58 replies to this topic

[abelard lindsay]

I posted this over on my Methylene Blue anti-viral thread, but I thought many of you might have methylene blue lying around and would find this useful:

 

https://guerir-du-ca...methylene-blue/

 

 

 

We report the case of a cohort of 2500 French patients treated among others with methylene blue for cancer care. During the COVID-19 epidemics none of them developed influenza-like illness. Albeit this lack of infection might be by chance alone, it is possible that methylene blue might have a preventive effect for COVID-19 infection. This is in line with the antiviral activity of Chloroquine, a Methylene blue derivative.

Both Chloroquine and Methylene blue have strong antiviral and anti- inflammatory properties probably linked to the change in intracellular pH and redox state.

 

[Turnbuckle]

Presently, one in a thousand have the virus in France, so 2-3 not getting it in this group isn't very strong statistics.

[thompson92]

Presently, one in a thousand have the virus in France, so 2-3 not getting it in this group isn't very strong statistics.

 

The mere fact that a significant number of these cancer patients are probably in and out of a hospital and 0 of 2500 (N=2500) patients them have had any significant flu-like symptoms is basically statistical significance that is probably off the charts in terms of a p-value.  Nevermind the fact that against a comparative cohort of the normal population (i.e. people not going to hospitals/healthcare providers) would probably still be statistically significant against the broader population, not just those coming into contact with highly infectious locations like hospitals or outpatient clinics AND having a pre-existing condition.  

 

.

Edited by thompson92, 05 April 2020 - 03:00 PM.

[Turnbuckle]

The mere fact that a significant number of these cancer patients are probably in and out of a hospital and 0 of 2500 (N=2500) patients them have had any significant flu-like symptoms is basically statistical significance that is probably off the charts in terms of a p-value.  Nevermind the fact that against a comparative cohort of the normal population (i.e. people not going to hospitals/healthcare providers) would probably still be statistically significant against the broader population, not just those coming into contact with highly infectious locations like hospitals or outpatient clinics AND having a pre-existing condition.  

 

.

 

 

I see it as the exact opposite. This is a population that would be very careful with their health, thus I would want to see a larger group. In fact, it seems that one patient in the group may have had symptoms of the virus.

 

These patients were contacted by internet (video and e-mail). We had only one response regarding possible contamination by Covid-19 (a moderate flu syndrome). The limits of this type of retrospective investigation are obvious.

 

 

I don't see the response rate, which is also important, or if any actual testing for the virus was done.

Edited by Turnbuckle, 05 April 2020 - 03:58 PM.

[thompson92]

I see it as the exact opposite. This is a population that would be very careful with their health, thus I would want to see a larger group. In fact, it seems that one patient in the group may have had symptoms of the virus.

 

 

I don't see the response rate, which is also important, or if any actual testing for the virus was done.

 

Odds are low that you are going to see something in 10,000 patients, that you already don't see in 2,500 patients.

 

The problem is there is no serological testing for antibodies.  This is really the reason why there is no 'actual testing'.  If they tested these 2500 for antibodies for the virus and you saw they were exposed, then as compared to a population cohort, we might be able to know better.

[Turnbuckle]

Odds are low that you are going to see something in 10,000 patients, that you already don't see in 2,500 patients.

 

 

 

 

It really depends on the national incidence, doesn't it? If it is one in a hundred, then detecting only one possible case in 2.5k is meaningful. If one in a thousand as it is in France, then it's iffy. If one a million, it's meaningless.

[Daniel Cooper]

Presently, one in a thousand have the virus in France, so 2-3 not getting it in this group isn't very strong statistics.

 

 

Especially since presumably cancer patients might be more shut in and trying to avoid infectious exposure anyway, particularly if they are receiving chemotherapy.

[abelard lindsay]

Here's a brief summary of evidence from my other thread and a few new things:

 

"Methylene blue photochemical treatment as a reliable SARS-CoV-2 plasma virus inactivation method for blood safety and convalescent plasma therapy for the COVID-19 outbreak"

 

https://www.research...cle/rs-17718/v1

 

Grifols is running the convalescent plasma transfusion trials for SARS-COV-2 for the FDA; they are using methylene blue phototreatment to inactivate any coronavirus

 

https://www.grifols....geting-covid-19

 

 

Doctor reports case of delayed Hypoxia among covid patients.  Mhttps://twitter.com/...156584273084416

 

Methylene blue is used as an antidote for carbon monoxide and cyanide poisoning because it alleviates hypoxia.

