BACKGROUND AND OBJECTIVE: Clinical decision rules have reduced use of computed tomography (CT) to evaluate minor pediatric head injury in pediatric emergency departments (EDs). CT use remains high in community EDs, where the majority of children seek medical care. We sought to reduce the rate of CT scans used to evaluate pediatric head injury from 29% to 20% in a community ED.
METHODS: We evaluated a quality improvement (QI) project in a community ED aimed at decreasing the use of head CT scans in children by implementing a validated head trauma prediction rule for traumatic brain injury. A multidisciplinary team identified key drivers of CT use and implemented decision aids to improve the use of prediction rules. The team identified and mitigated barriers. An affiliated children’s hospital offered Maintenance of Certification credit and QI coaching to participants. We used statistical process control charts to evaluate the effect of the intervention on monthly CT scan rates and performed a Wald test of equivalence to compare preintervention and postintervention CT scan proportions.
RESULTS: The baseline period (February 2013–July 2014) included 695 patients with a CT scan rate of 29.2% (95% confidence interval, 25.8%–32.6%). The postintervention period (August 2014–October 2015) included 651 patients with a CT scan rate of 17.4% (95% confidence interval, 14.5%–20.2%, P < .01). Barriers included targeting providers with variable pediatric experience and parental imaging expectations.
CONCLUSIONS: We demonstrate that a Maintenance of Certification QI project sponsored by a children’s hospital can facilitate evidence-based pediatric care and decrease the rate of unnecessary CT use in a community setting.
- CI —
- confidence interval
- ci-TBI —
- clinically important traumatic brain injury
- CT —
- computed tomography
- ED —
- emergency department
- EMR —
- electronic medical record
- ICD-9-CM —
- International Classification of Diseases, Ninth Revision, Clinical Modification
- IQR —
- interquartile range
- LOS —
- length of stay
- MOC —
- Maintenance of Certification
- PA —
- physician assistant
- PECARN —
- Pediatric Emergency Care Applied Research Network
- QI —
- quality improvement
- SPC —
- statistical process control
- TBI —
- traumatic brain injury
Pediatric head trauma is a common reason to seek emergency department (ED) care in the United States, accounting for 650 000 visits per year.1 Patients with head trauma present a diagnostic challenge. Clinicians need to quickly identify serious traumatic brain injuries (TBIs) while limiting the radiation exposure, sedation risk, and cost from unnecessary head computed tomography (CT). Children are particularly susceptible to the carcinogenic properties of radiation; it is estimated that 1 case of leukemia results from every 5250 head CT scans performed on children <5 years old.2
The Pediatric Emergency Care Applied Research Network (PECARN) created a validated prediction rule to identify pediatric patients with blunt head trauma at very low risk of clinically important TBI (ci-TBI), who can safely be evaluated without a CT scan.3 The PECARN guidelines identify multiple risk factors for ci-TBI for patients with minor head trauma (eg, severe mechanism of injury, loss of consciousness, palpable skull fracture). Among patients with no risk factors, the prediction rule has a high negative predictive value for ci-TBI of 100.0% for children <2 years old and 99.95% for children ≥2 years old. Since the publication of the PECARN guidelines, multiple quality improvement (QI) projects have achieved significant reduction in head CT scan rates for pediatric head injury in academic pediatric EDs,4,5 but there have not been similarly reported efforts in community settings. It is important to address head CT use in general EDs because 89% of emergency visits in the United States for patients <14 years old are in general EDs6 and head CT scan rates are significantly higher among patients who present to general EDs (22%) compared with pediatric EDs (13%).7,8
Leaders in the study hospital’s pediatric department and ED identified pediatric head CT use as an area for improvement because the baseline head CT scan rate was significantly higher than the rate observed at other general EDs.8 The goal of reducing head CT scans aligned with the Washington State Hospital Association 100K Children Campaign, whose aim is to reduce pediatric radiation exposure in Washington hospitals.9 We established a multidisciplinary team, including nurses, general pediatricians, and ED physicians, to identify key drivers of CT use and implement decision aids to improve the use of head trauma prediction rules for pediatric TBI. We used the Maintenance of Certification (MOC) Multispecialty Portfolio Program from an affiliated academic children’s hospital to provide support and coaching for the QI project. The aim was to reduce the rate of head CT scans for the evaluation of pediatric head injury in a general ED from the current median rate of 29% to a goal median rate of 20% within 12 months.
