Once again: Ivermectin doesn’t work

February 20, 2022 • 10:00 am

The paper at issue today reports the most thorough and well controlled study of the effect of ivermectin on Covid-19 around, and it was just published in JAMA Internal Medicine. What it supports, contra the claims of Joe Rogan, Bret and Heather Weinstein, and a whole slew of Republican loons, is that the drug ivermectin—as already asserted by the FDA—neither prevents nor cures covid-19. Or rather, this study shows that once adults over 50 who get the virus and are hospitalized with comorbidities, Ivermectin doesn’t help them get better. (An earlier study published in BMC Infectious Diseases, both randomized and double-blinded, and including a placebo, showed that taking ivermectin had no significant effect on keeping people out of the hospital.)

The upshot is that every well controlled study shows that ivermectin is useless in helping you once you get the virus. Another meta-analysis of reasonably well done studies that included prevention concluded that there was no good evidence that the drug even prevented infection.  The only studies that may show value of ivermectin are those in which many participants have high loads of worms as comorbidities. In such studies (which don’t apply in the US or UK), the drug may, by helping you get rid of worms (see below), make recovery from covid more likely. But even in that case there’s no excuse not to get vaccinated.  And of course you wouldn’t take ivermectin unless there was evidence you had worms.

Ivermectin is used in humans to cure parasitic worms and head lice, but can be dangerous if taken in doses high enough to kill viruses in vitro. Nevertheless, in this age of conspiracy theories and general lunacy, even credentialed scientists like the Weinsteins have recommended ivermectin and criticized vaccines, even though it’s beyond doubt that the vaccines prevent severe illness and hospitalization. You’d have to be crazy or paranoid to pass up vaccination in favor of ivermectin But thousands do it, so there you are.

With luck, you should be able to get the new study by clicking on the screenshot below, especially if you have the legal Unpaywall app. There’s also a link to the pdf, and if all else fails, make a judicious inquiry.

First, the background: the authors summarize what’s know about Ivermectin in studies to date. As I and others have mentioned before

Although some early clinical studies suggested the potential efficacy of ivermectin in the treatment and prevention of COVID-19, these studies had methodologic weaknesses. In 2021, 2 randomized clinical trials from Colombia and Argentina found no significant effect of ivermectin on symptom resolution and hospitalization rates for patients with COVID-19. A Cochrane meta-analysis also found insufficient evidence to support the use of ivermectin for the treatment or prevention of COVID-19. [JAC: The meta-analysis is linked above.]

The new study was done in Malaysia (other studies are ongoing, I believe), and iswasrandomized with respect to patient condition and age, but there was no placebo control. That is, half the infected patients were given “standard of care” (none were vaccinated) and the other half were given “standard of care” plus Ivermectin. The outcomes were followed over time.  The result: not only did Ivermectin not work, but there was a slightly higher, though nonsignificant, risk of the ivermectin treated patients progressing to the phase that required supplemental oxygen. That is exactly the opposite of what you would expect if some patients were taking placebos, for you might expect that if there’s a placebo effect, then placebo-ingesting patients would do better than those not taking placebos (i.e., the patients in this study). But despite the absence of placebos, there was still no effect of ivermectin in this study using any measure of “disease progression”.

I’m not going to summarize the results in detail, because the authors do a good job of that themselves.

There are three summaries of the results. First, the “TL/DR” version:

Note below that the 490 patients observed were all over 50 and had documented comorbidities—factors that make them more susceptible to complications and death.  Here is the protocol (indented, bolding is mine except in headers).

The Ivermectin Treatment Efficacy in COVID-19 High-Risk Patients (I-TECH) study was an open-label randomized clinical trial conducted at 20 public hospitals and a COVID-19 quarantine center in Malaysia between May 31 and October 25, 2021. Within the first week of patients’ symptom onset, the study enrolled patients 50 years and older with laboratory-confirmed COVID-19, comorbidities, and mild to moderate disease.

Interventions  Patients were randomized in a 1:1 ratio to receive either oral ivermectin, 0.4 mg/kg body weight daily for 5 days, plus standard of care (n = 241) or standard of care alone (n = 249). The standard of care consisted of symptomatic therapy and monitoring for signs of early deterioration based on clinical findings, laboratory test results, and chest imaging.

