What does Botox actually do, and is it safe?

Short answer

Botox paralyzes muscles by cleaving a protein that nerves need to release acetylcholine, and it works. For dynamic facial wrinkles the effect is not controversial. For chronic migraine, the PREEMPT trials found 8.4 fewer headache days per month on Botox versus 6.6 on placebo, a real but modest 1.8-day edge. The doses used are absurdly small: a 100-unit vial holds about 0.73 nanograms of toxin, and you would need roughly 100 vials injected intravenously to kill a 70-kg adult.

Botulinum toxin is the most poisonous substance known, gram for gram. It is 15 million times more toxic than arsenic, 6 million times more toxic than cyanide, 100,000 times more toxic than sarin gas, and 200 million times more toxic than uranium. The only thing that comes close is tetanus toxin, which is half as potent and comes from a cousin bacterium in the same genus. Those clostridia know what they are doing.

And we inject it into our faces. On purpose. Roughly $13 billion a year worth of it, in one of the more remarkable markets in medicine.

It started with bad sausage

Germany, the 1820s. A physician named Justinus Kerner worked out through some old-school epidemiology that a cluster of deaths traced back to poorly made blood sausage. The Latin for sausage casing is botulus. So botulism is, literally, sausage-ism.

Kerner guessed there was a poison involved. He had no way to isolate it or explain it. But he wrote that someday this paralyzing poison might be useful for overactive muscles and nerves. He was about 150 years early.

The organism is Clostridium botulinum, a rod-shaped anaerobe that lives in soil and is essentially everywhere. You are probably touching some right now. It exists as spores, which survive boiling and are inert, like seeds. They need the right environment to sprout and produce bacteria that make toxin, and that environment has to be free of oxygen. Oxygen kills these bugs. Our world is full of it. Put the spores in an improperly canned jar, let other bacteria consume the available oxygen, and the botulinum has its chance. That is the main route to adult botulism.

Infants are a different story, which is where the honey advice comes from. The standard line is no honey before age one, because the infant gut cannot handle the spores the way an adult gut can. Adults have stronger stomach acid and a competing population of gut bacteria that infants lack. That part is real. The honey link specifically is more tenuous than people think. It was a suspected source, and the follow-up has not been especially compelling. It is perfectly reasonable to skip honey, but when a mother-in-law gives the baby half a graham cracker, nobody needs to panic. Infant botulism is rare, a couple hundred cases a year in the US, and not every cracker is loaded with spores. Formula has also caused cases in the past year, which again is a problem for infants and not for the adults who might drink the same bottle.

What the toxin actually does to you

Think of botulism as a poisoning, not an infection. There is no fever, no sepsis, no delirium. What you get is a descending symmetric flaccid paralysis. It starts at the head and works down. You are wide awake the entire time. When it reaches the nerves running to your diaphragm, you stop breathing. In the 1820s that meant suffocating while fully conscious. Today we put you on a ventilator and wait, and anyone who has had Botox knows how long that wait is. Months, sometimes, until you can breathe on your own. There is an antitoxin, and with timely treatment both infants and adults typically recover.

The mechanism explains both the duration and the limits of treatment. A nerve tells a muscle to contract by releasing acetylcholine onto a receptor on the muscle. Botulinum toxin gets inside the nerve terminal and cleaves a specific protein required for that release. The signal stops. The muscle, no longer being told to do anything, goes slack.

That cleavage is essentially irreversible. The way function returns is that the nerve grows a new terminal process out to the muscle, and nerve cells are slow-growing. That is why both the disease and the cosmetic injection last as long as they do. The antitoxin can mop up toxin still circulating in the blood before it reaches nerves, so it limits further damage. It cannot undo the paralysis already in place.

The potency is why this keeps coming up as a weapon. Back-of-the-envelope, four pounds of the stuff could kill everyone on earth if you could aerosolize it and get people to inhale it. In practice that seems hard. The Aum Shinrikyo cult tried to weaponize it and failed, then switched to sarin, which is vastly less toxic but apparently easier to deploy. Iraq produced about 19,000 liters of concentrated toxin sized for bombs and missiles. There is no evidence it was used.

The dose makes the poison, and these doses are tiny

A standard 100-unit vial of Botox contains about 0.73 nanograms of toxin. A nanogram is a billionth of a gram. A lethal intravenous dose runs around 1 nanogram per kilogram, so killing a 70-kg adult by IV would take roughly 100 vials pushed straight into a vein. Treating the lines on your forehead takes about 20 units, injected into muscle, not vein. It is a vanishing fraction of a lethal dose.

