Article, Neurology

Multicenter prevalence of opioid medication use as abortive therapy in the ED treatment of migraine headaches

a b s t r a c t

Despite a range of therapeutic options for treating acute migraine headaches, the use of opioids is still reported to be common practice. This study describes Treatment practices in regards to migraines in the ED. It characterizes the prevalence of opioid orders during visits in three different settings, an academic medical center, a non-aca- demic urban ED, and a community ED.

Fourteen months of consecutive migraine visits were identified. All medications ordered were separated into first-line and Rescue medications. Number of visits, length of stay, door to provider time, and total provider time were compared.

A total of 1222 visits were identified. Opioids were ordered in 35.8% of these visits. By facility, opioids were or- dered in 12.3% of academic medical center visits, 40.9% of urban ED visits, and 68.6% of community ED visits. This ranged from 6.9% of first-line therapies in the academic center to 69.9% of rescue therapies in the community ED. Of those who received opioids, 36.0% versus 25.1% required rescue medications. Patients who received opi- oids had more repeat visits, 1.79 versus 1.30. The academic center and urban ED both found greater than 30% de- crease in length of stay in visits where opioids were not given.

In the face of evidence against opioids for migraines, over one third of patients received them. There was a higher prevalence in the community setting. There were no significant benefits in overall throughput time, however, opioid visits required more rescue medications, increased length of stay, and resulted in more repeat visits.

(C) 2017



Headaches account for a significant portion of emergency depart- ment (ED) visits. As of 2010, they were among the top 20 primary coded diagnoses representing almost 2 million visits yearly in the US [1]. A significant portion of these are further specified as migraine head- aches. In 2008, there were over 1.5 million visits to EDs for migraines [2]. When compared to other clinical settings, the ED provides almost 20% of all migraine care [3].

Migraine sufferers seek emergency care for acute reduction in their pain. Many medications have been shown to be effective in this scenario. The most efficacious options include NSAIDs (ketorolac, ibuprofen, naproxen), antiemetics (metoclopramide, promethazine, prochlorperazine), triptans (rizatriptan, sumatriptan), and ergotamines (dihydroergotamine [DHE]) [4-6]. These tend to be mainstays of abortive therapy.

* Corresponding author.

E-mail address: [email protected] (N. Young).

opioid medications are not indicated for Migraine therapy. In head to head studies, ketorolac, antiemetics, and DHE have all been shown to be superior to opioids [7]. They can lead to chronic migraines and re- bound headaches [8]. In addition to being considered ineffective, they have adverse pathophysiologic and psychologic effects and are general- ly counterproductive as an abortive agent [9]. The American Academy of Neurology has determined that reducing opioid usage in migraine care is significant enough to be a primary goal in their choosing wisely cam- paign [10]. In this campaign a taskforce of neurologists has determined that abortive opioid use is one of the top 5 correctable actions to be ad- dressed within their specialty.

Despite a wide range of therapeutic options, ordering opioids still appears to be a fairly common practice. The prevalence of opioid use for migraine headaches in the emergency department over a span of over a decade appears that there has been no change in this practice [10]. In 1998 and 2010 almost half of visits for acute migraine had opi- oids ordered. These numbers were extracted from the National Hospital Ambulatory Medical Care Survey data which examines a sixth of the nation’s emergency departments, looking at 100 randomly chosen visits over a 4 week period of time. The validity of the NHAMCS data set has been put into question [12,13,14]. In reviewing the litera- ture, there are no recent studies using other methodologies aimed at

0735-6757/(C) 2017

analyzing the Prescribing habits of emergency providers in treating mi- graine headaches.


The purpose of this study was to accurately explore and describe provider treatment habits in three different settings: an academic med- ical center, a non-academic urban hospital, and a community hospital. Basic demographics of patients and throughput measures related to these visits were examined.


Study design and variables

This was a cross-sectional analysis of consecutive adult emergency department patients in three diverse EDs who received treatment for migraine headache. Expedited approval by the Institutional Review Board that encompasses all three facilities was obtained. Investigation into the rates of utilization of migraine abortive medications, attributes of patients that receive these medications, and factors that are associat- ed with usage was explored. Analysis of patient characteristics and ED courses were conducted.