Note again: no placebo pills were given. And here’s how they measured outcome:

Main Outcomes and Measures  The primary outcome was the proportion of patients who progressed to severe disease, defined as the hypoxic stage requiring supplemental oxygen to maintain pulse oximetry oxygen saturation of 95% or higher. Secondary outcomes of the trial included the rates of mechanical ventilation, intensive care unit admission, 28-day in-hospital mortality, and adverse events.

The results are below. Note that slightly more patients in the ivermectin-dosed group (4% more) progressed to severe disease (i.e., requiring supplemental oxygen than those in the group treated the same but without ivermectin. Incorporating still other measures of “progression to severe disease,” there was again no significant difference, although there was slightly more deaths (nonsignificantly more) in the ivermectin versus control group. Finally, the most common side effect, diarrhea, was found more often in the ivermectin versus control group,but I can’t find the statistics for that difference.

Results  Among 490 patients included in the primary analysis (mean [SD] age, 62.5 [8.7] years; 267 women [54.5%]), 52 of 241 patients (21.6%) in the ivermectin group and 43 of 249 patients (17.3%) in the control group progressed to severe disease (relative risk [RR], 1.25; 95% CI, 0.87-1.80; P = .25). For all prespecified secondary outcomes, there were no significant differences between groups. Mechanical ventilation occurred in 4 (1.7%) vs 10 (4.0%) (RR, 0.41; 95% CI, 0.13-1.30; P = .17), intensive care unit admission in 6 (2.4%) vs 8 (3.2%) (RR, 0.78; 95% CI, 0.27-2.20; P = .79), and 28-day in-hospital death in 3 (1.2%) vs 10 (4.0%) (RR, 0.31; 95% CI, 0.09-1.11; P = .09). The most common adverse event reported was diarrhea (14 [5.8%] in the ivermectin group and 4 [1.6%] in the control group).

And the upshot:

Conclusions and Relevance  In this randomized clinical trial of high-risk patients with mild to moderate COVID-19, ivermectin treatment during early illness did not prevent progression to severe disease. The study findings do not support the use of ivermectin for patients with COVID-19.

Here’s the entirety of the discussion (bolding mine); note that the authors, as is proper, point out the limitations of the work.


In this randomized clinical trial of early ivermectin treatment for adults with mild to moderate COVID-19 and comorbidities, we found no evidence that ivermectin was efficacious in reducing the risk of severe disease. Our findings are consistent with the results of the IVERCOR-COVID19 trial,17 which found that ivermectin was ineffective in reducing the risk of hospitalization.

Prior randomized clinical trials of ivermectin treatment for patients with COVID-19 and with 400 or more patients enrolled focused on outpatients.16,17 In contrast, the patients in our trial were hospitalized, which permitted the observed administration of ivermectin with a high adherence rate. Furthermore, we used clearly defined criteria for ascertaining progression to severe disease.

The pharmacokinetics of ivermectin for treating COVID-19 has been a contentious issue. The plasma inhibitory concentrations of ivermectin for SARS-CoV-2 are high; thus, establishing an effective ivermectin dose regimen without causing toxic effects in patients is difficult.27,28 The dose regimens that produced favorable results against COVID-19 ranged from a 0.2-mg/kg single dose to 0.6 mg/kg/d for 5 days2932; a concentration-dependent antiviral effect was demonstrated by Krolewiecki et al.29 Pharmacokinetic studies have suggested that a single dose of up to 120 mg of ivermectin can be safe and well tolerated.33 Considering the peak of SARS-CoV-2 viral load during the first week of illness and its prolongation in severe disease,34 our trial used an ivermectin dose of 0.4 mg/kg of body weight daily for 5 days. The notably higher incidence of AEs  [“adverse effects”] in the ivermectin group raises concerns about the use of this drug outside of trial settings and without medical supervision.


Our study has limitations. First, the open-label trial design might contribute to the underreporting of adverse events in the control group while overestimating the drug effects of ivermectin. Second, our study was not designed to assess the effects of ivermectin on mortality from COVID-19. Finally, the generalizability of our findings may be limited by the older study population, although younger and healthier individuals with low risk of severe disease are less likely to benefit from specific COVID-19 treatments.

Note as well the “limitatation” that placebos were not given. Another limitation is that this study didn’t assess the chance of getting infected in the first place when you take ivermectin, or of being hospitalized if you get infected, though other work (see above) has suggested no effect of ivermectin on either of these measures.

