• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information

PubMed Health. A service of the National Library of Medicine, National Institutes of Health.

Sumamo Schellenberg E, Dryden DM, Pasichnyk D, et al. Acute Migraine Treatment in Emergency Settings [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2012 Nov. (Comparative Effectiveness Reviews, No. 84.)

Results

This chapter reports on the results of the literature search and evidence synthesis. First, the results of the literature searching, selection process, and a summary of the study characteristics and methodological quality of the included studies are described. The results of analyses are presented by Key Question (KQ). We present the results of the comparative effectiveness of parenteral pharmacological interventions versus placebo, standard care, or active agents (KQ 1 and KQ 2). These results are organized by drug class (e.g., neuroleptics, opioids) and then are grouped by placebo-controlled studies or direct head to head comparisons of drugs or combinations of drugs. The adverse effect results (KQ 3) are organized by categories of adverse effects (e.g., sedation, nausea/vomiting) and then subgrouped by drug class. This is followed by results for the specific side effect, akathisia (KQ 4). Results related to subpopulations (KQ 5 and KQ 6) appear at the end of this chapter.

Metagraphs and tables reporting the strength of evidence for key outcomes are presented within each applicable section. Within each metagraph, the studies that provided data are identified by the name of the first author and year of publication. A list of acronyms is provided at the end of the report.

Literature Search

The search identified 3,138 citations from electronic databases. Screening based on titles and abstracts, grey literature searches, and hand searching identified 231 potentially relevant studies that were evaluated for inclusion. Using a standardized inclusion–exclusion form (Appendix B), 71 studies (and three companion studies) were included, and 157 were excluded (Figure 2). Prospective cohort studies were screened for potential inclusion; however, none met the inclusion criteria. There are 69 randomized controlled trials (RCTs) and 2 nonrandomized controlled trials (NRCTs) in the review. One of the included studies had three associated publications. Three studies were published in non-English language journals; the articles were translated and data extracted by third party translators.

Figure 2 is a flow chart that summarizes study retrieval and selection. It begins with the total number of citations retrieved from the literature searches and ends with the number of studies that met all inclusion criteria of the report. This figure is described in the Section entitled “Literature Search” as follows: “The search identified 2,136 citations from electronic databases. Screening based on titles and abstracts identified 205 potentially relevant studies. We identified 21 additional studies by hand searching the reference lists from included studies, relevant systematic reviews, and conference proceedings. The full text of five articles could not be retrieved through the university interlibrary loan service. Therefore, the full texts of 226 potentially relevant reports were evaluated for inclusion. Using a standardized inclusion–exclusion form (Appendix B), 69 studies were included and 157 were excluded (Figure 2). We screened prospective cohort studies for potential inclusion; however, none met our inclusion criteria. There are 67 randomized controlled trials (RCT) and 2 nonrandomized controlled trials (NRCT) in the review. Three studies were published in non-English language journals; these articles were translated and data were extracted by third party translators. The most frequent reasons for exclusion were: ineligible intervention (53), ineligible study design (44), and ineligible population (18). Forty-two studies were excluded for other reasons (Figure 2). A complete list of excluded studies and reasons for exclusion can be found in Appendix C.

Figure 2

Flow diagram of study retrieval and selection.

The most frequent reasons for study exclusion were: ineligible intervention (53), ineligible study design (44), and ineligible population (18). Forty-two studies were excluded for other reasons (Figure 2). A complete list of excluded studies and reasons for exclusion can be found in Appendix C.

Description of Included Studies

There were 71 unique studies (69 RCTs and 2 NRCTs) that met the eligibility criteria. Nine different classes of drugs were investigated: antiemetics (metoclopramide), neuroleptics, ergotamines, nonsteroidal anti-inflammatory drugs (NSAIDs), opioids, corticosteroids, triptans, magnesium sulfate (MgSO4), and anithistamines. In addition, there were several studies that examined combinations of active agents compared with other active agents. For the mixed treatment analysis, we identified a group of drugs that were not easily classified and were infrequently studied (i.e., hydroxyzine (Atarax), lidocaine, MgSO4, sodium valproate, tramadol, and octreotide). We collectively referred to these drugs as “orphan agents”.

Most trials were of parallel design; three used a cross-over design. Most trials (60, 85 percent) had two study arms. Seven trials (10 percent) had three study arms, and four (6 percent) had four study arms. One study41 described a five armed trial where the efficacy of metamizole, diclofenac, chlorpromazine, MgSO4 and placebo were compared. Since this publication did not provide any extractable data, we included three associated publications that compared the placebo arm with diclofenac,42 chlorpromazine,43 and MgSO4.44 We did not include the publication on metamizole (dipyrone) since this drug is banned in the United States. In the body of this review, we only cite the three publications from which data were extracted.

Evidence tables that describe the studies in more detail are presented in the results section. The studies were published between 1986 and 2011 (median = 2001 [interquartile range (IQR), 1993 to 2004]). The majority of studies were conducted in the United States (62 percent). The rest were conducted in Canada (13 percent), Turkey (8 percent), and other countries (15 percent). The most commonly reported measure of pain was the visual analogue scale (VAS). While there is no consensus on the minimally clinically important difference, a summary of the research suggests that a change in score between 1 and 2 cm (10–20 mm) on the VAS is considered clinically significant.4551 In 43 studies (61 percent) migraine was classified using the criteria established by the International Headache Society.

Methodological Quality of Included Studies

A summary of the risk of bias assessments is presented in Figure 3; the detailed consensus assessments are presented in Appendix D. Overall, 60.6 percent (n = 43) of the trials had an unclear risk of bias, 28.2 percent (n = 20) had low risk, and 11.0 percent (n = 8) had high risk of bias. Risk of bias was generally low for incomplete outcome data, selective reporting, and other bias. This means that these methodological sources of bias were uncommon in this body of evidence. Approximately 50 percent of studies were assessed as unclear risk of bias for sequence generation and allocation concealment.

Figure 3 displays a Risk of Bias summary chart. This figure is described further in the section “Methodological quality of included studies” as follows: “The risk of bias assessments for the 69 RCTs and 2 NRCTs are presented in Appendix D. A summary of the risk of bias for the 71included studies is presented in Figure 3. Risk of bias was mostly generally low for incomplete outcome data, selective reporting, and other bias. The two domains that were mostly split between low and unclear were sequence generation and allocation concealment. There were marginal amounts of high risk of bias in three domains (sequence generation, allocation concealment and incomplete outcome data). The domain where high risk of bias was the most frequent was blinding of subjective outcomes. Overall, 60.6 percent% of the studies trials had an unclear risk of bias, 28.2 percent% had low risk, and 11.0 percent% had high risk of bias. Eleven studies were funded by industry, seven studies were funded by associations and foundations, eleven studies were funded by industry, one received government funding, and two had other sources of funding. Funding was not reported by 47 studies.”

Figure 3

Risk of bias summary for acute migraine headache trials.

Twelve studies were funded by industry,5263 seven were funded by associations and foundations,19,20,6468 one received government funding,69 and two had other sources of funding.70,71 Funding was not reported by 47 (68 percent) studies.21,22,29,32,42,43,72112

Key Question 1. Effectiveness of Parenteral Pharmacological Interventions Versus Standard Care, Placebo or an Active Treatment

The findings for KQ 1 are presented by drug class, comparing the drug class with placebo, if applicable, and then with other active agents. Note that some studies included both head to head and placebo comparisons and appear in both sections. For studies that assessed antiemetics, all but one79 examined metoclopramide. Therefore, we titled the section “Metoclopramide”. As appropriate, we highlight the outcomes that include results from the study that assessed the combination of trimethobenzamide and diphenhydramine.79

Metoclopramide

Key Points

  • Patients who received metoclopramide had greater improvement in pain intensity as measured by VAS (mm) compared with those on placebo based on five RCTs (moderate strength of evidence).
  • Single trials assessed headache recurrence and headache relief for patients who received metoclopramide compared with placebo (insufficient strength of evidence).
  • There was insufficient strength of evidence for improvement in pain intensity (VAS) for patients who received metoclopramide in combination with either dihydroergotamine (DHE) or dexamethasone.
  • In general, neuroleptics were more effective than metoclopramide for pain relief based on four trials (low strength of evidence).
  • There was no statistically significant difference in change in pain intensity (VAS) for patients receiving metoclopramide compared with prochlorperazine based on two RCTs (low strength of evidence).
  • For all other head to head comparisons, single trials compared metoclopramide with another active agent for headache relief, pain free response, headache response, and headache recurrence at various timepoints (insufficient strength of evidence).

Results

The results for the metoclopramide studies are summarized below. Table 4 and Table 5 and provide the strength of evidence grades for all key outcomes. See Table 6 for details on study and patient characteristics.

Table 4. Strength of evidence for metoclopramide versus placebo.

Table 4

Strength of evidence for metoclopramide versus placebo.

Table 5. Strength of evidence for metoclopramide versus active agents.

Table 5

Strength of evidence for metoclopramide versus active agents.

Table 6. Patient and study characteristics of trials comparing metoclopramide with placebo or active agents.

Table 6

Patient and study characteristics of trials comparing metoclopramide with placebo or active agents.

Metoclopramide Versus Placebo
Description of Included Studies

Six RCTs83,91,92,95,107,113 assessed the effectiveness of metoclopramide compared with placebo. One three-armed trial107 compared a combination of metoclopramide plus dihydroergotamine (DHE) with placebo and metoclopramide plus dexamethasone with placebo. The studies were all conducted in the ED. The mean ages of participant groups ranged from 32.1 to 40.0 years. Participants were predominantly female, and no study reported the race or ethnicity of study participants. All studies reported pain relief or severity as the primary outcome. Timepoints measured in the ED ranged from 30 to 60 minutes. Post-ED followup timepoints ranged from 4 to 48 hours. In all but one study,91 the secondary outcomes were adverse effects or ability to function.

Two studies had a low risk of bias;83,91 the remaining four 92,95,107,113 had an unclear risk of bias (Appendix D).

Effectiveness Results

The detailed analyses of results are provided below. Results are presented by outcome. Studies in which metoclopramide monotherapy was compared with placebo are presented first, followed by studies in which metoclopramide was administered in combination with another drug and compared with placebo.

Metoclopramide Monotherapy Versus Placebo

Five RCTs83,91,92,95,113 assessed metoclopramide monotherapy compared with placebo. In each study, participants were administered 10 mg of metoclopramide.

Change in Pain Intensity (VAS)

The change in pain intensity was measured by change in VAS (mm). The pooled results (Figure 4) showed that those who received metoclopramide experienced a statistically significant, homogeneous decrease in pain intensity compared with those who received placebo (MD = −21.88; 95% CI: −27.38, −16.38; I2 = 0%).

Figure 4 displays a forest plot of studies reporting change in pain intensity using the Visual Analogue Scale, metoclopramide vs. placebo. Five studies reported pain intensity. The pooled results showed a significant decrease in pain intensity in those receiving metoclopramide (MD = −21.88, 95% CI: −27.38, −16.38; I2 = 0%).

Figure 4

Change in pain intensity (VAS) in trials comparing metoclopramide and placebo. VAS = visual analogue scale

Headache Relief

One study95 measured relief of headache using a questionnaire given to patients 1 hour after treatment. The difference in headache relief between the two groups was statistically significant in favor of the metoclopramide group (RR = 3.47; 95% CI: 1.50, 8.01).

Relief of Nausea and Vomiting

One study assessed the relief of nausea and vomiting91 and reported that significantly more patients receiving metoclopramide experienced relief compared with those who received placebo (RR = 4.19; 95% CI: 1.35, 13.03).

