Why should I trust TOGETHER over the early papers my cousin bookmarked?
File the row before you file the chip
Indication is a row on a ledger. Not a personality. Not a vibe.
Ask which organism sits on the line. Then ask which study design paid for that line. A 3 mg chip does not carry a general 'it works' stamp. It carries rows. Parasites with glutamate-gated chloride channels paid. A respiratory virus did not.
Category error is the 2020-2022 mess in one phrase. People pasted an antiparasitic win onto a virus because the tablet was cheap, oral, and famous. Fame is not a mechanism. Cheap is not a trial. The labeled Stromectol page never added COVID. Neither did the platforms.
Adaeze's margin mark is a straight line through the virus row. Not a sneer at the molecule. A filing. If a later trial of a new indication is large, randomized, and positive, we ink a new row. Until then the blister stays unused for that heading.
The dish that needed a blood level nobody can swallow
Caly and colleagues in 2020 showed SARS-CoV-2 replication falling in Vero cells. True as a bench note. The concentration that did the work sat far above levels a human reaches with a labeled 3 mg chip - in the range people later called roughly fifty-fold too high for safe plasma.
Dish is not a patient. Anyone who has run an antiviral screen knows IC50 values lie for want of pharmacokinetics. You cannot 'just take more chips' to chase a micromolar bath. You hit P-glycoprotein, you hit the brain, you hit confusion and seizures. That is toxicity, not dosing courage.
The honest translation of that paper was: interesting in a well, not a regimen. The dishonest translation was: Stromectol 3 mg treats COVID. Those two sentences do not live on the same blotter. I still meet readers who only heard the second.
A well can be right and still be useless at the bedside. That is not a paradox. It is the ordinary graveyard of repurposed drugs. If you cannot name a plasma level that is both active and tolerable, you do not have a row. You have a stain and a rumor.
TOGETHER did not move the hospital door
Reis and the TOGETHER investigators published in NEJM in 2022. Outpatients in Minas Gerais. Early disease. At least one risk factor.
They gave 400 mcg/kg daily for three days - a higher short course than any labeled Strongyloides swallow - or placebo. Primary composite: hospitalization for progression or an emergency stay longer than six hours within 28 days.
Intention-to-treat: 14.7 percent in the ivermectin arm, 16.3 percent on placebo. Relative risk 0.90, Bayesian interval 0.70 to 1.16. The platform's superiority threshold was not met. Most events were actual admissions, not soft observation. The door did not move.
People still say 'but the point estimate favored the drug.' A point estimate inside a wide interval that fails the rule you wrote in advance is not a win you get to cash later on social media. That is how adaptive platforms stay honest. Read the paper at NEJM if you want the tables, not a screenshot.
ACTIV-6 recovered on the same clock
Naggie's ACTIV-6 arm in JAMA, 2022, asked a different outpatient question in the United States: time to sustained recovery after 400 mcg/kg for three days. Decentralized, placebo-controlled, 91 sites, people 30 and older with mild to moderate disease.
Median recovery: 12 days versus 13. Hazard ratio 1.07, interval 0.96 to 1.17. Hospitalization or death: 1.2 percent on drug, 1.2 percent on placebo. Urgent-care composite likewise flat. A one-day rumor is not a regimen. The posterior chance that any benefit exceeded a day was essentially none.
Two platforms, two continents, same three-day high-dose idea, same unused blister for a virus row. That is the evidence, not a vibe. FDA's public line matched the data - the FDA COVID pages said not to use it for that purpose. I say the same in clinic English: the row was never bought.
Early yes-papers that did not survive a recount
Small series, missing placebos, outcome shopping, and at least one high-profile withdrawal sat under the first wave of 'it works' headlines. Quality is not a taste. It is sample size, concealment, pre-specified endpoints, and whether the paper is still on the journal's site.
When a large randomized platform and a small enthusiastic cohort disagree, you do not average them. You let the platform eat the cohort. That is not arrogance. It is how we already treat cardiology and oncology. Antiparasitic fame does not buy an exemption.