From Jan 1, 2014 through Feb 27, 2015 opioid prescribing habits were assessed at three EDs within the state of Connecticut of varying ac- ademic and geographic basis: an academic medical center, an urban non-academic hospital, and a community hospital. The academic center is an academic, tertiary referral, Level I Trauma Center ED with a volume of approximately 110,000 patients annually. The urban hospital has a fast-paced city emergency department that sees approximately 85,000 patients a year. The community hospital is a smaller facility with ap- proximately 19,000 ED visits a year. Emergency medicine residents con- tribute to care at the academic center only. There was no known provider that practiced at multiple sites.


All patients presenting to a participating emergency department during the 14 month study period were assessed for inclusion. Eligible patients were 18 years of age or older. Those with primary or secondary diagnosis of migraine headache or one of its variations (ICD9 code 346 or 346.*) had information extracted for data collection. Patients who Left without being seen, left prior to receiving medications, or did not re- ceive medications associated with migraine abortive therapy were ex- cluded (see Table 1 for list of medications). Patients with the conflicting neurologic diagnosis of stroke or intracranial hemorrhage were excluded. Patients who received nitrates for treatment of acute coronary syndrome or decompensated heart failure were also excluded.


The main outcome was the portion of patients who received opioids for migraine treatment. The time in which these medications were given was assessed. All initial orders were considered first line therapy. Orders placed 60 min or more from the initial order(s) were considered Rescue therapy. Several factors were compared including number of visits per patient, age, and gender. Throughput measures were assessed at each facility including door to provider time, door to medication time, and length of stay. Finally, all orders were examined for the prevalence of the therapies used.

Table 1

List of migraine-centered medications analyzed.

Category Medication

Opioid Codeine (and combinations)

Fentanyl Hydrocodone Hydromorphone Morphine

Oxycodone (and combinations)

Antiemetics Ondansetron

Meclizine Metoclopramide Prochlorperazine Promethazine

NSAIDS Ibuprofen

Ketorolac Naproxen

Muscle relaxants Alprazolam

Clonazepam Cyclobenzaprine Diazepam Lorazepam Methocarbamol

Other Acetaminophen

Acetaminophen/butalbital/caffeine Aspirin

Dexamethasone Diphenhydramine Hydroxyzine

IV fluids Magnesium Sumatriptan Tramadol

Data sources

Patients were identified from database maintained by d2i (formerly Emergency Medicine Business Intelligence [EMBI]). This database auto- matically extracts information from the electronic medical records of all ED visits and was available for each of the included EDs. Patients ful- filling the inclusion criteria had the following information assessed: medical record number, account number, visit date, age, gender, length of stay, time from arrival to when first seen by provider, time from arriv- al to when first given medication, and all migraine-centered medica- tions administered with time markers.

Study size and statistical methods

A sample size of 783 individual patient encounters was required to determine a 5% difference in opioid use from an average of 50% as discussed in the literature with an ? = 0.05 and ? = 0.2.

Total migraine visits and average migraine visits per patient were

assessed as descriptive data. Medications ordered during each visit were separated into drug class and further divided by specific agent; they were separated into first-line and rescue therapy categories. If a medication was reordered during a single visit within either of these time categories, it was counted only once within that time category. Medications were represented as a proportion of visits in which the agent was used.

Comparisons of opioid and non-opioid Treatment regimen groups were made. The portion of opioid use by facility was assessed using Chi Squared testing. Median age, average number of visits during study period, average door to provider time, average door to medication time, and average length of stay was compared using both t-testing and Wilcoxon testing for statistical significance. Gender was compared using Chi Squared analysis for statistical significance. These calculations were conducted using JASP (Version statistical software.

Table 2

Characteristics of migraine visits to the three participating facilities.

Table 3

Throughput time assessing average additional time in minutes between visits where opi- oids were given versus those without [95% CI].

Academic Urban Community Total

Unique visits







Length of stay

Unique patients






15.9 [-2.3 to 34.1]

8.4 [-14.1 to 34.1]

111.0 [65.5 to 156.5]

Average visits/patients






Median age (yrs)

37.5 (29, 48)

35 (28, 45)

42 (33, 51)

36 (28, 47)

Urban ED

-1.7 [-8.4 to 4.9]

-1.5 [-8.4 to 5.4]

43.2 [28.9 to 57.6]

Sex Male (%)

79 (20.2)

100 (15.2)

32 (18.3)

211 (17.3)

Community ED

-0.1 [-7.8 to 7.7]

-4.8 [-13.0 to 3.4]

12.1 [-12.1 to 36.3]

Female (%)

311 (79.8)

557 (84.8)

143 (81.7)

1011 (82.7)



by 43.2 min [95% CI 28.9 to 57.6] in the urban ED. There was a trend to- wards a 12.1 min [95% CI -12.1 to 36.4] reduction at the community ED.

