14

So according to Pfizer, ~39,000 people received a second dose of vaccine/placebo. So about 20,000 got the placebo, and out of that group only 85 got infected in the span of 3.5 months, during the second wave. That's 0.425%.
So they either:
* Proved Covid is not infectious
* Manipulated the shit out of their data
How that fuck is this considered valid analysis?!
If I tried to publish a paper based on such unreliable, statistically meaningless data, I would be rejected without a blink.

Comments
  • 10
    Tldr: control group failure
  • 4
    "How that fuck is this considered valid analysis?!"

    I'll let the scientific community weigh in on that.

    Nothing targeting you here Nicky, but this whole pandemic has had a lot of folks taking rando numbers and running mental simulations and playing armchair epidemiologist and coming up with way big conclusions.

    And then inevitably there's some clarification or the numbers aren't what we thought they were.

    Science is hard....
  • 4
    Let me get this straight, do the results show that the placebo vaccine works? Because that's not new. Placebos have weird behavioral effects, some of which may very well be in line with sensible precautions to prevent infection.
  • 2
    People who want to participate in the test are probably more aware and careful about the virus than average. That's why you need the control group who receive placebo.
  • 0
    Although @SortOfTested 's explanation is much more feasible.
  • 0
    @homo-lorens

    I wonder how you control for 'folks who would participate in a trial'. That's gotta be a group with some commonalities / behaviors that aren't in 'folks who would not participate in a trial'.
  • 5
    @electrineer They actually encouraged the people to socialize more - it was stated. There is no point in picking people that shelter at home to test efficiency of a vaccine.
  • 5
    @N00bPancakes Hi, I am member of the scientific community. Also, since Moderna's results were identical, with a completely different solution, a lot of relevant people started raising eyebrows.
    This has nothing to do with epidemiology - it has everything to do with statistics and sampling bias.
    And if you want to be specific about epidemiology, seasonal vaccines have efficiency of 40-60%. Their number of 90-95% are extremely strange.
  • 0
    @NickyBones

    Is their number super high ... for their group?

    Like the grand numbers form seasonal flu vaccines have played out over huge sample rates so far right?

    Not sure we've seen it shake out for the proposed vaccine yet. I would expect some variation ... let alone that this isn't the flu...
  • 3
    @N00bPancakes Their numbers are super high for a virus that mutates frequently.
    Flu vaccines are administrated to millions of patients every year, thus statistics about its efficiency has more solid ground.
    Declaring efficiency over a sample of 94 people (85 placebo, 9 vaccinated) is a bloody joke.
    Especially when the data distribution itself is not representative.
  • 2
    @homo-lorens I am not sure you can deduce that placebo works - that would be correlation and not casualty.
    What is fairly clear is that the data distribution is problematic, as they conducted the experiment in countries that had a dramatic increase in cases during the months of the trial (e.g US).
  • 1
    @NickyBones do seasonal vaccines have a lower effectiveness because the virus mutates faster? Maybe that's taken into account in that number?
  • 0
    @NickyBones it mutates....but enough to impact a given vaccine that is also new?

    I don't know how we could say that it should or shouldn't do that or how much?
  • 2
    Finding factual information about covid is a monumental challenge in itself. Almost everything I’ve heard and read is contradictory or blatantly wrong, like:
    • “10 million people have died in the US alone!” from our senile would-be president.
    • “There are many cases of covid reinfections!” vs 6 months later: “First reported case of covid reinfection”
    • “Wearing this mask protects me more than anything else” vs “masks do not protect you, they protect others” (both from CDC) vs “masks proven largely ineffective at blocking spread of COVID-19”
    • “Covid can last for up to 6 weeks on hard surfaces” vs “Covid can live for up to 9 minutes on hard surfaces”
    • “Covid can spread up to 900 feet away in the wind” vs “Covid can travel up to 20 feet in the wind” vs “Crowded outdoor areas are safe from airborne transmission”

    The stated mortality rates are also pretty suspect: I’ve seen rates between 2% and 20%, and yet all the case/death statistics I can find put it significantly lower, even in the older age groups. It really feels like overhyping.

    And the worst part: the scariest “facts” I’ve heard have all come from news stations, with nary a retraction.

