Treatment of Children With Migraine in Emergency Departments: National Practice Variation Study
OBJECTIVE: Children with migraine may present to an emergency department (ED) when outpatient management has failed; however, only limited research has examined migraine-abortive medications among children.
METHODS: A retrospective chart review of ED presentations for migraine or headache between July 1, 2004, and June 30, 2005, in 10 Canadian pediatric EDs was conducted. A priori, evidence-based treatments were defined as any treatment that was based on high-quality evidence and an absence of opioids as first-line agents.
RESULTS: A total of 2515 records were screened, and 1694 (67.4%) met inclusion criteria. The average age of patients was 12.1 years, 14.5% (95% confidence interval [CI]: 12.1%–17.2%) of patients experienced headache >15 days per month, and 62.6% (95% CI: 55.7%–68.9%) had already used migraine-abortive therapy. Significant variations in practice for all classes of migraine-abortive medications were observed. Dopamine receptor antagonists (prochlorperazine, metoclopramide, or chlorpromazine) (39% [95% CI: 28.4%–50.8%]) and orally administered analgesics (acetaminophen and ibuprofen) (24.5% [95% CI: 23.9%–46.8%]) were prescribed most commonly. Predictors for the use of evidence-based treatment included older age (odds ratio: 1.15 [95% CI: 1.07–1.24]) and a discharge diagnosis of migraine (odds ratio: 1.84 [95% CI: 1.11–3.05]).
CONCLUSIONS: Children presenting to EDs for treatment often have frequent attacks and have experienced failure of outpatient, migraine-abortive efforts. Practice variations were impressive for the care of children with migraine in these Canadian EDs.
WHAT'S KNOWN ON THIS SUBJECT:
There has been 1 randomized, controlled trial of acute migraine therapy for children in EDs.
WHAT THIS STUDY ADDS:
This study examines current practices in a national sample of Canadian tertiary, pediatric EDs.
Migraine headache is a common condition in children and adults1 and may cause significant morbidity, with time lost from school, sports, and other activities.2 Children with migraine headache may present to the emergency department (ED) when outpatient management has failed; however, only 1 study has examined migraine-abortive medications in children.3 Although there are published narrative reviews of the treatment of children with migraine in EDs,4,5 definitive, evidence-based guidelines are lacking. Not surprisingly, in a regional migraine headache study in Canada, significant variation between mixed-population and pediatric EDs was observed.6 The pediatric EDs adhered more closely to available evidence regarding the use of dopamine receptor antagonists and the less-frequent use of opioids. To our knowledge, however, there are no practice variation studies comparing tertiary, pediatric EDs. Our objectives were (1) to describe the characteristics of children being treated for migraine in Canadian pediatric EDs, (2) to document the diagnostic approaches to these cases, and (3) to examine the treatment practices of Canadian pediatric emergency physicians.
Pediatric Emergency Research Canada Sites
Pediatric Emergency Research Canada is a federally funded, research network of all 15 tertiary, pediatric EDs in Canada. In this study, 10 Pediatric Emergency Research Canada sites, representing 6 Canadian provinces, provided data. Both English- and French-speaking EDs, as well as East, Central, Prairie, and West Coast regions, were represented. The participating EDs were all academic, pediatric emergency centers, generally staffed by full-time, certified, emergency physicians, and functioned as regional referral sites for trauma, surgery, and complex congenital or acquired diseases in patients <17 years of age (some variation in the upper age limit exists across the country).
Study Design and Case Selection
A retrospective chart review of pediatric ED presentations for migraine or headache was conducted by adhering to the suggested guidelines for chart reviews in emergency medicine research.7 Abstractors were trained and their performance was monitored. Variables were defined and a standardized, Internet-based, electronic form was implemented (OpenClinica, Boston, MA). The charts of children 5 to 17 years of age who were seen in EDs between July 1, 2004, and June 30, 2005, with discharge diagnoses of all migraine types or headache, on the basis of the International Classification of Diseases, Ninth or 10th Revision, were identified in hospital data systems and screened for inclusion and exclusion criteria. Institutional ethics approval was obtained at each site. The study was conceived and designed after the patient encounters had occurred; at the time of the patient encounters, physicians were unaware that a study would be conducted.
Inclusion criteria were (1) the presence of headache at the time of physician assessment, (2) age of 5 to 17 years, and (3) diagnosis of migraine by the emergency physician or sufficient findings for a diagnosis of at least probable migraine (on the basis of the International Classification of Headache Disorder, Second Edition) for ED admissions coded only as headache. The following were required for a diagnosis of migraine or probable migraine when the ED admission was coded only as headache: presence of headache and 2 of nausea, emesis, photophobia, or sonophobia. Exclusion criteria were (1) leaving before assessment by a physician, (2) the possibility that another disorder (eg, intracranial shunt, intracranial mass, intracranial hemorrhage, or intracranial infection) might explain the headache, and (3) an ED visit within 1 week (7 days) after the initial presentation.
