BACKGROUND: Epilepsy or seizure care is the most common neurologic condition that presents to an emergency department (ED) and accounts for a large number of annual cases. Our aim was to decrease seizure-related ED visits from our baseline of 17 ED visits per month per 1000 patients to 13.6 ED visits per month per 1000 patients (20%) by July 2014.
METHODS: Our strategy was to develop a quality improvement (QI) project utilizing the Institute for Healthcare Improvement model. Our defined outcome was to decrease ED utilization for children with epilepsy. Rate of ED visits as well as unplanned hospitalizations for epilepsy patients and associated health care costs were determined. A QI team was developed for this project. Plan do study act cycles were used with adjustments made when needed.
RESULTS: Nineteen months after implementation of the interventions, ED visits were reduced by 28% (from 17 visits per month per 1000 patients to 12.2 per month per 1000 patients) during the past year. The average number of inpatient hospitalizations per month was reduced by 43% from 7 admissions per month per 1000 patients to 4 admissions per month per 1000 patients. For both outcome measures, a 2-sample Poisson rate exact test yielded a P value < .0001. Health care claims paid were less with $115 200 reduction for ED visits and $1 951 137 reduction for hospitalizations.
CONCLUSIONS: Applying QI methodology was highly effective in reducing ED utilization and unplanned hospitalizations for children with epilepsy at a free-standing children’s hospital.
- ED —
- emergency department
- EHR —
- electronic health record
- PDSA —
- plan do study act
- QI —
- quality improvement
- SAP —
- seizure action plan
- UEC —
- urgent epilepsy clinic
At Nationwide Children’s Hospital, over 600 patients with epilepsy are seen in the emergency department (ED) each year. Throughout the United States, epilepsy or seizure care is the most common neurologic condition that presents to an ED, accounting for more than 1 million visits annually.1 The majority of these cases are for children younger than 5 years.2 Many such ED visits are preventable for children with epilepsy.3 ED visits are costly and can consume time for caregivers and patients.4
Attempts to improve the management of asthma in children and to decrease ED utilization by using quality improvement (QI) have been made with some success.5–9 To date, no similar studies have been performed in pediatric epilepsy. A previous study suggested variables that if changed may prevent an ED visit in the first instance.3 Therefore, we developed a comprehensive QI project utilizing the Institute for Healthcare Improvement model10 in hopes of decreasing ED utilization for children with epilepsy. Our goal was to decrease ED visits and unplanned hospitalization to improve the quality of care for the patient, improve patient and caregiver experience by providing improved access to specialty outpatient care and education, and to decrease health care costs as a result.
Our primary aim statement defined our targeted improvement goal: to decrease ED visits from our baseline of 17 ED visits per month per 1000 patients to 13.6 ED visits per month per 1000 patients (20%) by July 2014 and sustain the decrease (Fig 1). In addition, a secondary aim was to decrease unplanned hospitalizations as defined by a patient being admitted after being seen in the ED from 7 admissions per month per 1000 patients to 5.6 admissions per month per 1000 patients by 20% from our baseline. The balancing measures for our project included readmissions to the hospital within 30 days, overall mortality of our epilepsy population, and patients seen again in the ED within 72 hours of an ED visit.
Nationwide Children’s is a freestanding pediatric hospital with 427 licensed beds and the sole pediatric tertiary care facility in Central Ohio. The ED has 62 beds and is a level 1 American College of Surgeons certified pediatric trauma center with 86 060 visits in 2014. Nationwide Children’s serves over 3000 children with epilepsy. Epilepsy patients are seen by a pediatric neurology service that includes 29 neurology physicians with 9 specifically trained in epilepsy care. In addition, 6 advanced practice nurses and 1 physician assistant provide patient care. The neurology division completes over 7000 epilepsy-related patient care visits in the outpatient setting each year. Outpatient clinics occur 5 days a week (Monday to Friday) from 8 am to 5 pm. A comprehensive multidisciplinary outpatient epilepsy center exists and serves ∼50% of the active epilepsy patients and occurs 2 days per week. This center is staffed by the 9 epilepsy physicians and 3 advanced practice nurses. In addition, the center consists of social work and other related care providers to assist epilepsy patients and their families.