 

For example:  "After Treatment with Methylene Blue is Effective against Delayed Encephalopathy after Acute Carbon Monoxide Poisoning"

 

https://onlinelibrar...1111/bcpt.12940

 

 

"COVID-19: Attacks the 1-Beta Chain of Hemoglobin and Captures the Porphyrin to Inhibit Human Heme Metabolism" .  Carbon Monoxide also attacks Hemoglobin.

 

https://chemrxiv.org...phyrin/11938173

 

Doctor discusses above study on twitter: https://twitter.com/...026602771251201

 

Edited by abelard lindsay, 06 April 2020 - 05:17 PM.

[gamesguru]

?? So it's mediated through a carbon monoxide mechanism?  Seems a bit far-fetched.

 

I'm with Turnbuckle that this is highly speculative at this point.  The statistics aren't convincing.  Meta-analysis hasn't been done.  It's just a white paper.

[Daniel Cooper]

And a point of fact -

 

Carbon monoxide does not "attack hemoglobin".  It binds with hemoglobin, just like oxygen binds with hemoglobin.  The problem is that the CO - Hgb binds far more tightly than O2 - Hgb (over 200 times more tightly).  Once carbon monoxide binds to hemglobin it does not unbind, rendering that hemoglobin useless for carrying oxygen.  This is in no way similar to anything covid-19 does to hemoglobin.

 

This is no more an an "attack" than two hydrogen atoms "attacking" an atom of oxygen.

 

 

 

 

 

[thompson92]

?? So it's mediated through a carbon monoxide mechanism?  Seems a bit far-fetched.

 

I'm with Turnbuckle that this is highly speculative at this point.  The statistics aren't convincing.  Meta-analysis hasn't been done.  It's just a white paper.

 

It's not mediated by carbon dioxide.  Methylene blue is a known to facilitate the cycling of redox reactions and enhance electron transport chain function.  It acts as an electron donor/receiver and facilitates mitochondrial function.  It becomes particularly useful in cases where there is some kind of respiratory insult like cyanide. 

 

Well what do we see in hypoxia/hypoxemia?  An insult to mitochondrial function that can be recovered partially through methylene blue.  It's not about carbon dioxide. 

[Hip]

Presently, one in a thousand have the virus in France, so 2-3 not getting it in this group isn't very strong statistics.

 
You can't go by the official figures for the number infected in a country, as these are vastly under reported.
 
Tomas Pueyo devised a mathematical formula to estimate the number presently infected, which is calculated by multiplying the death toll so far by 800. 
 
So the death toll in France is 10,869 today, and thus by this formula, the number infected will be 800 x 10,869 = 8.7 million, or about 1 in 8 people infected.
 
 
This then makes it very, very unlikely that by chance alone,  nobody in the 2500 cohort of cancer patients would not have the infection.
 
Of course, it depends on the dates that this study took place, because if we go back a few weeks, then there would have been less infected people in France. The study was published on 28 March, so if we go back in time to one week before the study date, on 21 March, we find the death toll was 562, which means the number infected at that time was 800 x 562 = 0.45 million, or about 1 in 150 people infected.

Edited by Hip, 08 April 2020 - 07:34 PM.

[Daniel Cooper]

Anyone can devise any mathematical formula they care to arrive at any number of infected that they wish. They are all irrelevant.  The only way you will ever arrive at a realistic estimate of the total number infected is to do a truly random statistical sampling.  The outcome of all these mathematical approaches are all determined by their front end assumptions, which if we've learned anything over the last three months it is that they are all over the map and almost always wrong.