Data were analyzed with Stata 10 (Stata, College Station, TX). Categorical variables were described with proportions and 95% confidence intervals (CIs), whereas continuous variables were described with means and 95% CIs. The primary unit of sampling was clustered with respect to site; therefore, the data were analyzed as panel data. Clustering was considered in the variance estimation by using the survey methods in Stata (Taylor series linearization) and population weights based on the average number of ED admissions per year.
The primary outcome variable was an aggregate marker of any evidence-based treatment. Migraine-abortive treatments with some evidence for children or use in EDs included the following: acetaminophen, ibuprofen and other nonsteroidal antiinflammatory drugs (NSAIDS), serotonin 1b/1d receptor agonists (triptans), dihydroergotamine, and dopamine receptor antagonists (metoclopramide, prochlorperazine, and chlorpromazine), with no use of opioids. Two secondary outcome variables, namely, no use of medication and use of a dopamine receptor antagonist, also were analyzed. Oxygen, diphenhydramine, dimenhydrinate, opioids (codeine, meperidine, morphine, and oxycodone), steroids (prednisone, dexamethasone, and methylprednisolone), benzodiazepines, muscle relaxants, and ondansetron were not included in the evidence-based group.
Practice variation was assessed by using the χ2 test, and results were displayed as a figure (Fig 2). Predictive factors for the use of evidence-based treatment (with or without opioids) and secondary outcome variables were assessed by using logistic regression analyses with clustering with respect to site, with a conditional, fixed-effects model. Predictive factors were included at a 5% level of significance and included all recorded factors, that is, age, gender, presence of aura and migraine-associated symptoms (nausea, emesis, photophobia, and sonophobia), duration of current headache, diagnosis of migraine, headache frequency, season, date of admission, and previous use of migraine-abortive or -prophylactic medications.
A total of 2515 hospital records were screened, in 10 tertiary, pediatric EDs in 6 Canadian provinces, and 1694 (67.4%) met the inclusion criteria. The most-common reasons for exclusion were no headache at the time of assessment by a physician (n = 509) and insufficient criteria for a diagnosis of migraine (n = 213). Other reasons included intracranial hemorrhage (n = 7), intracranial shunt (n = 64), and intracranial mass (n = 28). The ages (12.1 vs 9.4 years) and gender distributions (57.4% vs 89.7% female) of the included and excluded groups were different.
Demographic Features and Headache Description
The mean age was 12.1 years (95% CI: 11.4–12.9 years), with a range of 5 to 17 years; 57.4% (95% CI: 51.8%–62.9%) of subjects were female. On average, children presented 2.19 days (95% CI: 1.62–2.75 days) after the onset of their migraine attacks. An aura was present in 27.2% (95% CI: 19.9%–35.9%) of cases, and 14.5% (95% CI: 12.1%–17.2%) of children reported migraine >15 days per month. The time of year was equally distributed between seasons, and the average length of stay was 4.4 hours (95% CI: 3.7–5.1 hours).
Medications Used Before ED Visit
Among all subjects presenting to EDs, 62.6% (95% CI: 55.7%–68.9%) had already used ≥1 migraine-abortive medication. The most-commonly used medications were orally administered analgesics (52.6% [95% CI: 46.3%–58.9%]), including acetaminophen and ibuprofen; only 2.2% (95% CI: 1.5%–3.1%) of subjects had used a triptan (serotonin 1b/1d receptor agonist) medication. Overall, 4.7% (95% CI: 3.3%–6.7%) of subjects had used an opioid medication (eg, codeine, meperidine, morphine, or oxycodone).
Use of migraine-prophylactic medications was observed in only 6.1% (95% CI: 3.3%–11.3%) of cases. Among subjects with migraine >15 days per month, 82.1% (95% CI: 58.5%–93.7%) were not receiving prophylactic treatment. The most-common prophylactic treatments were tricyclic antidepressants (eg, nortriptyline and amitryptiline), which were used in 38.9% (95% CI: 11.7%–75.3%) of cases. Alternative treatments (predominantly vitamin B2) were used in 24.4% (95% CI: 4.6%–68.4%) of cases, and topiramate was used in 8.9% (95% CI: 3.1%–22.9%) of cases. Cyproheptadine, pizotifen, propranolol, valproate, feverfew (medicinal herb), and other medications each accounted for ≤5% of cases in which a prophylactic medication was used.
