Understand that I say "first" because I don't know that I won't just volunteer my time to sort through the rest of their data when there is a good moment.
First congratulations for this nugget of gold. Thank you for sharing for free. It's the protocol that I've been sharing with everyone since Dr Peter McCullough first published in Annuals of Cardiovasular Medicine. Truthforhealth.org I am so appreciative for all the frontline covid 19 treating physicians like Tyson and Fareed. They rocked the Imperial Valley Health Supervisory Commitee meeting. I also very grateful for peeps like you for the statistical analyses. Thank you. Quick question- what's up with the rhino horn in the table?
The point of the rhino horn was to help people understand that the failure of an absurd protocol call (like one using rhino horn...or the WHO's protocols) is entirely meaningless in the big picture of whether or not a medicine is effective. If one protocol that employs a medicine succeeds, then the medicine is effective and all discussion of its efficacy should center around optimal protocols, or candidate protocols that are good enough to compete for such a standard.
A lot of people still talk about the WHO trials for HCQ as if they should be taken seriously, and even include them in meta-analysis. It's baffling.
Jan 13, 2022·edited Jan 13, 2022Liked by Mathew Crawford
Rhino horn, lol. I thought so. Good one. Also, No doubt on the point you make. Completely bogus and outright fraudelent South American study that intentional overdosed the human patients and killed them. Then they let it go viral and influence all of western and most likely international medicine and then basically secretely retracted it. They are pathetic and I hope and pray the sheep wake up to this nefarious malfeceance. PS Brilliant Idea for New Journals. McCullough, Fleming, et al have editorial and publishing experience. They should all get together and JUST DO IT.
The problem is that they are all so busy that asking them to either pass through several months of education regarding cryptographic (Bitcoin/cryptocurrency) publishing is a nonstarter, and asking them to trust technologists they don't know isn't exactly an easy task.
Jan 13, 2022·edited Jan 13, 2022Liked by Mathew Crawford
This is a really interesting paper, not just for the science, but for what it represents.
Treat this as a preprint, which is to say that changes can be made where commenters convince us to make changes. We may soon upload it to a preprint server, though publication is less the point than presentation for the public.
Has anyone discussed setting up an alternative / non-censorable preprint server?
If it catches on, this is a way to bypass the old system entirely.
There are floating discussions of creating new journals. I have tried to inject Web 3.0/cryptocurrency/cryptographic standards into the discussion on a basic level and have been ignored so far.
Part of the problem is that prior to the pandemic, these problems all existed and all our institutions were languishing, but there was no discussion anywhere among the old institutions that was possible. Now, as X proportion of them are standing up and saying, "This is all wrong," they are entirely uneducated as to the discussions of technological solutions that have been bubbling up for the past decade. Even worse, trying to engage the conversation sometimes gets me cut out of discussion circles and conversations. The "new good guys" are sadly disorganized in terms of technological solutions and they're facing a firehose of information that they're trying to battle just to get their messages out.
The other day I was in a meeting with somebody whom I'll call one of the "generals" in this battle, and I recommended organizing action as in something like a corporate hierarchy, and then I was surprised that the idea was dismissed out of hand.
I'm not in a position to command or demand leadership among those fleeing the ruins of places that I actively rejected and disrespected when I was young and making my own decisions of where I wanted to stake ground in this world. I have a small realm of influence and tried to build it without actively taking leadership, eventually decided to try to take leadership among those I could influence (while giving them space to work independently). But I am too nonstandard in too many ways to have influence among those who are trying to battle the old system using *primarily* the rules of the old system. Sometimes it's tough to watch.
Another worry that I have is that political action is the old tool remaining for those escaping the old system wanting to battle the old system. Will that approach lead to civil war? It might. Will anyone have that discussion with me? Not so far.
I completely understand. It's very difficult to influence the minds of cogs in the wheel. They seem to roll forward a bit and then roll right back, but they never seem to be able to get out of the tracks. I'm on your team and I am an avid researcher so keep me in your loop of your successes please. I 'd love to help if I can.
If you are motivated to participate, Operation Uplift is always taking new team members. There is so much to do. Email operationupliftteam@gmail.com, please.
Oh yes. Thanks for the reminder. My friend just requested I lighten up on the stream of factual info that I provide to my loved ones so that is a great option for me.
We are not living in the world we knew. The proverbial nuclear bomb has already gone off. The dead are gone, the injured know it, the uninjured don't know and are angry. Then there's the small group who didn't take the shots.
The point is the old institutions are dead, so new ones need to be established to regain public trust and support. Its thinking and acting outside the box time (nonstandard as you say).
We will survive, the question is how quickly do we recover from the old?
I know you were a trader, so treat 2020 and 2021 as end of year tax losses and move forward.
I wish it was like sports where we could take an L but then rebuild for next season and start fresh. But turns out the winners get to keep the spoils and continue to dominate Yankees style.
New journals and publishing infrastructure is a worthy long-term goal... https://www.datprotocol.com and https://ipfs.io have laid a lot of the technical groundwork already... but why not just put your preprint on arxiv.org like everybody does now? Even if they were to censor it, that outrage would be valuable in itself.
I like IPFS, but haven't gone deep into it or used it yet. My hope is to get a group in Operation Uplift at some point that will start testing use of such protocol systems.
Thank you for sharing this study! I’m looking forward to telling everyone in my circle of influence to view it.
I found a typo in the penultimate paragraph of the discussion section. The word “that” appears twice: “ We believe that that the case for early ambulatory care for COVID-19 patients...”