I was prepared to admit that my criticism of ivermectin was wrong had these properly-designed studies shown an effect, but of course because ivermectin is usually suggested as a substitute for getting vaccinated, the real thing you want to know is whether, compared to getting the jabs, you’re better off swallowing ivermectin.  Given the efficacy of the vaccines, which has now been conclusively demonstrated (and yes, the effects wane over time, so we may need a yearly booster), I was already pretty sure that those who touted ivermectin as a better substitute for vaccination were wrong.  But this study does nothing to convince me that I was wrong.

More studies will come out, and eventually we’ll have a pretty solid conclusion. And I’ll bet any reader $100 that it will show that ivermectin is no substitute for vaccination or the other new drugs that are being used to relieve symptoms and combat the virus.

The question now is whether people like Joe Rogan or, especially, Bret and Heather Weinstein will admit that ivermectin—at least in this study—has no effect. Ideally, scientists will admit when they’re wrong, and, as Richard Dawkins has emphasized, that is an admirable trait. Such admissions move science along faster than waiting for a generation wedded to an idea to die off and be replaced by those who have different ideas. Now I haven’t been wrong about ivermectin yet, but when I see a study showing it’s more efficacious than vaccines in keeping you out of the hospital, or alive, I’d like to think I’d say whose three words. Will the Weinsteins say them?

I predict no. We will see a lot of hemming and hawing frothe Quacksters, and perhaps qualifications like “well, the study showed X but didn’t show the real prediction, which is Y.” In fact, I don’t remember hearing anybody pushing quack remedies admit that they were wrong.

The reason I’m so dogged about this is because people who tout quack remedies when there are good ones can do harm. And it’s far worse if they push the quackery while wearing the mantle of science.



h/t: Alex, Leslie

34 thoughts on “Once again: Ivermectin doesn’t work

  1. You’ve got an “of of” in the first paragraph. I suspect you meant “host of”.

    Thanks for the post although it’s hardly surprising. 🙂

  2. I think you have a missing word in the second paragraph. I think it should say “…not to get vaccinated” instead of “ But even in that case there’s no excuse to get vaccinated”

    Not looking to have this comment posted.

    1. So, you want billions of people who routinely take Ivermectin to stop taking it and suffer from parasites just because you hate Trump? Quite the humanitarian there…

      1. Good god, I’ve seen a lot of idiotic comments here, but this about takes the cake. No, Neuronless One, I want people to get covid shots instead of taking ivermectin to prevent or cure covid. Can’t you even read? People who have parasites should take ivermectin if their doctor recommends it.

        I don’t insult readers often, but I have to say that you’re quite the ignoramus. You can’t even grasp the content of a simple post.

  3. I thought this Ivermectin lunacy was sorted out a few months ago. In areas where worm infestation is endemic, it is absolutely beneficial in the treatment of Covid. Worm infestation (well established for strongiloides|) reduces one’s immune response. The ‘good’ RCTs in worm infested areas showed a great clinical benefit, in ‘wormless’ areas not so much.
    I’m surprised there is still some kind of controversy.

  4. “ I’m going to take this drug, does it work for that?”
    “No, there is no indication that it works for that.”
    “My second cousin‘s dentist’s math teacher says it works, does it work now?“
    “No there is no proof that it works.”
    “How about now? Does it work?“
    “Many comprehensive studies say it does not work.“
    “How about now? Does it work?“

  5. Many of us used to have a high opinion of Weinstein and Heying. They could restore our esteem if
    they were to issue a well-reasoned statement about the negative conclusions in regard to ivermectin treatment of this JAMA study and the 2021 papers—either acknowledging the correctness of their design and conclusions, or explaining reasons to question them if they can point to any.

    1. I listened to most of their first 40 podcasts – they were pretty good. Theeeennn…. they went all screwy and I’ve avoided them ever since. That kind of stuff: anti-vax, quack cures etc are my line in the sand. It is predictable idiots like Marg Taylor Green etc boost garbage but when people who should know better do it it is upsetting.

  6. In the study, 3 of the ivermectin patients died, versus 10 of the control group. See Table 2, All-cause in-hospital mortality. But “our study was not designed to assess the effects of ivermectin on mortality”. This does not convince me that ivermectin is bad.