A total of 1234 visits was assessed for inclusion. Twelve were re- moved due to various reasons. Five patients had acute coronary syn- drome and was given nitroglycerine, 3 were ischemic strokes given tPA, and 4 were various other primary diagnoses without pain or mi- graine centered medications provided during their visit. The remaining 1222 visits was included for final analysis (Table 2). The urban ED re- ceived most visits at 657, followed by the academic center at 390, and the community ED at 175.

There were 931 unique patients within the total visit dataset. On av- erage, patients had 1.31 visits. The visits per patient were highest at the community ED with an average of 1.57, followed by 1.38 at the urban ED, and 1.13 at the academic center. The average age was 35 years for females and 37.5 years for males. Patients were predominantly female, averaging 82.7% of visits across all 3 centers.

Main results

Of the 1222 visits for migraine headaches, 35.8% had opioid medica- tions ordered (Graph 1). Overall, opioids were given in 68.6% of visits in the community ED, 40.9% in the urban ED, and 12.3% in the academic center. Opioids were used as a first-line agent in 29.5% of visits on aver- age and as a rescue agent in 49.4% of visits where additional medications were required. Opioids were given 58.2% of the time as first-line agents in the community ED, compared to 35.3% in the urban ED and 6.9% in the academic center. As a rescue medication, opioids were given 69.9% of the time at the community ED, 63.9% at the urban ED, and 20.6% at the academic center.

Throughput and other variables

In assessing throughput measurements, there was no statistical dif- ference in the time to being first seen by a provider or the time to receiv- ing the first medication when comparing visits that involved opioids versus those that did not (Table 3). Length of stay was significantly re- duced by 111.0 min [95% CI 65.5 to 156.5] at the academic center and

Patients who were given opioids as part of the treatment regimen had a 37.7% increase in visits over the study period than those who were not given opioids (1.79 visits/patient versus 1.30 visits/patient [95% CI 1.207 to 1.617]) (Table 4). Of those who received opioids, 36.0% required further treatment in the form of rescue medications compared to 25.1% in the Standard therapy group (95% CI 0.053 to 0.167). The median age was 41 years in the opioid group and 35 years in the non-opioid group (95% CI 1.11 to 1.30). Visits with female pa- tients were significantly more likely to have opioids ordered than those with male patients, 38.2% versus 24.2% (95% CI 0.072 to 0.201).


In assessing the ordering practices from all 3 facilities, it was found that medications from the antiemetic category were ordered most fre- quently as first-line therapy at 35.3% of all orders, followed by NSAIDs at 16.0%, IV fluids at 13.3%, and opioids at 12.6% (Table 5). Opioids were ordered more frequently as rescue therapies at 29.9%, followed by antiemetics at 23.4%, NSAIDs 12.8%, and non-NSAIDS analgesics 10.0%. Medications included in the other category included antihista- mines, magnesium, and steroids.

Of all medications given the most frequently administered was metoclopramide at 141 orders per 1000 orders, followed by ketorolac at 82 per 1000 orders (Table 6). Included in the top 10 order medica- tions was intravenous fluids (79 per 1000 orders), diphenhydramine (70 per 1000 orders), ondansetron (69 per 1000 orders),

hydromorphone (48 per 1000 orders), acetaminophen/butalbital/caf-

feine (38 per 1000 orders), morphine (28 per 1000 orders), oxyco-

done/acetaminophen (22 per 1000 orders), and dexamethasone (15 per 1000 orders).


There are several limitations to this study. Observational studies can convey associations but are unable to determine causality. While this study was formatted to look only at migraines and migraine variants,

Graph 1. Portion of visits where opioid medications were ordered in percent separated by facility and when the medication was provided.

Table 4

Comparison of patient characteristics between visits where opioids were ordered and those without opioids ordered.

Table 6

Top 10 ordered medications listed in rate per 1000 orders.