So, this study would receive a "revise and resubmit" from any peer reviewed journal. The main problem is that the appropriate control group is never spelled out. It is true that given the Imperial county, CA data the expected number of deaths from the sample is much higher than that observed. If the patients presenting to All Valley Urgent Care constituted a random sample, this would be a knock down argument for the protocols. However, there is no way that patients presenting to a specific urgent care facility can be considered a random sample. This constitutes a convenience sample. Thus, to make specific claims one has to match the control patients to the treatment patients using something like a propensity score or some other means. To do that, we would need far more demographic information about the treatment groups, especially comorbidities, than is given; we only have ages and sexes in the study. How can I rule out the null hypothesis that this group of patients would likely have gotten better without any significant treatment or with standard treatment because they are generally healthy with few to no comorbidities? The data given in the study does not allow us to rule this out. Thus, this study gives us no reason to trust the hypothesis that the protocols do anything.
Loren, you miss the point of the study because you're stuck with the epidemiologist's tool kit, which is mathematically sparse and often unjustified.
What I have done is such a wide sensitivity analysis that if you aren't imagining that there is a highly perfect comparison group within it (and a statistically significant result), then you're missing the point.
It's not this analysis that is lacking, it's the toolkit if epidemiology.
I'm probably only understanding 80% of this but the question that jumps to my mind is how was the CDC able to cite a study declaring vaccine efficacy 5 times greater than natural immunity based on random patients showing up with covid like symptoms to a hospital. How is this different?
I presume you are talking about this report [1]. With vaccines and natural immunity, there are two things to distinguish from each other. The initial immune response and the memory cell generation. The initial response will involve t-cells and b-cells. The latter produce antibodies to the viral particles in the blood. The former do two things: helper t-cells produce chemicals to elicit other immune cells to respond; killer t-cells initiate program cell death (apoptosis) in infected cells. Antibodies eventually get reabsorbed by the body and the b-cells that produced them cease doing so. However, the memory cells stay with you for a long time, if not your entire life.
The antibodies will decrease the chance of subsequent infections. The memory cells don't doing anything until there is another infection; when you are infected your immune system will work faster on account of the memory cells. So, the paper I cited is comparing recent vaccination to infection that is more than 90 days past. It is looking at how these two affect the chances of subsequent infection and illness. If one currently has antibodies in one's system, infections will be both less likely and milder. The paper says that recent vaccination does better than natural infection that is more than 3 months past. Effectively, we can conclude that a recent vaccination engenders the body to produce more antibodies than will be in the blood from the natural infection 3 or more months ago.
So, it is not so much that the vaccine is 5 times better at protecting you than a natural immunity. It is that something recent (vaccine or natural immunity) is better at protecting you than distant past immunity. So, still makes sense to get the vaccine even if you have been infected and thus have natural immunity.
The CDC offers a summary article of that study that states vaccine immunity is 5 times better...https://www.cdc.gov/media/releases/2021/s1029-Vaccination-Offers-Higher-Protection.html. CNN covered it that way as well I believe. My question isn't the mechanism at all, but how they got their numbers via different cohorts in the hospital. As it strikes me to be open to the same type of definitional issues in terms of a control group as I think you were mentioning as concern with this study.
That's reasonable objection. I did not give the paper an intense read. I only looked at the abstract to get the details. Not sure I shall have time to read it for a week or so; I am moving presently. I shall try to remember to give it a read when the storm settles down.
I need to read your post more closely and try hard to be sure I understand the objections, how the control group is selected for, and do the same for that study. Or I guess I could just listen to the experts.
I think listening to experts and developing an ability to understand what they are saying and why they are drawing the conclusions they are is a valuable tool. Just listening to experts and deferring to them can pose problems.
I think the CDC has shot it's credibility a bit over this pandemic because they were not being completely forthcoming in why they were making the recommendations they were making. Several times, they were trying to be "too clever" by using reverse psychology or trying to anticipate people who might use their recommendations to push an agenda, etc...
For example, the mask issues in the beginning. They said masks don't work, but what they were really trying to do was make sure that hospitals had enough masks. They should have just said that hospitals are a priority for masks and consumers should wait until hospitals had the PPE they needed before ordering masks. They have done many such "too clever" things over the course of this pandemic. It is not good for their credibility.
Likewise, lots of Experts have catastrophized things because they think it is the only way to get people to do things. In catastrophizing they have ignored all the trade-offs that need to be made that are not so simple. It has been very disappointing that our expert institutions and political institutions have assumed the population is full of infants that cannot think reasonably.
It's in the book, but I'll talk about it more here when I get the chance. Loren is wrong, and has done no math to justify his stance. It is by fiat of methodology, which is why epidemiology is a field stuck in 1980, sadly.
First, You have accused me of not having done the work before and called me to publish my work. I did so. You remained silent on my return challenge to you. Why are you making the same mistake again?
