Comprehensive Migraine Initiative in the Pediatric Emergency Department Improves Treatment Outcomes

Objectives

The objective of our study was to determine if a comprehensive initiative to protocolize migraine management in the emergency department improved headache outcomes prior to discharge. Our primary outcome for this question was reduction in headache pain score, with goal of headache freedom (pain score of 0). Secondary outcomes included total time spent in the emergency department, time to first treatment, admissions to the hospital, and placement of neurology consults.

Methods

Our comprehensive initiative was implemented in July 2016, and consisted of a lecture series for emergency department residents and attendings discussing evidence-based migraine treatment, a standardized combination migraine treatment protocol for the pediatric emergency department, and the creation of an electronic note template to facilitate physician documentation and adherence to the protocol. Our standardized combination migraine treatment protocol consisted of a stepwise approach to treating headache using evidence-based therapies. Medications included intravenous ketorolac, prochlorperazine, normal saline fluid bolus, valproate sodium, and magnesium sulfate (Figure 1). At each phase of the protocol several medications were listed, and it was suggested that all medications be given, as close together as possible (given the different times for infusions, this typically took an hour for the patient to receive all medications). We emphasized to the emergency department physicians that our treatment protocol was meant to serve as a guideline rather than a requirement.

OPEN IN VIEWER

Figure 1.

Recommended institutional pediatric migraine clinical pathway. This figure was posted in the pediatric emergency department and sent to all providers via email. A box outlining appropriate weight-based doses of all medications was included with this diagram. An asterisk (*) notes that this pathway does not routinely recommend diphenhydramine in the headache cocktail, unless there is a history or presence of a dystonic reaction.

A template for a migraine-focused history and physical was developed for the electronic medical record system used at our hospital. The history section asked focused questions about the patient’s headache, and the examination section defaulted to a normal neurologic examination, with the option for the provider to free text any abnormal findings. There was a dropdown menu in the Assessment/Plan section that included the medications in our suggested combination treatment protocol. The note also had a series of assessment sections to prompt the provider to document the treatments given, time given, and the patient’s pain score after the intervention.

Lectures were given at the beginning of the academic year (July 2016). E-mail reminders outlining key points from the lectures, the recommended combination treatment protocol, and instructions on how to use the suggested note template were sent at the beginning of each new emergency department rotation block (every 2 weeks) to ensure all residents had the information.

We performed a retrospective chart review of patients seeking acute migraine treatment in our pediatric emergency department, both before and after our migraine initiative was implemented. A search using “headache” and/or “migraine” as the reported chief complaint in children aged 4-20 years identified a subset of patients presenting to our pediatric emergency department for headache treatment over our study periods—the preintervention group presenting between November 2011 and April 2012, and the intervention group presenting between February and September 2017 (Figure 2). Data abstractors (a pediatric neurology resident and senior medical student) then screened the charts to include only patients who had a final diagnosis of migraine (with or without aura) according to the most recent International Headache Society (IHS) guidelines (beta version). ⁹ For the preintervention group (who presented before the development of the most recent IHS guidelines), these same beta version guidelines were used and the diagnosis was made retrospectively using emergency department documentation in the medical record. Exclusion criteria included patients with fever or infectious symptoms, cerebrospinal fluid shunts (if there was malfunction or presumed malfunction), pregnancy, or patients who had neurosurgery within the past 4 weeks. Equal numbers of individuals were compared from the intervention and preintervention groups (n = 110).

OPEN IN VIEWER

Figure 2.

Study enrollment distribution.

In both groups, data were extracted from the primary resident’s history and physical examination and the electronic flowsheet. Variables of interest included initial and final headache pain scores, time spent in the emergency department, and time elapsed until first treatment given. Headache pain severity was measured at the beginning and completion of the emergency department visit using a numeric pain scale of 0-10, with 0 being absence of pain and 10 being the most severe pain. In younger children, the Wong-Baker FACES scale was used to approximate pain score. The use of this scale is standard practice in the emergency department for children who are unable to understand or articulate a numerical score. It is administered by the bedside nurse and asks the child to point to the face that best indicates their pain intensity, and the corresponding numeric value is recorded.

After data collection, Stata statistical analysis software was used to analyze results. In the analysis of outcomes pertaining to pain scores as well as total emergency department visit time, we excluded patients admitted to the hospital, as our primary focus in this study was to evaluate treatment in the emergency department prior to discharge. For the remainder of outcomes (patient demographics, initial pain scores, whether or not the child was admitted to the hospital from the emergency department, whether or not a neurology consult was placed, and time to first treatment), admitted patients were included in the analysis. Quantitative variables (patient age, pain scores, time elapsed to first treatment, and total emergency department visit time) were compared between the intervention and preintervention groups using independent t tests with Levene’s test of equal variances. Qualitative variables (race, gender, request for neurology consult) were compared between the 2 groups using Pearson chi-square.