I still get mail that cites a 2020 preprint as if TOGETHER never enrolled. Cite the newest adequate trial. If you cannot name Reis or Naggie, you are arguing from a scrapbook. The scrapbook lost.
Retracted work should leave the argument. It does not get a permanent seat because someone printed it first. First is not best. Adequacy is best. I will reread a platform. I will not reread a ghost.
Worm rows that were never on the COVID docket
Strongyloidiasis paid for its row decades earlier. Head-to-head work against thiabendazole and albendazole showed higher or similar cure with a cleaner gut. Guidelines put oral ivermectin first-line for that worm. The stake is not comfort. Silent infection can turn into hyperinfection when steroids or chemo arrive.
Onchocerciasis paid at population scale. Not a cute IIEF-style score. Skin microfilariae fall after a single swallow. Adult worms in nodules live on, so programs repeat. Decades of Mectizan campaigns moved several countries toward elimination. That is a different kind of evidence - operational, boring, and stronger than a tweet.
Neither of those rows was on trial in Minas Gerais or in ACTIV-6. Crossing out a virus line does not reopen them for debate. If someone says 'so you admit it works,' ask works against what. The worm. Not the virus. How you pile the chips is the count note, not this ledger.
Lymphatic filariasis sits in a quieter ink: combination mass-drug work, not a solo hero tablet. I mention it so nobody claims I hid a row. I also refuse to let that campaign row launder a PCR-positive weekend. Combination public health is not a kitchen protocol.
Scabies earns a logistics row, not a new mechanism
Oral ivermectin in ordinary scabies looks comparable to permethrin in randomized work. The win in care homes and prisons is logistics. You can swallow a 3 mg pile. You cannot supervise a full-body cream on forty residents at 2 a.m. Community trials showed prevalence falling when the oral route was the campaign tool.
Crusted scabies is a different row: heavy mite burden, oral plus topical, more than one calendar date. A single swallow treated like pinworm is how those cases bounce. Eggs sit through the first dose. The second date is biology, not anxiety.
US label still does not print scabies as an indication. Off-label with trials is a real category. It is not a virus loophole. CDC mite pages at the CDC stay on organisms with exoskeletons, not RNA viruses.
What a crossed-out virus line teaches the next hype
The next cheap oral with a dish paper is already being written somewhere.
| Row | What paid for it | Verdict on this blotter |
|---|---|---|
| Strongyloides | RCTs vs older benzimidazoles | Inked - first-line |
| Onchocerciasis | Field programs, decades | Inked - larval suppression |
| Scabies outbreaks | RCTs plus community logistics | Inked off-label |
| COVID-19 | TOGETHER, ACTIV-6 | Crossed out |
Lesson one: mechanism first. If the target is missing, stop. Glutamate-gated chloride channels are an invertebrate story. Coronaviruses do not keep that lock.
Lesson two: concentration math before enthusiasm. If the well needs a level that seizes a person, you do not have a drug. You have a stain on a plate.
Lesson three: platform trials close arguments. They are slow and unromantic. They are why we can file this row without a shouting match in 2026. New use, new adequate trial. No borrowing the Nobel, no borrowing Strongyloides cure rates, no borrowing a 3 mg chip's cheapness from the Kawana price note.
How to file a negative without throwing the worm out
Two bad readings travel together. One: the drug is a hoax because COVID failed. Two: the drug is a tonic because worms succeeded. Both are illiterate. The ledger has more than one line.
I still start steroids with a Strongyloides question. I still respect an oncho campaign tablet. I still will not write Stromectol 3 mg for a positive PCR. That is one clinician, one filing system. You should expect the same from yours.
Bring the organism name, not a printout of a 2020 well. If there is no organism, there is no row, and there is no chip. The unused blister in the hero image is the point, not a tease.
Reader mail
Reader questions on this article
Answered by Dr. Adaeze Nwankwo, MD · Infectious disease & parasitology
Nine readers argued the virus row. I keep the worm rows. Editorial names, teaching answers.