Second, your sensitivity analysis, via your synthetic cohorts, is akin to assuming a null hypothesis about hospitalization and death rates in a similar population that does not receive your protocols. This is a reasonable method (it is basically a data imputation method), which is standard in applied statistical and probabilistic modeling; so not sure what you are talking about concerning the epidemiologist's tool kit. However, it is only reasonable if your null hypothesized hospitalization and death rates are reasonable. They are not. Your minimum death rate is 0.1%, which would entail, as you correctly point out, that 21 of your study's patients would be expected to die. However, what if we assume none of your patients have any comorbidities? We know that 92.8% of all covid deaths occur to patients with comorbidities [1]. Thus, the probability of dying if one lacks any comorbidities Is significantly less than if one has comorbidities. Back of the envelope calculations, using the death and case data from Imperial County covid dashboard [2] and the 92.8% rate cited above, the expected number of deaths in a population of similar age to the one in your study, assuming no patients have any comorbidities, is less than 1, 0.50 deaths, this is a 0.01% death rate, an order of magnitude less than your minimum (R code to get this is added below). Since in your patient population, you have more deaths than that we must conclude that the protocols are not causally efficacious, unless we have information about the comorbidities of your patient population in order to collect a more appropriate control.
I hope you will stop accusing me of not doing the math to justify my comments. This is the second time I have presented my math to you. This is what I do for a living. You can either choose to respectfully respond to objections that are offered in the spirit of scientific progress or offer ad hominem, unjustified rebuttals ( for example: "epidemiologist's tool kit, which is mathematically sparse and often unjustified"). However, if you do the latter, you lose credibility and convince serious statisticians that you are a quack.
Thank you Matthew for this work. THIS will be how history will record the absolute failure of public health and the vast majority of those in the medical profession who simply followed orders from on high instead of making patient care their highest priority. It’s just shameful. Thank God for the brave doctors who kept their oath and saved so many lives.
Anyone know yet. . . What cures those? Hopefully the same easily acessable over the counter and herbal products if we can still get them. Let's start thinking about this. Thanks for the heads up.
Tyson and Fareed should look at treatment with 25OHD in moderate and severe patients in order to tamp down inflammation. With adequate levels of 25OHD, inflammation ordinarily gets tamped down by the immune system.
This study should reference and discuss the study by Accinelli, which found treatment initiation with HCQ after 72 hours from symptom onset to be the strongest correlation with mortality.
I preordered this book and just got a notice from Amazon saying, "The release of the item in the order below has been canceled by the publisher and we have canceled your pre-order." Do you know anything about this, Mathew? Censorship?
One question I have is on the hospitlization rate of the untreated. 20% sounds high to me. But is that due to the patients being those that seeked treatment via a Doctor, vs just those that had a positive test?
One thought I have on this is that this strikes me as exactly the kind of work Senator Ted Kennedy was fighting for back in the HIV pandemic days per RFK Jr's book. The idea of formalizing and leveraging Doctor's like Tyson, or groups like the FLCCC. It seems to me the answers to what needs to be changed, needs to be done where figured out but somehow not implemented. Or implemented but then compromised. I don't have a point here other than how do we keep making progress in this direction? How do we implement what was envisioned and ensure it stays in place?
What was the method of diagnosing covid? Distinguishing mild covid from flu is a fool's errand. I think that this problem with any study, including covid vaccine trials, makes the inclusion description very sketchy. For an accurate study inclusion description, I think that you have to look at anyone with ILI symptoms, no matter whether their PCR test is positive or not. False positive percents (1%) are lower than false negative percents (>20%). (False positive percents don't include unculturable positive results, which can be huge.) I think that you have to throw out PCR for clinical diagnosis altogether.
"I think that you have to look at anyone with ILI symptoms"
1. You have begun with a fallacy. There is no "COVID-like illness". Unlike influenza, COVID is a disease, and thus defined by symptoms.
2. AVUC invested $100k in a machine to run PCR. Ultimately, patients were diagnosed the same way the are everywhere else, and that establishes a basic relative match. But some patients may have received positive PCR tests prior.
3. "False positive percents (1%)" This is very far off from all false positive research. It would take pages to explain all this, but false positive rates are not static, and should be expected to surge to high levels when less virus is circulating. Here is some of the math behind that:
On the other hand, false negative rates were likely low early during the pandemic before the viral lineage had exploded and so many new assays were needed to differentiate.
4. "I think that you have to throw out PCR for clinical diagnosis altogether." While I agree that PCR is not great for diagnosis, group comparison automatically corrects for a lot of that challenge because the problem spans all groups. However, AVUC physicians were also using a lot of chest x-rays and symptom lists in addition to testing. Sometimes precision is almost as good as accuracy, but adding an additional approach vector improves every aspect.
5. Note that these critiques apply a large body of research that doesn't necessarily correct as well as we have.
1. Mild covid is indistinguishable symptomatically from influenza. This is a fact attested to by numerous primary care physicians.
2. Mild covid can progress to moderate and severe covid, which have very different symptoms from mild covid and PCR and RAT aren't needed to confirm a diagnosis. But the aim is to prevent progression from mild to moderate covid, so there is a diagnostic problem when there is no progression.
3. PCR and RAT cannot distinguish between mild covid and influenza. It's quite possible to have symptoms due to an active flu infection and test positive for covid due to previous exposure or due to a false positive. Or the symptoms could be caused by covid. There is no way to eliminate or reduce the uncertainty that I know of.
4. False negatives hit a floor when testing occurred 3 days post symptom onset (20%). See "Variation in False-Negative Rate of Reverse Transcriptase Polymerase Chain Reaction–Based SARS-CoV-2 Tests by Time Since Exposure" by Kurcika, et. al.
5. Raoult found that viral load maxed a median of 3 days post symptom onset for mild cases. We would expect this to confirm false negative floor on day 3 post symptom onset.
6. False positive percent...I specifically excluded unculturable positives, which left false positives because of contamination, misreporting, etc. Unculturable positive rates will vary with spread in a population, of course, but that isn't relevant.