Results

Our preintervention population was derived from 360 patients presenting to the pediatric emergency department between November 2011 and April 2012 with chief complaint of headache and/or migraine. Of these, 154 had a diagnosis of migraine per IHS criteria. Our intervention group was derived from 345 patients presenting to the pediatric emergency department between February and September 2017 (after implementation of the comprehensive migraine initiative) with a chief complaint of headache and/or migraine. Of these, 137 had a diagnosis of migraine. Incomplete or missing variables reduced our preintervention and intervention groups to 110 unique patients each.

We found no significant differences in various patient characteristics (age, gender, and initial headache pain score) of the 2 groups. Average age in the preintervention group was 13.7 ± 3.6 years compared with 13.9 ± 3.6 years in the intervention group (P value .654). The preintervention group consisted of 61% females compared with 70% females in the intervention group, but this was not statistically significant (P value .169). We did find a significant difference in racial breakdown of the 2 groups, with substantially more African American patients and fewer Hispanic and Caucasian patients in our preintervention group (P value .002); however, this was not significant in our sub-analysis, when hospital admissions were excluded. The average initial headache pain score on arrival to the emergency department was 7.5 ± 2.2 in the preintervention group and 7.7± 2.2 in the intervention group (P value .43) (Table 1).

OPEN IN VIEWER

Table 1.

Patient Characteristics.

	Preintervention group(n = 110)		Intervention group(n = 110)
Characteristic	Mean ± SD or %	95% CI	Mean ± SD or %	95% CI	P value
Age, y	13.7 ± 3.6	13.0, 14.4	13.9 ± 3.6	13.2, 14.6	.654
Initial headache score	7.5 ± 2.2	7.0, 7.9	7.7 ± 2.2	7.3, 8.1	.432
Gender, %					.169
Male	39		30
Female	61		70
Race, %					.002
Caucasian	25		40
African American	73		50
Hispanic	1		5
Other	1		5

Abbreviation: CI, confidence interval.

Data analysis revealed that there was a significantly reduced time to receipt of first treatment in the intervention group compared with the preintervention group, with an average of 1.8±1.3 hours in the intervention group compared with 2.1±1.7 hours in the preintervention group (P value .046). It appeared that there was a higher frequency of hospital admissions in the intervention group compared with the preintervention group, 10% and 3.6%, respectively; however, this result turned out not to be statistically significant (P value .06). We also did not see a significant difference in the number of neurology consults (Table 2).

OPEN IN VIEWER

Table 2.

Outcome Measures.

	Preintervention group			Intervention group			P value
Outcome measures	Mean ± SD or %	95% CI	n	Mean ± SD or %	95% CI	n
Time to first treatment, h	2.1±1.7	1.8, 2.4	110	1.8±1.3	1.5, 2.0	110	.046
Final pain score	2.6±2.9	2.0, 3.1	106	1.8±2.3	1.3, 2.2	99	.018
Mean reduction in pain score	4.8±3.0	4.2, 5.4	106	5.9±2.9	5.3, 6.5	99	.006
Percentage pain reduction	63.3±50.9	53.5, 73.1	106	75.7±30.1	69.7, 81.7	99	.019
≥50% pain reduction, %	67		106	82		99	.039
Admission rate, %	4		110	10		110	.06
Neurology consulted, %	17		106	10		99	.34

Abbreviation: CI, confidence interval.

There was a notable reduction in pain scores in the intervention group compared with the preintervention group. Admitted patients in both groups were excluded from this analysis. Final average pain score on emergency department discharge in the intervention group was 1.8±2.3 compared with the 2.6±2.9 in the preintervention group (P value .018). This translated to a 75.7% pain reduction in the intervention group compared with 63.3% reduction in the preintervention group (P value .019). We also found a significantly larger percentage of patients achieving ≥50% pain reduction, a common benchmark for identifying successful response, in the intervention compared with the preintervention group, with 81.8% achieving this prior to emergency department discharge in the intervention group compared with 67% in the preintervention group (P value .039). Average time spent in the emergency department was not found to be statistically significant between the 2 groups (Table 2).

Discussion

Our results demonstrate that the implementation of a comprehensive migraine initiative in the pediatric emergency department at our institution improved headache outcomes, as measured by reduction in headache pain score prior to emergency department discharge, with significantly more patients achieving ≥50% pain reduction compared with the preintervention population. Our results also convey a mechanism for quality improvement of pediatric emergency department care. By creating a streamlined headache protocol and optimizing provider education surrounding the acute management of pediatric migraine, we showed not only significantly reduced pain scores but also reduced patient wait time to receive initial treatment. Interestingly, total time spent in the emergency department was not found to be significantly different in our postintervention group. We posit that this may be due to the fact that following our intervention, patients are less likely to be pre-emptively discharged with higher pain scores, and this is supported by our data showing reduced pain scores at discharge.