The units themselves have a grim origin. One unit is not a weight or a volume. It is the amount of toxin required to kill a mouse when injected into its belly. So 20 units in your forehead is, by definition, enough to kill 20 mice. We do not treat mice well.

Everything that makes the toxin dangerous is what makes it useful. It stays roughly where you put it. It is not reversible. It paralyzes a muscle completely. And it wears off in about three months. That combination gives you a temporary on-off switch for a single small muscle, which is exactly what Kerner imagined in 1820.

The modern version arrived in the 1970s with an ophthalmologist named Alan Scott, who wanted a nonsurgical fix for strabismus. If one muscle is pulling the eye out of alignment, paralyze it. He tested purified botulinum toxin, showed it worked, and sold the technology to Allergan for reportedly very little money. Sorry, Dr. Scott. FDA approval came in 1989 for strabismus and blepharospasm, the involuntary eyelid twitch. Both ophthalmologic, both sensible.

Once you have an on-off switch for a muscle, the obvious question is what else fires when it shouldn't. The answer turned out to be a lot: spasticity after stroke, cerebral palsy, overactive bladder, spasmodic dysphonia (the vocal cord condition RFK Jr. has), and excessive sweating. That last one is not a muscle at all. Sweat glands are also signaled by acetylcholine, so the same mechanism applies. Injecting a patient's armpits sounds unpleasant and works well.

Migraine, depression, and the blinding problem

The biggest medical use is chronic migraine, which is not an obvious fit. Migraine is a complex neurologic syndrome with clear brain-level phenomena on functional imaging. It does not look like a muscle problem. The data are reasonable anyway.

The PREEMPT trials, funded by the manufacturer, randomized 1,384 adults with chronic migraine to Botox or placebo injections. Clinicians injected seven pre-specified sites in the head and neck, with the option to add sites based on where the patient reported pain. The primary outcome was headache days per month at 24 weeks. Botox: 8.4 fewer headache days. Placebo: 6.6 fewer. The difference is 1.8 days, statistically significant.

Two things about that. The placebo response is enormous, which is typical of trials in pain and other perception-heavy outcomes. And the blinding here is close to impossible. They injected saline in the control arm, but anyone who has had Botox can tell whether their face is moving. If anything, that pushes the true drug effect below 1.8 days.

Then again, a patient getting the injection gets the whole 8.4, because they get the placebo effect and the drug effect together, and they would not get much of a placebo effect if nobody told them it worked. From a decision standpoint, for someone whose life is organized around chronic migraine, that is not nothing. Headache clinics across the country inject Botox for this. It is common practice, if not quite universal standard of care.

Depression is the more speculative use, and the hypothesis is facial feedback: if you cannot frown, maybe you feel less sad. That sounds silly until you remember there is a respectable body of social psychology suggesting forced smiling nudges mood upward. I find it plausible for the same reason beta-blockers help performance anxiety. They do nothing in the brain, but if you cannot feel your heart pounding, the cycle breaks.

The trial data are another matter. The pooled randomized controlled trials come to something like 150 to 160 people total, with a 54% response rate versus 10% on placebo. That effect is far too large to believe, from studies that are once again very hard to blind. Larger follow-ups show much smaller effects. This needs a real trial.

Other experimental uses are out there. Cardiac surgeons have injected Botox into the epicardial fat pads to prevent postoperative atrial fibrillation, with mixed results. And there are small trials for premature ejaculation, because once a drug exists someone will ask how it can be pointed at a penis.

The face: what works, what's marketing

Botox works for wrinkles. There is no real controversy here, and you can confirm it by looking around at anyone of a certain age. The important caveat is that it works on dynamic wrinkles, the lines that appear when muscles move. Static lines from skin aging do not respond. Those are a job for fillers, which is a separate conversation.

The cosmetic discovery was, as usual, an accident. In 1992 an ophthalmologist named Jean Carruthers was treating patients for blepharospasm, and they mentioned their frown lines looked better. Her husband happened to be a dermatologist. They published the observation, and about a decade later the FDA approved it for glabellar lines, the vertical creases between the eyebrows sometimes called the angry elevens. Those two did all right for themselves.