    1. Would you rather have ivermectin or a shot? The point is not that ivermectin is bad, the point is that it is not good at preventing or curing Covid 19. The difference is not statistically significant.

      1. I would be a bit more careful with the interpretation of statistically non-significant results. The lack of statistical significance means that the data are compatible with a lack of effect but also with the presence of effect.

        In this study the risk ratio for intervention vs control in the 28-day in-hospital mortality was 0.31, which is a rather strong effect (intervention had three times lower risk of mortality compared to control). The 95% confidence interval goes from 0.09 to 1.10, meaning that data are strobgly compatible with a lack of effect (RR=1) but also with a 10-fold decrease in mortality (RR = 0.09) in the intervention group. Thus, this study cannot exclude the existence of a strong beneficial effect of ivermectin on 28-day in-hospital mortality. Non-statistical considerations may help deciding whether this is plausible.

        In this study mortality was probably a secondary outcome. The sample size in a RCT is calculated to have the desired statistical power for detecting an, a priori quantified, desired or expected, effect in the primary outcome. Such calculations do not apply to secondary outcomes. Significant tests become irrelevant and, worse, may lead to wrong decision-making. As some statisticians say, significance tests in clinical research have probably killed people.

        Anyway, I am not convinced by ivermictin and in general I do not understand why people would prefer risking to get a disease and then cure it, rather than getting a vaccine that decreases the probability of needing a cure.

        1. Good point, Pierluigi.
          And “among the 13 deaths, severe COVID-19 pneumonia was the principal direct cause (9 deaths [69.2%]). Four patients in the control group died from nosocomial sepsis. None of the deaths were attributed to ivermectin treatment.”
          That’s 3 to 6 for “direct” covid’s death. That’s far more less significant, I presume.

          1. That would be called special pleading to make results fit your belief. The outcome was all-cause mortality— doesn’t matter that some deaths in the control group can be “explained away” as not due to Covid. Nosocomial sepsis is an obvious complication of severe Covid disease in this context. I could just as easily argue that ivermectin prevents deaths from nosocomial sepsis, since that occurred only in the controls, who didn’t get it.

            Anyway, the important thing is not to pay attention to non-significant secondary results that the trial did not have statistical power to be confident of detecting.

    1. This one happens to be one of the best studies, so I thought it was worth highlighting. I’m not expecting to convince anybody who hasn’t already been convinced, but I want the data to be known.

    2. Ivermectin is a wonderful drug. An outstanding dewormer with few side effects if taken in short courses -as it is for deworming.
      It basically eradicated river blindness (caused by the worm Onchocerca volvulus) in West Africa, whole river valleys were depopulated, because of this disease.
      As said above, the ‘good’ (eg properly randomised) RCTs in worm infested areas showed a great clinical benefit, not in the worm free or low incidence areas.
      IIRC it’s development even earned a Nobel prize.
      I would not describe it as “sheep drench” (or ‘horse paste’, for that matter), that does not do justice to this exceptional drug, which always was, and still is, used in humans.

  7. True and recent story:

    Someone I know got COVID-19; not sure if it was identified as Delta, Omicron, etc.

    This person is unvaccinated and, by his own description, had *never been sicker* in his whole life. He was unable to leave his house for two weeks and felt too weak to get out of bed for much of that time.

    I saw him after he had recovered, and he told me with a straight face:

    “It’s lucky I had the right medicine, or I would have died!”

    And what, I asked, was the “right medicine.”

    I don’t even need to answer, I’m sure, but I will: Ivermectin.

    He’s not dumb. He’s a nice guy who helps a lot of people. As far as I know, he’s not a political extremist of any kind.


      1. I know somebody who caught COVID19 without being vaccinated and only had a minor cough. They didn’t take ivermectin. Can we use them as a control and infer that ivermectin makes it worse?

        🙂 in case there’s any doubt

      2. Indeed, uncontrolled.

        Upon reading my comment, I realize that some may interpret it as an *endorsement* of Ivermectin (a problem of clarity in my writing, not their comprehension).

        So, to be clear: I’m a hard NO when it comes to Ivermectin as a preventative or treatment for COVID-19.

        I’m surely one of the few Americans with any extensive experience of this drug pre-COVID, having administered it to literally thousands of cattle and scores of horses during my years working on ranches in the 1980s. I recall so well how the animals drooled sheets of white saliva for hours and tried to get the taste of the paste out of their mouths.