Headache Recurrence

Patients in one study were contacted 24 hours after discharge from the ED to determine headache recurrence.83 A lower proportion of patients who received metoclopramide experienced recurrence of headache (16/37) compared with those who received placebo (21/40); however, the difference between groups was not statistically significant (RR = 0.82; 95% CI: 0.51, 1.32).

Metoclopramide in Combination Versus Placebo
Pain Improved by at Least One Unit

One study compared metoclopramide plus DHE and metoclopramide plus dexamethasone versus placebo.107 Participants were administered 5 to 10 mg of metoclopramide. Patients were asked to rate their headache on a scale from zero to three (three being the most severe headache). Comparisons of metoclopramide plus DHE versus placebo (RR = 3.89; 95% CI: 1.07, 14.10) and metoclopramide plus dexamethasone versus placebo (RR = 4.09; 95% CI: 1.15, 14.57) significantly favored the metoclopramide combination therapy (Figure 5).

Figure 5 displays a forest plot of studies reporting pain improvement by at least one unit on a four point scale, metoclopramide plus DHE and metoclopramide plus dexamethasone vs. placebo. One study reported pain improvement. Comparisons of metoclopramide plus DHE versus placebo (RR = 3.89, 95% CI: 1.07, 14.10) and metoclopramide plus dexamethasone versus placebo (RR = 4.09; 95% CI: 1.15, 14.57) significantly favored the metoclopramide combination therapy.

Figure 5

Pain improved by at least one unit (four-point scale) in trials comparing metoclopramide in combination with other active agents and placebo.

Return to Normal Functioning

In one study107 patients were asked to rate their ability to function on a scale from zero (normal functioning) to three (requiring bed rest) 30 minutes after injection. More patients who were administered metoclopramide plus DHE improved their function compared with those who were given placebo (RR = 9.90; 95% CI: 0.61, 161.73). Similarly, more patients who were administered metoclopramide plus dexamethasone improved their ability to function compared with those who were administered placebo (RR = 10.08; 95% CI: 0.63, 162.06). The differences in both comparisons were not statistically significant.

Metoclopramide Versus Active Agents
Description of Included Studies

Eight RCTs and one NRCT22,65,79,82,83,91,92,105,113 assessed the effectiveness of metoclopramide versus other active agents. Of these, four22,65,91,92 specifically compared metoclopramide with neuroleptics. All interventions were delivered in the ED with timepoints measured between 30 and 120 minutes. Post-ED followup timepoints ranged from 4 to 48 hours. For all trials, the number of participants who were randomized ranged from 40 to 342 (median = 78; IQR = 70, 91). The mean ages of intervention groups ranged from 31.6 to 40.0 years. All studies had a pain related primary outcome (e.g., pain relief, change in pain intensity, pain free status). The secondary outcomes were varied and included adverse effects, time in ED, and use of rescue medication. See Table 6 for details on study and patient characteristics.

Four trials65,79,83,91 had a low risk of bias, while five22,82,92,105,113 had an unclear risk of bias (Appendix D).

Effectiveness Results
Metoclopramide Versus Neuroleptics

Three studies65,91,92 assessed metoclopramide monotherapy compared with neuroleptics (i.e., prochlorperazine and chlorpromazine). In one study, patients who received metoclopramide or prochlorperazine were also administered 25mg of IV diphenhydramine.22 Participants were administered 0.1 mg/kg,65 10 mg,91,92 and 20 mg22 of metoclopramide.

Change in Pain Intensity (VAS)

All four studies reported change in pain scores as measured on the VAS (mm) (Figure 6).22,65,91,92 Results were consistent across studies in favor of neuroleptic agents. Two studies compared metoclopramide monotherapy with prochlorperazine.65,92 While both studies favored the neuroleptic, only one study reported statistically significant results (MD = 34.0; 95% CI: 19.68, 48.32; I2= 90%).92 Statistically significant results favoring the neuroleptic were found in the one study comparing chlorpromazine with metoclopramide (MD = 25.0; 95% CI: 12.14, 37.86).91 In the study where the antihistamine diphenhydramine was administered to both the metoclopramide and prochlorperazine groups, the differences in pain scores were not statistically significant.22 The pooled results are statistically significant in favor of the neuroleptic agents (MD = 16.45; 95% CI: 2.08, 30.83; I2 = 81%).

Figure 6 displays a forest plot of studies reporting change in pain using the Visual Analogue Scale, metoclopramide vs. neuroleptics. Four studies reported change in pain. One study comparing prochlorperazine and metoclopramide showed significant difference in favor of prochlorperazine (MD = 34.0; 95% CI: 19.68, 48.32). Another study comparing metoclopramide and chlorpromazine showed significant difference in favor of chlorpromazine (MD = 25.0; 95% CI: 12.14, 37.86).

Figure 6

Change in pain (VAS) in trials comparing metoclopramide and neuroleptics. VAS = visual analogue scale

Severe Headache Recurrence

In one study,65 patients were contacted by a nurse by telephone 48 hours post discharge to evaluate recurrence of headache. Patients who received metoclopramide had less headache recurrence compared with patients who received prochlorperazine; however, the results were not statistically significant (RR = 0.41; 95% CI: 0.11, 1.51).

Relief of Nausea and Vomiting

One study91 assessed relief of nausea and vomiting post-treatment and found no statisically signicant difference between metoclopramide and prochlorperazine (RR = 0.71; 95% CI: 0.44, 1.16).

Additional Outcomes

One study compared prochlorperazine and diphenhydramine versus metoclopramide and diphenhydramine.22 The study assessed whether patients could sustain a pain free state (achieving a pain-free state within 2 hours of medication administration and maintaining it for 24 hours), sustained headache relief (for 24 hours), sustained normal functioning, 2 hour pain free, and 2 hour headache relief. For every outcome measurement, the results were not statistically significant.

Metoclopramide Versus Other Active Agents (Excluding Neuroleptics)

Four studies investigated the efficacy of metoclopramide compared with other active agents including: MgSO4,83 ondansetron plus paracetemol,105 pethidine,113 and sumatriptan.82 In one study the dose of metoclopramide was 20 mg;82 in the other four studies the dose was 10 mg. In one study, trimethobenzamide plus diphenhydramine was compared with sumatriptan.79

Change in Pain Intensity (VAS)

Five studies79,82,83,105,113 measured pain or change in pain intensity using the VAS (mm) (Figure 7). In the figure, a negative number is a ‘change from baseline’ while a positive is a final pain score. Results were inconsistent across studies. While three studies reported nonsignificant pain or change in pain intensity,79,83,105 one study reported a statistically significant difference favoring metoclopramide versus pethidine.113 The results of the study that compared trimethobenzamine and diphenhydramide with sumatriptan showed that sumatriptan was more effective but the difference was not statistically significant.79

Figure 7 displays a forest plot of studies reporting change in pain intensity of less than 2 hours using the Visual Analogue Scale, metoclopramide vs. other active agents (excluding neuroleptics). Five studies reported pain intensity. One study had significant results when comparing metoclopramide to pethidine in favor of metoclopramide (−10.00; 95% CI: −19.21, −0.79). There were no differences found for any of the other comparisons.

Figure 7

Change in pain intensity (<2 hours) (VAS) in trials comparing metoclopramide and other active agents (excluding neuroleptics). VAS = visual analogue scale

Headache Recurrence

One study measured headache recurrence at 24 hours and found no statistically significant difference between metoclopramide and MgSO483 (RR = 0.82; 95% CI: 0.51, 1.33)

Other Outcomes

One study assessed the administration of paracetemol with both metoclopramide and ondansetron.105 The study measured the use of additional analgesia, mean duration of ED stay (minutes), and change in pain intensity at 24 hours (measured using a numerical rating system). There were no statistically significant differences between groups for any of the outcomes.

There were no statistically significant differences in the single study of sumatriptan versus trimethobenzamine plus diphenhydramine in the measurement of pain free response at 1, 2, and 24 hours, or for headache response at 1 and 2 hours.79 The same study assessed headache response at 24 hours, limitation to activities, and whether patients wanted the same medication in the future. There were no statistically significant differences for any of the outcomes.

Neuroleptic Agents

Key Points

  • Patients who received neuroleptic agents had greater improvement in pain intensity as measured by VAS (mm) compared with those receiving placebo based on four RCTs (moderate strength of evidence).
  • Patients who received neuroleptic agents had greater headache relief at 1 hour compared with those receiving placebo based on five RCTs (moderate strength of evidence).
  • Fewer patients who received neuroleptic agents experienced headache recurrence compared with those receiving placebo based on two RCTs (low strength of evidence).
  • 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 a neuroleptic agent with another active agent for headache relief, pain free response, headache response, and headache recurrence at various timepoints (insufficient strength of evidence).

Results

The results from studies that compared neuroleptics with placebo or with other active agents are presented below. Note that the studies that specifically compared neuroleptics and metoclopramide were described previously in the metoclopramide section. Table 7 and Table 8 provide the strength of evidence grades for all key outcomes. See Table 9 for details on study and patient characteristics.

Table 7. Strength of evidence for neuroleptics versus placebo.

Table 7

Strength of evidence for neuroleptics versus placebo.

Table 8. Strength of evidence for neuroleptics versus active agents.

Table 8

Strength of evidence for neuroleptics versus active agents.

Table 9. Patient and study characteristics of trials comparing neuroleptics with placebo or other active agents.

Table 9

Patient and study characteristics of trials comparing neuroleptics with placebo or other active agents.

Neuroleptic Agents Versus Placebo
Description of Included Studies

Seven RCTs43,63,68,80,91,92,97 and one NRCT90 evaluated the effectiveness of neuroleptics versus placebo. The neuroleptics included prochlorperazine,68,9092 chlorpromazine,43,97 haloperidol,80 and droperidol.63 Most trials took place in the ED; one took place in a headache clinic.63 The mean ages of the participant groups ranged from 29.6 and 41.0 years; age was not reported in one study.92 In seven studies, the majority of patients were female; in one study, 40 percent of the placebo group was female.90 Race or ethnicity was not reported in any of the studies. The primary outcomes were pain related,43,63,68,80,91,92 incidence of akathisia,90 and response to treatment.97 Secondary outcomes included therapeutic gain, nausea, vomiting, sedation, treatment failures, and successful treatment response or therapeutic gain. The timepoints measured in the ED ranged from 30 minutes to 4 hours. The followup timepoints after discharge ranged from 24 hours to 1 month.

Two studies had a low risk of bias,68,91 five43,63,80,92,97 had an unclear risk of bias, and one90 had a high risk of bias (Appendix D).

Effectiveness Results
Change in Pain Intensity (VAS)

The change in pain intensity was reported by change in VAS (Figure 8). The pooled result was statistically significant in favor of neuroleptics (MD = −46.59; 95% CI: −54.87, −38.32; I2 = 46%).43,80,91,92 In all but one study43 authors reported pain as change from baseline (negative numbers); when these data were not reported, end of study data were presented (positive numbers).

Figure 8 displays a forest plot of studies reporting change in pain intensity using the Visual Analogue Scale, neuroleptics vs. placebo. Four studies reported change in pain intensity. Two studies compared prochlorperazine to placebo and both significantly favored prochlorperazine (MD = −50.80; 95% CI: −70.39, −31.20; I2 = 76%). One study compared haloperidol to placebo and results favored haloperidol (MD = −45.00; 95% CI: −58.98, −31.02). One study compared chlorpromazine to placebo and there was a significant difference in favor of chlorpromazine (MD = −42.50; 95% CI: −51.14, −33.86). The pooled result of all of these studies was MD = −46.59; 95%CI: −54.87, −38.32.