7. Using PCR for diagnostic purposes adds noise to the signal which is unnecessary. Basically, you are adding noise, then using a statistical filter to remove that noise. Some signal will inevitably be lost. Just admit that there is a diagnostic problem for mild covid which no one has bothered to discuss yet.
8. Your statistical filter will work, but your problem will be justifying your certainty when using that filter to non-statisticians. "That's not how we do it in infectious disease" will be the response.
Can chest x-rays distinguish between mild covid and influenza?
"Mild covid is indistinguishable symptomatically from influenza. This is a fact attested to by numerous primary care physicians."
This is impossible, by definition, unless you're saying that influenza is an etiological cause of COVID. That neither you nor the unnamed numerous primary care physicians that generate "a fact" recognize this is one of the many why we're in an astoundingly difficult conversation about everything.
A disease is a set of symptoms. That doesn't change. That didn't change. Think it through. Words mean something.
Regardless, the fact does not affect the study in any way. If there is some form of incorrectness in the diagnosing on patients, it is still consistent between cohorts and does not affect the protocol level analysis. And in fact, if there are influenza cases mixed in both at AVUC and in the county population (which I would suspect there are), then we may have the added bonus of once again identifying quinine-derived drugs as treating influenza quite well!
This list isn't rising to the top of my priority list. You've given a citation for a study talking about false negatives in the extraordinarily specific context of incubation periods, though if you meant for that to be the conversation, you did not take the time to make that anything like clear, which is to say that you've set up a difficult waste of time to converse.
Several illnesses produce "influenza-like" symptoms. That's why they are called "Influenza-Like Illnesses", which term you can search in google scholar. So when someone is treating "mild covid", they are really treating an ILI with a positive test result, which test result can only confirm exposure, not the type of illness manifesting the clinical symptoms. Run this by Fareed and Tyson if you doubt me.
This issue isn't a killer for the study, but it will be in the minds of skeptics and should be addressed. Basically, there is noise that we have no filter to remove.
So what is your definition of covid? Is there such a thing as mild covid? What are its symptoms? Can it progress to moderate covid? What are the symptoms of moderate covid? Can moderate covid progress to severe covid? What are its symptoms?
Htf do you know if early treatment works or doesn't work? What do physicians do if mild covid progresses to moderate covid?
I never discussed pre-symptomatic incubation periods--merely progress of covid. Viral load is part of the progress either to recovery or to more severe disease. Once you have symptoms, the immune system has been activated and the virus no longer is in the incubation phase.
Title: "Clinical manifestation and disease progression in COVID-19 infection"
"About 80% of SARS-CoV-2 infections in ambulatory patients manifest as a mild respiratory illness and could usually be managed by outpatient care. About 15% of patients need inpatient care for moderate to severe pneumonia.18 Among the hospitalized patients, the median time from initial symptoms to the occurrence of dyspnea is five days (IQR, 1-10 days), and the median time to be hospitalized is 5 days (IQR, 4-8 days).13 Disease course may show rapid progression to multiple organ failure and even death in severely ill patients."
Typical covid progression in cases resulting in death is from flu-like symptoms (mild) to silent hypoxia (moderate) to dyspnea/ARDS (severe) to organ failure. Sometimes a patient gets a pulmonary embolism and dies instead of progressing to organ failure. Or sometimes cardiac arrest from myocarditis.
If you are unfamiliar with silent hypoxia, you can find it in google scholar by searching on "covid silent hypoxia".
“Among 4,385 individuals sorted in both protocols and three severity levels, combined or excluded from this study, the mean age was 40.5±18.2 years and 12.8% were greater than twenty years of age.”
So, the majority of those treated were (very) young and therefore at minimal risk from the ‘rona. Disappointing sample. Are we upping the tools of the enemy?
First congratulations for this nugget of gold. Thank you for sharing for free. It's the protocol that I've been sharing with everyone since Dr Peter McCullough first published in Annuals of Cardiovasular Medicine. Truthforhealth.org I am so appreciative for all the frontline covid 19 treating physicians like Tyson and Fareed. They rocked the Imperial Valley Health Supervisory Commitee meeting. I also very grateful for peeps like you for the statistical analyses. Thank you. Quick question- what's up with the rhino horn in the table?
The point of the rhino horn was to help people understand that the failure of an absurd protocol call (like one using rhino horn...or the WHO's protocols) is entirely meaningless in the big picture of whether or not a medicine is effective. If one protocol that employs a medicine succeeds, then the medicine is effective and all discussion of its efficacy should center around optimal protocols, or candidate protocols that are good enough to compete for such a standard.
A lot of people still talk about the WHO trials for HCQ as if they should be taken seriously, and even include them in meta-analysis. It's baffling.
The HCQ trials ought to be taken very seriously since they constituted a series of crimes, incl. premeditated murder.
Rhino horn, lol. I thought so. Good one. Also, No doubt on the point you make. Completely bogus and outright fraudelent South American study that intentional overdosed the human patients and killed them. Then they let it go viral and influence all of western and most likely international medicine and then basically secretely retracted it. They are pathetic and I hope and pray the sheep wake up to this nefarious malfeceance. PS Brilliant Idea for New Journals. McCullough, Fleming, et al have editorial and publishing experience. They should all get together and JUST DO IT.
The problem is that they are all so busy that asking them to either pass through several months of education regarding cryptographic (Bitcoin/cryptocurrency) publishing is a nonstarter, and asking them to trust technologists they don't know isn't exactly an easy task.