Opioid visits Non-opioid visits 95% CI


Order rate





















Average visits/patients



1.207 to 1.617

Rescue therapy needed



0.053 to 0.167

Median age (Q1, Q3)

41 (31, 49)

35 (27, 46)

1.207 to 1.617

Sex Male % (n)

24.2 (51)

75.8 (160)

0.072 to 0.201

Female % (n)

38.2 (386)

61.8 (624)

only the ICD-9 code 346 was used to identify cases. Patients with mi- graines who were coded alternatively, such as ICD-9 code 748 for head- ache at the direction of the diagnosing provider, would not have been captured. This could explain why the two EDs serving a similar urban population could have the differing rates of migraine visits as seen in this study.

Additionally, given the variation in how migraines present, such as pain intensity and therapies attempted prior to arrival, providers may initially have had concern for a different underlying pathology. It is pos- sible that some opioids were administered for analgesia prior to rule- out testing. An example would be utilizing opioids over NSAIDS in the setting of concern for intracerebral hemorrhage until imaging was attained. This can account for variation in first-line therapies, but would be less relevant in rescue therapies. Laboratory and radiological work up was not assessed in our study.

Throughput times could also be confounded by migraine intensity and therapies attempted prior to arrival. The majority of medications ordered where via intravenous route when not clearly an oral medica- tion such as oxycodone/acetaminophen. Depending on individual hos- pital protocol, some medications could be given as a bolus while others required a longer infusion. This however would have benefited the group receiving opioids as these are typically bolus dosing.

The majority of variables in this study were assessed utilizing indi- vidual visits despite having a substantial number of patients that had multiple visits. It can be assumed that if a patient had more than one visit, it was likely that they were treated by a different ED provider. This allows for the perspective of how ED providers treat migraine headaches in general rather than the perspective of how the migraine- patient is treated. Having prior visits with medical records describing prior Treatment decisions can bias providers during subsequent visits. Some migraine sufferers may also have an opioid treatment regimen or- dered by their neurologist and this type of documentation was not assessed. This again allows this study to view provider practices with real world influences.

Having a cutoff of 60 min to separate first line and rescue treatments

seems somewhat arbitrary, but it establishes a useful timeframe for dis- cussion. The ED experience at each individual facility was not examined beyond the data collected for this study. For example, it is not possible to know if physicians prescribing opioids at the community ED faced a larger patient burden, such as being a sole provider, and therefore turned to this well-known class of medications for fast relief of patient suffering and in hopes of increasing Patient throughput. It is also impos- sible to know specifically how provider views, training, or culture

Table 5

Portion of migraine therapy category types ordered as a percentage of all orders during migraine visits.

First line























Non-NSAIDS analgesic




Muscle relaxants








regarding opioid administration varied significantly between physicians staffing the community, urban, and academic centers.


This study was unique as it examined and compared the practices of ED clinicians in the treatment of migraine headache across multiple types of facilities over a year. Despite widespread recommendations and evidence against opioids for migraine headaches in neurology guidelines, over one third of patients received them. The highest preva- lence of opioids given was at the community ED where over two thirds of all visits received them. They were given less frequently in the urban ED, but still given to one third of patients as first-line and nearly two thirds of patients as rescue. This is in contrast with the academic center that only gave opioids in 1 out of 14 visits for first-line medications and 1 out of 5 visits as rescue medication.

In comparing these facilities, there was evidence of ineffective knowledge translation of evidence-based migraine guidelines creating a difference in prescribing practices. In places where there are dedicated education programs, the presence of residents and teaching facility seems to significantly improve adherence to recommended practices. This is contrary to suggestions made from NHAMCS data [11]. The Insti- tute for Clinical Systems Improvement provides evidence-based re- views and has made recommendations on improving how treatment of migraine headaches are treated [15]. They recommend an aim of de- creasing opioid medication below 5% of patients. Even at the academic facility 12.3% of all migraine visits had opioids ordered, well above the 5% threshold.

There evidence in this study that reflect why opioids are not favored. In visits where opioids were prescribed, the patients had almost 40% more repeat visits. During visits where opioids were used, patients re- quired rescue medications almost 40% more than their counterparts who did not receive opioids. In this study, non-opioid treatment lead to less visits and a decrease need to add or reorder additional medications.

