Butterbur for migraine: why a once-endorsed herb lost its recommendation
Table of Contents
In 2012, a US neurology guideline gave butterbur its strongest rating for migraine prevention. Three years later, that guideline was retired. If you’ve seen the old endorsement on a supplement page, the missing date matters: the American Academy of Neurology (AAN) records its retirement as September 16, 2015, and the National Center for Complementary and Integrative Health (NCCIH) says the Academy stopped recommending butterbur because of serious safety concerns.
That isn’t the same as saying the trials never happened. It’s saying a reduction in migraine attacks, measured over a few months with a particular extract, can’t answer whether the bottle someone buys now is free of hazardous compounds or safe to take over time. And this was retirement of the complementary therapies and NSAIDs guideline, not every migraine-prevention guideline.
The part the old recommendation got right #
Butterbur (Petasites hybridus) has a long medicinal history, but the migraine claim didn’t rest on tradition alone. The 2012 AAN/American Headache Society guideline rated Petasites effective for preventing migraine - a historical Level A efficacy recommendation. That rating is still easy to find online without its retirement notice attached.
The often-cited 2004 randomized trial enrolled 245 people and tested a specific butterbur root extract against placebo for four months. In the trial’s per-protocol analysis, migraine attack frequency fell 48% from baseline in the higher-dose extract group and 26% in the placebo group. Those are changes within each group, not a claim that butterbur beat placebo by 48 percentage points. The proportion with at least a 50% improvement at four months was 68% versus 49%. A lower-dose group didn’t show a statistically significant difference from placebo.
It’s a real signal, with boundaries. Four months isn’t long-term safety follow-up; a studied extract isn’t interchangeable with butterbur leaf tea or a retail supplement bearing the plant’s name. A 2020 systematic review described the butterbur evidence as positive but limited and called for better research. That review and the guideline draw on earlier trials - they aren’t independent new confirmations. Nor does the trial tell us what to take during a migraine attack: its endpoint was prevention of future attacks.
And the older evidence base isn’t as big as a stack of citations can make it look. A reanalysis of an earlier small Petadolex trial suggested a benefit but flagged shortcomings in the original protocol and analysis. It’s a second view of an existing dataset, not another fresh trial just because it has a later paper attached. If a sales page counts guidelines, reviews and reanalyses as though each recruited a new group of people, it’s making the evidence look sturdier than it is.
The problem in the bottle #
Raw butterbur contains pyrrolizidine alkaloids, or PAs. NCCIH warns that these compounds can damage the liver and lungs and may cause cancer. Processing can remove PAs, which is why a claim of PA removal matters. But “PA-free” is a starting safety condition if use is even contemplated, not a promise of clinical safety and certainly not an endorsement of a brand.
Consider one historical snapshot, not a survey of today’s shelves: in a 2012 analysis of 21 commercial butterbur products described by NCCIH, only seven contained petasins, the compounds sought in extracts, without detectable PAs; seven had toxic PA levels. It’s a small, dated sample. It can’t tell you the contamination rate in stores now (or certify the contents of any bottle in your hand). It does show why the name of the herb on a label isn’t enough to transfer the trial result to a product.
What if the label says PA-free? NCCIH notes rare liver injuries associated with products reported to be PA-free and says longer-term safety hasn’t been established. LiverTox makes an important distinction: carefully PA-depleted extracts haven’t been definitively linked to liver injury, and residual contamination is suspected in some commercial-product reports. We don’t know that the purified extract itself caused every case. We also can’t turn that uncertainty into a safety guarantee. Those two points belong together.
So the word “free” has to do a lot of work here. It describes a claim about processing or testing; it doesn’t tell you whether the product matches the extract studied in the migraine trial, how thoroughly a particular batch was checked, or what repeated use does over a longer stretch. The FDA’s general supplement guidance says dietary supplements aren’t preapproved for safety and effectiveness the way drugs are. A PA-free claim isn’t an FDA seal of approval.
Who needs extra caution #
For pregnancy or breastfeeding, don’t treat a PA-free label as a workaround. NCCIH says products containing PAs shouldn’t be used in these periods and that little is known about the safety of PA-free butterbur then. The short studies sometimes mentioned in children aren’t permission to self-treat a child’s headaches or evidence of long-term safety.
People with liver disease, abnormal liver tests, heavy alcohol use or medicines that can affect the liver should bring the question to a clinician rather than testing a product themselves. That’s a precaution based on butterbur’s liver hazard, not a universal formal contraindication established for every PA-depleted extract. And if you’re already taking butterbur and develop unusual fatigue, nausea, pain in the upper right abdomen, dark urine or yellowing of the skin or eyes, stop the supplement and seek medical assessment promptly; jaundice or severe illness needs urgent care. The reported liver cases warrant attention even though their precise causes aren’t settled (NCCIH; LiverTox).
There are more ordinary problems, too: NCCIH lists belching, diarrhea, stomach upset, drowsiness and rash. If you react to ragweed, daisies or related plants, mention that before considering butterbur. Bring a list of prescription drugs, over-the-counter medicines and other supplements to a clinician or pharmacist; the general interaction warning doesn’t establish a specific butterbur interaction with every migraine medicine.
What to do with the old Level A claim #
I’d read it as a piece of trial history, not a shopping instruction. There was evidence of benefit with a particular preparation. But the AAN retired the complementary guideline in 2015, and the NCCIH safety account leaves product purity and long-term use unresolved. There isn’t a dose, brand or lab test in these sources that lets us responsibly wave that problem away.
If migraines are frequent or disabling, ask a qualified health professional about prevention options without dropping prescribed acute or preventive treatment. The practical conversation is about your headache pattern, current medicines, liver history and what evidence actually applies to the option you’re considering. And if a headache is sudden, exceptionally severe or the worst you’ve had, seek emergency care - that isn’t a moment to try a supplement. New or changing headache patterns deserve medical assessment as well.
The awkward answer is the useful one: butterbur’s old efficacy rating and its later safety-based retirement can both be true. A promising short trial doesn’t make an uncertain retail product a sensible self-prescribed preventive.