Figure 8

Change in pain intensity (VAS) in trials comparing neuroleptics and placebo. VAS = visual analogue scale

Headache Relief (1–2 hours)

Five studies evaluated relief of headache at 1 hour (Figure 9).43,63,68,80,97 All studies reported a statistically significant result in favor of the neuroleptics; the pooled result was RR = 2.69 (95% CI: 1.66, 4.34; I2 = 76%). In two studies, the neuroleptic used was chlorpromazine.43,97 In one study43 patients were given an IV injection of 5.0 ml/kg 0.9 percent normal saline solution followed by IV chlorpromazine, 0.1 mg/kg diluted to 10 ml of 0.9 percent normal saline. In the remaining studies, patients were administered 50 mg/2ml of chlorpromazine97 or 2.75 mg droperidol.63

Figure 9 displays a forest plot of studies reporting relief of pain at one hour, neuroleptics vs. placebo. Four studies reported relief of pain at one hour. All studies showed a significant difference in favor of the neuroleptic drugs. One study compared haloperidol vs. placebo (RR = 5.33; 95% CI: 1.84, 15.49). Two studies reported chlorpromazine vs. placebo (RR = 3.66; 95% CI: 1.56, 8.57). One study reported prochlorperazine vs. placebo (RR = 1.96; 95% CI: 1.37, 2.81). One study reported droperidol vs. placebo (RR = 1.67; 95% CI: 1.21, 2.29). The pooled result of all these studies was RR = 2.69; 95%CI: 1.66, 4.34 which showed a significant difference in favor of neuroleptics.

Figure 9

Headache relief (1 hour) in trials comparing neuroleptics and placebo.

One study assessed headache response at 2 hours.63 Significantly more participants who received droperidol experienced relief of pain at 2 hours (RR = 1.51; 95% CI: 1.19, 1.92).

Pain Free (1 hour)

Four studies reported on pain free status of participants at 1 hour (Figure 10).43,63,68,97 The pooled results of two studies43,97 comparing chlorpromazine and placebo had statistically significant results favoring the neuroleptic (RR = 4.03; 95% CI: 1.02, 15.93; I2 = 78%). The different concentrations of chlorpromazine may explain some of the heterogeneity. The pooled result was statistically significant in favor of the neuroleptics (RR = 3.38; 95% CI: 1.16, 9.83; I2 = 90%).

Figure 10 displays a forest plot of studies reporting being pain free at one hour, neuroleptics vs. placebo. Four studies reported being pain free at one hour. Two studies compared chlorpromazine vs. placebo; one study was significantly in favor of chlorpromazine (RR = 7.76; 95% CI: 3.29, 18.30),. There was no difference in the other study. One study reported prochlorperazine vs. placebo and found a difference in favor of prochlorperazine (RR = 5.90; 95% CI: 2.55, 13.67). One study found significant difference in favor of droperidol over placebo (RR = 3.38; 9% CI: 1.16, 9.83).

Figure 10

Pain free (1 hour) in trials comparing neuroleptics and placebo.

Pain free status was also measured at 2 hours in one study.63 The results were statistically significant in favor of droperidol versus placebo (RR = 2.11; 95% CI: 1.37, 3.26).

Headache Recurrence (24 hours)

One study considered recurrence of pain to occur when patients stated that they were pain free any time after administration of the intervention, only to have the headache return within 24 hours.43 In this study, patients who received chlorpromazine had significantly lower rates of headache recurrence than those who were given placebo (RR = 0.28; 95% CI: 0.15, 0.55).43 Another study63 recorded the number of patients whose headache improved at 2 hours but recurred within 24 hours. There was no significant difference in the rates of headache recurrence for those who received droperidol and those who received placebo (RR = 0.69; 95% CI: 0.43, 1.12). The pooled results favor neuroleptics, however, the difference was not statistically significant (RR = 0.46; 95% CI; 0.19, 1.10; I2 = 78%).

Nausea and Vomiting

One study assessed relief of nausea and vomiting 60 minutes after the administration of prochlorperazine or placebo.91 Participants who received prochlorperazine experienced significantly greater relief than those who received placebo (RR = 5.89; 95% CI: 1.98, 17.57). One study reported the percentage of patients who experienced nausea and vomiting 2 hours post-treatment.63 The difference between the droperidol and placebo groups was not statistically significant for either nausea or vomiting (RR = 0.36; 95% CI: 0.12, 1.08 and RR = 0.33; 95% CI: 0.01, 8.03, respectively).

Patient Satisfaction

One study reported patient dis-satisfaction as the number of patients who asked for a second drug at the end of 1 hour.97 We used the inverse of this number to determine patient satisfaction. Significantly more patients who received placebo asked for more medication compared with those who received chlorpromazine (RR= 3.28; 95% CI: 1.10, 9.82).

Neuroleptic Agents Versus Active Agents
Description of Included Studies

There were 17 RCTs29,54,64,66,67,6971,7375,85,8789,96,114 that assessed the effectiveness of neuroleptics versus other active agents. The neuroleptics included prochlorperazine,29,64,6971,73,75,8789 chlorpromazine,54,67,96 haloperidol,114 droperidol,85 methotrimeptrazine,66 and olanzapine.74 One study was a three-arm trial that compared chlorpromazine, DHE, and lidocaine.96 The active comparators included anticonvulsants (sodium valproate and MgSO4),69,89 dexamethasone,114 DHE,96 neuroleptics (droperidol and prochlorperazine),29,70,74,87,88,96 NSAIDs (ketorolac and ketorolac tropethamine),54,71 opioids (meperidine),66,67,85 somatostatin analog,73 sumatriptan,64 and lidocaine.88

All studies took place in the ED with timepoints that ranged between 30 and 120 minutes. Post-ED followup ranged from 2 to 45 hours. Eight studies did not report any followup data after discharge from the ED.64,69,71,74,75,85,88,89 The number of participants who were randomized ranged from 29 to 168 (median = 64; IQR: 40, 82). The mean age of patients ranged from 27 to 35 years.

Every study but one had a pain related primary outcome. The one study measured akathisia as its primary outcome.29 While the VAS was the primary means to measure pain, one study used the Wong-Baker Faces Rating Scale to assess pain scores.54 Secondary outcomes varied across studies and included headache recurrence, patient satisfaction, nausea, and sedation.

Seven studies had low risk of bias,29,54,64,66,69,88,89 nine had an unclear risk of bias,67,70,71,7375,85,87,114 and one had a high risk of bias96 (Appendix D).

Effectiveness Results
Change in Pain Intensity (VAS)

Fourteen studies reported change in pain scores. Twelve studies specifically stated that pain was measured using the VAS (mm).54,66,67,69,71,85,89,64,7375,87 One reported using a headache scale ranging from 1 to 10,96 while another used the Wong-Baker Faces Rating Scale.54

Eight studies64,67,69,71,73,87,89,96 reported statistically significant results in favor of the neuroleptic agents (Figure 11). In four studies the neuroleptic was favored over the other active agent, although the differences were not statistically significant.54,74,75,85 In one study, the participants who received meperidine plus dimenhydrinate experienced more improvement in pain scores compared with those who received methotrimeptrazine; however, the difference was not statistically significant.66 We did not pool the results due to statistical and clinical heterogeneity.

Figure 11 displays a forest plot of studies reporting change in pain intensity using the Visual Analogue Scale, neuroleptics vs. other active agents. Fourteen studies reported pain intensity using this scale. There were four studies in which the neuroleptic was favored over the other active agent, though the differences were not statistically significant. In one study, the participants who received meperidine plus dimenhydrinate experienced more improvement in pain scores than those who received methotrimeprazine; however, the difference was not statistically significant. In one of the two studies comparing prochlorperazine with droperidol, the interventions were considered to be equally effective (MD = 0.00, 95% CI: −8.53, 8.53, I2 = 75%). However, in the other study, those administered prochlorperazine experienced significantly greater change in pain scores than those administered droperidol (MD = 18.10, 95% CI: 10.17, 26.03, I2 = 75%).

Figure 11

Change in pain (VAS) in trials comparing neuroleptics and active agents. VAS = visual analogue scale

Headache Relief

Headache relief was evaluated in four studies (Figure 12).70,73,87,96 One of the trials had three study arms in which chlorpromazine, DHE, and lidocaine were compared.96 In one study, significantly more participants in the prochlorperazine group experienced headache relief compared with those in the octerotide group (RR = 1.59; 95% CI: 1.08, 2.34).73 In the two studies comparing prochlorperazine with droperidol, one study did not report a significant difference between groups,70 while the other study showed a statistically significant difference favoring droperidol.87

Figure 12 displays a forest plot of studies reporting headache relief, neuroleptics vs. other active agents. Four studies reported headache relief. Significantly more participants in the octerotide group experienced headache relief than those in the prochlorperazine group (RR = 1.59, 95% CI: 1.08, 2.34). In the two studies comparing prochlorperazine with droperidol, one study did not report a significant difference between groups, while the other study showed a statistically significant difference favoring droperidol (RR = 0.81, 95% CI: 0.68, 0.98; I2 = 8%).

Figure 12

Headache relief in trials comparing neuroleptics and active agents.

Pain Free at 30 Minutes

Two studies reported the number of patients who were pain free 30 minutes after administration of the interventions (Figure 13).70,114 In one study, haloperidol was found to be more effective than dexamethasone,114 while in the other study more people in the droperidol group were free from pain at 30 minutes compared with those in the prochlorperazine group.70 Neither of these differences were statistically significant. At 120 minutes, haloperidol was signifantly more effective than dexamethasone (RR = 2.06; 95% CI: 1.21, 3.50) (metagraph not shown).114

Figure 13 displays a forest plot of studies reporting being pain free at 30 minutes, neuroleptics vs. other active agents. Two studies reported pain free status at 30 minutes. There was no difference at this timepoint.

Figure 13

Pain free at 30 minutes in trials comparing neuroleptics and active agents.

Headache Recurrence

Three studies assessed headache recurrence 24 hours after discharge (Figure 14).75,87,114 In the study comparing haloperidol with dexamethasone, no patients in either group reported a recurrent headache.114 There were no statistically significant differences between groups for proclorperazine versus promethazine,75 or prochlorperazine versus droperidol.87

Figure 14 displays a forest plot of studies reporting headache recurrence 24 hours after discharge, neuroleptics vs. other active agents. Three studies reported headache recurrence. There was no difference reported in any of the comparisons.

Figure 14

Headache recurrence in trials comparing neuroleptics and active agents.

Patient Satisfaction

One study measured patient satisfaction and found no difference between those who were administered prochlorperazine and those administered promethazine (RR = 1.00; 95% CI: 0.65, 1.54).75

Nausea

Three studies assessed the effect of a neuroleptic versus another active agent on nausea as a symptom of migraine (Figure 15).69,73,74 One study reported a statistically significant result in which the prochlorperazine group experienced a greater reduction in nausea than the sodium valproate group (MD = −33.5; 95% CI: −51.55, −15.45).69

Figure 15 displays a forest plot of studies reporting nausea, neuroleptics vs. other active agents. Three studies reported nausea. Only one study reported statistically significant results in which the prochlorperazine group experienced a greater reduction in nausea than the sodium valproate group (MD = −33.5, 95% CI: −51.55, −15.45).

Figure 15

Nausea in trials comparing neuroleptics and active agents.

