Efficacy of ketamine in the treatment of migraines and other unspecified primary headache disorders compared to placebo and other interventions: a systematic review

Primary headache disorders, including migraine, tension-type headache and trigeminal autonomic cephalalgias, are highly prevalent and cause substantial global morbidity and lost productivity. The introduction summarises the clinical burden of migraine specifically, notes common acute treatments (NSAIDs, triptans, dopamine antagonists and others), and describes ketamine as an N‑methyl‑D‑aspartate receptor (NMDAR) antagonist with multiple routes of administration and a range of dose‑dependent adverse effects. Earlier case series and mechanistic work suggesting ketamine can attenuate central sensitisation (the “wind‑up” phenomenon) have prompted interest in its use for refractory and acute primary headaches, but evidence from randomized trials has been sparse and inconsistent. Chah and colleagues set out to systematically review randomized controlled trials (RCTs) that compared ketamine (any route) to placebo or active comparators for migraine and other primary headache disorders. The stated aim was to assess ketamine’s efficacy across outcomes such as headache frequency and pain intensity, aura severity and duration, duration of relief, need for rescue medication, subsequent healthcare utilisation, quality of life or satisfaction, and functional disability, and to evaluate safety and applicability of the evidence.

This study is a systematic review and meta‑analysis conducted according to PRISMA guidance, with the protocol registered on PROSPERO (CRD42021232591). The authors defined a PICOS question: adults with migraine or other primary headaches (Population); ketamine by any route (Intervention); placebo or active comparators (Comparison); multiple clinical outcomes including pain intensity, aura measures, need for rescue medication and satisfaction (Outcomes); and settings spanning emergency departments, outpatient clinics and community/home care (Setting). Randomized controlled trials in humans were eligible; exclusions were non‑English reports, pilot or open‑label studies, reviews, editorials and abstracts. Database searches (Cochrane Library, PubMed/Medline, Web of Science and EMBASE) were first performed on 12 March 2020 and repeated on 1 February 2021, supplemented by hand‑searching reference lists. Screening and selection of studies was undertaken by three investigators, with a fourth consulted for disagreement. Data extraction (study design, demographics, interventions and outcomes) and risk‑of‑bias assessment were performed independently by three reviewers and adjudicated by a fourth, using Cochrane Handbook methods. For quantitative synthesis the authors converted reported medians/IQRs to means/SDs where necessary (the extraction contains a truncated formula). Treatment effects were expressed as difference in means for continuous outcomes (with 95% confidence intervals) and risk ratios for dichotomous outcomes. Heterogeneity was assessed with Cochran’s Q and the I2 statistic; a random‑effects model was planned if Q test P < 0.10, otherwise a fixed‑effect model was to be used. Analyses were performed using Comprehensive Meta‑Analysis version 3. Because only a small number of trials were identified, sensitivity analyses for low risk‑of‑bias studies and funnel plots for publication bias were not possible. The review ultimately included five RCTs (three double‑blind, one single‑blind, one double‑blind crossover).