This is a really interesting paper, not just for the science, but for what it represents.
Treat this as a preprint, which is to say that changes can be made where commenters convince us to make changes. We may soon upload it to a preprint server, though publication is less the point than presentation for the public.
Has anyone discussed setting up an alternative / non-censorable preprint server?
If it catches on, this is a way to bypass the old system entirely.
There are floating discussions of creating new journals. I have tried to inject Web 3.0/cryptocurrency/cryptographic standards into the discussion on a basic level and have been ignored so far.
Part of the problem is that prior to the pandemic, these problems all existed and all our institutions were languishing, but there was no discussion anywhere among the old institutions that was possible. Now, as X proportion of them are standing up and saying, "This is all wrong," they are entirely uneducated as to the discussions of technological solutions that have been bubbling up for the past decade. Even worse, trying to engage the conversation sometimes gets me cut out of discussion circles and conversations. The "new good guys" are sadly disorganized in terms of technological solutions and they're facing a firehose of information that they're trying to battle just to get their messages out.
The other day I was in a meeting with somebody whom I'll call one of the "generals" in this battle, and I recommended organizing action as in something like a corporate hierarchy, and then I was surprised that the idea was dismissed out of hand.
I'm not in a position to command or demand leadership among those fleeing the ruins of places that I actively rejected and disrespected when I was young and making my own decisions of where I wanted to stake ground in this world. I have a small realm of influence and tried to build it without actively taking leadership, eventually decided to try to take leadership among those I could influence (while giving them space to work independently). But I am too nonstandard in too many ways to have influence among those who are trying to battle the old system using *primarily* the rules of the old system. Sometimes it's tough to watch.
Another worry that I have is that political action is the old tool remaining for those escaping the old system wanting to battle the old system. Will that approach lead to civil war? It might. Will anyone have that discussion with me? Not so far.
I completely understand. It's very difficult to influence the minds of cogs in the wheel. They seem to roll forward a bit and then roll right back, but they never seem to be able to get out of the tracks. I'm on your team and I am an avid researcher so keep me in your loop of your successes please. I 'd love to help if I can.
If you are motivated to participate, Operation Uplift is always taking new team members. There is so much to do. Email operationupliftteam@gmail.com, please.
Oh yes. Thanks for the reminder. My friend just requested I lighten up on the stream of factual info that I provide to my loved ones so that is a great option for me.
We are not living in the world we knew. The proverbial nuclear bomb has already gone off. The dead are gone, the injured know it, the uninjured don't know and are angry. Then there's the small group who didn't take the shots.
The point is the old institutions are dead, so new ones need to be established to regain public trust and support. Its thinking and acting outside the box time (nonstandard as you say).
We will survive, the question is how quickly do we recover from the old?
I know you were a trader, so treat 2020 and 2021 as end of year tax losses and move forward.
I wish it was like sports where we could take an L but then rebuild for next season and start fresh. But turns out the winners get to keep the spoils and continue to dominate Yankees style.
New journals and publishing infrastructure is a worthy long-term goal... https://www.datprotocol.com and https://ipfs.io have laid a lot of the technical groundwork already... but why not just put your preprint on arxiv.org like everybody does now? Even if they were to censor it, that outrage would be valuable in itself.
I like IPFS, but haven't gone deep into it or used it yet. My hope is to get a group in Operation Uplift at some point that will start testing use of such protocol systems.
Thank you for sharing this study! I’m looking forward to telling everyone in my circle of influence to view it.
I found a typo in the penultimate paragraph of the discussion section. The word “that” appears twice: “ We believe that that the case for early ambulatory care for COVID-19 patients...”
Thanks for the heads up.
This is amazing, praise God for you and the other authors, thank you for sharing!
please pursue publication
I suspect that this will be published, and there is a long story as to why that hadn't happened yet that will take some days to tell.
So, this study would receive a "revise and resubmit" from any peer reviewed journal. The main problem is that the appropriate control group is never spelled out. It is true that given the Imperial county, CA data the expected number of deaths from the sample is much higher than that observed. If the patients presenting to All Valley Urgent Care constituted a random sample, this would be a knock down argument for the protocols. However, there is no way that patients presenting to a specific urgent care facility can be considered a random sample. This constitutes a convenience sample. Thus, to make specific claims one has to match the control patients to the treatment patients using something like a propensity score or some other means. To do that, we would need far more demographic information about the treatment groups, especially comorbidities, than is given; we only have ages and sexes in the study. How can I rule out the null hypothesis that this group of patients would likely have gotten better without any significant treatment or with standard treatment because they are generally healthy with few to no comorbidities? The data given in the study does not allow us to rule this out. Thus, this study gives us no reason to trust the hypothesis that the protocols do anything.
Loren, you miss the point of the study because you're stuck with the epidemiologist's tool kit, which is mathematically sparse and often unjustified.
What I have done is such a wide sensitivity analysis that if you aren't imagining that there is a highly perfect comparison group within it (and a statistically significant result), then you're missing the point.
It's not this analysis that is lacking, it's the toolkit if epidemiology.
I'm probably only understanding 80% of this but the question that jumps to my mind is how was the CDC able to cite a study declaring vaccine efficacy 5 times greater than natural immunity based on random patients showing up with covid like symptoms to a hospital. How is this different?