Interventions around each key driver were developed with 5 being implemented sequentially over the subsequent 12 months. The 5 major interventions, annotated in Figs 2 and 3, included the following: (1) beginning an established urgent epilepsy clinic (UEC) to improve access; (2) deploying a seizure action plan (SAP) to improve knowledge around home seizure management; (3) supporting proper dosing of abortive seizure medications via the electronic health record (EHR); (4) providing reminder magnets with instructions of use for abortive seizure medications; and (5) review of previous high utilizers of the ED for epilepsy care to address unique issues.
Our first and highest intensity intervention was the established UEC that developed the ability to access outpatient epilepsy care urgently. The UEC was not created de-novo, but instead was a reorganization of an existing clinic to allow for 2 additional appointments with additional time allocated. The UEC was staffed by an epilepsy nurse practitioner and an epilepsy social worker. The clinic was held 4 days a week, either morning or afternoon (Monday, Tuesday, Thursday, and Friday). Two such clinic appointments were available per each session with extended patient time (60 or more minutes) to ensure all issues were addressed properly and education about epilepsy care was given. Appointments were scheduled flexibly as add-on appointments as triggered by the neurology triage nurses who were given a list of “red flags” and by any neurology provider, including residents while taking night call for neurology.
To improve education and enhance communication among patients with epilepsy and providers, we developed a SAP. By analogy, an asthma action plan has been successfully used previously.5 However, minimal work in this area has been performed in epilepsy.11–14 The action plan was developed to mirror the appearance of the asthma action plan. Information about seizure baseline, when caregivers should contact neurology, and other important information were available on the plan in a color-coded scheme utilizing green, yellow, and red zones.
Abortive Seizure Medication Dosing and Magnets
One variable known to increase ED utilization is not having or having inappropriate dosing of abortive seizure medication.3 Therefore, utilizing the EHR, a dosing guide and alert system was developed to assist providers in prescribing the correct dose for abortive seizure medication. If incorrect doses were selected, an EHR alert would notify the prescriber of the recommended dose. Additionally, a dosing guide with recommendations on the basis of age and weight was attached to the order. To improve education and reminders for caregivers, a magnet was given for the prescribed abortive medication with detailed instructions on proper medication administration. It was given in conjunction with the SAP.
Review of High Utilizing Patients
During the last 9 months (November 2014 to July) of the project, we deployed extrapolation and review of data from EPIC by developing and utilizing a dashboard. We used Qlikview, which is a data processing and analysis software tool (Qlik, Lund, Sweden). This enabled the presentation of daily updates and trends of actionable information including no-shows to our outpatient epilepsy clinic, ED visits, and hospitalizations. Review of patients with high ED utilization occurred monthly by the QI team to determine if further action was needed.
Study of the Interventions
For each of the above 5 interventions, plan do study act (PDSA) cycles were used with adjustments made as necessary on the basis of the analysis of the cycle. By design, we initiated our highest intensity intervention, the UEC, first to evaluate the effect of that isolated intervention. The UEC was initiated in October 2013. Our PDSA cycles for each intervention consisted of obtaining feedback provided by staff and patients. For the UEC, feedback was obtained after the first 20 patients were seen. Feedback was obtained for all other interventions implemented. This approach was felt most useful as changes to work flow and caregiver engagement was the major focus of our interventions.
To identify patients for this project, International Classification of Diseases, Ninth Revision codes were used. Our population consisted of those with a completed office visit within a rolling 13 months who had a primary or secondary diagnosis of epilepsy (345.xx). Current Procedural Terminology codes were used to identify the completed office visit. An ED visit was included if the visit was related to a seizure (780.39) or epilepsy (345.xx) as either the primary or secondary reason listed for the visit. An unplanned hospitalization was counted if the ED visit resulted in the patient being admitted to the inpatient neurology service at Nationwide Children’s Hospital.
