Key Findings and Strength of Evidence

This comparative effectiveness review (CER) report provides a comprehensive synthesis of the evidence on the comparative effectiveness of parenteral pharmacological interventions versus standard care, placebo, or an active treatment in the treatment of acute migraine headaches in adults visiting the emergency department (ED) or an equivalent setting. The strength of the body of evidence for key effectiveness outcomes is summarized by intervention below.

For the majority of studies pain relief or severity was the primary outcome. There were nine different classes of drugs investigated in 71 studies. The interventions included metoclopramide, neuroleptics, ergotamines, nonsteroidal anti-inflammatory drugs (NSAIDs), opioids, corticosteroids, triptans, magnesium sulfate (MgSO₄), and antihistamines. There were several studies that examined combinations of active agents compared with other active agents. The mixed treatment analysis included a group of drugs collectively referred to as “orphan agents”.

Data were provided primarily from randomized controlled trials (RCTs). Risk of bias assessment showed that 28 percent of the trials had low risk of bias and 61 percent had unclear risk of bias. Sample sizes varied but they were generally small, with an overall median of 64 patients per study (interquartile range [IQR]: 40 to 100).

Generally, active interventions compared with placebo were more effective in relieving pain and reducing headache recurrence. In the mixed treatment analysis of pain relief (VAS), there was a clear indication that combinations of anti-migraine medications and monotherapy with neuroleptic agents out-performed other active agents. The pain relief data must be weighed carefully with the data on adverse effects, especially akathisia. The following is a summary of the evidence for the six Key Questions.

Key Question 1. Effectiveness of Parenteral Interventions Versus Placebo or an Active Treatment

The mixed treatment analysis showed that the most effective treatments were combination therapy (i.e., dihydroergotamine [DHE] added to either neuroleptics or metoclopramide) or neuroleptic monotherapy (low SOE) with a pain reduction of approximately 40 mm on the VAS. Metoclopramide monotherapy, opioids, and NSAIDs were the next most effective treatments with a pain reduction of approximately 24 mm (low SOE). Other agents (e.g., DHE, triptans, orphan agents) were less effective with a pain reduction of approximately 12–16 mm (low SOE).

Metoclopramide was compared with placebo in six trials and with other active treatments in nine trials (Table 45). Metoclopramide was significantly more effective than placebo for pain relief (moderate strength of evidence). Metoclopramide was generally less effective than neuroleptics for pain relief (low strength of evidence). Results for pain relief were inconsistent when comparing metoclopramide monotherapy with other active treatments (excluding neuroleptics). Single trials compared metoclopramide with MgSO₄, ondansetron plus paracetemol, pethidine, and sumatriptan (insufficient strength of evidence). The mixed treatment analysis demonstrated that as monotherapy, metoclopramide was similarly effective to opioids and NSAIDs for pain relief (low strength of evidence). There was insufficient strength of evidence for headache recurrence when comparing metoclopramide with other active agents including neuroleptics.

Table 45

Summary of the strength of evidence for the effectiveness of metoclopramide versus placebo or an active treatment (Key Question 1).

Neuroleptics were compared with placebo in seven trials and with other active treatments in 17 trials (Table 46). Neuroleptics were more effective than placebo for pain relief (moderate strength of evidence) and for headache recurrence (low strength of evidence). Neuroleptic agents were generally more effective than other active treatments for pain relief, but this wasn’t consistent across studies. More patients who received droperidol experienced headache relief compared with patients who received prochlorperazine based on two RCTs (moderate strength of evidence). For all other head to head comparisons, single trials compared different neuroleptics with anticonvulsants, corticosteroids, dihydroergotamine (DHE), other neuroleptics, NSAIDs, opioids, somatostatin analog, sumatriptan, and lidocaine (insufficient strength of evidence). Single trials compared a neuroleptic agent with another active agent for headache recurrence (insufficient strength of evidence). The mixed treatment analysis demonstrated that monotherapy with neuroleptic agents was one of the more effective treatment options (low strength of evidence).

