Botox, Nerve Blocks, Stimulators, and Surgery for Chronic Migraine: What a 17-Review Umbrella Study Found

A review of 17 systematic reviews on procedures for chronic migraine found Botox has the strongest evidence. Nerve blocks may help short term, while stimulators and surgery rest on weak data.

On this page · 13 sections
  1. Why a Review of Reviews
  2. What the Researchers Did
  3. OnabotulinumtoxinA (Botox)
  4. Greater Occipital Nerve Blocks
  5. Occipital Nerve Stimulation
  6. Acupuncture and Other Nerve Blocks
  7. Surgery
  8. Botox Plus a CGRP Antibody
  9. Evidence at a Glance
  10. Limits of This Review
  11. What This Means if You Have Chronic Migraine
  12. Key Takeaways
  13. References

Why a Review of Reviews

Chronic migraine means headache on at least 15 days a month for more than 3 months, with migraine features on at least 8 of those days. People living with it are often offered procedures as well as pills: injections, nerve blocks, acupuncture, implanted stimulators, and sometimes surgery. Each of these gets studied on its own, and the results are scattered across dozens of papers that use different methods and outcomes.

A team from Shandong University of Traditional Chinese Medicine set out to pull that evidence together. Their paper, published in October 2026 in Frontiers in Neurology, is an umbrella review. Instead of pooling individual trials, it gathers the existing systematic reviews on each procedure and grades how much those reviews can be trusted.

What an umbrella review does

An umbrella review sits one level above a meta-analysis. It asks two questions about each treatment: what do the published reviews say, and how solid is the evidence underneath them? The authors rated review quality with a tool called AMSTAR 2 and rated certainty of the evidence with GRADE.

What the Researchers Did

The team searched five databases for reviews published from March 2015 to July 4, 2026. They kept reviews that studied people with chronic migraine diagnosed by ICHD-3 criteria and that looked at onabotulinumtoxinA (Botox), acupuncture, peripheral nerve blocks, neuromodulation, or surgery. The protocol was registered in advance.

17
Reviews included
12 contained a meta-analysis. Sample sizes ranged from 32 to 3,565 people.
2 of 17
Rated high quality
By AMSTAR 2, 5 were moderate, 7 were low, and 3 were critically low.
202
Unique primary studies
Overlap between reviews was slight overall, but higher for the same pivotal trials.

For most procedures and outcomes, the certainty of evidence came out low or very low. The main reasons were small studies, risk of bias, wide confidence intervals, inconsistent results between studies, and patient groups or comparisons that did not match the question well.

The authors also kept different types of results apart. They did not combine controlled comparisons (procedure versus sham) with before-and-after changes, and they did not merge different outcome measures to produce one big number. This keeps weaker before-and-after data from being mixed in with results from controlled trials.

OnabotulinumtoxinA (Botox)

Botox had the largest and most established body of evidence of any procedure in the review, and the authors' evidence map placed it at moderate certainty with a consistent benefit.

  • Against sham injections, one review found a small reduction in headache days at 24 weeks (standardized effect -0.28). Another found a larger standardized effect at 12 weeks (-0.58).
  • Headache impact scores (HIT-6) improved at 24 weeks compared with sham in one review. A second review's 12-week estimate pointed the same way but was not statistically clear.
  • Side effects were reported differently across reviews. One found a higher rate of treatment-related adverse events with Botox than with sham (risk ratio 2.29), while another found no clear difference. Reported events included neck pain, upper respiratory infections, and nasopharyngitis.

Much of this evidence traces back to the same PREEMPT trials, so several reviews are measuring the same patients. The standardized effect sizes also can't be converted into an exact number of headache days saved. For a sense of what Botox looks like outside of trials, see what happened in 479 real-world patients treated with Botox.

Greater Occipital Nerve Blocks

A greater occipital nerve block is an injection of local anesthetic, sometimes with a steroid, near a nerve at the back of the head. The strongest data here came from two randomized controlled trials with 116 people in total that used weekly blocks on both sides with local anesthetic.

4.46
Fewer headache days vs sham
At 4 weeks, 95% CI 2.16 to 6.76 fewer days. Low certainty.
1.33
Points lower headache severity
At 4 weeks compared with sham, from the same 2 trials.

That headache-day difference is a mean difference between groups, so it reads in real days. The authors still rated it low certainty because it rests on just two small trials, and other reviews did not find a consistent or lasting benefit across different injection schedules. Evidence on duration of attacks and on disability scores came mostly from uncontrolled studies.