The most-common method of investigation was computed tomography (16.3% [95% CI: 12.2%–21.3%]), and 8.2% (95% CI: 5.1%–12.9%) of the scans demonstrated abnormal findings. None of the abnormal findings altered management. Abnormalities included arachnoid cyst, previous infarction, and cerebral malformation. A lumbar puncture was performed in 2.1% (95% CI: 1%–4.1%) of cases overall, and no results were abnormal. The 10 EDs varied significantly in the use of computed tomography (P < .001) and lumbar puncture (P = .002).
Figure 1 presents the most-common management choices for the ED treatment of migraine. The use of evidence-based treatment was observed in 64.2% (95% CI: 56.1%–71.9%) of cases overall. Figure 2 presents the variation observed among the 10 sites. In the adjusted, multivariate analysis, older age (odds ratio [OR]: 1.15 [95% CI: 1.07–1.24]) and a diagnosis of migraine (OR: 1.84 [95% CI: 1.11–3.05]) were associated with the use of evidence-based treatments. The same factors were observed when adjustment was made for the use of any opioids (ie, not considered evidence-based treatment) and the use of a dopamine receptor antagonist. Similarly, diagnosis of migraine by the treating physician (OR: 2.07 [95% CI: 1.23–3.48]) and older age (OR: 1.18 [95% CI: 1.09–1.27]) were the strongest predictive factors for prescription of any medication. The presence of an aura decreased the likelihood of a medication being prescribed (OR: 0.31 [95% CI: 0.14–0.70]).
Overall, there was significant variability between sites (P ≤ .002) for all medication classes except antibiotics, as indicated in Fig 2 and Table 1. The intravenous route was used for 48.4% (95% CI: 34.6%–62.4%) of subjects overall, and a bolus of intravenous fluid was given to 24.3% (95% CI: 14.8%–37.1%). Intravenously administered dopamine receptor antagonist medications were prescribed most frequently, with 82.4% (95% CI: 60.4%–93.5%) being metoclopramide, 12.7% (95% CI: 2.8%–42.0%) prochlorperazine, and 4.9% (95% CI: 1.0%–20.0%) chlorpromazine. NSAIDs often were prescribed in combination with a dopamine receptor antagonist (36.5% [95% CI: 21.3%–55.1%]). Diphenhydramine was prescribed for 14% (95% CI: 3%–46%) of subjects who received a dopamine receptor antagonist. Orally administered analgesics included ibuprofen (66.3% [95% CI: 59.8%–72.3%]) and acetaminophen. Other parenterally administered NSAIDs included predominantly ketorolac (70.9% [95% CI: 21.8%–95.5%]) and naproxen (20.9% [95% CI: 2.3%–74.5%]) (Fig 1). Among the opioid medications, codeine (39.1% [95% CI: 20.4%–61.6%]), acetaminophen/codeine combined medication (30.5% [95% CI: 15.5%–51.2%]), and morphine (27.6% [95% CI: 12.8%–49.7%]) were prescribed most frequently. Triptan medications were used infrequently and included eletriptan (n = 1), sumatriptan (n = 6), and zolmitriptan (n = 1).
No serious adverse events were reported. Minor adverse events included vomiting (7 cases), dizziness (3 cases), nausea (2 cases), and 1 case each of dystonia, tachycardia, agitation, hypotension, and paresthesias. Dystonia, agitation, hypotension, and paresthesias were observed in association with a dopamine receptor antagonist.
The most-frequently prescribed, migraine-abortive medications at discharge were ibuprofen (45.7% [95% CI: 34.2%–57.7%]), acetaminophen (17.2% [95% CI: 12.1%–23.8%]), and other NSAIDs (10.8% [95% CI: 3.9%–26.6%]). Triptans were prescribed in only 1% (95% CI: 0.5%–2.2%) of cases. Migraine-prophylactic medications (amitryptiline, topiramate, propranolol, cyproheptadine, flunarizine, pizotifen, carbamazepine, and gabapentin, in order of frequency) were prescribed in 10.2% (95% CI: 2.7%–32%) of cases overall. Of subjects who reported migraine on >15 days per month, only 30% (95% CI: 13.3%–44.6%) were given prescriptions for prophylactic medications, but 47.7% (95% CI: 35.3%–60.3%) were also referred to a neurologist or pediatrician. Opioid medications were prescribed in 5.4% (95% CI: 3.2%–9.1%) of cases, steroids were prescribed in 2.4% (95% CI: 0.5%–11.6%) of cases, and 2.2% (95% CI: 1.3%–3.6%) of all subjects were admitted to the hospital.