Table 46

Summary of the strength of evidence for the effectiveness of neuroleptics versus placebo or an active treatment (Key Question 1).

NSAIDs were compared with placebo in two trials and with other active treatments in nine trials (Table 47). NSAIDs were more effective than placebo for pain relief (moderate strength of evidence). There was insufficient strength of evidence for headache recurrence when NSAIDs were compared with placebo. Results were mixed for NSAIDs compared with other active agents for pain relief. Single trials compared NSAIDs with meperidine, sumatriptan, paracetamol, DHE, and tramadol (insufficient strength of evidence). The mixed treatment analysis demonstrated that NSAIDs were similarly effective to opioids and metoclopramide (low strength of evidence). There was insufficient strength of evidence for headache recurrence when NSAIDs were compared with active agents.

Table 47

Summary of the strength of evidence for the effectiveness of NSAIDs versus placebo or an active treatment (Key Question 1).

Opioids were compared with placebo in three trials and with other active treatments in 13 trials (Table 48). Opioids were more effective than placebo for pain relief (moderate strength of evidence). Results were mixed for opioids compared with other active agents for pain relief. Single trials compared opioids with hydroxyzine, other opioids (i.e., nalbuphine, meperidine), methotrimeprazine, metoclopramide, methotrimeprazine, neuroleptic agents, NSAIDs, and DHE (insufficient strength of evidence). The mixed treatment analysis demonstrated that opioids were similarly effective to NSAIDs and metoclopramide (low strength of evidence). There was insufficient strength of evidence for headache recurrence when comparing opioids and other active agents.

Table 48

Summary of the strength of evidence for the effectiveness of opioids versus placebo or an active treatment (Key Question 1).

DHE was compared with other active treatments in five trials (Table 49). Results were mixed for pain relief. Single trials compared DHE with meperidine, neuroleptics agents, sumatriptan, lidocaine, and lysine acetylsalicylic acid (insufficient strength of evidence). There was insufficient strength of evidence for headache recurrence when comparing DHE with other active agents. The mixed treatment analysis demonstrated that DHE monotherapy was similarly effective to orphan agents and anti-nauseants, but less effective than opioids, NSAIDs, and metoclopramide (low strength of evidence).

Table 49

Summary of the strength of evidence for the effectiveness of DHE versus placebo or an active treatment (Key Question 1).

Triptans were compared with placebo in eight trials and with other active agents in six trials (Table 50). Sumatriptan was more effective than placebo for pain relief (moderate strength of evidence), and more effective than placebo for headache recurrence in the ED setting (low strength of evidence). Results were mixed for pain relief when triptans were compared with other active agents. Single trials compared triptans with neuroleptics, metoclopramide, trimethobenzamide, DHE, and ketorolac (insufficient strength of evidence). The mixed treatment analysis demonstrated that sumatriptan was similarly effective to orphan agents and other anti-nauseants, but less effective than opioids, NSAIDs, and metoclopramide (low strength of evidence). There was insufficient strength of evidence for headache recurrence when comparing triptans with other active agents.

Table 50

Summary of the strength of evidence for the effectiveness of triptans versus placebo or an active treatment (Key Question 1).

MgSO₄ was compared with placebo in four trials and with other active agents in two trials (Table 51). MgSO₄ was more effective than placebo for pain relief (moderate strength of evidence) and headache recurrence (low strength of evidence). There was insufficient strength of evidence for pain relief and headache recurrence when comparing MgSO₄ with other active agents.

Table 51

Summary of the strength of evidence for the effectiveness of MgSO₄ versus placebo or an active treatment (Key Question 1).

Antihistamines were compared with placebo in one trial (Table 52). There was insufficient strength of evidence for pain relief.

Table 52

Summary of the strength of evidence for the effectiveness of antihistamines versus placebo or an active treatment (Key Question 1).

Eight RCTs compared eight different combination interventions with other active agents (Table 53). There was insufficient evidence to draw conclusions about the effectiveness of specific combination therapies for pain relief because single trials with low power investigated different pairs of interventions. The mixed treatment analysis demonstrated that DHE in combination with metoclopramide or neuroleptic agents was one of the more effective treatment options (low strength of evidence).