I presume you are talking about this report [1]. With vaccines and natural immunity, there are two things to distinguish from each other. The initial immune response and the memory cell generation. The initial response will involve t-cells and b-cells. The latter produce antibodies to the viral particles in the blood. The former do two things: helper t-cells produce chemicals to elicit other immune cells to respond; killer t-cells initiate program cell death (apoptosis) in infected cells. Antibodies eventually get reabsorbed by the body and the b-cells that produced them cease doing so. However, the memory cells stay with you for a long time, if not your entire life.
The antibodies will decrease the chance of subsequent infections. The memory cells don't doing anything until there is another infection; when you are infected your immune system will work faster on account of the memory cells. So, the paper I cited is comparing recent vaccination to infection that is more than 90 days past. It is looking at how these two affect the chances of subsequent infection and illness. If one currently has antibodies in one's system, infections will be both less likely and milder. The paper says that recent vaccination does better than natural infection that is more than 3 months past. Effectively, we can conclude that a recent vaccination engenders the body to produce more antibodies than will be in the blood from the natural infection 3 or more months ago.
So, it is not so much that the vaccine is 5 times better at protecting you than a natural immunity. It is that something recent (vaccine or natural immunity) is better at protecting you than distant past immunity. So, still makes sense to get the vaccine even if you have been infected and thus have natural immunity.
[1]: https://www.cdc.gov/mmwr/volumes/70/wr/mm7044e1.htm?s_cid=mm7044e1_w
The CDC offers a summary article of that study that states vaccine immunity is 5 times better...https://www.cdc.gov/media/releases/2021/s1029-Vaccination-Offers-Higher-Protection.html. CNN covered it that way as well I believe. My question isn't the mechanism at all, but how they got their numbers via different cohorts in the hospital. As it strikes me to be open to the same type of definitional issues in terms of a control group as I think you were mentioning as concern with this study.
That's reasonable objection. I did not give the paper an intense read. I only looked at the abstract to get the details. Not sure I shall have time to read it for a week or so; I am moving presently. I shall try to remember to give it a read when the storm settles down.
I need to read your post more closely and try hard to be sure I understand the objections, how the control group is selected for, and do the same for that study. Or I guess I could just listen to the experts.
I think listening to experts and developing an ability to understand what they are saying and why they are drawing the conclusions they are is a valuable tool. Just listening to experts and deferring to them can pose problems.
I think the CDC has shot it's credibility a bit over this pandemic because they were not being completely forthcoming in why they were making the recommendations they were making. Several times, they were trying to be "too clever" by using reverse psychology or trying to anticipate people who might use their recommendations to push an agenda, etc...
For example, the mask issues in the beginning. They said masks don't work, but what they were really trying to do was make sure that hospitals had enough masks. They should have just said that hospitals are a priority for masks and consumers should wait until hospitals had the PPE they needed before ordering masks. They have done many such "too clever" things over the course of this pandemic. It is not good for their credibility.
Likewise, lots of Experts have catastrophized things because they think it is the only way to get people to do things. In catastrophizing they have ignored all the trade-offs that need to be made that are not so simple. It has been very disappointing that our expert institutions and political institutions have assumed the population is full of infants that cannot think reasonably.
yes, while i (unlike you?) found this really compelling, i too would have liked to see more in-depth discussion of the control population
It's in the book, but I'll talk about it more here when I get the chance. Loren is wrong, and has done no math to justify his stance. It is by fiat of methodology, which is why epidemiology is a field stuck in 1980, sadly.
First, You have accused me of not having done the work before and called me to publish my work. I did so. You remained silent on my return challenge to you. Why are you making the same mistake again?
Second, your sensitivity analysis, via your synthetic cohorts, is akin to assuming a null hypothesis about hospitalization and death rates in a similar population that does not receive your protocols. This is a reasonable method (it is basically a data imputation method), which is standard in applied statistical and probabilistic modeling; so not sure what you are talking about concerning the epidemiologist's tool kit. However, it is only reasonable if your null hypothesized hospitalization and death rates are reasonable. They are not. Your minimum death rate is 0.1%, which would entail, as you correctly point out, that 21 of your study's patients would be expected to die. However, what if we assume none of your patients have any comorbidities? We know that 92.8% of all covid deaths occur to patients with comorbidities [1]. Thus, the probability of dying if one lacks any comorbidities Is significantly less than if one has comorbidities. Back of the envelope calculations, using the death and case data from Imperial County covid dashboard [2] and the 92.8% rate cited above, the expected number of deaths in a population of similar age to the one in your study, assuming no patients have any comorbidities, is less than 1, 0.50 deaths, this is a 0.01% death rate, an order of magnitude less than your minimum (R code to get this is added below). Since in your patient population, you have more deaths than that we must conclude that the protocols are not causally efficacious, unless we have information about the comorbidities of your patient population in order to collect a more appropriate control.
I hope you will stop accusing me of not doing the math to justify my comments. This is the second time I have presented my math to you. This is what I do for a living. You can either choose to respectfully respond to objections that are offered in the spirit of scientific progress or offer ad hominem, unjustified rebuttals ( for example: "epidemiologist's tool kit, which is mathematically sparse and often unjustified"). However, if you do the latter, you lose credibility and convince serious statisticians that you are a quack.
