q and A session provided the ED medical director with individual provider HCT rates.
Site visit 7	June 2014
Educational
►  Q and A session.
►  Summary of study.
Data feedback
►  Presented preintervention and intervention data.

DOT, dot-phrase (EHR template); ED, emergency department; EHR, electronic health record; EM, emergency medicine; GCH, general community hospital; HCT, head CT; HIG, head injury guidelines; PECARN, Pediatric Emergency Care Applied Research Network; Q and A, question and answer; QI, quality improvement.

Figure 1.

P-chart demonstrating preintervention and intervention head CT rates. A refers to the oval which denotes special cause variation. Please refer to table 1. LCL, lower control limit; UCL, upper control limit.

The QI interventions began in January 2013. Throughout the next 17 months, an emergency medicine researcher met with the GCH ED staff to reinforce the use of the HIG and discuss the risk of unnecessary imaging radiation. At the first site visit (January 2013), the ED staff requested the development of a handout outlining the risk of radiation for children, discussing points regarding the HIG and ciTBI, and EHR templates (dot-phrase). The templates would allow ED staff to easily insert prewritten statements and document medical decision-making based on PECARN rules.¹⁴ We were impressed by the amount of suggestions and were encouraged to keep engaging with the ED staff (online supplementary material 1).

Two balancing measures were agreed on by the ED medical director and staff: (1) missed ciTBI, as defined by the PECARN study,¹⁴ after any initial visit, and (2) any HCT demonstrating a skull fracture or any intracranial bleed (subdural, epidural, intraparenchymal haemorrhage or cerebral venous sinus thrombosis) at initial evaluation or children returning to the ED or from clinic follow-up. The latter would unlikely result in a ciTBI; however, the ED staff requested the outcome variable.

In August 2013, the preintervention data on HCT rate were shown to the ED staff. In March 2014, the medical director was provided with preintervention and 6-month intervention data on HCT rate, along with individual ED provider HCT rates. A funnel chart was also presented to the medical director showing the intervention HCT rate of individual staff.

Preintervention and intervention QI project dates

The preintervention group included children who presented to the GCH ED from 1 January 2012 through 31 December 2012. The intervention date was from 1 January 2013 through 30 June 2014.

Abstraction of data

Patients were eligible for study inclusion if they were 3 months to 17 years of age and presented to the GCH ED for evaluation of an acute head injury or concussion within 24 hours of injury. We used the International Classification of Diseases-9 discharge codes for blunt head injury or concussion (959.1, 850.0–850.9) to identify eligible patients. A corresponding medical record identifier associated with the encounter date was provided by the EHR Medical Record Department. Based on the date of encounter, every patient’s EHR was reviewed and data variables were manually abstracted and entered into a spreadsheet.

Children who presented with a GCS score <15 and intoxicated patients, defined as a serum alcohol level of equal to or greater than 0.08 mg/dL, or a positive urine sample for illicit drugs, were excluded. We also excluded infants less than 3 months of age. This population of patients is more difficult to clinically assess, apply a GCS and is more likely to be worked for non-accidental trauma, and the PECARN study,¹⁴ at the time of our QI study, had not been externally validated by other researchers.

Each initial ED encounter was followed for 7 days for revisits to the ED or clinic (provided the patient was within the electronic medical record (EMR) healthcare system). The GCH Radiology Department was queried for children undergoing HCT during the intervention period and their data were compared with the EHR. This process ensured that cases were not missed from any initial or subsequent ED visit, and allowed verification if a primary care physician ordered an HCT for any follow-up visit. Data abstracted included demographic information, date of ED encounter, any return visits or clinic follow-up, mechanism of injury, patient intoxication, GCS score or documented mental status of ‘alert’ or ‘acting normally’, and HCT results. Criteria for ciTBI, defined as death from TBI, intubation for more than 24 hours for TBI, neurosurgery (procedure) or hospital admission of 2 nights or more,¹⁴ were also abstracted. A missed ciTBI was defined as a patient who returned to ED after initial visit and subsequently had an HCT, or as a return ED visit or ordered from the clinic and then found to have a ciTBI.¹⁴ All admissions were followed until disposition. Data were entered into an Excel spreadsheet (Microsoft, Redmond, Washington) and analysed using statistical process control (SPC) charts (QI Charts, V.2.022, Scoville Associates, 2009).