The specific aim of the project was to reduce the head CT scan rate among pediatric patients with head trauma from 29% to 20% within 12 months. We chose 20% because it is near the published national average for general EDs.8 Primary drivers identified for implementation of evidence-based care were ED provider knowledge of head trauma prediction rules, standard decision process for CT imaging, and effective caregiver education about head trauma and imaging (Fig 1).
Providence St Peter Hospital is a community hospital in Olympia, Washington, 60 miles from Seattle, with 12 000 pediatric visits per year. It is staffed by 24 emergency medicine and family medicine trained physicians and 7 physician assistants (PAs). The PAs see less acute patients. They see patients independently and consult with the ED physicians if they need guidance. Pediatric hospitalists contracted with Seattle Children’s Hospital provide 24-hour in-hospital consultation. The closest pediatric neurosurgical consultant is 28 miles away.
Seattle Children’s American Board of Medical Specialties Multispecialty MOC Portfolio Program was accredited in 2012 and maintains a portfolio of approved MOC projects aimed at improving health outcomes for children in the Pacific Northwest. The program provides MOC credit to >200 physicians per year from multiple specialties.
The Providence Institutional Review Board approved this study and waived participant consent.
Planning the Intervention
We convened a multidisciplinary team that included leaders (nurses and physicians) from the ED, trauma team, pediatric department, and a regional pediatric hospital MOC portfolio program. The project was led by a pediatric hospitalist and a general ED physician. Seattle Children’s MOC Portfolio program provided QI consultation, and participants were eligible for MOC credit. The team evaluated the impact of the interventions on patients <18 years old who presented to the ED with head injury from February 2013 to October 2015. Improvement performance, overall and by individual clinician, was evaluated monthly via annotated statistical process control (SPC) charts.
The team met and developed interventions, or secondary drivers, to address the primary drivers. The team created an evidence-based decision support tool, or clinical protocol, that was adapted from the PECARN prediction rule. This protocol was finalized through feedback with key stakeholders at multiple planning meetings in July 2014. The QI initiative was then launched for the ED physicians during an initial educational meeting in August 2014, where the PECARN prediction rule was reviewed and the new clinical protocol was introduced. The clinical protocol was posted on a laminated sheet at each physician workstation in the ED. Laminated pocket-sized cards were distributed to all physicians and PAs. To address caregiver education, we discussed possible scripts to use when discussing head injuries, concussions, and the role for CT scans with patients and caregivers.
Baseline CT scan rates were reviewed and compared with national averages at the stakeholder and ED provider meetings. The project’s progress was reviewed every 3 to 4 months at ED provider meetings, and barriers to implementation were addressed. An annotated run chart was updated monthly, publicly posted in a common area of the ED, and e-mailed to providers. Semiannually, clinicians were privately provided with their personal CT use rates, benchmarked to other providers in the group on a deidentified chart. Only providers who had treated ≥5 patients with head injuries in both the preintervention and postintervention periods were included.