One study reported no statistically significant difference between methotrimetprazine and meperidine plus dimenhydrinate for residual nausea and vomiting (RR = 0.80; 95% CI: 0.36, 1.80).66 In another study, resolution of nausea while in the ED was measured.75 More patients who received prochlorperazine experienced nausea resolution compared with those receiving promethazine; the results were not statistically significant (RR = 1.34; 95% CI: 0.99, 1.83).

Sedation

Three studies assessed the reduction of migraine-related sedation (Figure 16).29,69,73 One study favored octerotide over prochlorperazine (MD = 22.4; 95% CI: 3.23, 41.57).73 In another study,29 patients who received prochlorperazine experienced a significant reduction in sedation compared with those who received prochlorperazine plus diphenhydramine (MD = −21.0; 95% CI: −30.85, −11.15).

Figure 16 displays a forest plot of studies reporting sedation, neuroleptics vs. other active agents. Three studies reported sedation. One study significantly favored octreotide over prochlorperazine (MD = 22.4, 95% CI: 3.23, 41.57). In another study, those who received prochlorperazine experienced a significant reduction in sedation compared to those who received prochlorperazine plus diphenhydramine (MD = −21.0, 95% CI: −30.85, −11.15).

Figure 16

Sedation in trials comparing neuroleptics and active agents.

More Than One Dose Required

One study reported no significant difference between those receiving chlorpromazine or meperidine and dimenhydrinate when comparing the need for another dose of medication (RR = 1.18; 95% CI: 0.80, 1.74).67

Other Outcomes

In one study, patients were contacted at home 1 day after discharge to determine rates of home drowsiness and agitation.75 There was no significant difference in agitation between those who received prochlorperazine and those who received promethazine. When home drowsiness was reported, those in the prochlorperazine group experienced significantly less drowsiness.

Nonsteroidal Anti-Inflammatory Drugs

Key Points

  • More patients who received nonsteroidal anti-inflammatory drugs (NSAIDs) were pain free at 1–2 hours compared with those on placebo based on two RCTs (moderate strength of evidence).
  • There was insufficient strength of evidence for headache recurrence for patients receiving NSAIDs versus placebo based on one RCT.
  • For all head to head comparisons single trials compared NSAIDs with another active agent for change in pain (VAS), pain response, pain free at 1–2 hours, need for additional analgesia, and headache recurrence at various timepoints (insufficient strength of evidence).

Results

The results for studies that assessed nonsteroidal anti-inflammatory drugs (NSAIDs) are summarized below. Table 10, Table 11, and Table 12 present results or the strength of evidence grades for all key outcomes. See Table 13 for study and patient characteristics.

Table 10. Strength of evidence for NSAIDs versus placebo.

Table 10

Strength of evidence for NSAIDs versus placebo.

Table 11. Pain response after treatment in trials comparing NSAIDs and other active agents.

Table 11

Pain response after treatment in trials comparing NSAIDs and other active agents.

Table 12. Strength of evidence for NSAIDs versus active agents.

Table 12

Strength of evidence for NSAIDs versus active agents.

Table 13. Patient and study characteristics of trials comparing NSAIDs with placebo or active agents.

Table 13

Patient and study characteristics of trials comparing NSAIDs with placebo or active agents.

NSAIDs Versus Placebo
Description of Included Studies

Two RCTs assessed the effectiveness of NSAIDs compared with placebo in the treatment of acute migraine headaches.42,100 The NSAIDs included lysine clonixinate100 and diclofenac.42 One study100 was conducted in a headache clinic and one42 was conducted in a public health clinic. The mean age of participant groups was 32 years in one study.100 The participants were predominantly female and neither study reported the race or ethnicity of participants. Both studies reported pain relief or severity as the primary outcome at 60 to 120 minutes after administration. Post-ED followup timepoints ranged from 2 to 24 hours. The secondary outcomes included recurrence, use of rescue medication, and analgesic efficacy.

One study42 had an unclear risk of bias, and the other100 had a high risk of bias (Appendix D).

Effectiveness Results
Change in Pain Intensity (Pain Free)

The change in pain intensity was measured as pain free at 1–2 hours in two studies42,100 (Figure 17). The pooled results show that those who received NSAIDs experienced a greater decrease in pain intensity compared with those who received placebo (RR = 2.74; 95% CI: 1.26, 5.98; I2= 47%).

Figure 17 displays a forest plot of studies reporting being pain free at 1–2 hours, NSAID vs. placebo. Two studies reported pain free status. One study showed a significant difference in favor of diclofenac when compared to placebo (RR = 4.20; 95% CI: 1.70, 10.41). When pooled, the overall risk ratio was in favor of NSAIDs (RR = 2.74; 95% CI: 1.26, 5.98; I2 = 47%).

Figure 17

Pain free at 1–2 hours in trials comparing NSAIDs and placebo.

Analgesic Efficacy at 1 and 24 Hours

One study measured analgesic efficacy at 1 hour and then again at 24 hours for diclofenac versus placebo.42 The authors found that diclofenac was superior to placebo at 1 hour (RR = 3.11; 95% CI: 1.61, 6.02); however, no difference was found at 24 hours (RR = 1.14; 95% CI: 0.93, 1.39).

Headache Recurrence

One study42 reported headache recurrence, defined as return of pain within 24 hours after administration of the drug, and found that there was a statistically significant difference in favor of diclofenac (RR = 0.32; 95% CI: 0.17, 0.62).

NSAIDs Versus Active Agents
Description of Included Studies

Nine RCTs5456,71,81,84,94,101,111 assessed the effectiveness of NSAIDs versus other active agents. The NSAIDs included ketorolac,55,94,101 diclofenac81,111 and lysine acetylsalicylic acid.56 Comparators included meperidine monotherapy or in combination with other agents,55,94,101 sumatriptan,84 paracetamol,111 ergotamine,56 and tramadol.81 Two studies have been described in another section of the report (neuroleptics) and compared NSAIDs with prochlorperazine71 and chlorpromazine hydrochloride.54

All interventions were delivered in the ED, and assessments occurred between 60 and 180 minutes following administration. Followup ranged from 2 to 48 hours after patient discharge. The number of participants who were randomized ranged from 29 to 112 (median = 47; IQR = 37, 68). The mean ages of intervention groups ranged from 18 to 56 years. All studies had a pain related primary outcome. The secondary outcomes varied and included use of rescue medication, adverse effects, and assessment of clinical disability. See Table 13 for study and patient characteristics. Table 9 reports study and patient characteristics for the studies described previously.54,71

Two studies54,81 had a low risk of bias, while the remaining seven studies had an unclear risk of bias (Appendix D).

Effectiveness Results
Pain Intensity (VAS)

Five studies54,55,71,84,94 reported pain intensity using the VAS (mm) (Figure 18, Table 11). Table 8 describes the two studies that were analyzed in the neuroleptics section. All studies compared ketorolac with an active agent. One study84 showed a significant difference in favor of ketorolac compared with nasal sumatriptan (MD = −48.53; 95% CI: −65.54, −31.51). One study showed a signficant difference in favor of prochlorperazine (MD = −19.00 (95% CI: −34.97, −3.03).71 There was no difference when comparing ketorolac with meperidine plus hydroxyzine, ketorolac with meperidine plus promethazine (Figure 20), or ketorolac tropethamine with chlorpromazine hydrochloride (Table 8).

Figure 18 displays a forest plot of studies reporting pain intensity using the Visual Analogue Scale, NSAID vs. other active agents Two studies reported pain intensity. One study reported a difference in favor of ketorolac when compared to sumatriptan (MD = −48.53; 95% CI: −65.54, −31.51).

Figure 18

Pain intensity (VAS) in trials comparing NSAIDs and other active agents. VAS = visual analogue scale

Figure 20 displays a forest plot of studies reporting being pain free at 1–2 hours, NSAID vs. other active agents Three studies reported pain free status. One study favored diclofenac sodium when compared to paracetamol (RR = 5.08; 95% CI: 2.57, 10.03).

Figure 20

Pain free at 1–2 hours in trials comparing NSAIDs and other active agents.

Pain Response

Four studies reported a pain response after treatment (Figure 19).55,56,81,101 One study 56 comparing lysine acetylsalicylic acid and ergotamine significantly favored NSAIDs (RR = 1.92; 95% CI: 1.10, 3.36). There was no statistically significant difference between NSAIDs and the other three active agents.55,81,101 One study81 also reported a pain response at 48 hours and found no difference between diclafenec and tramadol (RR = 0.92; 95% CI; 0.57, 1.49).

Figure 19 displays a forest plot of studies reporting pain response after treatment, NSAID vs. other active agents. Four studies reported pain response. Only one study was significant and it favored lysine acetylsalicyclic acid when compared to ergotamine (RR = 1.92; 95% CI: 1.10, 3.36).

Figure 19

Pain response after treatment in trials comparing NSAIDs and other active agents.

Pain Free at 1–2 Hours

Three studies reported being pain free at 1–2 hours (Figure 20).81,101,111 One study111 showed a significant difference in favor of NSAIDs when comparing diclofenac sodium and paracetamol (RR = 5.08; 95% CI: 2.57, 10.03). There was no statistically significant difference in the other two studies.81,101 One study81 reported being pain free at 48 hours and found no difference between diclofenac and tramadol (RR = 1.33; 95% CI: 0.57, 3.14).

Headache Recurrence at 48 Hours

One study81 reported the recurrence of headache at 48 hours and found no difference between diclofenac and tramadol (RR = 1.50; 95% CI: 0.28, 8.04).

Additional Analgesia

One study94 reported the need for additional analgesia and found no difference between ketorolac and meperidine plus hydroxyzine (RR = 1.29; 95% CI: 0.57, 2.91).

Disability at 1 Hour

One study101 reported disability at 1 hour and found no difference between ketorolac and meperidine (RR = 0.64; 95% CI: 0.31, 1.32).

Opioids

Key Points

  • Patients who received opioids had greater improvement in pain intensity as measured by VAS (mm) compared with those receiving placebo based on three RCTs (moderate strength of evidence).
  • For all head to head comparisons, single trials compared opioids with other active agents for pain intensity, pain free, and headache recurrence (insufficient strength of evidence).

Results

The results for studies that assessed the effectiveness of opioids are summarized below. Table 14, Table 15, and Table 16 provide results or strength of evidence grades for all key outcomes. See Table 17 for details on study and patient characteristics.

Table 14. Strength of evidence for opioids versus placebo.

Table 14

Strength of evidence for opioids versus placebo.

Table 15. Opioids versus active agents in acute migraine.

Table 15

Opioids versus active agents in acute migraine.

Table 16. Strength of evidence for opioids versus active agents.

Table 16

Strength of evidence for opioids versus active agents.

Table 17. Patient and study characteristics of trials comparing opioids with placebo or active agents.

Table 17

Patient and study characteristics of trials comparing opioids with placebo or active agents.

Opioids Versus Placebo
Description of Included Studies

Three RCTs assessed the effectiveness of opioids versus placebo in patients with acute migraine headache.77,98,113 One study was a four-arm trial that compared nalbuphine monotherapy, nalbuphine plus hydroxyzine, hydroxyzine monotherapy, and placebo. The opioids included pethidine,113 nalbuphine,98 nalbuphine plus hydroxyzine,98 and tramadol.77 All studies were performed in the ED. The mean age of patient groups ranged from 37 to 40 years. The participants were predominantly female. None of the studies reported the race or ethnicity of participants. All studies reported pain relief or severity as the primary outcome at a range from 45 to 60 minutes after administration of the drugs. Followup occurred 4 hours to 7 days after ED discharge. Secondary outcomes included headache recurrence and adverse effects.

