35 Comments
User's avatar
тна Return to thread
Reggie VanderVeen's avatar

Math isn't precise?

Expand full comment
Kelly Green's avatar

Reading is fundamental

Expand full comment
Reggie VanderVeen's avatar

If reading is fundamental, the reread that sentence. Science is not precise then?

Expand full comment
Kelly Green's avatar

Science =/= math

Math = precise

Science =/= precise

Expand full comment
Reggie VanderVeen's avatar

https://www.cyberdefinitions.com/definitions/equals_forward_slash_equals.html

=/= means "Not Equal to"

Reminds me of the "the absence of evidence is not equal to evidence of absence of something" truism.

Expand full comment
Nobody's avatar

Science is quite precise with the exception of quantum physics where everything is calculated in probabilities. If science wasn't precise then planes fall out of the sky, bridges collapse and your computer stops working. Or people start dying with mysterious symptoms from dubious medication. Think about the billions upon billions of transistors that had to work just right to get your simple post broadcast around the globe. That's some amazing precision.

Expand full comment
Kelly Green's avatar

Experimentation and comparison of control groups to treatment groups involve probabilities as well. In many fields and situations, one can use sufficient numbers of subjects (n) to get a very precise and reliable result. In other fields, there are practical limitations that don't allow one to do that. Clinical studies are absolutely one of those areas, as the costs of clinical studies often result in compromises. p values from 0.01 to 0.05 are common, implying a 1-5% in the conclusion being wrong. Power calculations usually use 80-90% power to choose an affordable number of subjects (and still result in costs in the tens of millions per study), meaning a 10-20% chance of seeing no difference even if one truly exists, an obviously imprecise result.

Science is further made unreliable by the significant proportion of scientists who fail to understand that statistics don't allow you to accept the null hypothesis, only to fail to reject it. I understand this concept very well and can't explain why it is so difficult for so many, but nonetheless it is a persistent and broad problem. Not to mention that some portion of scientists can't even calculate statistics well at all, and I regularly see statistical errors in the pages of JAMA, NEJM, etc. I'm not talking arguable points, just fundamental misunderstandings of how the stats work.

As Thomas Kuhn showed in the classic text "The Structure of Scientific Revolutions", science is also marked by dogged dedication to dogma. As he showed, the history of science is the history of wrong suppositions firmly held by the scientific community not until a reasonable set of data outweighs the old assumptions, but until the weight of new data becomes so unbearable by the old paradigm that it comes crashing down years after it should. This is now known as "The paradigm paradigm". And by its very nature there is a ton of science today, as has always been the case in every era, that is not just imprecise, it is precisely wrong.

Expand full comment
norstadt's avatar

Do you have any thoughts on ivermectin for COVID-19? The recent ACTIV-6 study, which seemed adversarial but still found slight benefit made me lean pro-ivermectin. I suppose the small result could also be explained by effects like partial unblinding. According to clinicaltrials.gov, ivermectin pills were labeled "123". Placebo pills were supposed to "look similar", but were they labeled "123"? Maybe they weren't, and some enrollees found this info on the web.

https://clinicaltrials.gov/ct2/show/study/NCT04885530

Expand full comment
Kelly Green's avatar

You are my new favorite person!

I was already reasonably pro-ivermectin on the basis that dozens of researchers globally and the public health leaders of Uttar Pradesh are not likely all completely off base simply because they are brown and their countries are less well off, especially in the face of a complete NIH failure to run repurposing studies (until eventually getting around to ACTIV-6 as a 6th priority). "There's no data to suggest that it wor... oh that data, well, there's no GOOD data to suggest it works, and yes, only we can generate good data and we're not running a trial".

And then they did run a trial and it worked. So I think you are right on that the slight benefit definitely should make one lean pro-ivermectin and there are many things that could be dampening the effect including those you mention. Another one is that the study design delays treatment initiation, because this is a remote study where they ship drug/placebo to you. I am talking about the Duke/Vanderbilt study, to make sure we're discussing the same part of ACTIV-6. So, the median day of drug treatment start after start of symptoms was 5 or 6 days. Paxlovid requires no more than four days, and, for influenza A, Tamiflu in their clinical trials found that one had only 48 h from *exposure* to start treatment to receive the benefit.

Anyway, happy to find someone who knows this study!

Expand full comment
norstadt's avatar

Thanks for the comment. I wrote a longer note here, also considering the delay issue:

https://norstadt.substack.com/p/ivermectin-seems-to-have-benefits

Expand full comment
Kelly Green's avatar

Great post.