 

ETA: The 1 in 8 number you arrive at implies that there are a truly astounding number of asymptotic people out there. That may or may not be true.  It is however certainly irrelevant to this conversation of whether people taking methylene blue don't get SARS-COV-2.  What you're really discussing is not the number of people infected, but rather the number of people that are diagnosed with covid-19 disease.  The number of infected asymptomatic people aren't counted on either side of the equation - ie. the %of people taking methylene blue that get covid-19 versus the %of people not taking methylene blue that get covid-19.  Asymptomatic people aren't counted as covid-19 patients in either group.  It is only the people with symptoms that are diagnosed with covid-19 that matter.  So Tomas Pueyo's formula, in addition to being almost certainly wrong, is also irrelevant to this discussion.

 

 

 

 

 

 

 

Edited by Daniel Cooper, 08 April 2020 - 07:29 PM.

[Hip]

Anyone can devise any mathematical formula they care to arrive at any number of infected that they wish. They are all irrelevant.  The only way you will ever arrive at a realistic estimate of the total number infected is to do a truly random statistical sampling.  The outcome of all these mathematical approaches are all determined by their front end assumptions, which if we've learned anything over the last three months it is that they are all over the map and almost always wrong.

 

Yeah, well that's not going to happen until cheap and reliable antibody tests are devised. And even when they do arrive, we will not be able to backdate very easily, ie, we will not be able to go back in time to the period this study was conducted, and determine the number infected in the past.

 

So estimation formulas are all we have.

[Daniel Cooper]

Yeah, well that's not going to happen until cheap and reliable antibody tests are devised. And even when they do arrive, we will not be able to backdate very easily, ie, we will not be able to go back in time to the period this study was conducted, and determine the number infected in the past.

 

So estimation formulas are all we have.

 

 

See my edit to my prior post.  The number infected is not material to this discussion.  We're truly discussing how many people come down with covid-19, not the number of people infected by sars-cov-2.

 

The asymptomatic infected are invisible all the way around in this discussion.  They aren't measuring who was infected, but rather who was or wasn't diagnosed with covid-19.

Edited by Daniel Cooper, 08 April 2020 - 07:33 PM.

[Daniel Cooper]

Yeah, well that's not going to happen until cheap and reliable antibody tests are devised. And even when they do arrive, we will not be able to backdate very easily, ie, we will not be able to go back in time to the period this study was conducted, and determine the number infected in the past.

 

So estimation formulas are all we have.

 

 

Not true.  You could take a cross section with the virus test today and get a sample of #infected versus #symptomatic.  Those relative weights aren't going to vary much with time and will be far more reliable than any hand waving you do without that information.

[Hip]

Asymptomatic people aren't counted as covid-19 patients in either group.  It is only the people with symptoms that are diagnosed with covid-19 that matter.n.

 

The data from the Diamond Princess infected cruise ship showed that the number of asymptomatic people is approximately equal to the number of symptomatic. So there is not a great pool of asymptomatic out there.

 

Thus the we can divide the estimated figure for the number of infected by 2 to get the number of symptomatic infected patients.

 

 

 

Tomas Pueyo's calculation assumes only two figures: that the death rate is 1%, and that the rate of exponential growth of infected people is such that the total number infected people doubles every 6.2 days. From these two assumptions, you can arrive at the result that the total number infected people at any given point in time is 800 times the number of deaths at that point in time.

 

It works like this: from the 1% death rate, you know that whatever the death toll, the number actually infected must be 100 times more.  

 

But, the average time to death is 17.2 days, so people who died today will have caught the infection 17 days ago, on average. And during those 17 days, the number infected will grow by the normal rules of exponential growth. So using an exponential formula, you can estimate the number infected today. Very simple maths. 

Edited by Hip, 08 April 2020 - 07:47 PM.

[Turnbuckle]

 
You can't go by the official figures for the number infected in a country, as these are vastly under reported.
 

 

 

Same thing with people you call up and ask if they have the virus. They don't know. They haven't had the test. And those that didn't reply, did the researchers wonder why? Maybe they were in the hospital. Also, the national rate I quoted wasn't valid for weeks earlier when they were making the calls. It was lower then, much lower. So arguments like this aren't scientific. Not even a little. Real data is needed.

Edited by Turnbuckle, 08 April 2020 - 07:45 PM.