    It’s hard to say what is real, but I also find it hard to believe this thing is as scary as everyone makes it out to be — especially considering I can only find a few less-scary statistics that corroborate one another. The whole thing feels like a hypetrain wreck.
  • 4
    E.g. the US have reached 11 mio total cases of Corona since the beginning of the year, which is 11.5 months. That would be 3.3 mio cases in 3.5 months. With 330 mio inhabitants, that's 1%, which isn't out of line with the 0.425% from the study, especially when considering the large error range due to small sample group size.

    Or counting Germany and taking only the last 3.5 months, that's 600k cases out of 80 mio people, or 0.75%. Also in line with the 0.425%.

    I don't see the issue with the number that the study reported.
  • 1
    @electrineer It depends not only on the speed of mutation, but also how much each mutation affects the efficiency of the vaccine. Seems like a very challenging variable to quantify.
  • 0
    @Root

    I feel like the caveat needs to be

    "This is a new virus, all the other certainties about viruses are about viruses we understand / know well comes from decades of study ... this is not that".

    And there ya go.

    Like the scientist who comes up with numbers that indicate a death rate is X also the guy who sees someone else's numbers that are Y "hey that's interesting"... meanwhile and they go on doing science to investigate further.

    Anyone who wants certainties will have to wait for like 3 to 5 years I expect.
  • 1
    @Fast-Nop But the spread was not equally fast in the beginning of the year. 7 million out of the 11+ happened since the end of July (when the trial began). So that would put your calculations at 2% already.
    And you are missing the most important factor - the participants in the experiment were all tested, while the population of the US was not. The % of infected among the experiment participants should have been remarkably higher.
  • 4
    @Root I actually agree about the overhype. Panic sells papers, and unfortunately attention-starved "researchers" are using this to gain their 15 minutes of glory, fueling the panic further.
  • 5
    @N00bPancakes The docs havn't been lazy either and found out quite a lot. With the benefit of hindsight knowledge, a lot of patients were mistreated.

    40% of the death cases were caused by pulmonary embolism. Administering anti-thrombosis medicaments preventively would have been effective, e.g. the well-established heparin.

    It also turned out that attaching respiration machines made matters worse, and a lot of patients were attached way too early.

    With such improvements in progress, the death rate numbers from March aren't what we should expect today. It's also dubious to compare that to other diseases where we already have the experience what the best treatment is.

    For severe diseases in general, this knowledge has always been gained from the corpses of patients who had to be treated before we had that knowledge. That's just how science works - make assumptions, test them, improve your assumptions.
  • 2
    @NickyBones The numbers from Germany, the 0.75%, were from the last 3.5 months, i.e. including the second wave. I subtracted the total count from beginning of August from the total count now and divided by the German population number. And they've ramped up the tests here a lot because we actually have a health care system in place.
  • 1
    @Fast-Nop

    Indeed. It's a moving target and will be to some extent for a long time. Even if / after a magical vaccine would do away with it.
  • 4
    @Fast-Nop My mom (a doctor) says the recommendations issues initially by the WHO were contradicting to everything they usually did for ICU patients, and her hospital decided to go against the new guidelines and stick with they know. Saved a lot of lives....
  • 4
    @NickyBones One big thing is that you lie on your back when the respiration machines are attached. Problem: most of your active lung area is on the backside of the lung. Lying on the back basically suffocates you in your own lung liquid from the infection. That's why you better lie on the front.

    There was a video from an Australian doc early this year where he explained that, along with a breathing technique that is useful for lung infections in general:

    https://youtube.com/watch/...

    None of that will work when attached to a respirator.
  • 0
    @NickyBones

    That doesn't seem like a bad thing for a new disease... that's always how experimentation works historically with medicine.

    Problem is we (understandably) want certainties ... that's just not how it works.
  • 1
    @Fast-Nop Yes, but the trial data is not based on Germany. 30% of it is based on the US, which as I mentioned experienced a dramatic peak in this months. The other countries are Brazil, SA and Argentina - all which suffered a huge increase of cases since July.
  • 4
    Did you see the phd mathematician looking to get his 15 minutes by proving election fraud, but his entire team failed to run the same analysis for the antipodal position?