Two other things: the study is not the last word. The study team saw enough benefit to start an arm with a higher dose and longer duration of treatment (but not start treatment earlier). https://corporate.dukehealth.org/news/study-finds-no-benefit-taking-ivermectin-covid-19-symptoms

The study had two positive findings. One, in the regression model that attempted to control for several factors there was the 0.49 days benefit shown. Secondly, even before controlling for various factors the severe population showed an 80% benefit. They try to wave it away saying the severe population is small (about 50 per arm), but you and I know that the stats control for that and if they are significant, they are exactly as persuasive as a larger sample with equal p value.

And what's an 80% benefit? If the mean time unwell was 11 days, severe patients would have taken longer. Did the treated severe patients get better 10 or more days sooner than the untreated severe? Hard to pinpoint from this data, but something ten days or greater seems to be likely.

Expand full comment
Nobody's avatar

You said that science was not precise enough to prevent drugs from being released which cause more good than harm. Okay, double blind studies may not be 100% effective at keeping the public safe, no argument there. But the studies were good enough to catch problems with Vioxx years before the FDA approved that drug, and I submit that had the study period been longer for the covid vaccines that they would have caught some of the side effects that Pfizer, the FDA and MSM are desperately trying to ignore right now.

Expand full comment
norstadt's avatar

In the randomized trials, the Pfizer and Moderna vaccines caused several times more severe adverse events than cases of severe COVID-19 prevented. That was the most significant harm/benefit conclusion that could be drawn due to statistical limits. The vax seemed like a loser from the start, but science is no match for finance and politics. https://norstadt.substack.com/p/severe-adverse-events-vs-severe-covid

Expand full comment
Nobody's avatar

The vaccines probably provide some benefit to at risk groups. The data in your post is a single study, and the numbers aren't all that large, but even if those findings hold true in more studies you would need to see how the risk/reward ratio works out for different population subgroups.

I believe forcing the vaccine on young healthy adults and children was immoral and driven purely by profit seeking. Covid poses little risk for those groups. And I bet in time it will be proven that forcing young and healthy people to take the shots caused more harm than good.

Expand full comment
norstadt's avatar

Those randomized trials are the only ones there will ever be. I wrote about Pfizer, but Moderna numbers are even worse.

"probably provide some benefit to at risk groups" is based on much lower quality observational data from MDs who don't want to nuke their own profession.

Expand full comment
Kelly Green's avatar

I think the SAE's are probably on-target myocardia like I was describing, but this is a SWAG with <50% certainty. But I thought this in Jan 2022 shortly after the data came out and have the Signal threads to prove it.

If they are myocardia, one would still vaccinate. A clinical trial is a short term study, but in the real world, large numbers of the placebo folks will eventually get actual COVID and get the myocardia that way.

Expand full comment
Kelly Green's avatar

COVID vaccine side effects largely overlap with things that COVID itself does. H1N1 vaccine problems, severe narcolepsy caused in a subset of treated people, were shown to be that way - the virus also caused the same narcolepsy problem. Vaccines replicate sections of the amino acid sequence of the virus and can thus sometimes have the same effects.

Unfortunately, some mRNA vaccines cause the creation of full length spike protein. Spike protein has activity in the human body, causing things like the cardiovascular issues we see. They are caused by the virus AND also caused by the vaccine, because vaccines replicate portions of viruses.

H1N1 vaccine was trouble because it was given to 50% of some countries' populations, when only 10% of those people were going to get H1N1 that year. So it caused broad problems while preventing little. With COVID, pretty much everyone is going to get it so still a net benefit from vaccination even with the side effect profile.

Even autism after vaccines may be an on-target side effect related to what I am talking about. MMR uses attenuated or live rubella - rubella is a known, undisputed cause of autism. It is entirely possible that giving MMR vaccine can BOTH reduce overall population autism and yet cause individual cases right after administration. Even when a scientific data set is very precise - autism rates go down in vaccine-treated populations - society's conclusions from such data may not be. It does not necessarily follow 100% that the vaccine doesn't cause autism in rare cases, it could simply prevent more autism than it causes.

Expand full comment
norstadt's avatar

Mechanistic safety arguments are unconvincing, especially in light of the real-life reports of safety problems.

The distribution in the body of SARS-CoV-2 from natural infection would not be similar to the vax. Then why should effects be similar?