Side effects in the trials were mostly mild and short-lived: pain at the injection site, dizziness, nausea, and fainting-type reactions. No serious adverse events were reported in the randomized trials. When a steroid is added, local hair loss, lighter skin, and skin thinning at the site are possible. The authors place occipital nerve blocks as a short-term or add-on option, and the Cerebral Torque guide to migraine nerve blocks covers how they are done.

Occipital Nerve Stimulation

Occipital nerve stimulation uses a surgically implanted device that sends electrical pulses to the occipital nerves. It has mainly been studied in people whose chronic migraine has not responded to many treatments.

  • Compared with sham or subthreshold stimulation, it was linked to about 3 fewer headache days at 12 weeks (mean difference -3.06, from 2 studies) and to lower disability scores.
  • Headache severity and the 50% responder rate did not differ significantly from sham. The responder estimate was very imprecise, with a confidence interval running from 0.50 to 8.55.
  • Single-arm studies without a control group reported about 50% drops in headache days and severity, but these cannot separate the device from placebo effects and regression to the mean.

Because it involves surgery and hardware, the authors say it should be reserved for highly refractory chronic migraine after established preventives have failed or can't be used. Their review did not include non-invasive devices, which are covered in this look at whether migraine devices work.

Acupuncture and Other Nerve Blocks

Acupuncture

Across 18 studies with 1,412 people, acupuncture was linked to higher odds of a "response" (odds ratio 3.82). The problem is that "effective rate" was defined differently from study to study and does not match the standard 50% responder outcome used in migraine trials. The comparison groups and follow-up times were also unclear.

For headache frequency, the result showed extreme heterogeneity (I² of 97%), meaning the studies disagreed sharply on how big the effect was. For headache days, two studies with 389 people showed a small benefit. The authors call these findings possible benefits that should be read with caution. For a closer look at sham-controlled data in migraine without aura, see what sham-controlled acupuncture trials show.

Other peripheral nerve blocks

Reviews of nerve blocks in general reported modest drops in headache severity at 12 weeks and in headache frequency over 1 to 6 weeks. These were before-and-after comparisons with no control group, and the pain and frequency scales were not clearly reported. Side effects were mostly brief, including local pain, dizziness, neck and shoulder discomfort, and near-fainting.

Surgery

Surgery had the weakest evidence of any procedure in the review, and the details give little reason for confidence. The single review on this topic covered 17 primary studies: 14 case series and 3 randomized trials. A case series follows people after an operation and reports how they did, with no comparison group. The operations varied widely and included nerve decompression, cutting nerves, removing arteries, and nasal surgery, often combined differently for each patient, so the studies were not even testing one procedure.

14 of 17
Studies with no control group
Case series that only compared patients with themselves before and after surgery
11.6%
Reported complication rate
Ranged from 0% to 38%. Only 10 of 18 reports gave any safety data.

Why the before-and-after numbers mislead

Across the case series, average pain scores fell from 8.0 to 3.9 on a 0 to 10 scale, and attacks fell from 15.6 to 5.1 a month. The authors describe these as descriptive changes, not treatment effects. People tend to seek a procedure when their migraine is at its worst, so some improvement afterward is expected from regression to the mean alone. Operations can also produce a large placebo response, and none of these averages was adjusted for it or for different follow-up times. The studies were too different to combine, so no meta-analysis was done.

Safety data are incomplete

Only 10 of the 18 study reports gave any safety information, so the 11.6% complication rate likely misses events that were never recorded. Reported complications included numbness or altered sensation, itching, hair loss, uneven eyebrows or drooping eyelids, bleeding under the skin, nerve injury, neck stiffness or weakness, wound infection, thick scars, and wounds reopening. A nerve that has been cut or removed can't be put back, which sets surgery apart from an injection that wears off.

Where headache experts stand

The authors rated the certainty of benefit as very low and concluded that the evidence does not support a routine role for surgery in chronic migraine. That matches the American Headache Society, whose Choosing Wisely list advises against recommending surgical deactivation of migraine trigger points outside a clinical trial. If a surgeon offers an operation, ask which sham-controlled trials support that specific procedure, and consider it only within a registered trial. The recent push from plastic surgeons is covered in this breakdown of the campaign for migraine surgery.

Botox Plus a CGRP Antibody

One review looked at adding a CGRP monoclonal antibody, such as erenumab (Aimovig), fremanezumab (Ajovy), galcanezumab (Emgality), or eptinezumab (Vyepti), for people already on Botox. Headache days, disability, and response rates improved from baseline, but every study was an uncontrolled observational cohort. The authors treated this as background context and say it does not show that the combination beats either treatment alone.

They also note that CGRP-targeted drugs are now recognized as first-line preventive options, while Botox remains an established treatment for chronic migraine.