[1]: https://pubmed.ncbi.nlm.nih.gov/34449622/
[2]: https://www.arcgis.com/apps/dashboards/684c52c01a0c4dbda380d7b905ef0b46
Again, here is my R code:
library(tidyverse)
cases_by_age <- c(2852, 5063, 6712, 6616, 5516, 5249, 3772, 1854, 820, 207) # from [2]
deaths_by_age <- c(1, 0, 3, 11, 33, 104, 174, 208, 201, 76) # from [2]
age <- c("0-9", "10-19", "20-29", "30-39", "40-49", "50-59", "60-69", "70-79", "80-89", "90+")
deaths_by_age_noComorb <- (1-.928)*deaths_by_age # from [1]
county <- tibble(age = age,
cases = cases_by_age,
deaths = deaths_by_age,
percent = deaths/cases,
deaths_noComorb = deaths_by_age_noComorb,
percent_noComorb = deaths_noComorb/cases)
study_mild_covid <- tibble(age = age,
cases = c(132,418, 722, 828, 682, 587, 368, 172, 49, 4),
deaths = c(0,0,0,0,0,0,0,0,0,0),
expected_deaths = county$percent*cases,
expected_deaths_noCom = county$percent_noComorb*cases)
expected_deaths_noCom_mildCovid <- sum(study_mild_covid$expected_deaths_noCom*study_mild_covid$cases)/sum(study_mild_covid$cases)
study_mod_covid <- tibble(age = age,
cases = c(2,7,44,58,92,87,74,35,13,1),
deaths = c(0,0,0,0,0,1,2,0,0,0),
expected_deaths = county$percent*cases,
expected_deaths_noCom = county$percent_noComorb*cases)
expected_deaths_noCom_modCovid <- sum(study_mod_covid$expected_deaths_noCom*study_mod_covid$cases)/sum(study_mod_covid$cases)
total_expected_deaths_noCom <- expected_deaths_noCom_mildCovid + expected_deaths_noCom_modCovid
total_expected_deaths_noCom/sum(study_mild_covid$cases, study_mod_covid$cases)
Matthew you are obviously an intelligent and well intentioned guy. However I think Loren makes some good if not great points.
Your opportunity (and in fact your aim) is to produce a paper that is bullet proof to even the most harshest critic.
Take the advice in good faith, if you can produce a paper beyond reproach you may change the world
Thank you Matthew for this work. THIS will be how history will record the absolute failure of public health and the vast majority of those in the medical profession who simply followed orders from on high instead of making patient care their highest priority. It’s just shameful. Thank God for the brave doctors who kept their oath and saved so many lives.
Well done! 👍🏼
Get Ready For The Next One. The Next Plandemic: Smallpox, Marburg, or both?
https://lionessofjudah.substack.com/p/get-ready-for-the-next-one
Anyone know yet. . . What cures those? Hopefully the same easily acessable over the counter and herbal products if we can still get them. Let's start thinking about this. Thanks for the heads up.
Tyson and Fareed should look at treatment with 25OHD in moderate and severe patients in order to tamp down inflammation. With adequate levels of 25OHD, inflammation ordinarily gets tamped down by the immune system.
This study should reference and discuss the study by Accinelli, which found treatment initiation with HCQ after 72 hours from symptom onset to be the strongest correlation with mortality.
I preordered this book and just got a notice from Amazon saying, "The release of the item in the order below has been canceled by the publisher and we have canceled your pre-order." Do you know anything about this, Mathew? Censorship?
One question I have is on the hospitlization rate of the untreated. 20% sounds high to me. But is that due to the patients being those that seeked treatment via a Doctor, vs just those that had a positive test?
One thought I have on this is that this strikes me as exactly the kind of work Senator Ted Kennedy was fighting for back in the HIV pandemic days per RFK Jr's book. The idea of formalizing and leveraging Doctor's like Tyson, or groups like the FLCCC. It seems to me the answers to what needs to be changed, needs to be done where figured out but somehow not implemented. Or implemented but then compromised. I don't have a point here other than how do we keep making progress in this direction? How do we implement what was envisioned and ensure it stays in place?
What was the method of diagnosing covid? Distinguishing mild covid from flu is a fool's errand. I think that this problem with any study, including covid vaccine trials, makes the inclusion description very sketchy. For an accurate study inclusion description, I think that you have to look at anyone with ILI symptoms, no matter whether their PCR test is positive or not. False positive percents (1%) are lower than false negative percents (>20%). (False positive percents don't include unculturable positive results, which can be huge.) I think that you have to throw out PCR for clinical diagnosis altogether.
"I think that you have to look at anyone with ILI symptoms"
1. You have begun with a fallacy. There is no "COVID-like illness". Unlike influenza, COVID is a disease, and thus defined by symptoms.
2. AVUC invested $100k in a machine to run PCR. Ultimately, patients were diagnosed the same way the are everywhere else, and that establishes a basic relative match. But some patients may have received positive PCR tests prior.
3. "False positive percents (1%)" This is very far off from all false positive research. It would take pages to explain all this, but false positive rates are not static, and should be expected to surge to high levels when less virus is circulating. Here is some of the math behind that:
https://roundingtheearth.substack.com/p/the-chloroquine-wars-part-xviii
On the other hand, false negative rates were likely low early during the pandemic before the viral lineage had exploded and so many new assays were needed to differentiate.
4. "I think that you have to throw out PCR for clinical diagnosis altogether." While I agree that PCR is not great for diagnosis, group comparison automatically corrects for a lot of that challenge because the problem spans all groups. However, AVUC physicians were also using a lot of chest x-rays and symptom lists in addition to testing. Sometimes precision is almost as good as accuracy, but adding an additional approach vector improves every aspect.
5. Note that these critiques apply a large body of research that doesn't necessarily correct as well as we have.
These are my facts:
1. Mild covid is indistinguishable symptomatically from influenza. This is a fact attested to by numerous primary care physicians.