Methods of evaluation and data analysis

The primary QI outcome measure was to determine whether our series of interventions reduced overall HCT rates when compared with the preintervention baseline period. Only initial ED visits and HCT use were included in the final analysis. ED revisits or clinic visits were abstracted on a separate spreadsheet. We developed run and SPC (Shewhart Process Control) charts, mainly the P-chart, and followed established rules for determining special cause variation.^28,29 The PECARN HIG categorised two age groups: <2 years of age and ≥2 years of age. Each age group pertained to unique clinical risk factors for ciTBI.¹⁴ We will demonstrate our main outcome of HCT rates and demographic data to parallel the PECARN guidelines.

The overall preintervention and intervention ED HCT rates were compared using a proportion χ² test and segmented logistic regression. Segmented logistic regression was used to estimate the change in HCT rates in the preintervention versus the intervention period. The model included terms for preintervention slope, preintervention intercept, postintervention slope and postintervention intercept. The estimated difference in slopes (or intercepts) was calculated as the difference of the model-estimated coefficients (R Core Team 2017; R: A language and environment for statistical computing, R Foundation for Statistical Computing, Vienna, Austria; http://www.R-project.org/). SPSS V.24.0 was used to compare HCT rates, differences in means of patients’ age, medians with interquartile (IQ1) 25% and (IQ3) 75%, gender proportions, and 95% CI.

Data reliability

Data abstraction was performed by a trained research assistant and by a GCH ED registered nurse, serving as the site trauma coordinator, both of whom were trained by one of the study authors. To determine data agreement, a Cohen kappa statistic was performed (SPSS V.24.0). One researcher (JPL) extracted 100% of children with any ciTBI and randomly selected 50% of subjects who underwent HCT in the preintervention and intervention period to verify demographics, HCT utilisation and associated results, and GCS or mental status, and were appropriately selected to meet the inclusion and exclusion criteria. A kappa value of 0–0.20 indicates no agreement, 0.21–0.40 fair, 0.41–0.60 moderate, 0.61–0.80 substantial and 0.81–1 almost perfect agreement.

Patient and public involvement

This research was done without patient involvement. Patients were not invited to comment on the study design and were not consulted to develop patient relevant outcomes or interpret the results. Patients were not invited to contribute to the writing or editing of this document for readability or accuracy.

RESULTS

A total of 588 paediatric patients presented for evaluation of head injury to the GCH ED during the HCT study period. Twelve patients were excluded from analysis: six teenagers who all had elevated alcohol levels above legal limits of 0.08 mg/dL and had GCS score <15; two teenagers presented with GCS score <15 and were involved in motor vehicle collisions (MVCs), with multiple injuries; two children had severe head injuries (GCS score <8) related to MVC; and two infants were less than 3 months of age, and both children presented to the ED intubated from the field and were involved in MVC (one infant had a penetrating head injury). The final study population consisted of 576 children. The monthly median number of patients presenting to the GCH ED for head injury was 19 (mean 19.4, SE 0.91, range 12–31). The median number of HCT performed per month was 6.0 (mean 6.67, SE 0.57, range 0–12).

All children were followed for 7 days after initial ED encounter to determine if they either returned to the ED for an HCT, a repeat HCT for worsening symptoms or an outpatient HCT ordered by a clinic physician. We felt 7 days of follow-up was clinically a reasonable time span for most ED encounters, but we acknowledge that if any child had a missed ciTBI after initial examination they would likely represent within 24–48 hours. Children who had an initial HCT and not found to have ciTBI did not return to the ED. Three children clinically evaluated without an HCT on initial visit returned to the ED within 7 days. They were reassessed and noted to be neurologically stable and discharged home without head imaging.