[Daniel Cooper]

The data from the Diamond Princess infected cruise ship showed that the number of asymptomatic people is approximately equal to the number of symptomatic. So there is not a great pool of asymptomatic out there.

 

Thus the we can divide the estimated figure for the number of infected by 2 to get the number of symptomatic infected patients.

 

 

 

Tomas Pueyo's calculation assumes only two figures: that the death rate is 1%, and that the rate of exponential growth of infected people is such that the total number infected people doubles every 6.2 days. From these two assumptions, you can arrive at the result that the total number infected people at any given point in time is 800 times the number of deaths at that point in time.

 

It works like this: from the 1% death rate, you know that whatever the death toll, the number actually infected must be 100 times more.  

 

But, the average time to death is 17.2 days, so people who died today will have caught the infection 17 days ago, on average. And during those 17 days, the number infected will grow by the normal rules of exponential growth. So using an exponential formula, you can estimate the number infected today. Very simple maths. 

 

 

What are you not understanding here?  Asymptomatic people don't matter in that initial article.  They looked at people with "reported sars-cov-2 infections", in other words people that reported symptoms and were diagnosed as being infected.  They didn't take all 2,500 people and give them a sars-cov-2 test.  Had they then you and Mr. Puevo's calculation would matter, but they didn't and it doesn't.

 

For something like this, you must compare apples with apples.  So of that 2,500 cohort that took methylene blue, how many were diagnosed with having a sars-cov-2 infection versus how many would be expected to have been so diagnosed in a similar population that didn't take methylene blue.  That's what you're comparing.  These asymptomatic people you're trying to estimate don't matter since they aren't being counted.  So Turnbuckel's "1 in 1,000" number is basically right, except that is for the French population at large.  What you'd really want to compare with is French cancer patients not taking methylene blue.  So his 1 in 1,000 might be high or low, we don't know.  But your assertion that 125 in 1,000 might be infected, but most not know it, is of no relevance whatsoever because that's not what the initial study was looking at.

 

 

[Hip]

What are you not understanding here?  Asymptomatic people don't matter in that initial article.  They looked at people with "reported sars-cov-2 infections", in other words people that reported symptoms and were diagnosed as being infected.  They didn't take all 2,500 people and give them a sars-cov-2 test.  Had they then you and Mr. Puevo's calculation would matter, but they didn't and it doesn't.

 

For something like this, you must compare apples with apples.  So of that 2,500 cohort that took methylene blue, how many were diagnosed with having a sars-cov-2 infection versus how many would be expected to have been so diagnosed in a similar population that didn't take methylene blue.  That's what you're comparing.  These asymptomatic people you're trying to estimate don't matter since they aren't being counted.  So Turnbuckel's "1 in 1,000" number is basically right, except that is for the French population at large.  What you'd really want to compare with is French cancer patients not taking methylene blue.  So his 1 in 1,000 might be high or low, we don't know.  But your assertion that 125 in 1,000 might be infected, but most not know it, is of no relevance whatsoever because that's not what the initial study was looking at.

 

You are not understanding what I said. I agree that it's only the number of symptomatic patients in France at the time of the study that we are interested in.

 

That's why I said we should divide the number of infected patients calculated by Tomas's formula by two, to arrive at the number of symptomatic patients in France.

[Daniel Cooper]

You are not understanding what I said. I agree that it's only the number of symptomatic patients in France at the time of the study that we are interested in.

 

That's why I said we should divide the number of infected patients calculated by Tomas's formula by two, to arrive at the number of symptomatic patients in France.

 

We need to do no such thing.  We have a reported number of how many people are diagnosed with covid-19.  We don't need a calculation, because they are enumerated.  Those are reported cases.  Just like the number of reported cases in the methylene blue cancer cohort.  

 

This is straight forward.  There need be no estimates here, because we are only considering reported cases. We must compare like with like.  Not "reported cases versus estimated cases".

[Daniel Cooper]

It's so frustrating what's being done with mathematical models and the estimates they generated.