2. Mild covid can progress to moderate and severe covid, which have very different symptoms from mild covid and PCR and RAT aren't needed to confirm a diagnosis. But the aim is to prevent progression from mild to moderate covid, so there is a diagnostic problem when there is no progression.
3. PCR and RAT cannot distinguish between mild covid and influenza. It's quite possible to have symptoms due to an active flu infection and test positive for covid due to previous exposure or due to a false positive. Or the symptoms could be caused by covid. There is no way to eliminate or reduce the uncertainty that I know of.
4. False negatives hit a floor when testing occurred 3 days post symptom onset (20%). See "Variation in False-Negative Rate of Reverse Transcriptase Polymerase Chain Reaction–Based SARS-CoV-2 Tests by Time Since Exposure" by Kurcika, et. al.
5. Raoult found that viral load maxed a median of 3 days post symptom onset for mild cases. We would expect this to confirm false negative floor on day 3 post symptom onset.
6. False positive percent...I specifically excluded unculturable positives, which left false positives because of contamination, misreporting, etc. Unculturable positive rates will vary with spread in a population, of course, but that isn't relevant.
7. Using PCR for diagnostic purposes adds noise to the signal which is unnecessary. Basically, you are adding noise, then using a statistical filter to remove that noise. Some signal will inevitably be lost. Just admit that there is a diagnostic problem for mild covid which no one has bothered to discuss yet.
8. Your statistical filter will work, but your problem will be justifying your certainty when using that filter to non-statisticians. "That's not how we do it in infectious disease" will be the response.
Can chest x-rays distinguish between mild covid and influenza?
"Mild covid is indistinguishable symptomatically from influenza. This is a fact attested to by numerous primary care physicians."
This is impossible, by definition, unless you're saying that influenza is an etiological cause of COVID. That neither you nor the unnamed numerous primary care physicians that generate "a fact" recognize this is one of the many why we're in an astoundingly difficult conversation about everything.
A disease is a set of symptoms. That doesn't change. That didn't change. Think it through. Words mean something.
https://roundingtheearth.substack.com/p/the-efficacy-illusions-part-i-the
Regardless, the fact does not affect the study in any way. If there is some form of incorrectness in the diagnosing on patients, it is still consistent between cohorts and does not affect the protocol level analysis. And in fact, if there are influenza cases mixed in both at AVUC and in the county population (which I would suspect there are), then we may have the added bonus of once again identifying quinine-derived drugs as treating influenza quite well!
https://roundingtheearth.substack.com/p/the-chloroquine-wars-part-i
This list isn't rising to the top of my priority list. You've given a citation for a study talking about false negatives in the extraordinarily specific context of incubation periods, though if you meant for that to be the conversation, you did not take the time to make that anything like clear, which is to say that you've set up a difficult waste of time to converse.
Several illnesses produce "influenza-like" symptoms. That's why they are called "Influenza-Like Illnesses", which term you can search in google scholar. So when someone is treating "mild covid", they are really treating an ILI with a positive test result, which test result can only confirm exposure, not the type of illness manifesting the clinical symptoms. Run this by Fareed and Tyson if you doubt me.
This issue isn't a killer for the study, but it will be in the minds of skeptics and should be addressed. Basically, there is noise that we have no filter to remove.
So what is your definition of covid? Is there such a thing as mild covid? What are its symptoms? Can it progress to moderate covid? What are the symptoms of moderate covid? Can moderate covid progress to severe covid? What are its symptoms?
Htf do you know if early treatment works or doesn't work? What do physicians do if mild covid progresses to moderate covid?
I never discussed pre-symptomatic incubation periods--merely progress of covid. Viral load is part of the progress either to recovery or to more severe disease. Once you have symptoms, the immune system has been activated and the virus no longer is in the incubation phase.
Title: "Clinical manifestation and disease progression in COVID-19 infection"
"About 80% of SARS-CoV-2 infections in ambulatory patients manifest as a mild respiratory illness and could usually be managed by outpatient care. About 15% of patients need inpatient care for moderate to severe pneumonia.18 Among the hospitalized patients, the median time from initial symptoms to the occurrence of dyspnea is five days (IQR, 1-10 days), and the median time to be hospitalized is 5 days (IQR, 4-8 days).13 Disease course may show rapid progression to multiple organ failure and even death in severely ill patients."
https://journals.lww.com/jcma/Fulltext/2021/01000/Clinical_manifestation_and_disease_progression_in.2.aspx
An NIH article also may be helpful because it discusses clinical manifestations of covid--mild, moderate, severe, and critical.
Title: "Clinical Spectrum of SARS-CoV-2 Infection"
https://www.covid19treatmentguidelines.nih.gov/overview/clinical-spectrum/
Typical covid progression in cases resulting in death is from flu-like symptoms (mild) to silent hypoxia (moderate) to dyspnea/ARDS (severe) to organ failure. Sometimes a patient gets a pulmonary embolism and dies instead of progressing to organ failure. Or sometimes cardiac arrest from myocarditis.
If you are unfamiliar with silent hypoxia, you can find it in google scholar by searching on "covid silent hypoxia".
“Among 4,385 individuals sorted in both protocols and three severity levels, combined or excluded from this study, the mean age was 40.5±18.2 years and 12.8% were greater than twenty years of age.”
So, the majority of those treated were (very) young and therefore at minimal risk from the ‘rona. Disappointing sample. Are we upping the tools of the enemy?
That was a typo that has been corrected now.