Preintervention and intervention patients were compared with regard to HCT rate, median ages and gender proportions, and the results are shown in table 2. The overall HCT rate in the preintervention group was 41.8% and decreased to 27.7% after the intervention. The decrease in HCT rate of 14.1% was significant (p=0.0004, 95% CI 6.2% to 21.9%). Preintervention and intervention patients were grouped by age category to parallel the PECARN guidelines.¹⁴ There were no significant differences among the mean age groups: preintervention <2 years compared with intervention <2 years, preintervention ≥2 years compared with intervention ≥2 years, and overall preintervention and intervention (p=0.51, p=0.12, and p=0.54, respectively). Gender proportions were also provided. The overall male proportion preintervention of 60.3% (95% CI 54% to 66.4%) and at intervention of 67.7% (95% CI 62.4% to 72.3%) was not significant (p=0.076, 95% CI −15.2% to 0.76%). The overall female proportion preintervention was 39.7% (95% CI 33.7% to 46.0%) and at intervention was 32.5% (95% CI 27.7% to 37.6%).

Table 2.

Comparison of preintervention and intervention groups: head CT rates and demographics grouped by age category^*

	Preintervention (n=237)			Intervention (n=339)
	<2 years old	≥2 years old	Total	<2 years old	≥2 years old	Total
Head injury (n)	31	206	237	58	281	339
Head CT(n)	8	91	99	10	84	94
Head CT rate, % (95% CI)	25.8 (13.7 to 43.3)	44.1 (37.6 to 51.0)	41.8 (35.7 to 48.1)	17.2 (9.64 to 28.9)	29.9 (24.8 to 35.5)	27.7 (23.2 to 32.7)
Age, median (IQ1-IQ3)	1.0 (0.69–1.0)	9.29 (4.0–14.0)	8.0 (3.0–14.0)	0.86 (0.74–1.0)	11.0 (5.23–14.0)	9.0 (3.0–13.0)
Gender
Male, n (%) (95% CI)	17 (54.5)	126 (61.2)	143 (60.3)	33 (56.9)	196 (69.6)	229 (67.6)
	(37.8 to 70.8)	(54.4 to 67.6)	(54.0 to 66.4)	(44.1 to 68.8)	(64.1 to 74.8)	(62.4 to 72.3)
Female, n (%) (95% CI)	14 (45.2)	80 (38.8)	94 (39.7)	25 (43.1)	85 (30.3)	110 (32.5)
	(29.2 to 62.2)	(32.4 to 45.6)	(33.7 to 46.0)	(31.2 to 55.9)	(25.2 to 37.6)	(27.7 to 37.6)

^*Subgrouped according to the Pediatric Emergency Care Applied Research Network.

IQ1. interquartile 25% IQ3. interquartile 75%.

A segmented logistic regression model was used to estimate HCT trends over time. Preintervention odds of needing an HCT changed by a factor of 1.01 per month (95% CI 0.94 to 1.08, p=0.86), indicating a stable trend throughout the preintervention period with just over 40% of patients getting an HCT. In the month immediately following the intervention, the odds decreased by a factor of 0.74 (95% CI 0.37 to 1.48, p=0.40). Intervention odds of needing an HCT changed by a factor of 0.96 per month (95% CI 0.92 to 1.00, p=0.07). The estimated OR for the postintervention change (slope) relative to preintervention change was 0.95 (95% CI 0.94 to 1.08, p=0.26).

The characteristics of children with abnormal HCT on initial evaluation or found to have ciTBI are outlined in table 3. The preintervention cohort included two children with ciTBI defined by the PECARN criteria¹⁴ and two children with abnormal HCT findings. No children in the intervention group met the ciTBI criteria, but two children with abnormal HCT findings did. These children were hospitalised for observation and discharged home in less than 24 hours.

Table 3.

