I talk about medical practices that expire, and probe failed pandemic policy This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.drvinayprasad.com/subscribe
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
SPEAKER_01Well, thank you so much for coming, and it's my pleasure to be here to speak to you all there in Mount Sinai in the critical care department. I'm going to talk a little bit about when medicine comes with an expiration date. And by way of background, I'm a professor here at the University of California, San Francisco. I'm a practicing He-Monk doc. You'll find me there every week in clinic at the San Francisco General Hospital, our county hospital, and I attend on service there a few months a year. I'm also primarily a faculty in the Department of Epidemiology, where I teach a number of courses like on publishing and presenting research and designing clinical trials.
SPEAKER_01And I do a lot of research in epidemiology methods and health policy. All right, let's get started. So, you know, we've been interested in this topic for quite some time. This was a paper we wrote a few years ago called Should Evidence Come With an Expiration Date? And this was a paper I wrote with Palmer Green and Adam Sifu, who's my longtime collaborator. And basically, the theme of this paper was that there are many medical practices that appear to be beneficial, but only in the context of the surrounding medical care of the time.
SPEAKER_01The example we used, of course, was a daily aspirin which seemed to work very well in a population that was enriched with smokers that had a lot of hypertension that was untreated but then the same preventive aspirin primary prevention aspirin studies when done again in the modern age in a population that was more obese that had less smoking that had more statin use those studies were negative and so our question was should medicine and medical evidence that guides our practice come with an expiration date I think that's particularly relevant for critical care medicine,
SPEAKER_01where a number of our best practices are based on randomized trials from a few decades ago, and it's particularly relevant in this modern world. And, you know, as was alluded to earlier, there's a great new example of this. This is beta blockers after myocardial infarction. This is the reduced AMI study that came out from the Swedish group. This is the New England Journal of Medicine paper. And many of us grew up in an era where we were trained that after myocardial infarction, one of the most important mainstays of therapy was beta blockade. In fact, it was even endorsed as a quality metric, i.e.
SPEAKER_01that hospitals and doctors that didn't put their patients on a beta blocker after MI, even in the absence of reduced ejection fraction, they could be dinged for that. They could face a penalty. This REDUCE study was run in, there's some CME comment in the chat. This REDUCE AMI study was run in Sweden, Estonia, New Zealand, but principally was enrolling in Sweden. They enrolled patients with acute myocardial infarction and left ventricular ejection fraction over 50%. The difference here is that unlike the earlier studies,
SPEAKER_01we are now in an era of routine revascularization. So the question is, and there's a whole bunch of other medications we use, including more powerful platelet antagonists. So the question is, does beta blockade versus no beta blockade continue to exert a benefit in this modern era? The primary endpoint of this study was death from any cause or new MI, and the study has now reported results. I'll put one more caveat here. Some of the patients in this study, a small fraction, They were already on beta blockers when they enrolled in this study. The protocol of this study actually asked for those people to stop taking the beta blocker.
SPEAKER_01So not only did they randomize people to not initiate or initiate beta blockade, they also randomized people to stop taking a beta blocker. And they aimed for sort of a moderate dose of beta blockade in this study. The primary endpoint of this study, I think, is stone cold negative. Beta blockers versus no beta blockers, absolutely no difference in the primary endpoint, the hazard ratio 0.96. Some people look at that confidence interval and they say, oh, it goes from 0.79 to 1.16. P-value is totally negative. We'll talk more about that confidence interval. This is death from any cause or new myocardial infarction, this primary endpoint of the study.
SPEAKER_01People fault this study and they say, well, they didn't include, you know, ejection fraction 40 to 50. And I think that is a reasonable criticism of this study. The authors were scared that if they enrolled people with ejection fraction less than 50, they would be randomizing people who are benefiting from beta blockers to not have beta blockade. This was paired with a lower than anticipated event rate. So the event rate was lower than anticipated. What that means is the conference interval is wider than you'd want. And both of these problems I think would have been solved if they had just enrolled
SPEAKER_01down to 40, ejection fraction down to 40. Here's what the editorial in the New England Journal says. This person says, quote, a critical axiom in evidence-based medicine is the absence of evidence is not evidence of absence. Since the confidence interval for the beta blockers includes a potential 21% lower risk, close to the old risk that was established in older studies, an actual benefit with beta blockers cannot be ruled out. And I call this a silly thing to say because I hear over and over in medicine that the absence of evidence is not evidence of absence.
SPEAKER_01And that is logically true. In other words, that simply because you fail to find a benefit doesn't mean a benefit doesn't exist beyond the threshold of what you can detect. But the reason I say it's a silly thing to say is that in medicine, most things don't work. I think when you say the absence of evidence is not evidence of absence, the reader may be under the belief that there are all these things out there in the universe of potential medical practices that would save lives if only we could do the right study and document that. The reality is the vast majority of things don't work.
SPEAKER_01They don't improve outcomes in medicine. And I'll give you a few ways of thinking about this. One, probably in the history of humanity over the last 4,000 years, we have attempted 10 to the power of 8, 10 to the power of 9, or 10 to the power of 10 different medical interventions from trepanating skulls to phlebotomy for George Washington's pneumonia. If you look at all that... 10 to the power of 10 interventions we've tried over millennia. The vast majority did not make people better off. They probably made people worse off.
SPEAKER_01Phlebotomy was in vogue for hundreds of years. It turns out that they were right in two conditions, hemochromatosis and polycythemia vera. But all the other people who were phlebotomized, they got it wrong. They got it wrong. So most things don't work throughout human history. I think that's one fact. The second fact, even in the era of targeted precision medicine, precision oncology, genomic medicine, the vast majority of candidate compounds that are pursued by pharmaceutical firms that are incentivized to find successful products,
SPEAKER_01the vast majority don't work. We're talking about maybe one in 10,000 compounds tested in mice, make it all the way forward to humans. We're talking about from phase one to clinical approval, maybe one in 25. compounds clears that hurdle from phase one testing to clinical approval. And that's even with a low clinical approval bar. If you ask what percent of compounds entering phase one studies actually extend survival of cancer more than four months, you're talking about maybe one in a hundred or one in 200 compounds.