        I realize there are human formulations of the drug, but my experience with the veterinary formulation made me skeptical from the get-go. The original studies suggesting its efficacy for COVID-19 were quickly debunked by the likes of Eric Topol, and I’m glad to see that further trials continue to drive a stake in the heart of that silly notion.

  8. I’m curious about Dr. John Campbell’s take on this new study.

    He’s massively popular and influential and had been a steady voice of reason and caution through the pandemic, interpreting data for the laymen viewer. Unfortunately it seems he hopped on to the Ivermectin train. Not with abandon or lack of caution, but he does seem to now be on the side of saying data suggests it works and also implying some nefarious motives in suppressing it’s use.
    It will be interesting how he digests this one.

    He’s also gone a bit too gung-ho on “natural immunity” these days with Omicron. For some reason he has now been spending more time highlighting “concerns about the vaccines” over “concerns about the dangers of being infected with Omicron.” And naturally the anti-vaxxers/vaccine-resistant are eating it up in his comment sections.

    1. The mistake generally made is to compare exposure (‘natural immunity’ -coming at a horrendous price of more than 900.000 dead in the US alone) with vaccination.
      One should compare exposure with vaccination + exposure.

  9. Thanks for this article, PCC. Of all the fake news and free speech drama these days, this was one of the most egregious because it reliably sent some proportion of believers in sham COVID medicines to their graves. Hard to get worse advice than that!

    Bret Weinstein…is “a quackster.” And he should shut his bill on everything COVID, admit his stupidity, and leave the vaccines and medical science to the experts, which he is not.

  10. TL;DR: Studies showing a drug doesn’t work have to be surprisingly large to come close to proving what is in essence a negative proposition. Few are, because large studies are expensive and funders usually launch them in hopes that a drug will prove to be effective, not to prove that it doesn’t work. All negative studies show merely that if a treatment has any effect it all, that effect is smaller than some value set by the design of the study.

    A caution to be taken in interpreting any negative study (i.e. one where no effect of the drug is demonstrated) is that the study can only detect (or rule out) an effect of the size it was powered for. A negative study fails to reject the null hypothesis of equivalence. It never accepts the null hypothesis.

    In this study, the authors chose a sample size of 490 patients to have an 80% chance of detecting a reduction due to ivermectin of 50% or greater, if it existed, in the risk that a patient admitted to the mandatory isolation hospitals in Malaysia would deteriorate and need to be given oxygen. This has two important implications:

    1) The chance of missing an effect of 50% or greater, if such exists, was 20%. (This is the type-2 error, or beta.) A 20% chance of missing an effect seems high, but this is conventional in biomedical statistics. It has been criticized as leading to abandonment of actually effective treatments (long before ivermectin came along.)
    2) A true effect smaller than 50% reduction would be even more likely to be missed.

    Minimum detectable effect sizes and type-2 errors depend heavily and strikingly on sample size, as here:


    Reducing type-2 errors and, especially, trying to pick up smaller effect sizes both expand the required sample size dramatically and expensively for sponsors funding the study (the Malaysian government in this case.)

    Using a baseline risk of 17.5% (from their paper) that a new patient would need oxygen under usual conditions, you can play around with the sample-size calculator to find that:

    1) detecting, with 80% likelihood, a 10% reduction in risk (to 15.75%) conferred by ivermectin (if it truly does) would require a sample size of 7104 patients in each arm.

    2) if you are only going to be sold on IVM if it reduces the risk of death by at least 10% (from 40 per 1000 to 36 per 1000 using the death rate in the control group from this study), you need 35,864 patients in each arm. (I can tell you that very few drugs we use in medicine, antibiotics mostly, reduce death rates in any disease by as much as 10%.)

    3) living with the total sample size of 490 patients that the Malaysian government did fund, what is the likelihood that this study would have detected a risk reduction of 10% from ivermectin, if that risk reduction truly existed? Remember it had an 80% likelihood of finding a 50% reduction, and didn’t find it. Using the sample-size calculator in reverse, you can move the “Power” slider until it gives (close to) 490 patients in the sample size. I get between 5 and 10%. In other words, if the impact of ivermectin is in truth a 10% reduction in risk of deterioration, this study had about 1 chance in 15 of finding it. (If you like, you can unhide the formal equations and solve for beta to get an exact result.)