Table 53

Summary of the strength of evidence for the effectiveness of combination interventions versus an active treatment (Key Question 1).

Key Question 2. Corticosteroids in the Prevention of Migraine Relapse

Seven studies assessed the effectiveness of dexamethasone compared with placebo in the prevention of migraine relapse. Patients receiving dexamethasone plus standard care were less likely to report recurrence of pain or headache up to 72 hours after discharge compared with placebo plus standard care (moderate strength of evidence; Table 54). Of all patients with migraine, the subgroups most likely to benefit from dexamethasone are discussed under KQ 5 and 6.

Table 54

Summary of the strength of evidence for corticosteroids in the prevention of migraine relapse (Key Question 2).

Key Question 3. Safety of Parenteral Interventions Versus Placebo or an Active Treatment

We did not conduct a traditional pair-wise meta-analysis of adverse effects because we did not identify multiple studies testing the same medications and reporting common adverse effects (insufficient strength of evidence). We present a summary of adverse effects that provides an overall picture of which interventions had high rates of specific adverse effects.

The main side effect of neuroleptic agents was akathisia; the odds of experiencing akathisia was in the range of 10 times greater than with placebo and was similar to metoclopramide. There were few short-term side effects reported for NSAIDs. For patients receiving DHE, several side effects were reported—the most common were skin reactions (29 percent), local reactions (22 percent), sedation (20 percent), digestive issues (12 percent), nausea or vomiting (11 percent), and chest symptoms (9 percent). There were few short-term side effects reported for opioids. While the risk of dependence and the association with headache relapse are important long-term side effects, these were beyond the scope of this review. Short-term side effects were infrequent for patients receiving triptans. The most common side effect was local reaction in 39 percent of patients; however, this is not surprising since these agents were all delivered subcutaneously. Chest symptoms (5 percent) were relatively infrequent. Due to the select populations in trials, the potential for adverse effects of the triptans might be higher, especially for patients with vascular risk factors. In patients receiving MgSO₄, high rates of skin flushing (10 percent) and local reactions (43 percent) were reported.

Key Question 4. Akathisia

Akathisia is a perplexing adverse effect associated with the use of several effective acute migraine headache treatment options. While self-limited, this symptom complex creates patient discomfort and distress, as well as provider anxiety. The mixed treatment analysis indicates that metoclopramide and neuroleptics (e.g., prochlorperazine) are the anti-migraine agents most likely to cause these symptoms. Though other agents were associated with akathisia in the mixed treatment analysis, lack of precise diagnostic criteria may limit these results.

Clinicians commonly co-administer antihistamines (e.g., diphenhydramine, hydroxyzine) or anticholinergic agents (e.g., promethazine) with neuroleptics and metoclopramide to prevent akathsia. However, this review failed to identify convincing evidence to support this practice (Table 55). The small number of studies and small sample sizes of the included studies produced imprecise point estimates.

Table 55

Summary of strength of evidence for the development of akathisia when anticholinergic agents are added to metoclopramide or phenothiazines.

Key Questions 5 and 6. Subpopulations

This review cannot comment on variability in response to anti-migraine treatment due to sex, race, or duration of headache because included studies often did not report results based on these variables. In one study where sex was reported as a subgroup, sex did not predict headache relapse.²¹

In one trial, dexamethasone was less effective at preventing relapse in patients who had more residual pain at discharge (VAS scores >2). In three trials,^19–21 dexamethasone was more effective in patients with prolonged headaches. In one published systematic review,²⁸ authors found that higher doses (≥15 mg) of IV dexamethasone were more effective than lower doses (<15 mg). These dose comparisons were repeated in this current review and, while similar trends were observed, the differences were not statistically significant.

Findings in Relationship to What Is Already Known

Clinicians treating acute migraine headaches use a wide variety of parenteral agents.¹¹⁹ Research on practice patterns in adult patients with acute migraine headaches demonstrates considerable variation as well as the use of non-evidence based treatments.^16,120 Consequently, this CER is timely.