We performed a retrospective medical record review on eligible subjects during the preintervention period (February 1, 2013 to July 31, 2014) and postintervention period (August 1, 2014 to October 31, 2015). We included patients <18 years old at the time of presentation to the ED with an International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) discharge diagnosis code indicating head trauma (head injury [959.01], concussion [850.xx], skull fracture [800.xx–804.xx], intracranial hemorrhage [851.xx–853.xx], or other brain injury [854.xx]).7 For patients discharged in October 2015, the International Classification of Diseases, 10th Revision discharge codes of S06.0X0A, S06.0X1A, S06.9X1A, S09.8XXA, and S09.90XA were used. Patients with an ICD-9-CM discharge code indicating comorbid conditions that might increase risk for bleeding or brain injury were excluded. Excluded ICD-9-CM discharge codes were hemophilia (286.xx), thrombocytopenia (287.xx), and ventriculoperitoneal shunt (V45.2). For October 2015, the International Classification of Diseases, 10th Revision discharge codes of D65–D69.XXX, P54, P61.0, and Z98.2 were used.
The primary outcome of interest was the rate of cranial CT imaging among eligible patients. Covariates included race, age, sex, and insurance status (public versus private). Balancing measures were readmissions within 72 hours to the index ED to assess missed cases of ci-TBI, and length of stay (LOS) to assess changes in resource utilization. Patients with an LOS >12 hours were excluded from the LOS analysis if they were evaluated by the crisis counselor for psychiatric comorbidity after their initial evaluation for head trauma. The provider was considered the assigned attending provider listed in the electronic medical record (EMR) encounter. If a patient was evaluated by both a PA and a physician, the physician was assigned the role of attending provider.
We summarized sample characteristics by using appropriate descriptive statistics for quantitative (mean and SD or median and interquartile range [IQR]) and categorical (counts and percentages) variables. We calculated preintervention and postintervention proportions for CT head imaging. We estimated 95% confidence intervals (CIs) for CT scan proportions by using a logit transform and Huber–White sandwich variance estimates to account for within-subject correlation due to repeat admissions (<5% of patients) and performed a Wald test of equivalence for preintervention and postintervention proportion estimates.10,11 In addition, we carried out an interrupted time series analysis based on autoregressive integrated moving average models as a sensitivity analysis to account for potential secular trends.12
SPC charts with 1-month time intervals were constructed to assess the effect of the intervention on CT scan rates. We set control limits at 3 SD from the mean and used standard criteria to identify special causes.13,14 Centerline shifts were made when 8 consecutive points fell above or below the centerline. SPC charts were constructed with QI Macros (KnowWare International, Inc, Denver, CO).
We compared preintervention and postintervention LOS by using the Wilcoxon rank-sum test. For our other balancing measure, 72-hour readmissions, we calculated preintervention and postintervention monthly readmission proportions.
The baseline period included 695 patients, and the postintervention period included 651 patients. Four patients were excluded because of comorbid conditions (2 in the preintervention period and 2 in the postintervention period). The characteristics of our patient population were similar in the preintervention and postintervention groups (Table 1).
The proportion of head CT scans declined from 29.2% (95% CI, 25.8%–32.6%) at baseline to 17.4% during the postintervention period (95% CI, 14.5%–20.2%, P < .01) (Fig 2). These results were consistent with results from the interrupted time series sensitivity analysis, which estimated that monthly CT scan rates dropped by 12.0% (95% CI, 7.4%–16.5%, P < .01). Multiple barriers were identified and addressed by the team with new or modified interventions (Table 2).
The median LOS increased from 1.5 hours in the preintervention period (IQR 0.9–2.5) to 1.9 hours in the postintervention period (IQR 1.0–2.8, P < .01).
There were 13 readmissions within 72 hours during the study period, 5 in the preintervention period and 8 in the postintervention period. None of these readmissions were diagnosed with a ci-TBI; 1 patient in the preintervention period was diagnosed with a linear, nondisplaced skull fracture.
During the baseline period, individual provider rates of head CT scan usage varied from 4% to 92%. The majority of providers (22 out of 28) demonstrated a reduction in head CT scan rate during the postintervention period (Fig 3). Among physicians, the average CT scan rate declined from 47.0% (95% CI, 41.9%–52.1%) at baseline to 31.4% (95% CI, 26.3%–36.9%) after intervention, whereas among PAs the average CT scan rate fell from 6.6% (95% CI, 4.1%–10.1%) to 2.5% (95% CI, 1.1%–4.9%).