For the financial analysis, health care utilizations costs required estimation, as the outcome was a lack of ED visit or hospitalization. Therefore, these costs were estimated by previous average claims paid data available during the study period at our institutions, which were as follows: $640 per visit per ED visit and $18 066 per unplanned hospitalization. We considered the additional costs attributed to the UEC in our final total savings calculations for the level 5 office visit, which amounts to $133 in each claim paid for each visit. These add-on cases were felt to be a portion of the full effort for the nurse practitioner and social worker and not additional utilization of resources. SAP implementation was monitored during the project.
The Institute for Healthcare Improvement model for QI11 was used as the foundation of this initiative and study design. Baseline ED visits were measured over a retrospective period of 18 months (from January 2012 to June 2013) before initiating any interventions. The baseline period established that ED visits (Fig 2) shows a process in control, with no special cause variation trends and no seasonality associated with epilepsy patient care activity. The same held true for the baseline for unplanned hospitalizations (Fig 3). The improvement time, after the first successful intervention, was a 19-month period (from November 2013 through July 2015). A 20% reduction was chosen on the basis of our group’s consensus of feasibility. We used mortality and hospital readmissions as balancing measures for this project.
Key drivers or barriers for ED use were identified by using several techniques such as process maps and an affinity diagram from brainstorming sessions. In addition, previous literature and best practices were used for key driver development.3 Major drivers identified were as follows: system and communication issues within and outside of neurology; lack of access to resources by family and other health care providers; patient and caregiver comorbidities and beliefs; and a need to enhance education about epilepsy care for patients and caregivers.
For the purposes of our analysis, we defined a shift in data as 8 consecutive data points below our baseline.15 The primary data were generated from our EHR system (EPIC). Two analysts performed an independent review and validation to verify accuracy of our data queries. Control charts (a Statistical Process Control tool)15 were employed to monitor the outcome metrics: monthly baseline ED visits and unplanned hospitalizations, with the interventions recorded as implemented over time (Figs 2 and 3). The particular type of chart used to monitor the ED and inpatient processes (the U chart) displays the number of random events occurring during an opportunity “window” as measured by calendar days, patient days, etc, as appropriate for the type of events being measured. We used a U chart to control for the increase in unique patients with epilepsy seen at our institution, which started at 3167 patients and consisted of 3426 patients at our last data point. Epilepsy ED visits per 1000 patients per month is our illustration. The Poisson statistical distribution, the distribution assumed by the U chart, is typically well suited for representing such events. Accordingly, tests for statistical significance made use of the 2-sample Poisson rate test. For SAP monitoring, tracking of implementation was monitored when used within the EHR.
This QI project was discussed and approved by the Director of Quality Improvement at Nationwide Children’s Hospital. Updates were provided monthly via written reports and presented quarterly to the QI administration team consisting of the Division Chief of Neurology, Director of Quality Improvement, and Chief Medical Officer at Nationwide Children’s Hospital. Institutional review board approval at our institution is not required, nor is a letter of exemption needed to perform QI as the work performed was for QI purposes.
Nineteen months after implementation of the interventions to reduce ED utilization, ED visits were decreased from 17 visits per month per 1000 patients to 12.2 per month per 1000 patients (28%; Fig 2), P < .0001. Additionally, the average number of inpatient hospitalizations per month was reduced by 43% from 7 admissions per month per 1000 patients to 4 admissions per month per 1000 patients (Fig 3) per year, P < .0001.
Improvements occurred over time with implementation of each of the 5 interventions (Fig 1). The greatest impact was associated with use of the UEC. After implementation of the established UEC in October 2013, a shift in our baseline was noted (Fig 2). Two hundred ninety-one patients were seen in this clinic during the time of the project. Overall, 93% of the scheduled patients attended the clinic. The average show rate for neurology patients at our center is 80%. The estimated time spent per patient was 1.5 hours for the nurse practitioner for visit time and documentation, 1.75 hours per patient for social worker, and minimal additional administrative assistant support. A PDSA after the first 20 patients revealed that families felt more comfortable with the epilepsy care after the visit. Most families have expressed appreciation of the quickness the appointment was scheduled and the time spent.