To our knowledge, this is the first national and the largest migraine headache study conducted in pediatric EDs. There are only 15 pediatric EDs in Canada, and the 10 involved in this study treat similar populations of children with migraine, with a small female predominance, in all seasons. Of importance, children present to EDs an average of ≥2 days after the onset of migraine, and most (60%) of the children have already used orally administered analgesics at home. Therefore, children who present to EDs exhibit relative treatment resistance, a fact that should guide management approaches.
Confidence in the diagnosis of migraine may help to improve management choices for children in EDs, because older children and children for whom the physician made a diagnosis of migraine were more likely to be treated with evidence-based care. In the absence of >1 randomized trial of migraine-abortive treatment in pediatric EDs, it is not surprising that significant practice variations among the 10 pediatric EDs were observed. Although dopamine receptor antagonists (predominantly metoclopramide) were the most-frequently prescribed agents and are effective migraine-abortive medications in adults,8 studies with children are lacking. Only 1 randomized, controlled trial of intravenously administered prochlorperazine versus ketorolac has been performed.3 At least 1 adult study found that prochlorperazine was superior to metoclopramide,9 and an uncontrolled pediatric study of prochlorperazine indicated a very high response rate.10 Whether prochlorperazine and metoclopramide are comparable with respect to safety and efficacy in children requires further evaluation.
Orally administered analgesics and NSAIDs were the next most-commonly prescribed medications in EDs. Ibuprofen is superior to placebo but not clearly superior to acetaminophen in the relief of headache associated with migraine.11,12 Ibuprofen and other NSAIDs are nonspecific cyclooxygenase inhibitors, and these medications also may have unique properties of interest for patients with treatment-resistant conditions presenting to EDs.13 The combination of NSAIDs with prochlorperazine or metoclopramide was observed for one-third of subjects. Although the practice of combining medications (eg, sumatriptan and naproxen) was evaluated in some outpatient studies,14 no studies examined the benefits in EDs. Given its relatively common use across Canada, further examination of the efficacy of combined therapy in EDs seems warranted.
Migraine is a chronic disorder, and children may not be receiving effective outpatient care. Approximately 15% of children who presented to EDs experienced headache >15 days per month, but only 1 in 5 was being treated with prophylactic medications, and an orally administered analgesic had already failed for >60%. Presentation to an ED represents a unique opportunity for intervention not only for the current migraine attack but also after discharge. Specific migraine management strategies after discharge may reduce disability and improve outcomes; these interventions have not been well studied. Prospective evaluation of discharge plans, in terms of reduced ED visits, improved quality of life, or reduced headache disability, warrants further evaluation.
There are several limitations to this study. First, because many Canadian children present to nonpediatric EDs and Canadian territories do not have pediatric EDs, the results of this study are not generalizable to all children who present to Canadian EDs. Given the representation of the sites, however, our study is representative of a national sample of pediatric EDs. Second, the retrospective nature of the study precluded examination of effectiveness, and missing data were identified. The main objective, however, was to study variations in management, which were well documented in medical record reviews.
Presentation to an ED with migraine represents a failure of outpatient therapy, and most children have already used orally administered analgesics. Improved diagnostic tools for migraine for emergency physicians may secondarily improve the care of children with migraine and may be an appropriate target for future research. Of the evidence-based medications, dopamine receptor antagonists were the most-commonly used and included predominantly metoclopramide and prochlorperazine, but these medications have not been compared in children. The practice of using combinations of medications is common and warrants formal evaluation. Emergency physicians are uniquely positioned to intervene for this vulnerable population of patients, who generally present late in the course of a migraine attack, and a program of research is urgently required.
The study was cofunded by the Stollery Children's Hospital Foundation (Edmonton, Canada) and the Canadian Institutes of Health Research (Ottawa, Canada). Dr Rowe's research is supported by a 21st Century Research Chair in Emergency Medicine from the Government of Canada.
We thank Annie-Claude D'Anjou, Janie Williamson, and Marie-Robert André for their contributions.
- Accepted February 25, 2010.
- Address correspondence to Lawrence Richer, MD, FRCPC, University of Alberta, Department of Pediatrics, Room 7312, 11402 University Ave, Edmonton, Alberta, T6G 2J3, Canada. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- ED =
- emergency department •
- NSAID =
- nonsteroidal antiinflammatory drug •
- OR =
- odds ratio •
- CI =
- confidence interval
- Powers SW,
- Patton SR,
- Hommel KA,
- Hershey AD
- Colman I,
- Brown MD,
- Innes GD,
- Grafstein E,
- Roberts TE,
- Rowe BH
- Trottier E,
- Bailey B,
- Dauphin-pierre S,
- Gravel J
- Hämäläinen ML,
- Hoppu K,
- Valkeila E,
- Santavuori P
- Lipton RB,
- Dodick DW,
- Adelman JU,
- et al
- Copyright © 2010 by the American Academy of Pediatrics