SPEAKER_01And that's with all that preclinical science to make sure that it hits the target, that has really adequate safety, that is animal testing. That's even with all that. So I think we forget that most things in biomedicine certainly in critical care, most things don't work. And I'm going to show some sobering data there. Reduce AMI is not going to be the final word on this topic. We've got a bunch of other studies coming, and these studies are randomizing people to beta blocker versus no beta blocker, and they're dropping the ejection fraction down to 40. They have sort of different enrollment criteria. You can read that here in the editorial.
SPEAKER_01I think the final word is yet to be determined, but I think this is a very damning indictment of one thing in particular, which is this should never have been a quality metric. These quality metrics that are imposed on doctors place an undue burden on doctors. And very rarely, I can't even think of a single quality metric that has robust level one evidence that the institution of the quality metric actually makes people better off. Door to balloon time actually has some caveats. Antibiotics within four hours of having pneumonia diagnosis in the emergency room, that backfired.
SPEAKER_01Many of these quality metrics make life very difficult for doctors. They create perverse incentives. And this was a quality metric that simply, Didn't work. I'll also add, I think quality metrics around cancer screening are misguided as well. In the field of critical care medicine, you all have witnessed a number of high profile, what we call medical reversals. And this was the topic of the book I wrote with Adam Sifu called Ending Medical Reversal. And what we mean by reversals are medical practices that were instituted for a period of time, years or decades that were done widely outside of clinical protocols,
SPEAKER_01outside of randomized studies, outside of registration studies. These were just implemented in the community. And not only were they found not to work, they were found to be no better than a prior or lesser standard of care. These were things that were wrong, not just in retrospect, they were wrong at the time. And we didn't study them often appropriately. There are some really interesting reversals in the field of critical care medicine that are based on single center randomized control trials that failed to be replicated in the multicenter setting.
SPEAKER_01One of which is most famously is of course, the Netherlands study on tight glycemic control in medical ICUs. When I was in training, uh 2009 to 2012 as a resident the mantra was tight glycemic control tight glycemic control until in 2009 the nice sugar study came out so based on the results from the netherlands we thought tight glycemic control had better outcomes for people in the intensive care unit nice sugar is a multi-center randomized control trial asking the same question and it failed to find that benefit now you may ask well maybe the multi-center study gets it wrong and the single
SPEAKER_01center study gets it right i think that's the that's an incorrect interpretation we want multi-center studies because we want medical practices and interventions that can be scaled beyond single centers we want things that don't just work in the netherlands or don't just work in manny rivers hospital in detroit we want things that work in our hospital And multicenter clinical studies are asking that question, does it work in your hospital? Does it work in a hospital that's different than the hospital where there is a charismatic and devoted leader who's interested in the success of the product?
SPEAKER_01Early goal-directed therapy, for many of us, it felt like common sense resuscitation, but it was a formal protocol. It initially had a positive result in the study by Manny Rivers from Detroit, and it was contradicted, I think, now in two randomized controlled trials that are run in the multicenter setting. I wish I had done this paper. This is one of the few papers I kick myself when I see that it's been published, and I've been thinking about it for years, and I've been saying it in my talks, and I never did it myself. I wish I'd done it, but hats off to these investigators for doing it. It's called Positive Single-Center Randomized Controlled Trials and Subsequent
SPEAKER_01Multi-Center Randomized Trials in Critically Ill Patients, a systematic review, and they did exactly what you'd want to do, they do an empirical analysis of clinical questions that were studied both in the single center setting and in the multi-center setting. And they ask, what is the concordance? What do they find? Which ones are getting it right? Which ones are getting it wrong? They found 19 single center randomized control trials in critical care medicine published in the creme de la creme of biomedical journals, New England Journal, JAMA, and The Lancet. They followed these studies out and to see how many of them had at least one
SPEAKER_01subsequent multicenter randomized control trial that was 16. In one case, it was confirmed. The results were validated in the multicenter study. In one case, it was profoundly refuted. And in 14 cases, the results were not statistically significant. They call that neutral. I think that is a contradiction to find a significant result that vanishes with further study. These were the topics they studied. You'll see intensive insulin therapies on there. You'll see early goal-directed therapies on there. Beta blockers and sepsis is on there. Oxygen targets. Timing of renal replacement therapy is on there.
SPEAKER_01Whether or not people in acute renal failure need high-intensity renal replacement therapy. Non-invasive ventilation in immunosuppressed patients is on there. I mean, this is really an indictment. I think, of what we had been doing. Intensive insulin therapy, of course, statistically significant mortality increase was seen by pursuing that strategy. So it actually was harmful. And only epinephrine during cardiopulmonary resuscitation was validated, I think. All right. So that's a little bit of a prelude, and we can talk more about that.
SPEAKER_01But I think it's important to keep in mind that what are the conditions where you'd want to reassess the medical evidence? You want to reassess medical evidence when the backdrop of medical therapies changes. like with acute myocardial infarction and routine revascularization, when there are many other novel entrants in the space, again, acute MI, and you want to reassess the evidence when the initial evidence was based on, you know, a unique single center study and a unique study design. So maybe instead of randomized controlled trials,
SPEAKER_01you could look at before and after studies by way of something you all work on. I mean, the... The practice of routine gown and glove precautions and universal screening for MRSA and VRE has been common in the ICU setting and is still done in my hospital and many other hospitals. But we have two multicenter cluster randomized controlled trials, STAR-ICU and BUG, both of which are profoundly negative studies. And a lot of that practice is based on before and after studies that come from the prior era.
SPEAKER_01Before and after studies in my opinion aren't as good as randomized control trials because they also incorporate the Hawthorne effect where people start to behave better merely because they know they're being studied. I think rapid response teams is another great example where if you look at whether or not rapid response teams move people from having codes on the floor to the ICU setting, the answer is sure, they move those people. But whether or not they actually improve mortality is absolutely uncertain, and Merritt suggests that they don't improve mortality. They just merely change the room in which you die.