This study demonstrates that community EDs can implement evidence-based pediatric care, particularly when coached by a regional children’s hospital. Specifically, we were able to significantly reduce the rate of head CT scans in pediatric head injury patients in a community ED from 29.2% to 17.4% and sustain this change over 15 months.
Our project was aided by a multidisciplinary QI team, with coaching from an academic center with more QI expertise. Formal mentoring has been shown to help successfully implement evidence-based recommendations and reduce variation in care in the community setting.15 The educational outreach by a pediatric ED physician from a site that had successfully implemented the PECARN rules probably helped in the adoption of the intervention.16
The majority of providers improved their rate of CT use, although variation persisted across providers. Among providers who had a higher CT scan rate in the postintervention period, there were not identifiable reasons for the increase. We found that providing feedback to individual providers about how their performance compared with others within our organization was beneficial. This feedback was particularly useful in a community setting. The pediatric-specific resources available in community EDs are often not as robust as in academic pediatric EDs, and patient populations may differ. Our providers preferred feedback within their peer group, instead of being benchmarked to external providers.
We had marked improvement in ordering rates by PAs. Our PAs see less acute patients independently, and we encouraged them to consult the ED physicians for more acute patients who may need a CT scan. PAs can play a critical component in quality care endeavors and in fact are now required to incorporate QI in their MOC process.17 Our study demonstrates the importance of involving nonphysician care providers in QI processes.
There is a significant lag between when research is first described and when it is finally adopted into clinical practice.18 Numerous QI projects in tertiary care pediatric EDs have sought to increase the use of evidence-based guidelines.19–21 Yet the majority of pediatric patients are seen at community hospital EDs.6 Hospitals that see a smaller volume of pediatric patients are less likely to have adopted guidelines to help guide imaging decision-making processes for pediatric head trauma patients.22 Indeed, head CT scan rates among pediatric head injury patients are significantly higher in community hospitals.7
There are many barriers to the adoption of evidence-based pediatric care in community EDs. Because the majority of patients seen by general ED physicians are adults, continuing medical education programs may not focus on pediatric care and there may be no impetus to create pediatric-specific protocols. The Joint Commission has implemented multiple national hospital inpatient quality core measures, but nearly all these measures are specific to adult care.23 This discrepancy may lead some community hospitals to focus all or nearly all of their QI endeavors on adult care. When pediatric-specific quality measures are created, community hospitals may more readily focus on pediatric care. We found this to be the case at our own institution when the Washington State Hospital Association included imaging for pediatric head trauma among their improvement efforts.9 Our community hospital was able to overcome these barriers with the help of hospital leadership support for our pediatric-specific QI initiative. In addition, because of the presence of pediatric hospitalists, we have hospital-based providers invested in improving pediatric quality measures.
Another barrier to the dissemination of pediatric-specific evidence encountered at community hospitals is that providers may use adult-specific decision tools for pediatric patients. We found that before our intervention, many of our providers used the Canadian CT Head Rule to decide whether to image pediatric patients, even though this decision tool included only patients ≥16 years old.24 Pediatric QI efforts at community hospitals should emphasize the unique clinical considerations of pediatric patients for providers who manage the full age spectrum. We demonstrate that the use of pediatric-specific evidence-based clinical protocols can improve the use of evidence-based medicine in community EDs.
The adoption of the PECARN guidelines is important because it decreases radiation exposure, which can lead to cancer,25 and is cost-effective. Given the lifetime risk of cancer, imaging is beneficial only with higher pretest probability of ci-TBI, such as the higher-risk patients in the PECARN guidelines.26 Unnecessary testing can lead to incidental findings; of the patients included in the PECARN study, 4% who underwent CT scans were found to have incidental findings on head CT scans.27 Incidental findings can result in additional parental anxiety, testing, cost, and procedures and often have unclear clinical significance.28 Because of the risks associated with unnecessary testing, many professional groups have focused on appropriate use criteria. During our postintervention period, our hospital performed 76 fewer CT scans than we would have if our preintervention CT scan rate had continued.