    Was hilarious, rookie mistake subtracting percentages which inverted the gradient of the plot.

    https://kabir-naim.medium.com/dr-sh...
  • 0
    @NickyBones So did Germany. We went from 211k cumulated cases on 2020-08-01 to 816k cumulated cases on 2020-11-16. That's a pretty dramatic peak.
  • 3
    @Fast-Nop My mom said in general, that respirators are used when the patient has a mechanical problem expanding and collapsing the lungs. Which was not the case in most Covid patients - their issue was poor gas metabolism since the virus damages the cells that carried oxygen and got rid of CO2. That's outside my area of expertise by a light year, so I am just iterating what she explained to me (according to the literature they rely on).
  • 0
    Shits weird eh?
  • 1
    @Fast-Nop Alright. But for the US, and other countries the % of infected in population during that time, is way higher than in the trial population, even though it should have been the opposite due to more intense testing.
    Therefore, trial data is suspicious.
  • 0
    @NickyBones wouldn't poor metabolism of oxygen lead to (easier) exhaustion and thus potential issues physically breathing?
  • 2
    @SortOfTested That's hilarious and depressing. In my family we always say, don't trust statistics you didn't fake yourself :)
  • 3
    @AlgoRythm Possibly, but putting that poor person on a respirator won't make him better since he still can't process the oxygen. It will just increase his chances to die from complications, or from the faulty treatment of inexperienced staff (the majority of nurses are not sufficiently trained for this).
    Oxygen mask does the trick less invasively.
    I think they gave the patients some blood products to make up for the damaged cells.
  • 1
    How accurate are the tests they are using ?

    I'm reminded of:

    --------------

    The new test looks for T-cells, a type of white blood cells that binds to and kills viruses, to determine if someone was infected with coronavirus in the past

    In a pre-print study, researchers tested at samples from 2,200 people in Vo, Italy, with the T-cell test and with the antibody test Of the 70 people who had confirmed cases of COVID-19, the T-cell test correctly identified 97% of cases and the antibody test correctly identified 77% of case Of the more than 2,000 people who tested negative for the virus, the T cell test also returned positive results for 45 people

    --------------
  • 1
    I find it very hard to find any real facts to base my hypothesises on.

    I know in one area for sure of 43,000 people, 14 died of covid-19, all in care homes.

    This is an area with a high tourist factor pre-covid 19, and a high everyone is running here to escape the cities, post-covid 19 factor..

    With plenty of overweight, ill folk, so how come it was just a few old folk in care homes that died ?

    And, how come no one tells us, anything about those folk, their ages, gender, etc. so we can begin to look for patterns..

    At the moment I can only imagine they was very low in vitamin D..

    Plus, very old, and fragile health wise, eg. almost at the end of their lifespan as it was..

    I have little data for other areas..
  • 3
    @Nanos Hospitals here in the States receive extra funding for each patient who tests positive for covid (or dies from it), so it makes financial sense for them to use tests that have a higher false-positive chance, and more moral sense to use those same tests instead of ones that are more accurate but can give false-negatives instead. It also makes financial sense for them to
    report covid as the cause of death instead of some underlying condition. So even “valid” statistics from hospitals are apparently quite skewed.
  • 0
    @Root even if they would want more money for that, there is no evidence of an fradulent reporting.
  • 1
    @stop It's not wrong reporting. It's deliberately using tests with more false positives and reporting these results correctly.

    And counting people who died with Corona - including people who also have cancer and whatnot.
  • 0
    @Fast-Nop thats standard procedure, since not every illness is the direct cause of death.
  • 0
    and with the test, there is a balance between the exactness, speed and work required. an pcr test has nearly 100%, but it needs time and work, fast tests are not exact, but ypu can test many people in a short time. an hospital that test with an fast test can use the exact test capacity without beeing overwelmed.
  • 0
    @stop “There is no evidence of fraudulent reporting” vs “Hospitals in New York have been using tests known to give false positives” vs “Tens of thousands of covid cases reported before tests were available.” I’ve seen countless articles, etc. claiming all of the above.

    So, who even knows what’s going on?

    @Root’s axiom: Laziness and corruption are always more likely than effort and honesty.