    There is no reason carved in stone to choose one effect size over another around which to plan your study — you can specify any effect size you think is therapeutically reasonable. Just consider carefully the impact on sample size and budget. The question is usually posed as, “How much better would this new drug have to be before I would start using it in addition to everything else I do?” If it is easy to use, few side effects, and cheap, your threshold for adoption could be quite low, if the effect was real. (IVM at the doses used did cause some nasty diarrhea though, in patients who are already a bit ropy.) There are many cancer chemotherapy drugs that have an average effect size of 8% on tumour shrinkage (zero on survival time) and patients clamour to receive them.

    The IVM advocates do have a point: no trials have been anywhere near large enough to exclude the magnitude of benefit they think would be worth having, and one reason for this is that Pharma won’t spend big money on large trials of old drugs.. The other side of the coin is that there is no strong reason to think the drug really has any benefit in Covid at all, in the first place. Since almost everyone under 85 recovers from Covid no matter what they do, they will attribute their improvement to whatever they swallowed. In a rational world, no one would have started using it in people given the pre-clinical evidence…but there you are and here we are. It’s like being challenged to prove there are no elephants in the room. Just because you can’t find any doesn’t meant there aren’t any there; maybe you just haven’t looked hard enough. But eventually you have to ask your challenger, “So why are you so sure there are elephants in this room that I am conspiring not to find?”

    This high-quality study is a good one to use to think about the issues because most others are pure crap. (I don’t hold it against them that they didn’t use placebos Only a drug company can manufacture dummy pills that look enough like the real pills that the nurses can’t tell them apart.)

    1. Seconded.

      I usually teach this sort of stuff as “how narrow do you need the error bars to be?” rather than power calculations per se, as its much easier and can be more useful.

    2. Great post. I’m one of those odd people who really try to be open minded. My biggest problem with the whole issue is that Western countries came across as completely trying to shut down any significant trials of the drug even though during the SARS days there were indications both Ivermectin and HCQ might potentially have some effect and even though both drugs are relatively harmless (no risk to try, no substantial cost), and they were so small that oftentimes any result at all was nearly at the level of noise as far as being statistically significant.

      You said: “The IVM advocates do have a point: no trials have been anywhere near large enough to exclude the magnitude of benefit they think would be worth having, and one reason for this is that Pharma won’t spend big money on large trials of old drugs..”

      I would submit this isn’t exactly true – the entire world posed a “natural study” for this disease. One chart (which can be re-created by almost anyone willing to look at the world stats) appears to be a glaring set of empirical data supporting the elephants in the room theory – that shows that in the countries where parasites are perennially a problem (and thus people routinely take Ivermectin, HCQ, or both) the incidence of Covid is practically nil. I freely acknowledge that correlation is not causation, but it certainly can be “probable cause” to conduct large (not small) studies. Given that this evidence was available early on in the disease, the inexpensive cost of the drug, and the relative safety of taking it/them, I would have thought that volunteer programs involving millions of people taking the drug(s) would have been started. This would have included a large enough sample to include people both with and without comorbidity issues across all age groups, and would have addressed your concerns about small sample sizes for negative studies. There should have been similar “studies” for virtually any cheap, inexpensive treatment for which there was ANY hint of promise.

      It is also curious that this study ONLY included people with comorbidity factors already infected. What assurance is there that the types of comorbidity were equally divided among the test and control groups? It was established definitively early on that once the cytokine storm commenced, the drugs were in fact useless. Is there something about this group’s comorbidity which predisposed them to experiencing a cytokine storm?

      We can’t go back and do it over. I would say, however, that our world health agencies definitely dropped the ball early on in that they literally should have had millions of people trying almost anything before we had treatments or cures, so long as the risk were acceptably low for the trial drug.

  11. Dinna trust the word of a Campbell. Remember Glencoe, where the Campbells, quartered as an army detachment among the people of Maclain MacDonald, massacred 30 odd of their clan host.

  12. Enthusiasts of so-called alternative medicine (SCAM) seem remarkably often those who deny the evidence related to the dire consequences of a COVID-19 infection. Consequently, they also deny the value of COVID vaccinations. More info about placebo effect – net-boss.org/the-power-of-the-placebo-effect-by-randy-baker

Leave a Reply