One study98 had a low risk of bias, one113 had an unclear risk of bias, and one77 had a high risk of bias (Appendix D).

Effectiveness Results
Change in Pain Intensity (VAS)

All three studies assessed pain intensity using the VAS (mm) (Figure 21). Pooled results demonstrated that opioids significantly decreased pain intensity compared with placebo (MD = −16.73; 95% CI: −24.12, −9.33; I2 = 0%).

Figure 21 displays a forest plot of studies reporting change in pain intensity using the Visual Analogue Scale, opioids vs. placebo. Three studies reported pain intensity. The two studies that significantly favored opioids were using pethidine and nalbuphine. The overall pooled estimate favored opioids when compared to placebo (MD = −16.73; 95% CI: −24.12, −9.33; I2 = 0%).

Figure 21

Pain intensity (VAS) in trials comparing opioids and placebo. VAS = visual analogue scale

Pain Free Response

One study77 reported “pain free after treatment” and found no significant difference between tramadol and placebo (RR = 2.50; 95% CI: 0.56, 11.16).

Opioids Versus Active Agents
Description of Included Studies

Thirteen RCTs assessed the effectiveness of opioids versus other active agents. The opioids included meperidine,55,94,101 pethidine,113 tramadol,81 nalbuphine,98 meperidine plus dimenhydrinate,66,112 nalbuphine plus hydroxyzine,98 butorphanol,110 and morphine.115 The other active agents included nalbuphine plus hydroxyzine,98 hydroxyzine,98 meperidine plus hydroxyzine,110 methotrimeprazine,112 metoclopramide, methotrimeprazine, droperidol, chlorpromazine ketorolac, ketorolac plus promethazine, and DHE. Nine studies 53,55,66,67,81,85,94,101,113 have been described in other sections of the report (metoclopramide, neuroleptics, NSAIDs, and DHE).

All interventions took place in the ED with outcomes assessed between 30 and 120 minutes after treatment. Post-ED followup ranged from 24 hours to 7 days. The mean age of intervention groups ranged from 29 to 46 years. See the following tables for details on study and patient characteristics: Table 6 (metoclopramide), Table 9 (neuroleptics), Table 13 (NSAIDs), Table 17 (opioids), and Table 19 (DHE).

Table 19. Patient and study characteristics of trials comparing DHE and active agents.

Table 19

Patient and study characteristics of trials comparing DHE and active agents.

Four studies had low risk of bias,53,66,81,98 seven studies55,67,85,94,101,113,115 had unclear risk of bias, and two studies110,112 had a high risk of bias (Appendix D).

Effectiveness Results
Change in Pain Intensity (VAS)

The four studies98,110,112,115 that have not been reported in other sections of the report used the VAS (mm) to measure pain intensity (Figure 22). Two studies110,115 showed a significant result in favor of opioids when comparing butorphanol versus meperidine plus hydroxyzine and morphine versus dexamethasone (MD = −17.00; 95% CI: −31.41, −2.59 and MD = −8.2; 95% CI: −12.58, −3.82 respectively). There was no statistically significant difference between opioids and other active agents in the other two studies. The studies that assessed pain intensity in other sections of the report are summarized in Table 15.

Figure 22 displays a forest plot of studies reporting pain intensity using the Visual Analogue Scale, opioids vs. other active agents. Four studies reported pain intensity. One study showed a significant result in favor of opioids when comparing butorphanol versus meperidine plus hydroxyzine (MD = −17.00, 95% CI: −31.41, −2.59). Another study reported significant difference in morphine over dexamethasone (MD = 8.20; 95% CI: −12.58, −3.82). There was no difference seen between opioids and other active agents in the remaining studies.

Figure 22

Pain intensity (VAS) in trials comparing opioids and other active agents. VAS = visual analogue scale

One study measured pain intensity (VAS) at 24 hours115 and found no statistically significant difference between patients who received morphine and those who received dexamethasone (MD = 1.30, 95% CI: −2.47, 5.07).

Pain Free Response

Three studies55,81,101 reported “pain free after treatment” and found no difference between opioids and other active agents. One trial81 reported “pain free after 2 hours” and found a statistically significant difference in favor of opioids (RR = 1.29; 95% CI: 0.60, 2.77). The studies that assessed pain free status in previous sections of the report are summarized in Table 15.

Headache Recurrence

One study81 reported the recurrence of headache at 2 days following the intervention and found no difference between diclofenac and tramadol (RR = 1.50; 95% CI: 0.28, 8.04). The study that assessed headache recurrence previously in another section of the report is summarized in Table 15.

Dihydroergotamine (DHE)

Key Points

  • For all head to head comparisons, single trials compared DHE with other active agents for pain intensity, headache relief, pain response, and headache recurrence (insufficient strength of evidence).

Results

The results for studies that assessed DHE are summarized below. Table 18 provides the strength of evidence grades for all key outcomes. See Table 19 for details on study and patient characteristics.

Table 18. Strength of evidence for DHE versus active agents.

Table 18

Strength of evidence for DHE versus active agents.

DHE Versus Active Agents
Description of Included Studies

Five RCTs,53,56,61,93,96 with six comparisons, assessed the effectiveness of DHE versus other active agents. Active agents included meperidine,53 diclofenac,93 sumatriptan,61 chlorpromazine,96 lidocaine,96 and lysine acetylsalicylic acid.56 One study56 was described in a previous section (NSAIDs) of this report (Table 13).

Three studies53,56,96 were conducted in the ED, and two61,93 were conducted in clinics that managed patients with acute headaches. Assessments occurred immediately after treatment to 2 hours after treatment; followup assessments ranged from 2 to 24 hours following patient discharge. The number of participants who were randomized ranged from 34 to 310. The mean age of intervention groups ranged from 32 to 42 years. All studies had a pain related primary outcome. Secondary outcomes included adverse effects, functional impairment, recurrence, vital signs, and physician global rating.

One study53 had a low risk of bias, three had an unclear risk of bias,56,61,93 and one96 had a high risk of bias (Appendix D).

Effectiveness Results
Change in Pain Intensity (VAS)

Change in pain intensity was reported in two studies at 30 minutes53 and 60 minutes.53,93 There was no statistically significant difference between DHE and meperidine at 30 minutes,53 nor was there a difference at 60 minutes between DHE versus diclofenac or DHE versus meperidine (Figure 23).

Figure 23 displays a forest plot of studies reporting pain intensity at 60 minutes using the Visual Analogue Scale, DHE vs. other active agents. Two studies reported pain intensity. There was no difference shown for either of the studies.

Figure 23

Pain intensity (VAS) at 60 minutes in trials comparing DHE and other active agents. VAS = visual analogue scale

Headache Relief

Headache relief was reported at 1, 2, 3, 4, and 24 hours in one study.61 At both 1 and 2 hours, sumatriptan was significantly more effective than DHE (RR = 0.73; 95% CI: 0.61, 0.86 and RR = 0.86; 95% CI: 0.76, 0.96, respectively). There were no differences at the 3 and 4 hour assessments. At 24 hours, DHE was more effective than sumatriptan (RR = 1.17; 95% CI: 1.05, 1.30).

Pain Response

One study56 comparing lysine acetylsalicylic acid and ergotamine showed a statistically significant difference that favored NSAIDs (RR = 1.92; 95% CI: 1.10, 3.36) (Figure 19).

Improvement of Functional Impairment

Two studies53,61 assessed improvement of functional impairment. One study53 found that patients receiving DHE had greater functional improvement compared with patients receiving meperidine (RR = 2.27; 95% CI: 1.20, 4.29). The second study61 found that patients receiving sumatriptan had greater functional improvement compared with patients receiving DHE (RR = 0.65; 95% CI: 0.53, 0.80).

Headache Recurrence

One study reported headache recurrence and found a statistically significant difference in favor of DHE versus sumatriptan (RR = 0.39; 95% CI: 0.26, 0.59).61

Nausea and Vomiting

Two studies reported nausea, and one reported vomiting. One study61 showed a difference in favor of sumatriptan when compared with DHE for nausea (RR = 1.60; 95% CI: 1.10, 2.32). There was no statistically significant difference when comparing DHE with meperidine (RR = 0.94; 95% CI: 0.66, 1.35).53 One study61 compared DHE versus sumatriptan for emesis and found no statistically significant difference (RR = 1.38; 95% CI: 0.49, 3.88).

Triptans

Key Points

  • Patients who received sumatriptan had greater headache relief at 60 minutes compared with those receiving placebo based on four RCTs (moderate strength of evidence).
  • More patients who received sumatriptan were pain free at discharge compared with those receiving placebo based on five RCTs (moderate strength of evidence).
  • Fewer patients who received sumatriptan experienced headache recurrence compared with those receiving placebo based on four RCTs (low strength of evidence).
  • For all head to head comparisons, single trials compared sumatriptan with other active agents for change in pain (VAS), headache relief, and headache recurrence (insufficient strength of evidence).

Results

The results for studies comparing triptans and placebo and active comparators are summarized below. Table 20, Table 21, and Table 22 present results or the strength of evidence grades for all key outcomes. See Table 23 for details on study and patient characteristics.

Table 20. Strength of evidence for triptans versus placebo.

Table 20

Strength of evidence for triptans versus placebo.

Table 21. Triptans vs. other active agents.

Table 21

Triptans vs. other active agents.

Table 22. Strength of evidence for sumatriptan versus other active agents.

Table 22

Strength of evidence for sumatriptan versus other active agents.

Table 23. Patient and study characteristics of trials comparing triptans and placebo.

Table 23

Patient and study characteristics of trials comparing triptans and placebo.

Triptans Versus Placebo
Description of Included Studies

Eight RCTs (in seven publications) compared the effectiveness of triptans versus placebo in the treatment of acute migraine.5760,62,106,116 Most studies were conducted in the ED; one study was conducted in neurology departments, pain clinics, and physicians’ offices.116 One publication reported the results of two separate trials;58 in all metagraphs and analyses these individual trials are labeled as Mushet (1) and Mushet (2). All of the triptans were administered subcutaneously. Six studies evaluated sumatriptan (4–6 mg) and one evaluated almotriptan (2–10 mg).62

Most participants were female. The mean age ranged from 38 and 41 years. Two studies reported the ethnicity of participants.57,58 Six studies evaluated participants at 120 minutes, while in one study patients were assessed at discharge.106 For one study,116 we extracted data for the 60 minute timepoint. In this study patients who still had headache at 60 minutes were randomized to receive either placebo or additional medication. Followup timepoints ranged from 12 hours to 5 days; patients were not contacted following discharge in one study.60 All studies had primary outcomes that were related to pain. Secondary outcomes included nausea, vomiting, disability level, mean duration of migraine attack, headache improvement, functional disability, and headache recurrence.

All RCTs had an unclear risk of bias (Appendix D).5760,62,106,116

Effectiveness Results
Headache Relief at 60 Minutes

Five trials reported the number of patients who experienced headache relief at 60 minutes (Figure 24).58,60,62,116 In the four trials involving sumatriptan, the pooled results demonstrated that significantly more patients who received sumatriptan achieved headache relief than those who received placebo (RR = 3.03; 95% CI: 2.59, 3.54, I2 = 0%). There was no statistically significant difference between patients who received almotriptan and those who received placebo.62

Figure 24 displays a forest plot of studies reporting headache relief at 60 minutes, triptan vs. placebo. Four studies reported headache relief. In the four trials involving sumatriptan, the pooled results demonstrated that significantly more patients who received sumatriptan achieved headache relief than those who received placebo (RR = 3.03, 95% CI: 2.59, 3.54, I2= 0%). Patients who received almotriptan did not have significantly different results compared with those who received a placebo.