Utilization of the SAP increased over time through dissemination mostly from the outpatient clinics in the Pediatric Epilepsy Center at Nationwide Children’s Hospital. SAP implementation improved from a baseline of 0 to ∼418 (of 3426 epilepsy patients served at our center) by July 2015. During our PDSA for the SAP, we sampled 10 families who received the SAP and surveyed utilizing providers. When we inquired during our PDSA cycles, families reported having a better comfort and understanding of what to do during an epileptic seizure. Additional information in the form of specific instructions on whom to call was added.
After implementation of the dosing reminder into our EHR, the proportion of seizure patients prescribed an accurate abortive medication for seizure care improved when implemented in September 2014 from a baseline of 0 to 100% after being built into our EHR. In addition, the instruction magnet for abortive medications was used for all patients. Feedback from epilepsy providers, nurses, and caregivers of patients were obtained during the initial PDSA and changes to the magnet design and instructions were made before full implementation was performed.
In regards to health care costs, given the reduction in ED visits and unplanned hospitalizations, we calculated a savings in health care utilization of $115 200 for ED visits and $1 951 137 for hospitalizations.
To attribute additional costs of the UEC, a level 5 follow-up visit was billed for these appointments with a claims paid associated with $133 per visit. Approximately 291 patients were seen in the established UEC through July 2015. Therefore, $39 867 was attributed to the cost of this intervention in generated claims. Other costs of supplies, material preparation, and changes to EHR modules could not be calculated because they are not separately collected at our institution and part of our overall structure.
Our balancing measures included seizure-related hospital readmissions, which decreased from a baseline of 22 in 2012 to 12 in 2014, whereas readmissions for other neurologic conditions remained virtually unchanged (18 to 21). A second balancing measure was mortality of patients with epilepsy in our population. It remained the same in the overall epilepsy population (10 in 2012 and 9 in 2014). A third measure was patients seen back in the ED within 72 hours where our baseline was 4.0% and with 1.7% noted during our intervention implementation period.
For this QI project, our team was able to decrease seizure-related ED visits by 28% and unplanned hospitalizations by 43% for children with epilepsy at Nationwide Children’s Hospital. Cost savings to the health care system was noted as a result of the reduction in health care utilization. Phased interventions for epilepsy patients to reduce ED visits and unplanned hospitalizations were employed with success by improving access to ambulatory outpatient epilepsy care and increased education on epilepsy management.
The intervention determined most successful was the development of the established UEC staffed by an epilepsy nurse practitioner and a social worker. We felt that the social work support was equally important as the medical decision-making. A significant amount of time spent in this clinic was for education, counseling, and addressing psychosocial risk factors. Given the psychosocial complexities of helping families manage seizures, including parental concerns and anxiety, as well as school- and work-related contingencies, the role of social workers may be important to the success of our model. Future work is needed to validate this finding.
Lack of proper knowledge about epilepsy emergency care was targeted via the SAP. It was also used as an educational tool for providers in discussing how to approach emergency care for acute seizures. The SAP appeared to be well received by our patients and their families with potential expansion of use to the school system on the basis of the informal feedback obtained. Further studies evaluating improvement on the impact of epilepsy are ongoing. Our SAP was different from a recent published study.14 This study discussed limitations that may have resulted in their lack of change noted. They note that their SAP was not color coded and had a lot of information about daily medications, which may have confused the patient. They implemented their plan in the inpatient setting only.14 We used color coding, simplified language, and implemented the plan in all care settings. The low rate of SAP implementation was felt secondary to the time needed for it to be built within our EHR and dissemination by providers. Therefore, it was not likely a major contributing factor to the ED and hospital reduction.
Previous literature has demonstrated that improper dosing or lack of an abortive medication when applicable may contribute to increased health care utilization.3 Often, the medical provider does not audit dosing to determine if changes are needed to account for patient growth. Reminders to providers using behavioral economic principles16,17 have been used previously with other medications and can be a successful technique when built into an EHR used for prescribing medications. In addition, the magnets served as reminders of how to give an abortive medication and allowed for additional caregivers not present for the initial education to have a basic understanding of abortive medication administration. Further, it served to remind caregivers on where the medication was located within the home setting. These reminder cues as a behavioral economic principle17 can also be helpful for caregivers.