Patients who did not receive opioid medications had a 30.3% reduc- tion in their length of stay. At the Urban ED, there was a 36.4% reduction. This represents a 1.8 h and 1.6 h reduction in stay on average per patient treated with a nonopioid regimen. This is consistent with a 1.1 h reduc- tion found in an analysis at a similar California academic center [16]. If opioid ordering was solely attributed to this extending the length of stay, this could lead to significant reductions in overall ED throughput. In conclusion, despite evidence against opioid abortive therapy, a significant proportion of patients still receive them. It is unknown why it is still so prevalent. In this study, opioid orders are associated with in- creasing repeat visits, needing to order rescue medications, and increas- ing length of stay. Previous work has shown that there is no improvement in patient satisfaction with ordering opioids [17]. Educat- ing providers and establishing facility-based opioid-free migraine treat- ment guidelines may lead to more appropriate migraine management

strategies and better outcomes.




New England Regional SAEM Meeting in Worchester, MA – March 30, 2016; SAEM Annual Meeting in New Orleans, LA – May 10-13, 2016.

Author contributions

NY, DS, HS, and JF conceived the study, and designed the data queries. NY prepared documents, study forms, and processed this through IRB. JF worked directly with the third party database agency (EBMI) to attain accurate information. NY, DS, and HS ensured data in- tegrity and statistical preparation. NY drafted the manuscript. DS, HS, and JF contributed to the revision of the manuscript. NY takes responsi- bility for the paper as a whole.

Conflicts of interest



  1. The Ambulatory and Hospital Care Statistics Branch. National hospital ambulatory medical care survey: 2010 emergency department summary tables. CDC; 2010.
  2. Lucado J, Paez K, Elixhauser A. Headaches in U.S. hospitals and emergency depart- ments, 2008. HCUP statistical brief #111. Rockville, MD: Agency for Healthcare Re- search and Quality; May 2011.
  3. Schappert S, Rechtsteiner E. Ambulatory medical care utilization estimates for 2006. Natl Health Stat Rep 2008;8:1-30.
  4. Kelley N, Tepper D. Rescue therapy for acute migraine, part 1: triptans, dihydroer- gotamine, and magnesium. Headache 2012;52(1):114-28.
  5. Kelley N, Tepper D. Rescue therapy for acute migraine, part 2: neuroleptics, antihis- tamines, and others. Headache 2012;52(2):292-306.
  6. Kelley N, Tepper D. Rescue therapy for acute migraine, part 3: opioids, NSAIDs, ste- roids, and post-discharge medications. Headache 2012;52(3):467-82.
  7. Gelfand A, Goadsby P. A neurologist’s guide to acute migraine therapy in the emer- gency room. Neurohospitalist 2012;2(2):51-6.
  8. Gilmore B, Michael M. Treatment of Acute migraine headache. Am Fam Physician 2011;83(3):271-80.
  9. Tepper S. Opioids should not be used in migraine. Headache 2012;52(Suppl. 1): 30-4.
  10. Langer-Gould A, Anderson W, Armstrong M, et al. The American Academy of Neurology’s top five choosing wisely recommendations. Neurology 2013;81:1004.
  11. Friedman B, West J, Vinson DR, et al. Current management of migraine in US emer-

    gency departments: an analysis of the National Hospital Ambulatory Medical Care Survey. Cephalagia 2015;35(4):301-9.

    Cooper RJ. NHAMCS: does it hold up to scrutiny? Ann Emerg Med 2012;60(6): 722-5.

  12. Schuur JD, Tibbetts SA, Pines JM. Pregnancy testing in women of reproductive age in US emergency departments, 2002 to 2006: assessment of a national quality mea- sure. Ann Emerg Med 2010;55(5):449-457.e2.
  13. Green SM. Congruence of disposition after emergency department intubation in the National Hospital Ambulatory Medical Care Survey. Ann Emerg Med 2013;61(4): 423-6.
  14. Beithon J, Gallenberg M, Johnson K, Kildahl P, Krenik J, Liebow M, et al. Institute for Clinical Systems Improvement. Diagnosis and treatment of headache. Available from: . [Updated January 2013].
  15. McCarthy LH, Cowan RP. Comparison of parenteral treatments of acute primary headache in a large academic department cohort. Cephalagia 2015;35(9):807-15.
  16. Schwartz TM, Tai M, Babu KM, Merchant RC. Lack of association between Press Ganey emergency department patient satisfaction scores and emergency depart- ment administration of analgesic medications. Ann Emerg Med 2014;64(5):469-81.