    If it’s easier or more beneficial (and within their morals) to use tests with higher false-positive rates or to assume causes, people probably did.

    Also: accurately reporting inaccurate numbers makes for inaccurate statistics. Doesn’t matter why they’re inaccurate, or if the underlying reason is beneficial. They’re still inaccurate, and muddy the data.
  • 0
    I can’t speak to national numbers. And wouldn’t trust them anyway.

    But my local hospital is above capacity with 110 COVID patients, they are turning COVID patients away, and 5 people died there from COVID last week alone.

    This is in a small Midwestern town of about 20,000 people.

    So the “real numbers” are “we are above capacity heading into winter and are officially fucked”.
  • 2
    And then we have reports like this one (German only): https://aerztezeitung.de/Wirtschaft...

    TL;DR translation:
    - clinic sends in tests
    - gets 60 positive results in one day
    - clinic grows suspicious
    - turns out: lab had screwed up, only two were actually positive
    - nobody knows for how long the lab had been screwing up the tests.

    From people who work at that specific lab chain, I happen to know that they are using PCR. While PCR is very exact, it's also very sensitive to mishandling.
  • 1
    Infection rates around the world have varied wildly between the July date that this trial started, and today, where large chunks of Europe are seeing extremely high infection rates.

    Depending on the timing and distribution, the number of infected reported is consistent with the infectivity of the virus.

    Depending on the testing schedule, there could well be a larger number of unconfirmed infections in the sample, but with the assumption that there are not total idiots running the trial, they'll be comparing within homogeneous populations.

    If they are claiming >90% effectiveness and doing Bayesian analysis, you could work backwards from assumptions about the credible interval associated with the ratio of the Beta distributions corresponding to the observed case rates in the treated and untreated populations. Short of fraud, the sort of relative risk claimed could easily be determined by a total of 85 cases, because the ratio is so skewed.
  • 1
    We'll see the actual data in time, when the trial finishes and everyone pours over it. But I wouldn't rush to claim fraud, when the claimed data is quite explicable.
  • 0
    @HiFiWiFiSciFi

    > 5 people died there from COVID

    > last week alone.

    Any details on the people that died ?

    I wonder if the new strain out in the wild (I think there are 3 or 4 now..) is more lethal than the first one ?
  • 0
    @halfflat

    > We'll see the actual data in time,

    We might not !

    I asked for similar data about flu, and I can't find out local death figures, and apparently, no one spends much money studying it, so no one can answer my questions on it !
  • 0
    I'm a little confused..

    We can't get a vaccine for all flu types, because its constantly changing..

    But isn't Covid-19 like that too ?

    And yet, we have several vaccines for it..

    Is that just for the first strain, or all of them ?

    I thought there was just 4 strains, but it seems there are at least 6..

    https://sciencedaily.com/releases/...

    > The six strains of SARS-CoV-2

    Related link:

    https://who.int/emergencies/...-(covid-2019)-and-the-virus-that-causes-it

    ---------------

    Disease

    coronavirus disease

    (COVID-19)

    -----

    Virus

    severe acute respiratory syndrome coronavirus 2

    (SARS-CoV-2)

    ----------
  • 0
    The Spanish Flu disappeared. The CDC had to go looking for samples of it in Inuit graves in Alaska so they could reconstruct and study it in isolated lab conditions.

    In 102 years, COVID will also have disappeared or we will have adapted immune systems that can deal with it just like all the other strains of Coronavirus.