This review provides a comprehensive and up-to-date review of the available evidence. This includes evidence from placebo-controlled trials and head to head trials. Although there are published systematic reviews of DHE,¹²¹ metoclopramide,¹²² meperidine,¹²⁰ and systemic corticosteroids,²⁸ this CER contextualizes each class of medication vis-à-vis every other class of acute migraine therapeutics. To our knowledge, there have been no mixed treatment analyses published on this topic. While we did not conduct a meta-analysis of adverse effects, the evidence that we present provides a comprehensive summary of adverse effects across studies and interventions for this patient population.

The methodological techniques of the current review are robust and comprehensive which should help to inform clinical practice guidelines and clinical decisionmaking in the future.

Applicability

The study populations included in this CER were relatively homogenous. Most patients were females, and the mean age was generally between 30 and 40 years. Few studies reported on race or ethnicity; however, race was not an inclusion or exclusion criterion for any of the trials. Therefore, it would appear that these results are generalizable to most patients with acute migraine seen in similar EDs based on sex and age. Results may not apply to patients seen in EDs that serve more culturally diverse populations. It is unknown whether males respond differently than females to the interventions included in this review. Similarly it is unknown whether the results of this review apply to older populations.

Headache severity on admission was reported in a variety of ways. In studies that reported a baseline VAS (mm), the mean scores ranged from 6.3 to 9.4, indicating moderate to severe headaches. In other studies, patients rated their headache as moderate or severe. Migraine headache was diagnosed using the International Headache Society criteria² in 61 percent of the studies; the remaining studies used other criteria (19 percent), or did not specify their criteria (20 percent). The median baseline headache severity (VAS = 8 mm) for studies that used other criteria or did not specify their criteria was the same as for studies that used the International Headache Society criteria. The results of this review may be generalizable to patients who present to the ED for treatment of moderate to severe acute migraine headache that has not responded to simple analgesics, and for whom IV agents are being contemplated.

The majority of trials took place in the ED (79 percent). For two comparisons more than 50 percent of the studies were conducted in a non-ED setting (NSAIDs versus placebo (2 of 2 studies) and MgSO₄ versus placebo (2 of 4 studies). The results for these interventions may not be generalizable to the ED setting.

The majority of trials took place in the United States or Canada (75 percent). Of the six studies investigating MgSO_4, four took place in either Brazil or Turkey. Of the nine studies that examined NSAIDs, five took place outside North America. The results of these studies may not be generalizable to acute migraine patients in the United States.

Limitations of the Existing Evidence

The strength of the evidence was insufficient for the majority of outcomes across the head to head drug comparisons. This is primarily due to single, relatively small trials comparing pairs of active treatments. Where there were multiple trials, the strength of the evidence was low to moderate. These low grades were driven by moderate risk of bias within individual studies and a lack of consistency across studies. Most of the lack of clarity arose from poor descriptions of the system of randomization and concealment of allocation; however, this may be a limitation in the reporting and not of the conduct of the trials.

There is a relatively small body of evidence for the parenteral treatment of acute migraine headache in the ED setting, and the evidence arises from small studies, usually from single centers. Consequently, unique features (e.g., dose of drug, addition of an anticholinergic) make comparisons difficult. In addition, the therapeutic versus subtherapeutic dosing variation may limit some comparisons. This results in infrequent pooling and unclear direction of effect. For example, although there were multiple studies that investigated neuroleptic agents, use of different specific agents, doses, and comparators, as well as variable use of anticholinergic or antihistamine agents makes it difficult to draw conclusions about this class of drugs. Conversely, the corticosteroid data on relapse demonstrate the power of having consistent comparisons since the results are robust, precise, consistent, and generalizable.

There was inconsistency in reporting the outcomes from the studies included in this CER, which hampered efforts to provide pooled evidence summaries. In the case of the main primary outcome of pain relief, the reporting of VAS scores, complete relief, ordinal scales, and other methods limited the number of studies included in the results, and may have biased estimates of effect. The direction of this bias is difficult to estimate.