SPEAKER_01And so I think that for health policy interventions, particularly around critical care medicine, we need cluster randomized control trials before we implement new measures, or at least concurrent with implementation, and certainly also before they're enacted as policy metrics. Now let's talk about COVID-19 therapeutics because I figured you all want a little bit of controversy in your lives. I'm going to argue that actually the data for every single COVID-19 therapeutic is expired. I mean, COVID-19 therapeutics have expired because the backdrop has changed so dramatically
SPEAKER_01and the virus and the way in which the virus acts in the body and the strain of the virus and the harms from the virus are fundamentally different than first quarter of 2020. In first quarter 2020, we had a certain IFR. I'm going to show you what that's happened over time. In first quarter 2020, we had no population level immunity from prior infection. We had no population level immunity against spike protein from vaccination. Now it's 2024. We have... Huge amounts of population-level immunity from prior infection,
SPEAKER_01huge amounts of double immunity from two prior infections, and huge amounts of vaccine-induced immunity and many people with hybrid immunity. And we have a virus that's fundamentally different in sequence than the original virus. And so the data for all of the COVID-19 therapeutics, I think, has expired and really warrants serious reinvestigation. I'll just talk about a couple examples. This is Paxlovid. In 2023 and 2022, we wrote a couple of articles where we called it a regulatory gamble. And the reason we said it was a regulatory gamble was that Paxlovid was initially
SPEAKER_01validated in a randomized control trial called Epic HR. which tested whether or not it improved outcomes and lowered the risk of hospitalization in adult patients with COVID-19 who hadn't gotten any vaccine doses and hadn't had COVID in the past. They hadn't had vaccine doses and they hadn't had COVID and they had higher risks. They had risk factors for progression to severe disease. And it was done at a time where the prevailing strains were prior to pre-Omicron strains.
SPEAKER_01And in the pre-Omicron strain era, for somebody who's never been vaccinated, never had COVID-19, and at high risk of bad outcomes, Paxlovid seemed to work. But what we had no idea was whether or not Paxlovid has any benefit in somebody who has been vaccinated or double vaccinated or triple vaccinated, and somebody who's had and recovered from COVID-19, and in the era of Omicron and beyond. We had no data for that. And yet... The current administration granted Pfizer a $5 billion contract even before Epic HR results were known. And the current administration pushed very hard,
SPEAKER_01gave another $5 billion, $10 billion in government spending on Paxlovid and pushed very hard on television. You'll remember Ashish Jha on TV every day saying, you know, got COVID, have Paxlovid. It's good for everyone. Talk to your doctor. This was administered in widespread fashion to low-risk populations. In my city of San Francisco, 20-year-old tech workers with no risk factors for progression who work at Google and Facebook were taking Paxlovid routinely. There was never any evidence to support that, and we call that a regulatory gamble.
SPEAKER_01The FDA shouldn't have even authorized it, is what we wrote about. We weren't the only ones. Todd Lee from McGill University and Andrew Morris and others, they wrote, what is the place in therapy for Paxlovid? And they outlined many of the same concerns I've just detailed to you. This is a figure that comes from the Financial Times. I think it's really illustrative. If anything, if it was updated to the modern, it would be even more sobering. But there's no doubt in my mind that COVID-19 was a serious threat, of course, in 2020.
SPEAKER_01And through the winter of 2020, it was a devastating threat. Thankfully, in 2021, because of the vulnerable groups, older people who had not previously had COVID-19 being vaccinated, we see a dramatic reduction in the infection fatality rate. We can debate whether or not to what extent boosters played a role here. I think that is debatable, particularly whether or not they were overused in low-risk populations. And certainly the rise of Omicron or strains of the virus that have much more replication in the upper airways rather than lower airways has contributed to a further decline in the IFR.
SPEAKER_01And now there's no doubt about it. The IFR, I think, currently is much less than seasonal influenza. And COVID-19 is not the threat that it once was. And our policy has to change. And I think most of the policy has changed around this, except for a few pockets of delusion that still persist. COVID-19 is less lethal. And what this means is that randomized control trials that occur earlier in the pandemic are not really relevant to clinical practice today, including remdesivir, including recovery in dexamethasone, many of these things, including convalescent plasma. They're not really relevant to the patients we're taking care of.
SPEAKER_01For many, many years, Pfizer suppressed the results of the Epic SR study. They didn't publish this study. This was the randomized control trial that was actually done in people who had had gotten some vaccination. It was the only randomized control trial relevant to the Biden administration's policy decision to put a sheesh jaw on television and tell vaccinated people to take this product. Over and over, they pushed that. And this was the only randomized trial relevant. It has finally now been published. It has been published in the New England Journal of Medicine, spin-off journal, Clinical Evidence, and it shows rather sobering results that there is an increase in adverse events
SPEAKER_01related to the product, 12% versus 4%, any adverse events. There's more grade three, four adverse events. There's more events leading to discontinuation of the product. So, I mean, this is something with adverse events, no doubt. The distinction between emergent events and related events is whether or not in the eyes of the investigator they're linked. And the primary endpoint of this study is the time to sustained alleviation of symptoms. And that is absolutely stone cold negative. Paxlovid versus placebo in people who have been vaccinated,
SPEAKER_01who have been vaccinated as no benefit, frankly. Here's a figure that I put together. It shows all the randomized control trials of Paxlovid and Molnupiravir. And it shows you the dates in which the trials enrolled. Okay, the dates the trials enrolled are shown here by the bars. These are the different studies, Epic, HR, Panoramic, which is still the biggest and largest study that's still outstanding. The color gray means that it hasn't yet resulted the outcome. Red means it's a negative study.
SPEAKER_01Green means it was a positive study. Over here, I've superimposed when the prevailing variants were. This is based on, and then I've also superimposed COVID-19 hospitalizations. And so I think what you see is, you see these products, both Molnupiravir and Paxilavid, had positive randomized control trial data in a world of Delta and at a time prior to Omicron variants. This blue line here shows you the percent of people with spike antibodies in their bloodstream over time on the scale on the right.