Evidence at a Glance

The authors stress that their evidence map shows how mature the evidence is for each procedure. It does not rank which procedure works best, because no head-to-head comparisons were analyzed.

Procedure Best evidence found Certainty Main risks reported Where the authors place it
OnabotulinumtoxinA (Botox) Multiple reviews of sham-controlled trials favoring Botox for headache days and impact Moderate Neck pain, infections, injection burden Established preventive for chronic migraine
Greater occipital nerve block 2 sham-controlled trials, about 4.5 fewer headache days at 4 weeks Low Mild, brief local reactions; skin and hair changes with steroids Short-term or add-on option
Occipital nerve stimulation About 3 fewer headache days vs sham at 12 weeks; responder rate unclear Low Surgery and device complications Highly refractory cases only
Acupuncture Nonstandard "effective rate" outcome; small effect on headache days Low Not pooled in this review Not enough evidence for a routine role
Other nerve blocks Before-and-after changes only, no control group Low Brief local pain, dizziness, near-fainting Not enough evidence for a routine role
Surgery Mostly case series; 3 randomized trials with mixed procedures Very low 11.6% reported complications, incomplete safety reporting No routine role; the American Headache Society advises against it outside a clinical trial

Limits of This Review

The authors list several limitations. Because they worked from published reviews, any errors or selective reporting in those reviews carry over. Many procedure and outcome pairs rested on only two studies or on uncontrolled designs. Patient details that would help predict who benefits, such as prior treatment failures or medication overuse headache, were inconsistently reported.

Several results were given as standardized effects, which can't be translated into an exact number of headache days or compared directly with thresholds for a meaningful change. The review also excluded non-invasive neuromodulation and did not assess cost, which varies a lot between health systems.

How to read "low certainty"

Low certainty does not mean a treatment fails. It means the true effect could differ substantially from the estimate, so new, better-designed trials could change the picture. Placebo effects from procedures tend to be strong, which is why sham-controlled trials carry so much weight here.

What This Means if You Have Chronic Migraine

If you are not sure whether your headache pattern counts as chronic, the chronic migraine calculator can help you check before your next appointment. Prevention choices for chronic migraine are laid out in the AAN and AHS prevention guideline, which includes CGRP-targeted drugs and Botox.

When a procedure comes up, these questions can help the conversation with your physician:

  • Is this procedure meant for long-term prevention or short-term relief?
  • What sham-controlled evidence supports it for chronic migraine?
  • What are the risks, and are any of them permanent?
  • Have I had a fair trial of established preventives first?
  • How will we measure whether it is working, and when do we stop if it isn't?

For most people, the review supports Botox as the procedure with the firmest footing, occipital nerve blocks as a short-term bridge, and implanted stimulators only after other options have been exhausted. Surgery does not have the evidence to be offered outside a clinical trial.

Key Takeaways

Key takeaways
  • This umbrella review gathered 17 systematic reviews on procedures for chronic migraine, and only 2 were rated high quality.
  • Botox had the most mature evidence, rated moderate certainty.
  • Greater occipital nerve blocks cut about 4.5 headache days versus sham at 4 weeks in two small trials, with low certainty and no proven lasting effect.
  • Occipital nerve stimulation showed about 3 fewer headache days versus sham, but it requires surgery and is meant for highly refractory cases.
  • Evidence for acupuncture and other nerve blocks was too limited to support routine use.
  • Surgery had the weakest evidence of all: mostly uncontrolled case series, very low certainty, incomplete safety reporting, and an 11.6% complication rate among the reports that tracked it.
Important Medical Disclaimer

This article is for educational purposes only and is not a substitute for professional medical advice, diagnosis, or treatment. Talk with your neurologist or headache specialist before starting, stopping, or changing any migraine treatment or procedure.

References

  1. Xia L, Yang D, Lu Y, et al. Procedural interventions for chronic migraine: an umbrella review intervention-specific evidence map. Frontiers in Neurology. 2026;17:1915292. doi:10.3389/fneur.2026.1915292
  2. Loder E, Weizenbaum E, Frishberg B, et al. Choosing wisely in headache medicine: the American Headache Society's list of five things physicians and patients should question. Headache. 2013;53(10):1651-1659. doi:10.1111/head.12233

Guided practice

Alternate nostril breathing

A 5-minute guided practice for migraine prevention, from Unraveling Migraine. Captions are built into the video.

  • Prevention, not a rescue. Keep your usual medicine for migraine attacks.
  • Daily practice is what counts. The study behind it used 3 short sessions a day for 3 months.
  • Skip it or stop if your nose is blocked or it starts to feel like a strain.
Read the full guide and the research