The regional pediatric hospital facilitated an effective QI project in a local community hospital by using MOC credit to encourage participation. Other studies have shown that offering MOC credit can increase participation for QI projects in community settings.29 MOC was established by the American Board of Medical Specialties to encourage ongoing improvement in physicians’ knowledge, and improvement in medical practice to evidence-based care.30 In part IV of the MOC requirements, physicians participate in approved QI projects.29 However, MOC requirements have been criticized because they do not always fulfill practice-specific needs, and they provide minimal benefit if not tailored to the practice of the physician,30 and so valuable MOC projects are needed.31 We demonstrate that regional children’s hospitals can use MOC requirements to help community hospitals adopt evidence-based best practices and improve important outcomes.
One challenge we encountered was that we were unable to standardize the CT ordering process through a “best practice alert” in the EMR. Because the community hospital is part of a larger health system, EMR changes were difficult to institute. These changes may have further decreased our rate of head CT scans.32,33
It is important to note that we did have a statistically significant increase in our LOS, 1 of our balancing measures. This increase in LOS is probably secondary to more patients who were observed after head injury instead of immediately imaged. The median LOS increased by only 24 minutes (from 90 minutes to 114 minutes), and so the clinical significance of this increased LOS is unclear. We think that the benefits of decreased CT use outweigh the disadvantage of a longer LOS.
This study has several important limitations. First, the partnership with pediatric hospitalists from the regional pediatric hospital helped engage and coach multiple stakeholders in change, and this assistance is not available at many community hospitals. Therefore, our results may not be generalizable to community hospitals that do not have pediatric hospitalist presence. Second, although we tracked readmissions within 72 hours to our hospital, patients may have been readmitted to other hospitals. There may be patients with ci-TBIs that were missed when we evaluated readmissions. However, with proper implementation of PECARN, the rate of missed ci-TBIs is low.3,25 In addition, we evaluated only absolute CT scan rates and did not evaluate whether PECARN was appropriately used with each encounter, because this analysis was beyond the scope of this study. Glasgow Coma Scale scores were not available in the EMR for the majority of the included patients, and so we were unable to limit our study population to patients with only minor head injuries, and we were unable to compare the head injury severity in the preintervention and postintervention patient populations. We have not yet demonstrated prolonged sustainability, but we plan to continue tracking CT scan rates, monitoring readmissions, and engaging the ED providers to maintain change.
We demonstrate that a multidisciplinary MOC QI project sponsored by a regional children’s hospital can decrease the rate of CT use in the evaluation of pediatric head injury in a community setting.
- Accepted October 24, 2016.
- Address correspondence to Rebecca M. Jennings, MD, M/S FA.2.115, PO Box 5371, Seattle, WA 98145-5005. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: All phases of this study were supported by Seattle Children’s Hospital Academic Enrichment Fund.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- Kuppermann N,
- Holmes JF,
- Dayan PS, et al; Pediatric Emergency Care Applied Research Network (PECARN)
- Gausche-Hill M,
- Schmitz C,
- Lewis RJ
- WA State Hospital Association
- Huber PJ
- Box GEP,
- Jenkins GM,
- Reinsel GC
- Benneyan JC,
- Lloyd RC,
- Plsek PE
- Benneyan JC
- The Joint Commission
- Gorzkowski JA,
- Klein JD,
- Harris DL, et al
- Sheehan B,
- Nigrovic LE,
- Dayan PS, et al; Pediatric Emergency Care Applied Research Network (PECARN)
- Copyright © 2017 by the American Academy of Pediatrics