 

You have someone taking a statistically irrelevant sample size (i.e. # people on a cruise ship) of a highly unrepresentative population (i.e. people that take cruises are not like a nation's population at large in all sorts of ways), then applying that data to a model and extrapolating to something the size of a country or even the world.  And on top of that, applying the results in situations where they are not relevant. 

 

This is how you get national death toll estimates ranging from 60,000 up to 2 million, and all reporting high confidence intervals.  We are not being served by the people generating these models (their confidence intervals are absolute bullshit) nor by the media that pick and choose the most sensational. 

 

Until you do a good random cross section, the bottom line is nobody knows much of anything.  It's all hand waving.

 

 But, once again, totally irrelevant to this conversation where we are comparing reported cases against reported cases.

 

 

[Daniel Cooper]

Ok, back to the original article that suggests that methylene blue is protective against sars-cov-2 infection.

 

Based on their logic, isn't it equally valid to conclude that cancer might also be protective against sars-cov-2 infection?

 

 

[Hip]

We have a reported number of how many people are diagnosed with covid-19.  We don't need a calculation, because they are enumerated.  Those are reported cases.  Just like the number of reported cases in the methylene blue cancer cohort.  
 
This is straight forward.  There need be no estimates here, because we are only considering reported cases. We must compare like with like.  Not "reported cases versus estimated cases".

 
What a total load of nonsense. The reported number of COVID-19 cases is a vast underestimate of the actual cases COVID-19, a fact which I am sure even a schoolboy would understand. 
 
You cannot rely on those reported COVID-19 case figures at all to represent the actual situation.



 

It's so frustrating what's being done with mathematical models and the estimates they generated.

 
Is it frustrating because you do not understand the mathematics behind them? Just wondered why the frustration. Those models are all we have at present.
 
Same with climate change, it is all done on models.

 

Until you do a good random cross section, the bottom line is nobody knows much of anything.

Getting such empirical evidence will be great, because we will be able to validate which of the models is the most accurate, by comparing the predictions of the model with the actual measurement of the number of infected and previously infected people.

Edited by Hip, 08 April 2020 - 11:49 PM.

[Hip]

Based on their logic, isn't it equally valid to conclude that cancer might also be protective against sars-cov-2 infection?

 

Unlikely, as preliminary evidence indicates underlying cancer may increase the risk of severe illness and death from SARS-CoV-2. Ref: 1  I guess that may be because cancer is an inflammation condition, and death from coronavirus results from an inflammatory attack on the lungs. So if you have got inflammation going on already, that might be a bad thing from the coronavirus perspective.

Edited by Hip, 08 April 2020 - 11:47 PM.

[Daniel Cooper]

 
What a total load of nonsense. The reported number of COVID-19 cases is a vast underestimate of the actual cases COVID-19, a fact which I am sure even a schoolboy would understand. 
 
You cannot rely on those reported COVID-19 case figures at all to represent the actual situation.



 

 
Is it frustrating because you do not understand the mathematics behind them? Just wondered why the frustration. Those models are all we have at present.
 
Same with climate change, it is all done on models.

 

Getting such empirical evidence will be great, because we will be able to validate which of the models is the most accurate, by comparing the predictions of the model with the actual measurement of the number of infected and previously infected people.

 

Once again, there is no need for models nor estimates for this comparison.  The paper quoted in first post in this thread is about cancer patients taking methylene blue and their reported cases of sars-cov-2 infections.  To do a comparative analysis you must compare like with like. You can not compare their reported cases against some estimate of actual or total cases as you are comparing dissimilar items.

 

Yes, it would indeed be a fine thing if we had a model that derived total cases from reported cases.  The way you do that is with a random sampling and some Bayesian analysis.  You can not do this in any other way as you have no other means to validate your model.  One can not simply conjure a mathematical model out of thin air and declare it to be the truth.  You must have a feedback mechanism to update and correct your model.  That feedback mechanism is random testing.  Anything else is hand waving.

 

I know a bit about this as I do it for a living.  Models without validation are useless.

 

 

[Daniel Cooper]

Unlikely, as preliminary evidence indicates underlying cancer may increase the risk of severe illness and death from SARS-CoV-2. Ref: 1  I guess that may be because cancer is an inflammation condition, and death from coronavirus results from an inflammatory attack on the lungs. So if you have got inflammation going on already, that might be a bad thing from the coronavirus perspective.