SPEAKER_01So in other words, as population immunity has gotten better, as the variants have shifted towards Omicron, the color turns from green to red. It's turned to sobering, sobering red. And I think, I'll bet you money that Panoramic is going to be absolutely negative study. You know a trial is going to be negative when there's multiple protocol amendments to continue to enroll and enroll and enroll because their event rate is so absolutely floored because of population level immunity. The event rate in the control arm is so low, it's going to be very difficult for Paxlovid to eke out a benefit.
SPEAKER_01And I think when this trial is ultimately published, it's going to be a negative study. And that's going to suggest that Paxlovid, worse than Tamiflu, might be the single greatest waste of government money globally for a respiratory virus because it was predominantly given to people who don't benefit who are vaccinated and at low risk of bad outcomes from the virus. Okay, shown another way, you know, you can see in the, I'll skip this figure just for the sake of time. All right, I'll skip the cost too. This cost is horrific. I mean, even assuming the best case scenarios, we're talking about $112,000 to prevent a hospitalization.
SPEAKER_01The dollar per quality was always unaffordable. It was the great waste of government resources. And just a great example of how resources during the pandemic were pushed towards the shareholders of for-profit entities and not towards policies that might've actually made a difference in communities that might've actually stood to benefit from that money. One more thing, some people wanna justify PaxLivid and they cite to me observational study data to support PaxLivid. I think this is a very foolish thing to do because the people who have the ability
SPEAKER_01to get a doctor to prescribe them PaxLivid are different than the people who don't have that ability. So when you do observational studies, not only are you looking at the receipt of the product, you're looking at the kind of person who seeks out the product the worried well, and the kind of person with the resources to get the product, the rich and entitled people. And so what that means is you see many, many observational studies that show results that look amazing. This is PaxLivid on the top. This is Moldopiravir on the bottom. Look at that. If you don't get treatment, the cumulative incidence of hospitalization is horrific. You know, it's not horrific.
SPEAKER_01The absolute values on the scales are very low, but the benefit is massive. Look at how massive that benefit is. Whenever I look at observational studies telling me to make a causal conclusion about a clinical intervention, I always look at them really closely in a couple of ways. And I compare them to randomized trials when possible. So the same study on the top that's used to justify Paxlovid also looked at Molnupiravir. And this was the result they found, a big benefit of Molnupiravir. The hazard ratio was 0.37%. That's pretty low.
SPEAKER_01I don't see that many hazard ratios that good. But at the same time, the panoramic investigators in the United Kingdom did a randomized controlled trial in a similar population, and their hazard ratio was 1.06. In other words, when you compare the observational study to a contemporary randomized controlled trial that was run with tens of thousands of participants in each arm, the hazard ratio goes from 0.37 to 1.06. I think that shows you how much bias can be baked into observational studies. This is the healthy ratio. the bias of being healthy enough to get the product. And also, if you deteriorate very quickly, you couldn't have gotten the product.
SPEAKER_01I'm gonna come back to that. There's a phenomenon in observational studies, this is a paper we published in JAMA Internal Medicine, where they work too fast. The intervention works too fast and it can't possibly be true because nothing could work that fast. And here, let me give you some examples. In the initial randomized control trial that reported results in November, I think November 8th, 2020. This is the initial randomized control trial of the Pfizer-BioNTech vaccine versus placebo. This was the primary endpoint of the study.
SPEAKER_01This is why we celebrated, why I rushed to get that first dose. This is symptomatic SARS-CoV-2. You have symptoms that require swabbing and then the swab tells you you have COVID-19. In this randomized control trial, you see very clearly that the vaccine is better than placebo. But I want to draw your attention to the first 10 days. In the first 10 days after receipt of the product, not only is there no difference, the curves are superimposable. Why are the curves superimposable for 10 days and then they start to split?
SPEAKER_01Maybe somebody could put that in the chat. Why don't they split immediately? You can shout it out or put it in the chat. This is a test if anyone's still awake.
SPEAKER_00Why don't they smoke?
SPEAKER_01Okay, okay. Well, I mean, obviously, I mean, what I'm hoping somebody would say is that it takes a while for it to start working. It takes a while for it to start working. Your vaccine doesn't work immediately. You got to wait for the antibodies to build up and then you got to, you know, it takes a while for it to start working. So we know it takes probably seven to 10 days for the vaccine dose to start working because it needs to rev up antibody production. So what am I to think when I read a paper in Nature Medicine that's an observational study comparing people who got three doses of vaccine, one booster, first booster, to people who got four doses?
SPEAKER_01This comes from the Israeli investigators. And this is literally the reason we don't have any randomized controlled trials for a four versus three that sought clinical endpoints. This is the basis of that recommendation. This is what all the experts on Twitter and social media and the administration are saying why you should get it. This is what I see. I see death from COVID-19, vax three doses versus four doses. And what do I see? When do the curves start to split? They start to split, in my opinion, maybe day 12, 13, they start to split.
SPEAKER_01They're already splitting by day 12 or 13. How is it possible that they can split on day 12 when it takes you at least a week to rev up antibodies? And then it takes you a while to avoid getting sick. And then it takes you a while to avoid having progression and respiratory failure and avoid getting on the vent and then avoid dying. It splits too fast. It splits way too fast. It's not possible that it could work this fast. And that tells you that the people who are getting four doses are fundamentally different than those who get three. In fact, they're more likely to be higher educated.