    Also, the CDC holds patents on this strain of Coronavirus. That puts them in shaky legal territory because a) you can’t patent natural organisms and b) if you do, and it causes harm, you’re liable under bioweapons laws.
  • 0
    @Nanos I guess they are using PCR?
  • 1
    @stop My in-house doctor says: PCR isn't false positive. It means that you have some amount of the virus in you. But if the viral load is very low then you are most likely to be asymptomatic, or it is leftovers from a disease you fought off.
    Basically, a positive PCR test doesn't mean you are currently sick. So automatically writing it as comorbidity is wrong in this case.
  • 1
    @NickyBones PCR can be false positive, see the mention about how the lab produced such test results. You need very thorough clean room conditions, and many commercial labs fail to maintain that standard consistently - and then you have some cross contamination. Though of course not always as drastic as in the article (which is why it's worth a news article in the first place).
  • 2
    @Fast-Nop Every test can be contaminated. You need to have some standards and not shit where you eat.
    PCR is extremely sensitive, and people confuse it with a false positive. The test works incredibly well if done correctly. You just need to be careful with interpreting the results. Positive doesn't necessarily mean sick.
  • 2
    @NickyBones That's the point - commercial labs need to compete on cost and do the bare minimum in maintenance because that doesn't generate direct income. I know some people in that domain and can tell you that there's quite some difference between the precision that the tech theoretically allows, and what's actually going on.
  • 1
    @Fast-Nop Yes, it makes sense that greed lads to negligence.
  • 1
    I'm reminded having worked in a place where we was supposed to wear gloves to prevent sample cross contamination, but the only time anyone really wore gloves was when the TV cameras was there to film us !

    And this was in a world beating institution..
  • 1
    @Nanos So you're famous?
  • 0
    @Nanos A navy admiral!
  • 2
    I have some interesting data on the virus from my own experience. Currently my wife's mother is on ventilator and has been for 3+ weeks. She had underlying health conditions mainly kidney issues. Plus I dont think florida's healthcare is remotely good.

    Almost my entire immediate family has been infected since last week. My sister 35yo no health problems lost sense of taste smell. My 1yo nephew has worst symptoms with breathing lung issues. Her husband has body aches. No one has needed to go to hospital in 10+ days of testing positive.

    My older brother 46 is in same house. He is either asymptomatic or just hasnt caught it even though he is around them 100% of time. 12yo neice same. Either asymptomatic or the virus isnt latching on. My 65 yo mother has been around them as well when they were symptomatic and has yet to show any symptoms 10+ days in.

    So it is very interesting.
  • 1
    Also, so far the people in family that haven't caught it or possibly asymptomatic all have type O+ blood type. The two with worst symptoms my nephew and my sister have type A- blood type. So seems the correlation data regarding blood type I've been reading seems legit.

    Albeit the outlier is my wife's mom who has type O+ is on ventilator with grim chance of making it. But i honestly think that has more to do with the horrendous healthcare system in the state she is in and underlying conditions.
  • 1
    @superposition The canton where I live is bordering on Lombardy (where all the covid actions was). We didn't hear of anyone from my uni that got infected, or at the institute where I work. However, they did an antibody test around June, and 11% of population they tested has been exposed. We had a second wave since then, so I guess the numbers now are easily 20%+.
    Hospitals are holding on so far, so I assume most people had it easy.
    The picture is painted so differently between countries that it's hard to grasp it's the same virus.
    Switzerland opted for no lockdown or extreme measures for the second wave, with the exception of Geneva who did a partial lockdown. The spread seems to be dying out in last 2 weeks. On the other hand, you have numerous experts swearing their lives that without lockdown we are all going to die.
  • 1
    My favorite plot is the R factor plotted against the measures, showing pretty much no correlation. The down trends started way before any restrictions were added, and often the R reached a value of 1 2-3 days after they enforced some harsher measures, which is clearly too early to actually be linked.
    https://ibz-shiny.ethz.ch/covid-19-...
  • 0
    @NickyBones At the beginning, there was some number of people taking measures themselves before formal restrictions were imposed.

    But that aside, that effective R model uses a 3-day (iirc) mean of the point estimate, and not one that attempts to incorporate longer correlations.

    Using the piecewise model does enforce that, of course, but at the expense of throwing away any transient effects. For the purpose of evaluating intervention measures, it seems to be the better approach.

    Regarding linkage between estimate R and measure introduction, the model already includes a component that accounts for infection delay: you can see the effect by selecting 'show estimated infection times (deconvolution)'.

    The correlation is not nearly as bad you claim.
  • 1
    @halfflat I know it includes the delay. Still it got to 1 before the lockdown started. And it got bellow 1 now without a lockdown at all.
    If people show personal responsibility and it works, then there is no need to enforce extra restrictions.
  • 0
    @NickyBones that's a pretty big if given some people
  • 0
    @electrineer Factually, in some countries people are responsible enough.
Add Comment