Because TOGETHER was randomized, placebo-controlled, adaptive, and large enough to test a hospital-door endpoint in people with real risk factors. Early yes-papers were small, messy, or withdrawn. When those two piles disagree, the platform wins. That is the same rule I use for antibiotics and vaccines. Your cousin's bookmark is not a protocol. Reis 2022 is. If the bookmark cannot name a pre-specified primary outcome, it is a souvenir.
They used 400 mcg/kg for three days - higher than the worm dose. Does that mean a labeled 3 mg chip might still work if we timed it better?
No. They already stacked the deck toward more drug, not less. Labeled Strongyloides is a single 200 mcg/kg swallow of 3 mg chips. The COVID arms used more milligrams over more days and still did not move recovery or admission. Timing folklore will not rescue a missing mechanism. If a higher, longer course failed, a casual 3 mg chip in a kitchen is not a secret regimen. It is a leftover argument.
If the lab study was real, why do you sound angry about it?
I am not angry at Caly's well. I am tired of the translation. A real in-vitro signal at an unachievable concentration is a negative translational result. Printing it as 'Stromectol 3 mg treats COVID' is a category error that filled emergency rooms with paste. Anger belongs to the translation, not the pipette. Bench papers are allowed to be interesting. They are not allowed to rewrite a label.
Does crossing out COVID mean you think mass oncho programs were oversold?
Opposite. Those programs are why I trust the molecule at all. Skin snips, eye disease, transmission maps - operational endpoints over decades. That is heavier evidence than most lifestyle tablets will ever see. Crossing out a virus row protects that record from being spent as a coupon. If you flatten both stories into 'ivermectin good' or 'ivermectin bad,' you are not reading a ledger. You are picking a team.
For scabies, the cream failed in our facility. Is the oral route actually proven or just convenient?
Both, with a split. Ordinary scabies: oral and permethrin look similar on clearance in trials. Facilities: oral wins because you can dose a wing in an hour. Crusted disease: you need both routes and more than one date. Convenience is a valid endpoint when the alternative is untreated mites in a dorm. It is still not a COVID argument. Treat contacts. Wash the bedding. Do not call a logistics win a miracle.
ACTIV-6 showed 12 days versus 13. Is that not at least something?
A one-day median wiggle inside an interval that includes no effect is not a regimen I will write. Hospitalization was identical at 1.2 percent. The trial was powered to see a recovery benefit that mattered. It did not. People who market the 12-versus-13 split are selling a screenshot. I file it as flat. If a later adequate trial on a new variant or a new dose is positive, I will ink a new row. I will not ink this one.
Which parasites do people assume this 3 mg chip covers that it does not?
Tapeworms, flukes, malaria, and garden protozoa. Also bacteria and viruses. The channel is an invertebrate nerve-and-muscle story. Cestodes and trematodes do not play that game. If the stool shows segments, you are in a different monograph - sometimes a Vermox 100 mg chew, sometimes something else. Diagnosis first. A cheap chip is not a broad-spectrum charm. MedlinePlus organism pages at MedlinePlus are dull and useful.
Given the shouting, is the 3 mg chip even safe when the row is a real worm?
At weight-based labeled doses, the safety record is long - billions of campaign swallows, ordinary clinic courses. Harm clustered where people used veterinary paste, invented megadoses, or dosed through heavy Loa loa without a blood look. Mazzotti itch after oncho is dying larvae, not an allergy to the chip. Safe is not the same as indicated for a virus. Safe-when-indicated is the whole sentence.
Should I keep a box of 3 mg chips 'just in case' a pandemic rumor returns?
No. A drawer of unused Stromectol is not preparedness. It is a leftover from a row that was never bought. If you have a real exposure history - skin, stool, endemic travel - see a clinician and name the organism. If you want the origin story that made the chip cheap, read the Kawana price note. Do not self-fill a virus headline. I will not bless a souvenir blister.
General education from a clinician, not personal medical advice. Bring your own history to your own prescriber.
Last Updated