SPEAKER_01They're more likely to have more wealth. They're more likely to take other precautions in their lives. This smells like confounded data. And yet it's something that's missed by many people. This is an example of Eric Topol. He says, look at this massive reduction in death and hospitalization from Paxlovid, but look at the curves. They split on day zero. Paxlovid starts to work literally before you even took the first dose. You didn't even have to fill the prescription yet, and Paxlovid's already reduced the risk of mortality on day zero. How is that possible? That's only possible because of guarantee time,
SPEAKER_01meaning that there's a certain amount of time guaranteed to somebody who took Paxlovid. They couldn't have gotten hospitalized before they filled the prescription. That's called a mortal or guarantee time. And the other reason is it speaks to confounding that people who pursue prescriptions are people who are more worried and more able than people who don't pursue those prescriptions. You know, when you look at these PaxLivid observational studies, you see this. This is impossible. PaxLivid cannot work before you take it. That's impossible. And so that's why I don't trust these observational studies.
SPEAKER_01Okay, that's why we need randomized trials. All right, then the next thing I'll argue, if you're all still interested, I'll argue that, you know, What year is it now? 2024? This fall, we're going to get another booster and then the hospital is going to have some booster program. I've had three doses of the vaccine and I've had COVID, I think twice. Once I've documented and once I'm pretty confident I've had it. Okay. Is it possible, is it credible that I have a benefit to myself or a benefit to anybody else by getting the annual booster in the fall?
SPEAKER_01And the answer is there's absolutely no credible data for that claim. There is a risk to me to get it. There's a risk of myocarditis. I'm still on the tail, not that old yet. I'm still on the tail edge of the higher risk of myocarditis. There's no proven benefit to my patients because there's no sustained reduction in transmission that's ever been demonstrated. And there's no proven benefit to me for getting it. So why is the FDA allowing this to be recommended year after year? This is Peter Marks. He's the FDA scientist who's making this call. And I would say my position is the US FDA has an obligation to demand that the
SPEAKER_01company that's made $100 billion from these products run a simple randomized controlled trial for boosters that's powered and designed to assess clinical endpoints. They got a hundred billion reasons why they can do it. So why don't they do it if they want to keep selling me the product? That's not even my idea. That's an idea that was proposed by, oh my God, is this Peter Marks, the same person who's approving it? He wrote an article in JAMA in 2022. What did he say in this article? He says, quote, although experience with mRNA vaccine platforms has enabled authorization of
SPEAKER_01updated versions of vaccines without large clinical trials, when more significant modifications are made to a vaccine, the clinical effects are often unexpected. Biological properties that may plausibly have benefits often have unanticipated consequences. Therefore, unless correlates of protection that are strongly associated with duration of protection against COVID-19 can be identified, and let's be clear, none have ever been identified, it is likely that rather than relying on immunobridging, which means antibody studies, to infer vaccine effectiveness, Peter Marks writes,
SPEAKER_01large randomized clinical trials, similar to the initial trials of the currently authorized or licensed vaccines, the 40,000-person studies, will be required to ascertain the effectiveness of these new vaccines." This was Peter Mark's idea before he fell under, I think, tremendous pressure to approve this year-over-year product. It was his idea that they need to do bigger randomized studies, powerful clinical outcomes. Other nations don't do what we do. Other nations are not going around asking seven months old to get an annual COVID-19 booster. They don't recommend it.
SPEAKER_01Their CDC does not recommend it. This is the NHS. You know, it's recommended for predominantly older people over 65 and up. Sweden recommended older people or high risk groups. Italy, again, older people, high risk groups. France, older people, high risk groups. Germany, older people, high risk groups. Portugal, New Zealand, blah, blah, blah. No country. in the European CDC recommends COVID-19 immunization for healthy children currently in the European CDC. And yet in the United States, it's recommended for everyone six months and older and every single year. It's completely outside of any global norm and not justified by any evidence and
SPEAKER_01done at a time when the actual harms of the virus are plummeting and population level immunity is soaring. It really defies logic, in my opinion. All right, I'll skip that. Another point, Paul Offit himself, he doesn't support the annual COVID-19 booster policy. Paul Offit, a man who's invented the vaccine himself. So I think surely he should be above the criticism that he could possibly be anti-vaccine because he's a vaccine developer. He says that the recommendation makes less sense even now. And then he says... Wouldn't it make sense to get an additional dose to prevent your chance of having
SPEAKER_01long COVID, which is characterized by fatigue, headache, brain fog, and muscle aches? And then he asked, quote, are there data to support the additional doses beyond what most people have actually gotten? Because most people have already gotten a few doses plus a natural infection, which is defined as hybrid immunity. Is there evidence that an extra dose makes a difference? And the answer is absolutely not. There's no evidence that the annual dose is going to lower your risk of long COVID. And Offit, who is 72, who's already had COVID-19 once, he didn't get the bivalent booster and he's not going to get the new one. Okay. And then finally, the FDA officials, Marion Gruber and Phil Krause, they resigned. These were two people who worked in government for 20 years.
SPEAKER_01They resigned under pressure to approve a COVID booster for all ages, including young people, even though that did not have strong evidence. They didn't resign in the prior administration. They resigned in this administration in the fall of 2021. They wrote an op-ed saying we don't need universal booster shots. Okay. American people don't want it. Of course, nobody's taking the shot. And of course, finally, the COVID-19 fall booster last year actually didn't target COVID-19 at all. It was for XBB1.5. And actually, XBB1.5, if you looked at the CDC's
SPEAKER_01sequencing data from the fall of last year. I highlight where XBB1.5, by the time they even debuted the vaccine, it was already diminishing and it was not a contributor. The majority of COVID-19 in the fall of last year was not in the booster at all. They missed entirely. And the last point I'll make, is that it was a net harm to some subgroups. There are definitely people who have suffered net harm from vaccines and vaccine mandates, particularly booster mandates. And that's in contrast with what Rochelle Walensky was saying as part of Transparent Public Health.