Figure 24

Headache relief at 60 minutes in trials comparing triptans and placebo.

Headache Relief at 120 Minutes

There were five comparisons that evaluated the number of patients who experienced headache relief at 120 minutes (Figure 25).57,58,62,106 The differences between the triptan and placebo groups were statistically significant for sumatriptan (RR = 2.61; 95% CI: 2.09, 3.26; I2 = 21%) and almotriptan (RR = 1.65; 95% CI: 1.15, 2.36).

Figure 25 displays a forest plot of studies reporting headache relief at 120 minutes, triptan vs. placebo. Five trials reported this outcome. The difference between the triptan and placebo groups were statistically significant when participants were given sumatriptan (RR = 2.61, 95% CI: 2.09, 3.26, I2 = 21%) or almotriptan (RR = 1.65, 95% CI: 1.15, 2.36).

Figure 25

Headache relief at 120 minutes in trials comparing triptans and placebo.

Headache Relief

One study measured headache relief on the VAS (mm) at 30, 60, and 120 minutes.57 A second study measured headache relief at 30 minutes. 59 Patients receiving triptans experienced more relief compared with those receiving placebo. The differences were statistically significant at all timepoints, and the differences increased at each timepoint: 30 minutes—MD = −15.45; 95% CI: −19.49, −11.41(I2 = 0%), 60 minutes—MD = −25.0; 95% CI: −29.32, −20.68, and 120 minutes—MD = −30.70; 95% CI: −35.02, −26.38.

Pain Free

Six studies measured pain free status at discharge,106 and at 30,59 60,57,60,62,116 and 120 minutes.57,60,62 In the studies that compared sumatriptan and placebo,57,59,60,106,116 the pooled results showed a statistically significant difference in favor of sumatriptan (RR = 4.73; 95% CI: 3.77, 5.94, I2 = 0%). In the study comparing almotriptan with placebo, there was no statistically significant difference between groups. (Figure 26).

Figure 26 displays a forest plot of studies reporting being pain free at discharge, triptan vs. placebo. Five studies reported pain free status. In the studies that compared sumatriptan versus placebo, the pooled results showed a significant difference between groups (RR = 4.73: 95% CI: 3.77, 5.94, I2 = 0%).

Figure 26

Pain free status in trials comparing triptans and placebo.

Headache Recurrence

In five comparisons, patients were contacted within 24 hours of discharge to assess recurrence of migraine headache (Figure 27).58,59,106,116 The results were inconsistent across comparisons. A subgroup analysis by study setting (i.e., ED vs. other settings116) reduced the heterogeneity. The four studies that took place in the ED showed statistically significant results in favor of sumatriptan (RR = 0.72; 95% CI: 0.57, 0.90) while the study that took place in neurology departments, pain clinics, and physicians’ offices showed a significant effect in favor of placebo (RR = 2.40; 95% CI: 1.45, 3.97).

Figure 27 displays a forest plot of studies reporting headache recurrence at 24 hours, triptan vs. placebo. Four trials reported this outcome. Two studies showed a significant difference in favor of sumatriptan and the overall pooled estimate favored sumatriptan (RR = 0.72; 95% CI: 0.57, 0.90; I2 = 23%).

Figure 27

Headache recurrence at 24 hours in trials comparing triptans and placebo.

Functional Disability

One study measured functional disability 60 minutes after injection of sumatripan or placebo.116 Significantly more patients who received sumatriptan experienced an improvement in their ability to function compared with those who received placebo (RR = 5.11; 95% CI: 2.69, 9.70).

Nausea

Three comparisons assessed the effectiveness of sumatriptan in decreasing nausea at 60 minutes (Figure 28).58,60 The pooled results demonstrated that sumatriptan significantly decreased nausea (RR = 0.52; 95% CI: 0.45, 0.60; I2 = 0%).

Figure 28 displays a forest plot of studies reporting nausea at 60 minutes, triptan vs. placebo. Three trials reported nausea. All studies favored sumatriptan (RR = 0.52; 95% CI: 0.45, 0.60, I2 = 0%).

Figure 28

Nausea at 60 minutes in trials comparing triptans and placebo.

Vomiting

Two trials58 assessed vomiting after the administration of sumatriptan versus placebo and found no statistically significant difference between groups (RR = 0.33; 95% CI: 0.03, 3.06; I2 = 0%).

Photophobia

Three trials examined the effect of sumatriptan versus placebo on photophobia.58,60 The pooled results show a significant difference between groups in favor of sumatriptan (RR = 0.57; 95% CI: 0.52, 0.62, I2 = 0%).

Phonophobia

Three studies compared sumatriptan and placebo and found a significant difference between groups for the occurrence of phonophobia in favor of sumatriptan (RR = 0.57; 95% CI: 0.42, 0.77, I2 = 0%).58,60

Clinical Disability

Two trials compared clinical disability rates between the sumatriptan and placebo groups at 120 minutes.58 Significantly more patients in the placebo group were still experiencing clinical disability 1 hour after administration of the interventions (RR = 0.38; 95% CI: 0.25, 0.57, I2 = 0%).

Other Outcomes

One study assessed the difference in time to relief, time to discharge, and headache severity at discharge for participants receiving sumatriptan compared with placebo.106 Each outcome was statistically significant in favor of sumatriptan (MD = −23.0; 95% CI: −36.33, −9.67; MD = −36.0; 95% CI: −53.58; −18.42; MD = −0.80; 95% CI: −1.40, −0.20, respectively).

Another study compared the duration of attack (hours), and time between dosing and attack (hours) for those who were administered almotriptan versus placebo.62 For both outcomes, the differences between groups were not statistically significant.

Patient satisfaction with medication was assessed in two studies58 in which participants were asked if they would “take the injectable form of medication again”. In both studies, significantly more patients who were given sumatriptan responded with “yes, definitely” and “probably” compared with those who were given placebo (RR = 1.53; 95% CI: 1.23, 1.89, I2 = 0%).

Triptans Versus Active Agents
Description of Included Studies

Six studies compared sumatriptan with other active agents. The active agents included prochlorperazine and diphenhydramine,64 metoclopramide,82 chlorpromazine and metoclopramide,32 trimethobenzamide and diphenhydramine,79 DHE,61 and ketorolac.84 These studies are described in other sections of the report (i.e., metoclopramide, neuroleptics, NSAIDs, DHE).

The interventions took place in the ED in all but one study.61 Outcomes were assessed in the ED between 60 and 120 minutes; the post-ED followup, if applicable, occurred at 24 hours. The mean age of the participant groups ranged from 28 to 42 years. Refer to the following tables for details on study and patient characteristics: Table 6 (metoclopramide), Table 9 (neuroleptics), Table 13 (NSAIDs), Table 19 (DHE).

Two studies had low risk of bias,64,79 three had unclear risk of bias,61,82,84 and one32 had high risk of bias (Appendix D).

Effectiveness Results
Pain Intensity (VAS)

Four studies reported on this outcome (Table 21). Two studies comparing sumatriptan with antiemetics (metoclopramide and trimethobenzamide) found no statistically significant difference. One study comparing a neuroleptic agent and sumatriptan reported a statistically significant difference in favor of the neuroleptic agent. One study comparing NSAIDs and sumatriptan reported a statistically significant difference in favor of NSAIDs.

Headache Relief

Headache relief was reported at 1, 2, 3, 4, and 24 hours in one study.61 At both 1 and 2 hours, sumatriptan was more effective than DHE (RR = 0.73; 95% CI: 0.61, 0.86 and RR = 0.86; 95% CI: 0.76, 0.96, respectively). There were no differences at 3 and 4 hour assessments. At 24 hours, DHE was more effective than sumatriptan (RR = 1.17; 95% CI: 1.05, 1.30).

Headache Recurrence

One study reported headache recurrence and found a significant difference in favor of DHE versus sumatriptan (RR = 0.39; 95% CI: 0.26, 0.59).61

Magnesium Sulfate

Key Points

  • Patients who received MgSO4 had greater improvement in pain intensity as measured by the VAS (mm) compared with those receiving placebo based on three RCTs (moderate strength of evidence).
  • There was no difference in headache recurrence for patients who received MgSO4 compared with those receiving placebo based on two RCTs (low strength of evidence).
  • For head to head comparisons, single trials compared MgSO4 and other active agents for pain intensity measured by the VAS (insufficient strength of evidence).

Results

The results of the studies that assessed magnesium sulfate (MgSO4) are summarized below. Table 24, Table 25, and Table 26 provide the results and strength of evidence grades for all key outcomes. See Table 27 for details on study and patient characteristics.

Table 24. Strength of evidence for MgSO4 versus placebo.

Table 24

Strength of evidence for MgSO4 versus placebo.

Table 25. Pain response in trials comparing MgSO4 and other active agents.

Table 25

Pain response in trials comparing MgSO4 and other active agents.

Table 26. Strength of evidence for MgSO4 versus active agents.

Table 26

Strength of evidence for MgSO4 versus active agents.

Table 27. Patient and study characteristics of trials comparing MgSO4 and placebo.

Table 27

Patient and study characteristics of trials comparing MgSO4 and placebo.

Magnesium Sulfate Versus Placebo
Description of Included Studies

Four RCTs44,72,83,99 assessed the effectiveness of MgSO4 compared with placebo. Two studies44,72 were conducted in headache clinics, and two83,99 took place in the ED. The mean age of participant groups ranged from 29 to 40 years. The participants were predominantly female. One study99 reported that participants were predominantly white. All studies reported pain relief or severity as the primary outcome. Timepoints ranged from 20 to 60 minutes. Post-ED followup was 24 hours. Secondary outcomes included headache response, recurrence, use of rescue medication, and adverse effects.

Two studies83,99 had a low risk of bias and two44,72 had an unclear risk of bias (Appendix D).

Effectiveness Results
Change in Pain Intensity (VAS)

Three studies reported pain intensity using the VAS (mm) (Figure 29).44,83,99 The pooled estimate demonstrated a statistically significant difference in favor of MgSO4 (MD = −9.73; 95% CI: −16.75, −2.72; I2 = 0%).

Figure 29 displays a forest plot of studies reporting pain intensity using the Visual Analogue Scale, magnesium sulphate vs. placebo. Three studies reported pain intensity. The pooled analysis showed a statistically significant difference favoring magnesium sulphate over placebo (MD = −9.73, 95% CI: −16.75, −2.72; I2 = 0%).

Figure 29

Pain intensity (VAS) in trials comparing MgSO4 and placebo. VAS = visual analogue scale

Pain Reduction

Two studies reported pain reduction.72,99 The results were inconsistent (Figure 30).

Figure 30 displays a forest plot of studies reporting pain reduction, magnesium sulphate vs. placebo. Two studies reported pain reduction. One study did show a difference in favor of magnesium sulphate (RR = 10.33; 95% CI: 2.25, 47.53) while the other showed no difference between groups.

Figure 30

Pain reduction in trials comparing MgSO4 and placebo.

Headache Recurrence

Two studies44,83 reported headache recurrence. The pooled results showed no significant difference between MgSO4 and placebo (RR = 0.68; 95% CI: 0.29, 1.63; I2 = 78%).

Other Outcomes

One study44 assessed headache response, and use of rescue medications. The results showed significant effect in favor of MgSO4 (RR = 2.78; 95% CI: 1.42, 5.44 and RR = 0.65; 95% CI: 0.53, 0.82, respectively).

Magnesium Sulfate Versus Active Agents
Description of Included Studies

One study compared the effectiveness of MgSO4 and prochlorperazine89 and one study compared MgSO4 and metoclopramide.83 These studies are described in other sections of the report (metoclopramide; neuroleptics).