The retrospective study by Leung et al similarly found improved pain scores following implementation of a standardized combination therapy for acute treatment of pediatric migraine. They however found a significantly reduced hospital admission rate, by close to 10-fold, following implementation of their intervention. ⁷ In contrast, the trend of our results appeared to show the opposite, with more admissions following implementation of our intervention by approximately 3-fold; however, following statistical analysis, this was not found to be significant (P value .06). We suspect that our small sample size and low percentage of admissions overall reduced the statistical power of this result. Although our goal in implementing our migraine initiative was to reduce pain scores, with goal of headache freedom (pain score of 0), another goal was to admit patients promptly for escalation of therapy (such as initiation of dihydroergotamine) if they were not adequately responding to migraine therapies in the emergency department. The flowsheet created for the emergency department providers recommended consideration of early inpatient admission if there was no or inadequate response to the first round of medications. Thus, conceivably our intervention could have led to more admissions, which seemed to be the trend of our data, albeit not statistically significant. The fact that we did not see a significant change in admissions may be attributed to a combination of more patients achieving headache reduction of ≥50% to allow for discharge from the emergency department, and a lower threshold for inpatient admission for patients who did not respond early to medications.

From a quality-improvement perspective, there are numerous studies analyzing the cost-effectiveness and time to treatment in other disease populations. A study by Zand et al ¹⁰ found that implementation of a clinical pathway with standardization of therapy for children with inborn errors of metabolism improved measures of timeliness as well as clinical effectiveness. Other studies report similar results after implementation of standardized clinical pathways for children with asthma, ¹¹ appendicitis, ¹² and neonatal sepsis. ¹³ However, to our knowledge, this is the first study to analyze the impact of a standardized pathway on time to treatment in a pediatric headache population.

Our study has several limitations. Incomplete documentation reduced our sample size considerably, thereby reducing the power of our study. The documentation from which data were collected was limited to the emergency department history and physical examination as well as the emergency department flow chart. We did not look into other documents in the medical history (such as neurologist notes) as many of the patients did not have neurologists or saw providers outside of our medical system. Therefore, if not enough information was included in the emergency department note in order to give a diagnosis of migraine by IHS criteria, that patient was not included in the study. Study populations were further reduced by the fact that sometimes emergency department documentation did not list an initial and/or final pain score, so we were unable to include those patients in the study.

Because we exclusively used emergency department documentation in obtaining our data, rather than perhaps more detailed neurology notes, this prevented us from controlling for baseline headache frequency and whether or not the child was on migraine preventive medication, which may have influenced our results with regard to response to treatment in the emergency department. Although emergency department notes would typically list medications, it was not always clear if the medication was being used for migraine or another indication; therefore, we did not include use of preventive medications in our analysis.

Additionally, our preintervention and intervention groups presented during relatively different times of the year, with our preintervention group containing more patients presenting in the mid-autumn and winter months and our intervention group containing more patients presenting in the spring and summer months. A study by Soriani et al ¹⁴ found a seasonal peak in migraine frequency between November and January, with a nadir observed in July, suggesting that school activities might represent an important trigger for migraines. A study by Yang et al ¹⁵ also found that migraineurs were more likely to have headaches in the winter months. Despite this, our data show similar initial pain scores between the 2 groups on presentation, making this limitation less likely to have influenced our results. Additionally, delaying our data collection in the intervention group allowed more time for the intervention to be adopted by emergency department providers. The data for our preintervention group was collected several years before our intervention group, coinciding just prior creation of a comprehensive pediatric headache center at our institution. This was intended to be used as comparison data for future quality improvement projects in the area of pediatric headache data at our institution, not limited to emergency department management. We do acknowledge the limitation that there is a substantial amount of time between the pre- and postintervention data.

Our results demonstrate that implementation of a standardized headache pathway and improved provider education can significantly improve headache outcomes prior to discharge from the pediatric emergency department; however, given the retrospective nature of the study we do not have information about headache recurrence following discharge. Some studies in adults suggests that use of corticosteroids prior to discharge reduces the likelihood of headache recurrence, ^16,17 although evidence is limited in children. A study by Legault et al evaluating factors predicting pediatric bounce-back to the emergency department found that no individual treatment, including the administration of steroids at discharge, was associated with decreased likelihood of return to the emergency department. ¹⁸ At our institution, we consider “bounce-back” as not just returning back to the same hospital but rather seeking any additional treatment for headache that recurs, such as visiting an urgent care center, primary care/neurologist office, or an after-hours phone line. The retrospective nature of our study, coupled with limited documentation from which to extract data, prevents us from thoroughly assessing this, though we hope to do so in the future. Further prospective studies assessing factors that influence headache recurrence in the immediate period following discharge are needed.

Conclusion

Our study demonstrates that a standardized treatment protocol coupled with improved provider education and ease of documentation improve both efficacy and efficiency of migraine treatment in the pediatric emergency department.