Characteristics of patients with ciTBI or abnormal head CT findings

Patient identification and age (years)	Mental status on presentation (documented as normal or according to GCS)	Mechanism of injury	Head CT report	ciTBI criteria*	Comments
Preintervention
Patient A, 1.1	Normal†	Fell off the couch, 0.9 m.	Subdural haematoma.	S/P craniotomy.	Abuse ruled out.
Patient B, 2	Normal	6 m fall out of the window.	Small intraparenchymal haemorrhage in the peripheral left parietal lobe and right orbit fracture.	Intubated for emesis; hospitalised for ≥2 nights.	Extubated in 7 hours. No neurosurgical procedures required.
Patient C, 4	15	Fell out of an electric children's car.	Left temporal bone fracture with suprasellar haemorrhage.	NA.	Discharged in <48 hours.
Patient D, 13	15	Fell from sitting in chair.	Non-displaced occipital skull fracture.	NA.	Discharged in <24 hours.
Intervention
Patient E, 0.33	Normal	Fell down the stairs in paren’s arms.	Linear skull fracture, left parietal bone.	NA.	Discharged in <24 hours.
Patient F, 9	15	In a tire swing and hit head against a tree.	Right temporal bone fracture, right small epidural.	NA.	Discharged in <24 hours.

^*ciTBI criteria: 1, intubated for head injury; 2, neurosurgical procedure; 3, hospitalised for ≥2 nights; 4, death.

^†Presented with dysconjugate gaze.

ciTBI, clinically important traumatic brain injury; GCS, Glasgow Coma Scale; NA, not appicable; S/P, status post.

Figure 1 shows a P-chart demonstrating the HCT utilisation rates and includes the preintervention months (January 2012–December 2012) and intervention months (January 2013–June 2014). The P-chart centreline represents the mean HCT rate per month and highlights the preintervention and intervention HCT rates of 41.8% and 27.7%, respectively. Two periods of special cause variation occurred during the intervention period. The first special cause variation occurred during the months of April 2013 through September 2013, with the six consecutive months showing a descending order.³⁰ The decreasing HCT rate corresponds to the emergency medicine researcher performing multiple site visits (see table 1). The second special cause variation occurred in April 2014, with a drop in the monthly HCT rate to 0, the lower control limit.³¹ Online supplementary table S1 and figure S1 demonstrate the HCT rate by month with 95% CI, and a funnel chart demonstrating the HCT rate by individual ED staff from January to December 2013 was created for the ED director, respectively.

Kappa statistics revealed 1.0 agreement with children meeting ciTBI criteria. For children who underwent HCT and were randomly selected, the agreement for GCS or mental status was 0.83 (95% CI 0.61 to 1.0), for meeting the inclusion and exclusion criteria was 0.87 (95% CI 0.89 to 1.0), for demographics was 0.94 (95% CI 0.82 to 1.0), and for use of HCT and HCT results was 1.0 (100% agreement, respectively).

DISCUSSION

We successfully used QI methodology to reduce the monthly HCT rate in children with blunt head injuries in a rural GHC ED. We noted a reduction from a preintervention HCT rate of 41.8% to 27.7% following our series of QI interventions, for a significant overall reduction of 14.1%. To our knowledge, our QI study is one of the first to collaborate between an academic university children’s hospital and a general rural community hospital ED to reduce paediatric HCT utilisation rates among children with normal GCS.

While 89% of paediatric visits occur in general hospital EDs,^3,19,20 previously published studies of reductions in paediatric HCT rates have generally been conducted in academic paediatric EDs.^7,18 The need to reduce HCT rates in general ED settings has been recognised.^3,6,32 In a recent article by Burstein et al³³ the authors found the national HCT rate was 32% and argued for increased QI initiatives focused at general (non-teaching and non-children’s) hospitals.