AFAIK lipid nanoparticles are also a new biotech with unproven safety.

Expand full comment
Kelly Green's avatar

The effects are not similar in all ways. For example, the vaccine does not cause shortness of breath and death within days.

However, they both cause myocarditis that is strongly believed to be mediated by interaction with the ACE2 receptor. The spike protein is known to engage the ACE2 receptor, and the spike protein is created intact by at least some forms of mRNA vaccine.

Expand full comment
Kelly Green's avatar

Tell it to OP, I'm translating for people unable to figure out what the message said.

Expand full comment
Kelly Green's avatar

I just realized you wrote "or people start dying woth mysterious symptoms from dubious medication". Familiarize yourself with Vioxx, thalidomide, trovafloxicin, etc.

Expand full comment
Nobody's avatar

Sorry for directing my response to the wrong person. But normally vaccines are developed over a much much longer timeframe (10-15 years) before being approved by the FDA. Science was used to establish a protocol for vaccine testing and release, and it was not followed. The covid vaccines were approved for emergency use, and because nobody knew exactly how dangerous the virus was it can certainly be argued that this was justified. But the continued defense of these impotent vaccines, the lockdowns, and mask mandates is inexcusable. As is the suppression and of information on dangerous vaccine side effects. Covid is now endemic, and even China with it's shall we say authoritarian response was unable to stop the spread.

Expand full comment
Kelly Green's avatar

Not sure why you're referencing COVID vaccines. I didn't bring them up.

Expand full comment
Nobody's avatar

Why am I referencing covid vaccines in a thread about covid vaccines?

Expand full comment
Kelly Green's avatar

No, why in response to "Familiarize yourself with Vioxx, thalidomide, trovafloxicin, etc."

Expand full comment
Nobody's avatar

Why was Vioxx approved and allowed to stay on the market for so long? It was due to a cozy relationship between regulators and industry. That and bad actors who suppressed data showing serious cariovascular risk to people who were taking the drug. One might even say that the science was not followed. Does any of this sound familiar to what's happening today? Don't fault science for the failings of humans.

"Merck also came under fire for promoting rofecoxib even after its scientists were aware of potential serious cardiovascular risks. Dr Gurkipal Singh, adjunct clinical professor of medicine at Stanford University, testified that as early as November 1996 Merck scientists тАЬwere seriously discussing a potential [heart attack] risk of Vioxx.тАЭ

....

Merck's failure to undertake a study of cardiovascular outcomes was a тАЬmarketing decisionтАЭ designed to minimise the possibility of finding cardiovascular adverse events, said Dr Singh. He added: тАЬIt would be better to kill the drug than to kill the patient.тАЭ"

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC534432/

Expand full comment
Kelly Green's avatar

You said, "If science wasn't precise then... people start dying with mysterious symptoms from dubious medication." I pointed out that people do die with mysterious symptoms from dubious medication. Now you have a new take on it, so please revise your original statement and try another phrasing that claims that medical science is precise.

Expand full comment
Nobody's avatar

The scientific method:

1. Define a hypothesis to investigate.

2. Make predictions.

3. Gather data.

4. Analyze data

5. Draw conclusions

This is essentially what happens during double blind placebo controlled studies. You have a drug someone dreamed up. They believe it will help by doing X. You take two groups of people, some get the drug, some get the placebo. To eliminate bias neither the studiy participants nor the people know who gets the drug and who gets the placebo.

Merck did this for Vioxx. Their data showed the drug could cause heart problems. But Merck ignored its own data and got their buddies at the FDA to not ask questions. The drug was approved and many people died unnecessarily. The double blind clinical trials worked but humans ignored the results.

For the covid vaccines, the studies were rushed, small in size, and there wasn't enough time to study long term effects. The established protocol for bringing a new vaccine to market has been used many times and has a good track record, i.e medical science is precise enough to prevent something like this from happening on a large scale or at least significantly reduce the probability. However standard protocols were not followed for political reasons.

Expand full comment
Kelly Green's avatar

I'm going to make that story into a book and call it "People start dying with mysterious symptoms from dubious medication."

Expand full comment
Reggie VanderVeen's avatar

Works for me!

Expand full comment
Kelly Green's avatar

"COVID testing is a scam" seems rather prominent in your recent past

Expand full comment
Reggie VanderVeen's avatar

Not sure if punctuation would have made a difference. Math or science? I'm guessing math is imprecise (which, as we know, cannot prove every theorem out there).

Expand full comment