We began reviewing patients with high ED utilization as an intervention. A high risk ED utilizing patient is defined currently as one that has used the ED for their epilepsy-related care 4 or more times in the previous year. Unlike asthma, where high ED utilization in 1 year does not necessarily predict subsequent high utilization,18 the same cannot be said about epilepsy. Recent publications highlighted that high ED utilization in 1 year is predictive of high ED utilization the following year.19–21
Applying QI methodology can be very effective in reducing unnecessary ED utilization and unplanned hospitalizations for children with epilepsy. Reducing such utilization can have a significant cost savings to the overall health care system. In previous studies, it was noted that such utilization accounted for the majority of health care costs for patients with epilepsy.16
Limitations exist in our ability to calculate a true cost savings because we could not calculate the exact cost of intervention implementation. Until models such as the University of Utah’s program to calculate true costs become available, this limitation will exist in any similar study.22 Costing out the professionals’ time provides the most conservative estimate of the costs of improvement, which would allow others who wish to replicate this effort a more accurate understanding of the percent effort required to duplicate the UEC locally. At our institution, we did not use additional resources. These appointments were in addition to the normally scheduled visits and considered a portion of the full effort used in salary determination for our nurse practitioner and social work positions.
Another limitation is how 1 specific intervention affects the entirety of the study period. To account for this limitation, interventions were phased in from our key driver diagram. Further, randomized control trials, which traditionally serve as the gold standard study method, were not performed and not easily possible in this setting. Patient care occurs in real time and interventions need to be applied actively and aggressively for active substantial improvement to be seen. Further, external factors unknown to the QI team may have an effect on the overall population of epilepsy patients seen at Nationwide Children’s Hospital. Our balancing measures performed well for the project, suggesting more harm was not performed elsewhere in the system.
As provisions of the Affordable Care Act become further implemented,23 focusing on value and quality, compared with quantity, will be increasingly crucial. Health care currently consumes a great portion of our gross domestic product with control of costs while maintaining or increasing quality of care desperately needed. Medical providers are best suited for controlling health care costs. One technique for cost reduction is by implementing QI methodology, which has been found successful in other models of care.24 Efforts to increase outpatient clinic utilization while decreasing unnecessary less effective utilization will continue to improve the patient experience, increase the quality of care, and reduce health care costs, thus achieving the Triple Aim of Healthcare.
- Accepted August 17, 2016.
- Address correspondence to Anup Patel, MD, ED533, 700 Children’s Dr, Columbus, OH 43205. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: Dr Patel has research support from Pediatric Epilepsy Research Foundation and GW Pharmaceuticals, he provided consultation for Health Logix and Cyberonics, and he participated in webinar development for the American Academy of Neurology; Dr Cohen has research support from Pediatric Emergency Care Applied Research Network; and Mr Wood has indicated he has no potential conflicts of interest to disclose.
- Adams RJ,
- Smith BJ,
- Ruffin RE
- Berwick DM,
- Nolan TW,
- Whittington J
- Davidoff F,
- Batalden P,
- Stevens D,
- Ogrinc G,
- Mooney SE; SQUIRE development group
- Ogrinc G,
- Davies L,
- Goodman D,
- Batalden P,
- Davidoff F,
- Stevens D
- Ogrinc G,
- Davies L,
- Goodman D,
- Batalden P,
- Davidoff F,
- Stevens D.
- Roundy LM,
- Filloux FM,
- Kerr L,
- Rimer A,
- Bonkowsky JL
- Benneyan JC,
- Lloyd RC,
- Plsek PE
- Manjunath R,
- Paradis PE,
- Parisé H, et al
- Stevens J
- Nash DR,
- Childs GE,
- Kelleher KJ
- Grinspan ZM,
- Shapiro JS,
- Abramson EL,
- Hooker G,
- Kaushal R,
- Kern LM
- Kawamoto K,
- Martin CJ,
- Williams K, et al
- Copyright © 2017 by the American Academy of Pediatrics