Children’s hospital EDs have also had published HCT rates similar to our community hospital ED, particularly our intervention rate of 27.6%. Mannix et al¹⁰ published a study of 40 paediatric hospital EDs using the Pediatric Health Information System database and found a median HCT rate of 36%, with no correlation between individual hospital ED HCT rates and a number of patient clinical head injury outcomes. Stanley et al⁸ in a secondary analysis of the original PECARN HCT database, which was gathered primarily from academic children’s hospitals, noted that HCTs for head trauma were more frequently performed at suburban and non-free-standing children’s hospitals, with hospital HCT rates ranging from 19% to 69%.⁸ They also noted that HCT was overused in all of the hospital EDs, and by all types of ED providers, particularly for low-risk patients, without an apparent relationship to ED-specific patient injury severity or ciTBI rates.⁸

Three studies have engaged community EDs using QI and successfully reduced HCT rates.^34–36 Two of the studies^34,35 provided Maintenance of Certification for participating physicians. Both had either general paediatricians in the hospital³⁴ or PEM physicians in the ED,³⁵ allowing inhouse or direct clinical support, and were affiliated with an academic children’s hospital, a resource lacking in smaller community hospitals across the country. A third study, by Puffenbarger et al³⁶ showed a decrease in HCT rate of 20.8% in a community ED (preintervention HCT rate of 37.7% and postintervention HCT rate of 16.9%). Similar to our community ED, there was no inhouse paediatric support. Trauma designation was not provided in the above studies.^34–36

Our study is unique for several reasons. We focused on the ED of a level III trauma centre located in a rural area. Our study provided a timeline, detailing each intervention which appeared to correlate with a reduction in HCT rate and was associated with two special cause variations. We described site visits with our three categories of interventions: educational, system change and data feedback. In addition, we documented how we used ED staff suggestions and incorporated their ideas into additional interventions, creating a more collaborative atmosphere. The Q and A sessions, we found, were also beneficial for ED staff and allowed them to discuss cases with a PEM physician. We also provided the medical director with each ED staff HCT rate.

We faced some challenges during the study. First, limited EHR resources made it exceedingly challenging to develop clinical decision support tools as other studies have documented.¹³ Second, the manual abstraction of data was time-consuming. Third, the inability to obtain data in a timely fashion was a barrier and prevented real-time data analysis. Without timely data, a priori, we understood we would not be able to gauge any impact from our initial series of interventions. One month of cleaned data took approximately 3–4 months of manual data abstraction, making it impractical to perform timely root cause analysis and initiate new and unique PDSA (Plan–Study–Do–Act) cycles. During the planning phase of the study, we realised that in order to strengthen the collaboration and encourage the GCH ED staff to use less HCT (without the benefit of timely data), we chose to perform site visits every 2–3 months, dependent on GCH ED monthly meeting agendas, the emergency medicine researcher’s schedule and the weather.

All of the planned series of interventions—educational, system change, data feedback—and having a site champion were based on prior reviewed QI literature.^37–39 Each site visit was an important conduit in developing a working collaborative relationship to change attitudes towards the use of HCTs. Accepting input from ED staff, incorporating these changes and developing new material at their request, we felt, were helpful in changing culture. Finally, having the medical director serve as the site champion was likely another factor for this study’s success.

Limitations

Our study has a number of limitations. As is true of most QI studies, the study is retrospective in nature and subject to the inherit source of misinformation or errors associated with electronic medical record documentation. Although unlikely, given that the GCH is the only trauma centre and hospital within a 48-km radius, a patient could have sought treatment in an ED outside of our healthcare system after the initial evaluation. The likelihood of a missed ciTBI using the PECARN guidelines, as demonstrated in a number of subsequently published studies, is extremely low.^39–41 We identified only two ciTBI cases in the preintervention data and none following our intervention. This difference in the number of ciTBI cases could have resulted from factors outside of our study, such as an independent change in patient referral patterns and regional emergency medical service practice. We were not aware of any such changes in regional referrals or EMS practice during the study. Although we demonstrated a significant decrease in the paediatric HCT rate, we did not assess ED provider or patient satisfaction as an outcome of this study. Thus, determining which intervention(s) was most agreeable or disliked is not known.

CONCLUSION

Our study is one of few published to date demonstrating a significant decrease in paediatric HCT utilisation in the evaluation of minor head trauma within a rural community hospital ED. Working in partnership with the ED medical director, we established the feasibility of sustainable practice change with implementation of paediatric guidelines using QI strategies.