Parenteral therapy for migraine in the emergency department: what the 2025 update found

The American Headache Society (AHS) published its 2025 update on injectable therapy for migraine in the emergency department (ED) online on 1 December 2025, with the print issue dated January 2026 ^[1]. The update repeated the search of its 2016 assessment through 10 February 2025 and expanded it to include nerve blocks and sphenopalatine ganglion blocks, which were not among the 28 injectable medications the 2016 assessment covered ^[1][2].

What changed from the 2016 assessment?

The 2016 AHS assessment identified 68 randomized trials of 28 injectable medications, of which 19 were rated class I for low risk of bias ^[2]. The 2025 search added 26 new randomized trials covering 20 injectable treatments, and of those 26 trials, 12 were rated class I ^[1].

Against 2016, prochlorperazine moved from level B to level A and hydromorphone from level C (may avoid) to level A (must not offer) ^[1][2]. The 2025 update lists dexamethasone at level C, may offer ^[1]; the 2016 assessment had rated it level B, should offer, specifically for preventing headache recurrence ^[2].

How does the 2025 update grade each therapy?

The 2025 AHS update assigns each injectable option a recommendation level: level A is must offer or must not offer, level B should offer, level C may offer or should not offer, and level U no recommendation ^[1]:

• Level A, must offer to eligible adults without contraindications: intravenous prochlorperazine and greater occipital nerve block ^[1].
• Level B, should offer: intravenous dexketoprofen, ketorolac and metoclopramide, subcutaneous sumatriptan, supraorbital nerve block, and intravenous eptinezumab only for patients matching the clinical trial population ^[1].
• Level C, may offer: intravenous chlorpromazine, dexamethasone and valproate ^[1].
• Negative ratings: intravenous hydromorphone at level A, must not offer, and intravenous paracetamol at level C, should not offer ^[1].
• Level U, no recommendation: caffeine, granisetron, ibuprofen, ketamine, lidocaine, normal saline, propofol, sphenopalatine ganglion block, and eptinezumab in an ED-specific population ^[1].

What do the key trials show?

A 2025 Bayesian network meta-analysis in Annals of Emergency Medicine (24 trials, n=2,361) ranked intravenous or intramuscular chlorpromazine highest for adequate pain relief at 2 hours, with a surface under the cumulative ranking curve (SUCRA) value of 87.3% ^[3]. The same meta-analysis placed intravenous ibuprofen and intravenous valproate among the lowest-ranked single agents for pain relief ^[3].

In a 2017 double-blind ED trial (n=127), sustained headache relief at 48 hours occurred in 60% of patients given prochlorperazine with diphenhydramine and in 31% given hydromorphone (difference 28 percentage points, 95% CI 12 to 45; number needed to treat 4, 95% CI 2 to 9) ^[4]. In a 2020 double-dummy noninferiority trial (n=99), bilateral greater occipital nerve block with bupivacaine improved pain by 5.0 points on a 0 to 10 scale at 1 hour against 6.1 points for intravenous metoclopramide, a between-group difference of -1.1 (95% CI -2.3 to 0.1), and noninferiority was not shown ^[5]. In a 2025 randomized open-label trial from Mexico City (n=42), adding an occipital nerve block to a three-drug intravenous regimen of ketorolac, paracetamol and metoclopramide raised the share of patients with at least 50% pain reduction at 2 hours from 47.6% to 95.2% ^[6]. The AHS update, whose search ended on 10 February 2025, before the 2025 occipital-block trial was published ^[6], judged greater occipital nerve block highly likely to be effective based on multiple class I studies and graded it level A ^[1]; the 2020 trial above did not show it noninferior to intravenous metoclopramide, which the update grades level B ^[1][5].

Where is the evidence uncertain?

The 2025 network meta-analysis rated certainty of evidence mostly low or very low and could not rank relative safety with sufficient certainty ^[3]. The network ranking of chlorpromazine does not match its AHS grade of level C, may offer, which applies to the intravenous route ^[1][3]. The 2025 occipital-block trial did not blind outcome assessors and was not registered in a public trial registry ^[6].

At the point of care: what to check

• Whether the presentation meets migraine criteria, since the trials enrolled adults with diagnosed migraine ^[1].
• Whether the patient used an opioid in the previous month, the exclusion applied in the 2017 hydromorphone trial ^[4].
• Which parenteral agents are in the local emergency department pathway, since the 2025 update rates intravenous hydromorphone level A, must not offer ^[1].
• Akathisia risk with intravenous prochlorperazine, the reason the 2017 trial paired it with diphenhydramine ^[4].
• Whether contraindications are present, since the level A must-offer statement applies to adults without them ^[1].