There are now several brands, and essentially nobody uses the names. Botox is Kleenex. Dysport, Xeomin, and Daxxify all exist, with slightly different unit conventions and marketing claims. Daxxify is pitched as longer-lasting, Dysport as faster to kick in, which matters because Botox takes a few days to a week to work. The data do not really support the differentiation. Tell someone a product works faster and you buy a few days of placebo effect, after which the drug kicks in anyway.

Preventative Botox, or "baby Botox," is where I get skeptical, and you only have to think about the business model for a moment. Cosmetic Botox has a natural customer window, maybe the years when someone has visible wrinkles and still cares. That is not a long runway. Convince a 25-year-old to start before anything has happened and you have expanded the market by decades. Add the claim that stopping makes things worse, and the customer is locked in. That claim appears to be made up.

The evidence always trotted out is a 2006 identical twin study from the Archives of Facial Plastic Surgery: one twin got 13 years of regular Botox, the other did not, and you can indeed pick out the treated twin from photographs. It is an n of 2. It is not randomized. Identical twins do not behave identically, and a woman committed enough to 13 years of injections is plausibly also the one using more sun protection and moisturizer. The study also tells you nothing about whether 13 years was necessary. Maybe six months before the photos would have done it. Botox works on wrinkles when you have them. I do not see an argument for starting before you do.

Off-label cosmetic uses keep expanding. Trapezius Botox, sometimes marketed as Barbie Botox, paralyzes part of the shoulder muscle that pulls up toward the neck, smoothing the sloped line some women dislike. Before-and-after photos show a real difference that I doubt anyone would notice in person. Masseter injections for jawline slimming relax the chewing muscle so the lower face looks less square, which is subtler still. There has been some evidence of thinning of the jawbone itself with this, which makes sense. Muscles that attach to bones and do real work are doing something for you. Chewing is one of those things.

The risks are mostly about technique and sourcing

Start with the generic risks of injecting anything: bruising, infection. Then add asymmetry if it is done badly. Because this stuff is so potent, a slightly misplaced injection can leave you with a drooping eyelid, and the fix is waiting three months. There is also Spock brow, where one side of the forehead stays paralyzed while the other can still lift. Injectors generally try to preserve your ability to raise your eyebrows, since expressing surprise is useful. I think the Spock look is cool. Not everyone agrees.

There is a more interesting possible downside. Part of how we read emotions in other people is by unconsciously mimicking their expressions. Freeze the face and you may interfere with that. Heavy Botox users rate other people's faces differently and have more trouble recognizing emotion, with some neuroimaging to back it up. The evidence is not strong. It is fascinating if true, and the relational flip side is worth a thought: if you are upset and your partner physically cannot produce the sympathetic face, something is lost.

The real safety issue is sourcing. There is a market in counterfeit and unlicensed product, and in people who will show up at your book club to inject it. The CDC documented 22 people across 11 states with harmful reactions, including ICU admissions, from discount-branded and home injections. This is the most toxic substance on earth. You want someone who knows what they are doing, product stored properly, and a supply chain you can account for. Hunting for the lowest price on Botox is not the move.

One practical question that comes up constantly: breastfeeding. The evidence is reassuring. Botox does not pass into breast milk at meaningful rates, so it is fine. Pregnancy is probably fine too, biologically, since botulism poisoning during pregnancy does not appear to affect the fetus. Good luck finding anyone willing to inject you. Doctors get skittish.

Bottom line

Botox is both the most toxic substance known and a genuinely useful drug, and those facts are the same fact. The doses are nanograms, the effect is local, and it wears off. For strabismus, blepharospasm, spasticity, overactive bladder, and excessive sweating, it does a specific job well. For chronic migraine the effect over placebo is modest and the blinding is shaky, but the total benefit patients experience is real. For dynamic facial wrinkles it plainly works. Preventative Botox in people without wrinkles is a marketing construct, not a finding. Get it from someone qualified, with real product, and you are dealing with a well-characterized drug rather than a poison.

Here's our discussion from the episode:

I covered this in depth on Wellness, Actually, listen below.

Wellness, Actually Podcast

"What's the deal with Botox?" — Listen to the full episode, including the week's health news and listener Q&A.

Related episodes

• Wellness, Actually Grabbag! Which wellness fads actually hold up?
• What can your poop actually tell you about your health?
• Do Cupping and Dry Needling Actually Work? What the Evidence Says

Topics

Biohacking & recovery trends  ·  Injectable & regenerative therapies