In both studies the interventions took place in the ED; outcomes were measured at 30 minutes following the intervention. One study83 also assessed participants at 24 hours post intervention. One study83 reported a mean age of 40 years. See the following tables for details on study and patient characteristics: Table 6 (metoclopramide), Table 9 (neuroleptics).

Both studies had a low risk of bias (Appendix D).

Effectiveness Results

Table 25 summarizes results for the studies that compared MgSO4 and other active agents. Two studies reported pain intensity (VAS). In one study metoclopramide was more effective than MgSO4 and the results were statistically significant. In the other study comparing a neuroleptics agent and MgSO4, the results were not statistically significant.

Antihistamines

Key Points

  • There was insufficient strength of evidence for improvement in pain intensity as measured by VAS (mm) for patients who received hydroxyzine compared with placebo based on one RCT.
Antihistamine Versus Placebo
Description of Included Studies

One RCT98 compared the effectiveness of hydroxyzine and placebo in the treatment of acute migraine headache. The study was conducted in the ED. Headache relief measured at 60 minutes was the primary outcome. Post-ED followup occurred at 7 days. No secondary outcomes were reported (Tables 28 and 29). The study had a low risk of bias (Appendix D).

Table 28. Strength of evidence for antihistamine versus placebo.

Table 28

Strength of evidence for antihistamine versus placebo.

Table 29. Patient and study characteristics of trials comparing antihistamine and placebo.

Table 29

Patient and study characteristics of trials comparing antihistamine and placebo.

Effectiveness Results
Pain Relief (VAS)

The authors found no statistically significant difference in pain relief comparing hydroxyzine with placebo (MD = 10.40; 95% CI: −7.38, 28.18).98

Active Combination Therapy Versus Active Therapy

Key Points

  • For all head to head comparisons single trials compared different combination interventions with other active agents for pain relief (insufficient strength of evidence).
  • A post hoc mixed treatment analysis found that combination therapy (metoclopramide plus DHE and prochlorperazine plus DHE) and neuroleptic monotherapy were most effective for pain relief (VAS) (low strength of evidence).

Description of Included Studies

Eight RCTs32,86,102,104,107,108,110,113 assessed the effectiveness of two active interventions versus one or more active interventions (Tables 30 and Table 31). None of the trials used the same combination of drugs. The studies were all performed in the ED. The mean age of patient groups ranged from 29 to 43 years. Five trials,32,86,108,110,113 with six separate interventions, reported pain reduction on the VAS (mm) measured between 30 and 120 minutes post-treatment. Two trials104,107 reported headache relief as a dichotomous outcome measured at 30 minutes and 4 hours. Risk of bias was unclear for five trials,86,102,107,108,113 and high for three32,104,110 (Appendix D).

Table 30. Summary of studies reporting active combination therapy versus active therapy for pain reduction (VAS).

Table 30

Summary of studies reporting active combination therapy versus active therapy for pain reduction (VAS).

Table 31. Patient and study characteristics of trials comparing active combination therapy and active therapy.

Table 31

Patient and study characteristics of trials comparing active combination therapy and active therapy.

Effectiveness Results

Three interventions102,110,113 showed a statistically significant result that favored metoclopramide plus DHE versus meperidine plus hydroxyzine, metoclopramide plus DHE versus ketorolac monotherapy, and metoclopramide plus pethidine versus pethidine monotherapy (Table 30). The strength of evidence was insufficient for all interventions because results were from single trials.

Mixed Treatment Analysis for Pain Relief (VAS)

We conducted a post hoc mixed treatment analysis of 36 studies that reported a pain score (VAS). In addition to neuroleptic agents, metoclopramide, NSAIDs, opioids, DHE, sumatriptan, and orphan agents (i.e., hydroxyzine (Atarax), lidocaine, MgSO4, sodium valproate, tramadol, and octreotide), we examined active combination therapy. The combination agents were metoclopramide plus DHE102,110 and prochlorperazine plus DHE.108 The results showed that both combination therapy and neuroleptic agents were most effective in pain relief, with a pain reduction of approximately 40 mm on the VAS (Figure 31). Metoclopramide, NSAIDs, and opioids reduced pain by approximately 24 mm. There were other, albeit less effective agents (e.g., DHE, triptans, and orphan agents) which reduced pain by approximately 12–16 mm. See Appendix F for the network diagram.

Figure 31 displays a mixed treatment analysis of studies reporting pain score (VAS). The results show that both combination therapy and neuroleptic agents were most effective in pain relief compared to placebo with a pain reduction of approximately 40 mm on the VAS. Metoclopramide, NSAIDs and opioids reduced pain by approximately 24 mm compared to placebo. Less effective agents were DHE, triptans and orphan drugs, which reduced pain by approximately 16-12 mm compared to placebo.

Figure 31

Mixed treatment analysis of studies reporting pain score (VAS). DHE = dihydroergotamine; NSAIDs = nonsteroidal anti-inflammatory drugs; PB = probability; VAS = visual analogue scale

The strength of evidence for the mixed treatment analysis was low. The overall risk of bias for these trials was assessed as moderate and the results were consistent. Since only one or two trials contributed data to some of the network nodes, we downgraded the strength of evidence to low.

Key Question 2. Effectiveness of Corticosteroids in the Prevention of Migraine Relapse

Key Points

  • Patients receiving dexamethasone plus standard abortive therapy were less likely to report recurrence of pain or headache up to 72 hours after discharge compared with placebo plus standard abortive therapy (moderate strength of evidence).

Description of Included Studies

Seven studies assessed the effectiveness of corticosteroids compared with placebo in the prevention of migraine relapse.1921,76,78,103,109 In every study, all patients were given standard abortive therapy after which they were administered either a placebo or intravenous (IV) dexamethasone prior to discharge. In the study by Fiesseler, participants were given either dexamethasone if IV access was obtained or oral prednisone if there was no IV access.103

All trials were conducted in the ED. The mean age of participant groups ranged from 32.6 to 38.0 years. The participants were predominantly female. All studies reported recurrence of headache or persistent pain free status post discharge. Three studies assessed participants at the time of discharge,19,21,78 one assessed patients at 120 minutes after administration of the intervention,20 and two studies did not assess patients in the ED.103,109 One study contacted patients at 3 and 30 days post discharge,76 and another assessed patients at 7 days after discharge.21 The post-ED followup timepoints for the remaining studies ranged from 24 to 72 hours. See Table 32 and Table 33 for strength of evidence grades and study and patient characteristics, respectively.

Table 32. Strength of evidence for corticosteroid versus placebo in prevention of headache recurrence.

Table 32

Strength of evidence for corticosteroid versus placebo in prevention of headache recurrence.

Table 33. Patient and study characteristics of trials comparing corticosteroid and placebo.

Table 33

Patient and study characteristics of trials comparing corticosteroid and placebo.

Three studies had an unclear risk of bias,21,76,78,109 and four studies19,20,78,103 had a low risk of bias (Appendix D).

Effectiveness Results

Headache Recurrence (24–72 hours)

We used the authors’ definitions of recurrence. In two studies,19,109 recurrence was classified by severity of headache pain. For these studies, we extracted data for patients who reported severe headache (defined as having provoked a repeat physician visit and precluded return to normal activity). All studies reported on recurrence of pain or headache between 24 and 72 hours after discharge from the ED (Figure 32). The pooled results were statistically significant in favor of the corticosteroids (RR = 0.68; 95% CI: 0.49, 0.96; I2 = 63%). Some of the heterogeneity resulted from the study by Baden, et al.,78 which was stopped early for benefit.

Figure 32 displays a forest plot of studies reporting recurrence of pain/headache between 24–72 hours, corticosteroids vs. placebo. The pooled results were statistically significant in favor of the corticosteroids (RR = 0.68, 95% CI: 0.49, 0.96, I2 = 63%).

Figure 32

Recurrence of pain/headache (24–72 hours) in trials comparing dexamethasone and placebo.

We conducted a post hoc subgroup analysis to investigate differences in headache recurrence based on dosage of dexamethasone. Studies that used less than 15 mg (n = 4) of dexamethasone reported a similar treatment effect (RR = 0.69; 95% CI: 0.40, 1.18; I2 = 65%) to those using 15 mg or more (RR = 0.65; 95% CI: 0.43 to 0.99; I2 = 37%). The difference between these two subgroups was not significant (χ2 = 2.01; df=1; p=0.16).

Severe Headaches (48–72 hours)

In one study, participants were contacted to determine whether the occurrence of severe headaches differed between those who received dexamethasone and those who received placebo.78 Fewer people in the dexamethasone group had severe headaches; however, the results were not statistically significant (RR = 0.39; 95% CI: 0.13, 1.13).

Recurrence of Pain (7 days)

One study looked at recurrence of pain at 7 days.21 While more individuals in the placebo group reported recurrent headache, the results were not statistically significant (RR = 0.70; 95% CI: 0.43, 1.14).

Recurrence of Pain (30 days)

One study compared headache recurrence at 30 days and found no statistically significant difference between dexamethasone and placebo (RR = 0.90; 95% CI: 0.58, 1.41).76

Key Question 3. Short-Term Adverse Effects of Parenteral Pharmacological Interventions

Key Points

  • No two studies reported the same adverse effects for the same pair of interventions. The strength of evidence is insufficient to conclude which active treatment for acute migraine results in more or less adverse effects
  • Adverse effects were examined for individual arms of the trials and rates of adverse effects reported. Strength of evidence was not graded for these comparisons.
  • All reported adverse effects were considered minor and self-limiting.
  • The risk of experiencing sedation following administration of metoclopramide and neuroleptic agents was common.
  • Short-term side effects were commonly reported for patients receiving DHE. The most common side effects were skin and local reactions, sedation, digestive problems, nausea or vomiting, and chest symptoms.
  • MgSO4 was associated with high rates of skin flushing and local reactions.
  • Adverse effects for triptans were infrequently reported; the most common adverse effect was local reactions.
  • There were few short-term side effects reported for NSAIDs and opioids.

This section addresses the short-term adverse effects of parenteral pharmacological interventions used to treat acute migraine headaches. Reporting of adverse effects was inconsistent across this body of evidence. As a result, no two studies reported the same adverse effects for the same pair of interventions. The strength of evidence is insufficient to conclude which active treatment for acute migraine results in more or less adverse effects.

As a post hoc analysis we analyzed adverse effects for individual arms of the trials. The results are presented by adverse effect categories (e.g., sedation, dizziness, vomiting). When an intervention had more than one study reporting on any adverse effect, the results were pooled using a standard inverse variance random effects meta-analysis. For this reason, the proportion calculated by simply pooling the data may not be identical to the point estimate computed from the meta-analysis.

Nausea or Vomiting

There were 26 unique studies that reported on the rates of vomiting, nausea, and emesis (Figure 33, Table 34).21,32,43,53,57,58,60,61,63,6567,75,76,79,81,91,94,98,100,101,104,106,109,110,113 When participants took the placebo, the risk of vomiting or experiencing nausea and emesis was 11 percent (95% CI: 6 to 14 percent). The risk for active agents ranged from 0 percent (95% CI: 0 to 4 percent) to 57 percent (95% CI: 41 to 72 percent).

Figure 33 displays a graph of the risk of nausea/vomiting across interventions. There were 22 different drug combinations, as well as placebo, that were associated with nausea and vomiting. The lowest risks were seen in chlorpromazine and droperidol while the highest risks were seen in prochlorperazine and promethazine.

Figure 33

Risk of nausea or vomiting reported in acute migraine trials.