 

What they've done is taken a cohort of cancer patients taking methylene blue, noticed that none of them reported having sars-cov-2 infection, and proposed that the reason they didn't report cov-2 infection is because they were taking methylene blue.  In other words, this correlation exists because the one caused the other. Now, I give them credit as they did propose a mechanism.

 

I was being tongue in cheek in noting that the other thing that is also correlated is the fact that they had cancer.

 

The fact is this is a very weak paper.  The sample size isn't large enough to reach any firm conclusions.  This is were Bayesian analysis would come in handy as it would give you a range of expected values for the number of reported sars-cov-2 infections for a cohort that size.  Once you realized that "0" is very likely in the expected range, you'd quickly conclude that this sample is statistically under powered. 

 

About the only thing that you can conclude from this paper is that the effect of methylene blue on cov-2 infection is likely worth further investigation.

[Hip]

The paper quoted in first post in this thread is about cancer patients taking methylene blue and their reported cases of sars-cov-2 infections.  To do a comparative analysis you must compare like with like. You can not compare their reported cases against some estimate of actual or total cases as you are comparing dissimilar items.

 

There's no reason why you cannot use a calculated estimate of the number infected. Why are you saying that you cannot compare calculated figures to empirical ones?

 

Sure, you will always worry that your calculation is out, because the calculation mathematics has not been validated against empirical data for the number infected. But given that we have no such empirical data, a calculated estimate is the best thing we have to go on at this point in time.

[gamesguru]

I find it unlikely there are more than 4x actual cases than reported ones. It's possible up to 5-10% are also false positives.

 

We know France had very few deaths prior to February and a disease does not just spread to everybody over night. Certain areas of New York are reporting nearly 2% of total residents are positive, and I think that is probably pretty close to the reality. People are scared in New York, and most people with symptoms are seeking testing.

 

So yeah, I'd believe the official numbers before some back of the envelop calculation that suggests ~50% of people have been infected.

 

Each country is also experiencing its own reality. Hard to generalize across the globe.

[Daniel Cooper]

There's no reason why you cannot use a calculated estimate of the number infected. Why are you saying that you cannot compare calculated figures to empirical ones?

 

Sure, you will always worry that your calculation is out, because the calculation mathematics has not been validated against empirical data for the number infected. But given that we have no such empirical data, a calculated estimate is the best thing we have to go on at this point in time.

 

Ok, I'm going to try one more time and then I think we've flogged this horse to death.

 

We have two populations to compare here - the general population (Gp) and the methylene blue cancer (Mbc) population.

 

We have a number for the reported cases in the general population.  These are people that were sick enough to go to a doctor/hospital and get diagnosed as having sars-cov-2 (i.e. covid-19).  Not counted are asymptomatic people who's symptoms were very mild or had no symptoms at all.  Hopefully we agree on this so far.

 

What you would have me do is take the reported cases in the general population and adjust them to reflect the total cases (diagnosed + asymptomatic) in the population.  Ok, we can do that. Let's not worry about the accuracy of the model because it has no bearing here.  I'll stipulate that the model is 100% accurate.

 

Fine, we've adjusted the reported cases in the Gp to be total cases.

 

Now, I can't compare total cases in the general population to reported cases in the Methylene Blue Cancer population because that would overstate the difference.  I have to have some way to adjust the reported cases in the Mbc group to reflect the total cases in that group.  If I had that then I could compare the total cases in the Gp group to total cases in the Mbc group.

 

Here's the rub - I don't have a model to do that.  I can't use the model I used for the general population because the assertion is that taking methylene blue altered the dynamics of sars-cov-2 infection in that population.  So that model doesn't apply here.  I don't have a model that takes reported cases for this group and returns total cases.  I could test everyone in the Mbc group for sars-cov-2 infection, but no one did that.

 

The only thing that I can compare is reported cases in the Gp group to reported cases in the Mbc group, because those are the only equivalent metrics that I have.

 

There's nothing more left to be said about this.