The lack of consistency in the reporting of adverse effects impaired the ability of the review to examine the relative safety of these agents. For example, the definition of adverse effects, the timing of assessment, and the scoring method used varied across studies. Still, serious or unexpected adverse effects were uncommon.

A small number of studies and overall small sample sizes contributed to imprecision. The nonsignificant differences between treatment comparisons reflect these weaknesses, and should not prompt conclusions about equivalence. Equivalence claims would require considerably larger sample sizes and 95 percent confidence intervals that did not include the minimally clinically important differences.

Mixed treatment analyses make an inherent assumption that the direct and indirect evidence estimate the same parameter. We checked the data for inconsistency and found that the number of inconsistent nodes was small. Therefore, inconsistency was not a major concern. We also had categories “active combination agents” and “orphan agents” that do not distinguish between possible heterogeneous treatments within these groups.

In addition to the issues identified above, this CER has several limitations. Due to the small number of studies for each comparison we were unable to formally assess the potential for publication bias. Nonetheless, a comprehensive search of the published and grey literature was conducted without restrictions on study design or language. Consequently, the risk of publication bias should be low. There is also the possibility of study selection bias. To address this, at least two independent reviewers identified potentially relevant studies and the authors are confident that the studies that were excluded were done so for consistent and appropriate reasons. Our assessment of the methodological quality on study publications was performed independently using the risk of bias tool, and we did not contact authors to verify the methods used. Some studies may have been adequately conducted; however, the methods were poorly reported.

Future Research

The following general recommendations for future research are based on the preceding discussion regarding the limitations of the current evidence:

• Since many of the trials demonstrated a benefit to treatment that exceeded placebo effect, placebo-controlled trials in this field should be replaced with comparative effectiveness research focusing on migraine-specific agents for the delivery of care.
• Since many clinicians provide combination agents when patients present with acute severe migraine headache, more efforts should be initiated to determine the effectiveness of combination agents compared with sequential administration of agents or monotherapy.
• Consensus on outcomes and outcome measures, including adverse effects, is needed to ensure consistency and comparability across future studies. Moreover, consensus on minimal clinically important differences is needed to guide study design and interpretation of results.
• Research in parenteral management of acute migraine is robust and ongoing. Consequently, updating this review should be a priority within 5 years.
• Future RCTs should investigate important subpopulations who may differentially respond to migraine treatment. This includes subgroup analyses by sex, race or ethnicity, age (e.g., older age groups), and duration of headache.
• Many trials included in this review were small and conducted in a single-center, which may have delayed the dissemination of evidence and knowledge more than necessary. A multi-centered acute migraine headache collaboration or consortium in emergency medicine would be an efficient method to answer the remaining important questions. The results from this review support calls for well-powered multi-center trials using standardized methodologies.
• Future RCTs should seek to minimize risk of bias by blinding study participants and outcome assessors, adequately concealing allocation, and handling and reporting missing data appropriately.
• Trials should be designed and conducted to minimize bias where at all possible. Investigators may find tools such as the CONSORT statements¹²³ helpful in designing and reporting on RCTs.

Conclusions

This report provides the most comprehensive synthesis of the comparative effectiveness of parenteral pharmacological interventions versus standard care, placebo, or an active treatment in the management of acute migraine headaches in adults presenting to the ED or an equivalent setting. Overall, there are several important conclusions from this work. First, many agents appear to be effective in the treatment of acute migraine headache when compared with placebo. Neuroleptic monotherapy or DHE in combination with either metoclopramide or neuroleptics appear to be the most effective options for pain relief (VAS). Second, several treatments reported here provide insufficient evidence for continued use (e.g., lidocaine, anithistamines, sodium valproate). Third, systemic corticosteroids effectively prevent relapses, especially in patients with prolonged headaches. Finally, the list of adverse effects is extensive, albeit they vary among agents and classes of drugs. Overall, the effectiveness of therapies described here must be weighed against their side effects to derive a strategy for treating patients with this common disorder. While the evidence collated here is an important step, more research is required in order to identify the most effective and safest parenteral medication for acute migraine.