Table 34. Vomiting, nausea, emesis reported in acute migraine trials.

Table 34

Vomiting, nausea, emesis reported in acute migraine trials.

Sedation or Somnolence

There were 25 studies that reported the development of sedation or somnolence including drowsiness and decreased levels of consciousness (Figure 34, Table 35).19,20,22,43,53,57,60,6568,75,79,8587,91,94,97,98,101,109,110,113 The risk of developing sedation or somnolence as a result of taking placebo was 8 percent (95% CI: 3 to 12 percent). The risk associated with active agents ranged from 3 percent (95% CI: 2 to 4 percent) to 84 percent (95% CI: 69 to 92 percent). The risk of experiencing sedation following administration of metoclopramide and prochlorperazine was common (17 percent each).

Figure 34 displays a graph of the risk of sedation across interventions. There were 18 different drug combinations, as well as placebo, that were associated with sedation. The lowest risks were seen in sumatriptan and metoclopramide plus magnesium sulphate, while the highest risks were seen in promethazine and methotrimeprazine.

Figure 34

Risk of sedation reported in migraine trials.

Table 35. Sedation/somnolence reported in acute migraine trials.

Table 35

Sedation/somnolence reported in acute migraine trials.

Dizziness

Twenty-three studies reported dizziness as an adverse effect. Included in this category is postural hypertension, syncope, relative hypotension, orthostatic hypotension, fainting, head rushes and dizzy spells (Figure 35, Table 36).19,20,22,43,53,57,58,60,6568,72,76,86,97,98,100,106,109,110,113 The risk of becoming dizzy in those who received placebo was 5 percent (95% CI: 2 to 8 percent). The risk in those who received an active agent ranged from 2 percent (95% CI: 1 to 8 percent) to 80 percent (95% CI: 63 to 91 percent).

Figure 35 displays a graph of the risk of dizziness across interventions. There were 17 different drug combinations, as well as placebo, that were associated with dizziness. The lowest risks were seen in DHE and prochlorperazine, while the highest risks were seen in methotrimeprazine and chlorpromazine.

Figure 35

Risk of dizziness reported in acute migraine trials.

Table 36. Dizziness reported in acute migraine trials.

Table 36

Dizziness reported in acute migraine trials.

Local Reaction

There were 14 studies that measured local reactions including pain or swelling at the injection site and IV site irritation (Figure 36, Table 37).21,53,57,58,60,61,67,76,86,89,98,100,109 The risk in those who received placebo was 19 percent (95% CI: 13 to 24 percent). For those who were administered active agents, the risk ranged from 3 percent (95% CI: 0 to 6 percent) to 43 percent (95% CI: 16 to 75 percent).

Figure 36 displays a graph of the risk of local reaction across interventions. There were 9 different drug combinations, as well as placebo, that were associated with local reaction. The lowest risks were seen in droperidol and dexamethasone, while the highest risks were seen in sumatriptan and magnesium sulphate.

Figure 36

Risk of local reaction reported in acute migraine trials.

Table 37. Local reaction reported in acute migraine trials.

Table 37

Local reaction reported in acute migraine trials.

Skin Reactions

Ten studies measured skin reactions to the interventions administered (Figure 37, Table 38).32,57,58,60,72,83,86,110,113 Included in this category was skin flushing or rash. The risk in those who received placebo was 3 percent (95% CI: 1 to 6 percent). For those who were administered active agents, the risk ranged from 2 percent (95% CI: 1 to 8 percent) to 48 percent (95% CI: 28 to 68 percent).

Figure 37 displays a graph of the risk of skin flushing across interventions. There were 8 different drug combinations, as well as placebo, that were associated with skin flushing. The lowest risks were seen in droperidol and metoclopramide plus pethidine while the highest risks were seen in DHE and metoclopramide plus magnesium sulphate.

Figure 37

Risk of skin reaction reported in acute migraine trials.

Table 38. Skin reaction reported in acute migraine trials.

Table 38

Skin reaction reported in acute migraine trials.

Extrapyramidal Symptoms

Six studies reported extrapyramidal symptoms as a result of treatment.58,66,79,82,83,87 Included in this category are dystonic reactions, stiff neck, abnormal movements, and muscle twitching. The symptoms varied across studies and included muscle cramps,58 dystonia,66,87 muscle twitching,66 stiffness or abnormal movements,82 and stiff neck.79 Results for akathsia are presented under KQ 4. See Table 39 for a summary of the results.

Table 39. Extrapyramidal symptoms reported in acute migraine trials.

Table 39

Extrapyramidal symptoms reported in acute migraine trials.

Chest Symptoms

Five studies assessed chest symptoms, which included palpitations, arrhythmia, and irregular heartbeat.32,57,58,61,106 See Table 40 for a summary of results.

Table 40. Chest symptoms reported in migraine trials.

Table 40

Chest symptoms reported in migraine trials.

Anxiety

Five studies reported anxiety and related adverse effects, including mood change, moodiness, agitation, and insomnia.57,60,75,76,78 See Table 41 for a summary of results.

Table 41. Anxiety reported in acute migraine trials.

Table 41

Anxiety reported in acute migraine trials.

Digestion Issues

Two studies assessed digestion issues that were attributed to the interventions.20,53 Included in this category were any reports on dyspepsia, heartburn, epigastric discomfort, and diarrhea. See Table 42 for a summary of results.

Table 42. Digestion issues reported in acute migraine trials.

Table 42

Digestion issues reported in acute migraine trials.

Emergence Reactions

Two studies reported emergence reactions that resulted from the administration of the interventions.32,58 Included in this category were unpleasant dreams and nightmares. See Table 43 for a summary of the results.

Table 43. Emergence reactions reported in acute migraine trials.

Table 43

Emergence reactions reported in acute migraine trials.

Key Question 4. Development of Akathisia

Key Points

  • No conclusions can be drawn regarding the development of akathisia when an anticholinergic is added to metoclopramide or phenothiazines (insufficient strength of evidence).
  • Based on a mixed treatment analysis, there is no statistically significant difference in the development of akathisia between neuroleptics and metoclopramide.

This section addresses the development akathisia following the administration of phenothiazines plus anticholinergic agents compared with metoclopramide plus anticholinergic agents. Different drugs are used to combat akathisia. While most are anticholinergics, some have antihistamine and anticholinergic properties. These agents have been classified as anticholinergics in this report.

One study22 examined the differences in the development of akathisia when metoclopramide or phenothiazines were used with anticholinergic agents (Table 6). In this study, participants were administered either prochlorperazine or metoclopramide, both accompanied by 25 mg of IV diphenhydramine.22 76 70 The difference in rates of akathisia between the two groups was not statistically significant (OR = 1.50; 95% CI: 0.24, 9.52) (Table 44).

Table 44. Strength of evidence for the development of akathisia when anticholinergic agents are added to metoclopramide or phenothiazines.

Table 44

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

In another study, participants were administered prochlorperazine plus diphenhydramine or prochlorperazine alone (Table 9).29 There was no statistically significant difference between groups in the development of akathisia symptoms (OR = 0.46; 95% CI: 0.17, 1.28) (Table 44).

We conducted a post hoc mixed treatment analysis of 15 studies that reported akathisia as an adverse event. In addition to neuroleptics and metoclopramide, other interventions included opioids, sumatriptan, and orphan agents (i.e., hydroxyzine (Atarax), lidocaine, MgSO4, sodium valproate, tramadol, and octreotide). The results show that there is no statistically significant increase in akathisia when using agents except neuroleptic agents and metoclopramide. The results also show that there is no statistically significant difference in the risk of akathisia between neuroleptics and metoclopramide. The odds of experiencing akathisia symptoms following administration of these drugs is in the range of 10 times greater than with placebo (Figure 38). See Appendix F for the network diagram.

Figure 38 displays a mixed treatment analysis of studies that report akathisia as a side effect. There were 15 studies that reported akathisia as a side effect. There was no statistically significant difference in the risk of akathisia between neuroleptics and metoclopramide, although both were significantly greater than placebo. The odds of experiencing akathisia after taking these drugs (i.e. neuroleptics, opioids, sumatriptan etc) was in the range of 2 to 10 times greater than placebo.

Figure 38

Mixed treatment analysis of studies that reported akathisia as an adverse effect. PB = probability

Key Question 5. Effectiveness and Safety of Parenteral Pharmacological Interventions in Different Subgroups

No studies presented results for the subgroups sex, race and duration of headaches. There were some data reported for the subgroup of patients who did not respond to treatment.

The detailed summary of the non-response data are available in Appendix E. Failure to respond was either defined by the authors (often in multiple ways), or described as not reaching a pain free status during the ED visit. The most commonly reported outcome was some measure of non-response; 32 studies (43 percent) reported both non-response and pain free status. There were variable definitions of non-response found in the acute migraine literature. The cut point for the reduction in pain indicating “response” varied widely (e.g., 90 percent, 45 percent). Time to assessment for response varied (e.g., end of treatment, 30–60 minutes, and up to 6 hours). Many studies failed to report the final scores in sufficient detail to determine which patients responded.

Few studies followed their patients after discharge, so it is difficult to determine the relationship between non-response and relapse outcomes. Several studies found that patients who achieved complete relief in the ED were less likely to have recurrence of headache within 48 hours.68,117 Another study specifically reported no difference in response between men and women at 24–48 hours after ED discharge; however, this study was focused on prevention, not the acute treatment.21 After multivariate adjustment, other investigators identified the following independent predictors of poor 24-hour outcomes: severe baseline pain, baseline nausea, screening positive for depression, and longer duration of headache.118

Key Question 6. Subpopulations in Studies Assessing the Effectiveness of Corticosteroids in Prevention of Migraine Relapse

One study reported no difference in response between men and women in the prevention of relapse at 24–48 hours after ED discharge.21 No studies presented results for by race or ethnicity.

Several studies conducted an a priori subgroup analysis based on duration of headache. In the first,20 the authors compared patients who had an acute migraine lasting longer than 72 hours (n = 45) and patients with headache duration of 72 hours or less (n = 160). The primary outcome was persistent pain free (i.e., pain resolved completely by 2 hours and not recurring through 24 hours followup). For patients with longer headache duration, more patients who received dexamethasone were persistently pain free compared with those receiving placebo (OR = 4.1; 95% CI: 0.9, 18). For patients with shorter headache duration, there was no difference between the groups (OR = 1.0; 95% CI: 0.5, 2.2).

In the second study, relapse was explored using the median headache duration (24 hours) from the study sample as the cut point.21 Among patients whose headache had lasted more than 24 hours prior to ED presentation, the odds of relapse for those treated with dexamethasone was 0.3 (95% CI: 0.1, 0.8); dexamethasone did not reduce relapses among patients whose headache had lasted less than 24 hours (OR=1.7; 95% CI: 0.5, 5.8).

Finally, using a post hoc regression analysis, a third study demonstrated an association between increased headache duration and severe recurrent headache, suggesting that the risk ratio of recurrent severe headache increases by about one percent per hour of headache duration.19 Overall, all authors concluded that a dose of IV dexamethasone administered in the ED may be more effective for patients with prolonged migraine headache.

One trial conducted a subgroup analysis based on residual pain at discharge (VAS >2) compared with patients with better response to therapy (VAS ≤2).21 After adjusting for experimental treatment, only residual pain as measured by the VAS was a significant predictor of relapse. Patients with a VAS score >2 at ED discharge were at a higher risk of relapse compared with those whose pain was assessed with a VAS ≤2 at discharge (adjusted OR=2.4; 95% CI: 1.1, 5.4).