SPEAKER_01Rochelle Walensky says, quote, April 27th, the US CDC has not seen has not seen the link between heart inflammation and COVID-19 vaccines. They say, we looked at 100 million cases. We didn't find any myocarditis. That has to be incorrect because, of course, the Israelis had already reported two months earlier that the risk was 1 in 3,500 for young men, and that's a risk that's sort of been zeroed in on by multiple population studies. We did a very simple paper led by Kevin Bardosh where we looked at universities that mandated a third dose for young men and we calculated the upper bound absolute
SPEAKER_01risk reduction in hospitalization from going from two to three doses versus the actual risk of myocarditis from going from two to three doses, which has now been published by Katie Scharf in the Kaiser Permanente Experience. And we compared the two. And so here we're going to show you the red bar that shows the risk of myocarditis and this is myocarditis that prompts medical care and then the smaller bars are going to show you the risk of hospitalization maximum risk of hospitalization proven with booster and what you see is the serious adverse events are just so much beyond what could possibly be averted uh myocarditis exceeds any potential benefit
SPEAKER_01from boosting strategies so it is very likely i think i believe that we did do net harm by mandating the third dose in young men particularly men in the target demographic of 16 to 24 the highest absolute risk One thing you may hear is that, well, myocarditis is worse from COVID-19 than the vaccine. A number of groups, including the Oxford Group, have published that. I think that's dishonest in two fundamental ways. One, when you look at myocarditis, we're not talking about myocarditis in an 80-year-old grandmother.
SPEAKER_01I'm not worried about that. I don't think she has a risk of myocarditis from getting booster after booster, and she's welcome to continue to get boosters for an 80-year-old in a nursing home. Continue boosting. Have at it. I'm not even going to be a stickler for strong evidence there if it's her choice. I'm talking about myocarditis. It's a risk in young men. And in this analysis done by Ben Knudson, a medical student, what we show is that there are many, many studies. Each bar here comes from a study. Many, many studies on myocarditis after vaccination. Some of them, on the rightmost, use zero stratifiers. By that, I mean they look to see of 100,000 people get the vaccine, how many got myocarditis?
SPEAKER_01And you can see it's pretty low. Look how low that number is, one to two. But some of them use one stratifier. They say, you know, let's just look at people in a certain age group. Let's just look at men instead of women. Let's just look at Pfizer instead of Moderna. And they start to get a little bit higher. Some of them look at two stratifiers. They just look at men in Moderna compared to everyone else or men in dose two. They start to get a little bit higher. Some look at three stratifiers. They look at men, dose two, and they look at certain age groups. Well, they start to get, now they're getting higher. And then finally, when you look at four stratifiers, just men, just Pfizer, just Moderna,
SPEAKER_01and just the target demographics that are at risk, you find massively increased risks. So of course, it's very possible that the risk of bad COVID outcomes is worse across the whole population. 80-year-olds have more benefit from vaccination than they do have harm, I have no doubt, but that's different than a 20-year-old man. And medicine has to have policies that separate 20-year-old men from 80-year-old women. If we cannot tell them apart, we're not doing a good enough job. So these studies I think are dishonest because they don't break apart the risk in the target groups.
SPEAKER_01Similarly, when we talk about myocarditis after COVID, All of the studies that are being talked about, the denominator for COVID is not everyone who had COVID. It's not everyone who had COVID. No, no, no. It's not a seroprevalence denominator. It's not even everyone who tested for COVID and tested positive. Oh, no, no, no. They don't have home tests. It's only the people whose COVID diagnosis was documented in the electronic health record. That's not only not everyone with COVID, that's a tiny fraction of people who've had COVID, and they're probably the sickest people. So if you look at what's the risk of death from food poisoning from eating McDonald's,
SPEAKER_01if you looked at everyone who ate at McDonald's, the risk of death from food poisoning is very low. But if you just looked at the people who were hospitalized in the intensive care unit with food poisoning after eating at McDonald's, well, then the risk of death might be much higher. Because you're not looking at the real denominator. And it just kills me that I can't, I can't believe people keep citing this statistic. The denominator is not the real denominator. So why would you say that? It would be like concluding that McDonald's is killing all these people because each year four people are hospitalized with food poisoning in the ICU and two of them die. So it is a 50% death rate from food poisoning. And it's worse than, you know, whatever.
SPEAKER_01But of course it's not. It's mostly, you know, of course McDonald's will kill you, but you know, not for the food poisoning. It's just that it's just terrible food. Okay, no, all right, anyway. medicine without evidence the last thing i'm going to say on this talk and then unless you all have questions and you want to keep going you can't tell you're all quiet um you know we're moving to a world where we talk about expiration dates we talk about reassessing the evidence but we're moving to a world where we're just so far past that we're just implementing medicine without evidence at all and no one seems to have any interest in generating evidence and i think
SPEAKER_01At some point, we have to ask ourselves, are we any better than the people who bloodlet George Washington and killed him? Are we any better? If we don't run any studies, we just set policy, we just make things up. And here's the policy at my university. I got this email. After four years of having to wear a mask in clinic, I was told in May 1st, I don't have to do that anymore. And so the patients and I, we celebrated, we threw our masks into the air. And then we picked them up because, of course, they're an environmental hazard. So we had to throw them away. Okay, we picked our masks up.
SPEAKER_01And then I got this email from the bosses. And the bosses say, congratulations, you don't have to wear masks anymore. But face masks are only required in the following time periods during the annual respiratory viral season from November 1st to April 30th. And I said, November 1st to April 30th. That's April, May, June, July. That's six months a year. What the heck are we talking about? We have a mask mandate at UCSF for the future, forever, six months a year. So six months without masks, six months with masks.
SPEAKER_01What evidence is there to justify this, I ask myself. There are three endpoints I think that the administrators may care about. One is nosocomial infections. We don't want to spread virus to our patients. That's terrible. One is we may want to decrease cumulative infections of people who walk in. Our policy is not just on the floor, not just seeing patients. Our policy is in the gift shop. Our policy is in the hallways. Our policy is in the cafeteria. Our policy is in the cafeteria line. Six months a year, you got to wear a mask while you're waiting in line for your food. Okay. The third reason maybe is that people in the community want it.
SPEAKER_01As far as I can tell, they have no robust evidence that it slows nosocomial spread. Two, they have no good evidence that it reduces cumulative infections when people pass through the hospital. And three, they actually have no idea what our patients want because we've never surveyed our patients. And finally I ask, why does the policy work on Thanksgiving but not on Halloween? The policy they've implemented is something that only works in some holidays. In November 1st it works, in October 30th it doesn't. That to me is ridiculous. If you wanna test it, test it in a way you can find out if it works and when it works. They're just making up rules.