Searches identified 554 database records plus 20 additional records; after duplicate removal 398 unique records were screened, 27 full texts were assessed and five RCTs met inclusion criteria. The extracted text does not clearly report the total pooled sample size in the Results section; individual trial sample sizes ranged from 17 to 54 participants and enrolled adults aged approximately 18–65 years. Trials were conducted in the USA, the UK and Italy. Two studies enrolled patients with defined migraine diagnoses and three enrolled adults presenting to emergency departments with unspecified primary headaches. Interventions and comparators varied by trial. Intranasal ketamine was compared with intranasal midazolam in one trial and to intranasal saline plus IV metoclopramide and diphenhydramine in another; IV ketamine was compared with IV saline in one trial and with IV prochlorperazine plus diphenhydramine in another (with ondansetron given alongside ketamine in that trial); a crossover trial compared subcutaneous ketamine to subcutaneous saline. Dosing strategies differed by route and trial: reported IV doses were in the sub‑dissociative range (≤0.3 mg/kg), intranasal doses included 25 mg/ml and 50 mg/ml regimens, and the subcutaneous dose reported was 80 μg/kg. Risk of bias was judged as unclear for two trials and high for three trials. Adverse events were reported in all trials; common events in the ketamine arms included feelings of unreality, fatigue and nausea, while control arms also reported fatigue, vomiting, nausea and, in some cases, hallucinations and restlessness. Key efficacy findings were inconsistent across trials and depended on route and comparator. In migraine‑focused trials, intranasal ketamine (25 mg/ml) reduced a composite aura severity score more than intranasal midazolam (P = 0.032) but did not shorten aura duration (P = 0.20). Subcutaneous ketamine at 80 μg/kg produced greater pain relief than subcutaneous saline in one trial, with reported relief of 45–100% versus an average 18% with saline (P = 0.0001). By contrast, low‑dose IV ketamine (0.2 mg/kg) did not outperform normal saline on a 0–10 numerical rating scale at 30 minutes (P = 0.5035). Among trials of unspecified primary headaches, intranasal ketamine plus IV saline was not superior to a standard combination of intranasal/IV therapy (saline, metoclopramide, diphenhydramine) for change in 0–100 mm VAS pain (reported changes −29 mm vs −22 mm; p‑value not reported). IV ketamine (0.3 mg/kg) given with ondansetron was inferior to IV prochlorperazine plus diphenhydramine for pain reduction on a 0–100 mm VAS (43.5 mm vs 63.5 mm, P = 0.03). Across studies there was no significant difference between ketamine and control groups in the proportion needing rescue medication. Patient satisfaction scores were generally high in both groups (reported as 9/10 in one comparison), although one trial found higher follow‑up satisfaction with prochlorperazine/diphenhydramine (8.3/10) than ketamine (4.9/10). Because of clinical and methodological heterogeneity (diagnostic criteria, routes, doses, comparators and outcome measures), most subgroup comparisons relied on single trials and the overall GRADE assessment was very low due to high/unclear risk of bias, few studies per comparison and small sample sizes.

The investigators interpret the pooled evidence as inconsistent and insufficient to support ketamine as a uniformly superior treatment for migraine or other primary headaches. They note that all included trials used sub‑dissociative regimens (lower than anaesthetic doses) and that dosing, route of administration and co‑administered medications varied across studies, complicating comparisons. The discussion highlights that higher sub‑dissociative doses may be more effective but carry a greater risk of adverse effects such as tachycardia and emesis; this trade‑off is mentioned as a rationale for further dose‑finding studies. In contextualising their findings, the authors align with some prior reviews that have reported potential benefits but emphasise that limitations in trial size and heterogeneity prevent firm recommendations. They contrast the mixed trial evidence with position statements and guideline panels: the American Academy of Emergency Medicine regards sub‑dissociative ketamine as a safe and effective acute analgesic, while an American Headache Society expert panel found insufficient evidence to recommend ketamine for acute migraine in the emergency setting. The reviewers also cite retrospective series and small case reports that suggest benefit in chronic refractory cohorts and in paediatric intranasal use, noting these as hypothesis‑generating rather than confirmatory. Limitations acknowledged by the study team include the small number of eligible RCTs, variable diagnostic criteria for migraine/primary headaches, heterogeneity of interventions and outcomes, and potential confounding from differing control regimens and co‑medications. They recommend further double‑blind controlled trials with longer follow‑up, evaluation of higher sub‑dissociative doses and of different routes (particularly intranasal and subcutaneous), examination of combination regimens to mitigate side effects, and exploration of longer or continuous infusion strategies for chronic refractory patients. The authors also point out applicability constraints: trials enrolled adults aged 18–65 and tended to include more women, so generalisability beyond those populations and settings (UK, USA, Italy) is uncertain.

Chah and colleagues conclude that current RCT evidence does not demonstrate consistent superiority of ketamine over placebo or active comparators for the treatment of migraine and other primary headaches. Some individual trials found benefits for specific routes or outcomes (subcutaneous ketamine for pain relief; intranasal ketamine for aura severity), while others showed no benefit or inferiority versus standard therapies. Overall evidence quality was very low. The authors call for further prospective, double‑blind trials with larger samples, standardised diagnostic criteria, varied dosing and administration routes, longer follow‑up and attention to adverse‑effect profiles, and they identify chronic refractory migraine as an important target population for future research.