SPEAKER_01And the last thing I'd say is these rules, I think, have nothing to do with science. They're fundamentally political, and I can prove it to you. Sarah Miller and I, we looked at all of the mask loosening in 2023 in the first quarter. We looked at all the hospitals and when they dropped mask mandates, and we did regression analysis, and we asked, Do mask mandates fall in hospitals when the cases drop, when the hospitalizations drop, or is there a third factor that explains why masks fall in some places and not others,
SPEAKER_01sooner rather than later? And the answer is, on COVID-19 cases, the odds ratio is one. The cases in your community have nothing to do with whether or not you extended the mask requirement. Hospitalizations, oh, it's 1.06 and a big, it has nothing to do with your hospitalizations, dropping the mask mandate in your hospital. But for every 10% increase in vote for the Republican candidate for president, you're much more likely to drop your mask mandate sooner. And that to me is the problem, that we have abdicated this policy decision, this medical decision to simple partisan politics.
SPEAKER_01And the reason, of course, San Francisco is going to continue to mask is that we don't have a 10% people voting for the Republican. We have a 0%. This is a democratic city. So we're going to mask until our earlobes fall off. All right. I'm going to pause there. I have two things I could talk about. I could talk about Entresto and why this is a very bad randomized trial paradigm HF and why Entresto probably doesn't work, which I think is very interesting. I think Entresto is the greatest scam ever perpetrated on the heart failure community.
SPEAKER_01Or I could talk about the science paper on Bangladesh, cluster randomized trial masking, a randomized control trial, and why actually, even though it looks like it's working, it actually is a really flawed randomized study that has this one problem of concealment bias that's such a devastating problem that in some ways it's actually not a randomized study. It looks like a randomized study, but it's not really randomized. So those are the two bonus things you can pick from. But I'm going to stop now and take questions and see if we can talk about either of those two studies or something else entirely.
SPEAKER_01So thank you for having me. Hey, this is Dr. Gopal, critical care from Einstein and Morningside, Breath of Fresh Air, wonderful talk.
SPEAKER_00You know, classic activated protein, you can see. Oh, yes, Zygris. Even now, balanced versus non-balanced solutions. We live in an algorithm. and residents are taught and it's introduced nice protocols when COVID came about, Plaquenil, Zetro, based on limited Wuhan studies. You become a pariah if you go against all this.
After all of that, what's your message?
SPEAKER_01That's a good question. I mean, I remember Zygris. I remember giving it and then not giving it. It all happened when I was a trainee. What's my message? My message is that, I mean, I think that you're absolutely right. People don't use their brain and follow guidelines. And so I get angry actually when the fellow tells me, I mean, why do you want to give this patient this chemo? Oh, NCCN says. Oh, NCCN says? No, that's not the answer. The answer is, why do they say, and do you agree? So I think we should all do that extra level of work, which is, why do the guidelines say to do it,
SPEAKER_01and do we agree with the guidelines? And I think that's not easy, and you can't do it every day for every question, but once a week, you pick one question and do it, and over a year, you're gonna be really good at sort of having a detector that says, okay, maybe this is an exuberant recommendation. so i think that's one thing we can do the next thing i think we can do is this is something that john mandrola and adam sifu and i we talk about um we call it the case for being a medical conservative and we don't talk about it's not political conservatism we just mean that when something new comes along when something new comes along in medicine
SPEAKER_01we have to be skeptical of it i mean this last month i see there's the first ever randomized control trial that shows impella improves outcomes and we've been doing impella for 10 years and i think it's easy to say well we've been doing impella and now finally we've been proven correct that it's that impella saves people's lives But that randomized trial has problems. One is the p-value is 0.04, and then the cutoff for significance is 0.048 because they took an interim look. Okay, so it's barely significant. That's one. Two, they exclude people without a hospital cardiac arrest,
SPEAKER_01which is the majority of the reason why we're putting the impeller in the first place. They're not even including those people. Three, they enroll over a very long period of time. Four, in the control arm, if you have hemodynamic collapse, they put you on ECMO, and you're getting ECMO in double digits, more people getting ECMO than the control arm, well, that ECMO that they're getting might not be a salvage, it might even be harmful. And so when I start to see like so much money, billions of dollars, and so many people being committed to a practice where the best available randomized study is P of 0.04 after many,
SPEAKER_01many years, and the pretest probability of IABP and ECMO was negative in a prior randomized study, the pretest probability is low. I think the totality of the evidence has to be skepticism, and the best answer is I think we should just do a bigger randomized study and get NIH to do it. Okay, so I mean, so my answer, yes, sorry. All right.
SPEAKER_02People want to hear about both of them. Now, I'm very familiar with both of these issues that you want to talk about. One is the Entresto and how the ASAR was not maximal and so on. So people want to know now, can you, in the next eight minutes, talk about, one, the fatal flaw in the masking RCT and the fatal flaw in the RNA trial?
SPEAKER_00All right. I'll do the masking RCT first. It's a quicker one. Hmm. Oh, wait. No. Am I have the right? Wrong slide deck. Okay. Sorry. Sorry. Wrong slide deck. Okay. Let's do the masking RCT. It's a fun one because I didn't see it initially. So I really it really bugs me if I don't see it right away.
SPEAKER_01Took me a while. Okay, so this is a cluster randomized control trial in Bangladesh. It appeared in the journal Science. And what they did was I think they took something like 300 villages and they matched them at baseline. So they say, you know, this village has a population of 40,000 people. This other village, 40,000 people, rural Bangladesh. This is a match. And this village is 10,000. This village is 10,000. This is a match. They create 300 pairs of villages. And so they know the villages in one arm and the villages in the other arm are roughly the same sample size. And then they randomize each of the two pairs. So they say one arm is gonna get the mask and one arm is not gonna get the mask.
SPEAKER_01It's done in Bangladesh. It's done in 2020. The baseline prevalence of COVID-19 is essentially zero. It's rural Bangladesh. Nobody had COVID at baseline. They go to the village and they randomize you to an intervention where they pass out free masks and tell you to wear masks. And in the control arm, they don't do that. They just go there and tell you, hey, we're doing a study. You want to participate. The primary endpoint of the study is symptomatic SARS-CoV-2. There's four endpoints. Symptomatic SARS-CoV-2 is, of all the people who signed up for this study, if any of you feel like you got a tickle in your throat,
SPEAKER_01a cough, runny nose, you call us up. We'll mail you a little kit. You prick your finger, put a drop. We'll look for antibodies to COVID-19. So this is antibody-confirmed symptomatic SARS-CoV-2. That's the primary endpoint of this study. All right, there's some other endpoints, but the primary endpoint of the study, here's what they found. The comparison villages, three quarters of 1% had COVID-19 by symptomatic seroprevalence. The cloth mask didn't do crap. Cloth mask, we know, is a useless mask. It's a shame we recommended it so hard. The surgical mask was a winner. 11% p-value, 0.043. It's below 0.05.
SPEAKER_01They popped the champagne. They're really celebrating, okay? Surgical masks is the winner. Cloth mask failed. All right, I'll skip that part. Here's one thing that caught people's eye. The villages were matched and it's a cluster randomized study. And if you looked at the people who consented to participate, it was 178,000 in the masking arm and 163,000 in the no masking arm, a difference of roughly 10%. Another way to put it is, for every 10 people in the control villages, there's an 11th person in the treatment villages.
SPEAKER_01But that's really interesting because that's an imbalance, and it is a randomized study, and we really did try to match that, the baseline sample size. If you look at all the other endpoints collected, the people with symptoms, that's about 13,000 similar. The people with blood samples collected, that's similar. The people who are tested, similar. Everything else is similar, except this. And in fact, some researchers very cleverly here, they create a sort of a distribution. The distribution is if we sample from the null distribution many, many times, what do we see in terms of symptomatic seropositive?
SPEAKER_01So this is sort of like really sort of another way to look at the p-value. The p-value for the primary endpoint of the study is over here. Like, it's something less likely to occur under the null assumption. Like, it's, you know, p of point one, you know, you can quibble about the p-value. But the p-value is over here. It's like, it looks significant-ish. It looks like it's unlikely that you would have seen that outcome under the null distribution. It only happens sort of some percent of the time. The rate of symptoms is very significant. Like the people in the mask arm, they're much less likely to report symptoms. They're slightly more likely to have symptomatic serocovid.
SPEAKER_01But look at the consenting population p-value. The p-value for the difference in the consenting population has got so many zeros in it, it's off the chart significant. In fact, the most significant thing about the study, the biggest statistical difference is the number of people who consent in one arm versus the other. And that's the fatal flaw. Why did that happen? Why in one arm of this study, the arm that gets the free mask, you're getting so many more people consenting than in the other arm. You're getting an 11th person for every 10 person.
SPEAKER_01You're doing this for 170,000 people. And the answer is it didn't have concealment. Now, you may remember in randomized studies, there's blinding, blinding, blinding. Blinding means the patient doesn't know they got it and the doctor doesn't know what they got and the statistician doesn't know what they got. Blinding is about you don't know what you're getting when you're in the study. Concealment means at the moment you're randomized, you don't know which arm you're being randomized into. And we don't have concealment problems anymore because now we use a digital phone number to call for randomization in clinic.
SPEAKER_01And you call the number and it says the patient's been randomized to whatever, they tell you, right, on the phone. But back in the day, we used to get these envelopes. And you remember doctors would get envelopes for randomization. And some doctors were obviously want to do the right thing for their patient. You get a young woman with breast cancer. I'm not doing taking a random envelope. I'm going to hold that envelope to the light. And if the envelope says control, I might get the next envelope and get treatment because I want to do everything for the 40-year-old woman with breast cancer. That's a concealment bias that I subverted randomization when I was doing the randomization. And they had it in this study.
SPEAKER_01And here's why. In the control villages, they pull up in a Toyota Yaris and a clipboard and they consent people to the study. In the masked villages, they pull up in a truck with boxes in the back and they consent people to the study. And people know in the masked villages, it looks like these American investigators, they got a lot of freebies in that back. And if I consent, maybe they're going to give me some free stuff. And in fact, that's true. And in the control arm, they know they're probably not going to get something free. And so there's an 11th person who agrees to participate because they think they're
SPEAKER_01getting something for free. And there's not that 11th person in the control arm. And that 11th person, if they have a tickle in their throat, if they feel a cough, they have to be equally likely to prick their finger and put the blood on the card and send that card in for this to be a real randomized study. But I would contend they're not equally likely because the kind of person that only enrolls in your study because they're getting a freebie is not the kind of person that wants to prick their finger and send you the blood. They don't care about your study. They just wanted that freebie. And that is, I think, the reason why all the results are the way they are.
SPEAKER_01They have an imbalance in people who are... not being ascertained for the endpoint, and the entire study is probably not a randomized study. In fact, it is not a randomized study. There's an imbalance in the population, and it's not at random. It is an imbalance based on knowing some characteristic of the study, which is whether or not you get something for free. All right, Entresto, you're going to go online to my YouTube channel. I got a couple of videos on Entresto because I burnt all my time. I'll walk you through that, but yeah, there's issues with the drug dosing, the run-in period, the double drug run-in period, and there's a little tidbit in the supplement that proves why the drug dosing is
SPEAKER_01very inadequate. In the supplement, they prove that you could have pushed that dose way higher, and it's buried, and we found it.
Live episode from Youtube This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.drvinayprasad.com/subscribe
Elizabeth Holmes/ AI researchers start smoking/ Moderna vax by Plenary Session This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit...
Pope has cancer; Phase 1 trials for AI; Can AI hack cancer trials__1 by Plenary Session This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit...