#COVID19 Through A Glass, Weekly

Guest Post by Willis Eschenbach

A couple of days ago, I got to looking at the daily record of US deaths from the coronavirus. It’s shown in Figure 1 below:

Figure 1. US daily deaths. Created on May 5, but shows May 4th data.

So … have the US deaths peaked, and if so when? Hard to tell. However, I looked at that graph in Figure 1 and I thought “It looks like the data might be reflecting lower counts on the weekends”. 

Now, my go-to method for determining the existence, period, and amplitude of underlying repeating cycles in a dataset is the curious method called “CEEMD”. That stands for Complete Ensemble Empirical Mode Decomposition. I discuss the method here. It is a way to decompose a signal into underlying signals. It’s called “complete” because when you add all the underlying signals back together, it gives you back the original signal.

Once all possible underlying cycles have been removed from the data, what remains is called the “CEEMD Residual”. This residual is an excellent indicator of the overall trend of the data. Here is an overview of the CEEMD decomposition of the daily deaths data shown in Figure 1.

Figure 2. CEEMD complete decomposition of the data shown in Figure 1. The top panel is the raw data. Panels C1 to C4 are the empirical modes. Finally, at the bottom is the CEEMD residual.

As you can see, two of the four empirical modes (C2 and C4) are weak, with very low amplitude. Modes C1 and C3, on the other hand, show a much stronger signal. We can see the periods and strengths of each of the empirical modes C1-C4 in Figure 3, which shows the periodogram of each of the empirical modes C1-C4.

Figure 3. Periodograms of each of the empirical modes shown in Figure 2. The strongest signal is the seven-day signal, showing that my guess about weekends was likely correct. There is also a significant amount of energy in the first overtone of the 7-day signal, with a period of 3.5 days.

So … how does this analysis work out in practice? Here is the same data as in Figure 1, along with the CEEMD residual.

Figure 4. US daily deaths, along with the CEEMD residual. Data from May 4th, analyzed May 6th.

Well, I’d have to say that that looks like good news … it would be excellent if we were indeed 20 days past the peak.

And here is a look with the underlying 7-day signal overlaid on the daily data.

Figure 5. As in Figure 4, but overlaid with the seven-day empirical mode signal (Mode C3). The overlaid empirical mode C3 is shown for illustrative purposes only. You can see that when the empirical mode is added to the residual it will be a good match to the data.

This is a most interesting result. It shows one of the reasons that I use the CEEMD analysis—it breaks the raw data down into meaningful underlying signals. In this case, early in the spread of the virus at the left-hand side of the graph, the 7-day signal (blue line) was quite small. But now that there are a large number of deaths the 7-day signal is much larger. It is this kind of a result that is unobtainable by say standard Fourier analysis.

Finally, I prefer the CEEMD residual method over say a Gaussian smooth because it goes all of the way out to both the start and finish of the data. Not only that, but the information out near the ends is meaningful. Here’s a comparison of the CEEMD residual with a Gaussian filter.

Figure 6. Daily US deaths, CEEMD residual, and 7-day Full-Width to Half Maximum (FWHM) Gaussian smooth of the data. This is data from May 4th, processed on May 6th. Treatment of the Gaussian smooth near the endpoints is discussed in the Appendix here.

As you can see, the Gaussian smooth is high at the start of the daily deaths data, and low at the end of the data. The Gaussian smooth is dropping at the right-hand end, and the CEEMD Residual is turning upwards.

And two days later, here’s the situation:

Figure 7. More recent data, from May 6th, daily deaths and CEEMD Residual

At the right-hand end of the graph, the CEEMD residual was already foreshadowing the turn from decreasing to increasing, at the same time that the Gaussian smoothing was wrongly indicating a further decrease (see Figure 6). As I said, the CEEMD residual contains important information out at the ends.

Conclusions? Well, my first one would be that attempting to analyze coronavirus death data without removing the repeating weekly variations is … well, I’ll call it “overly optimistic” and leave it there.

My next conclusion is that the CEEMD residual is an excellent indicator of the ever-changing and oft-deceptive central tendency in time series data. 

Next, about a week ago the CDC changed its guidance on the reporting of deaths involving the COVID virus. Rather than make an explicit distinction between deaths WITH coronavirus and deaths FROM coronavirus, they said to enter COVID-19 on the death certificate if the physician SUSPECTS that the coronavirus MIGHT have CONTRIBUTED to the death … “suspects the virus might have contributed” to the death??? Could they possibly be more vague?

The size of the effect of this change on the way the US reports the death count is unknown, but it can only increase the purported count, not decrease the count. As a result, we cannot be sure that the increase in deaths is real and not just a change in reporting

Finally, it appears that the US has peaked in terms of daily deaths. Might be another peak to come, might be two more peaks, might be no more peaks, but in any case but it appears we’ve passed the first peak.

Stay well, dear friends. When I was a young man, an old geezer (who was likely about my age now) told me “Son, when you have your health you have everything!”

But back then, I didn’t understand …

w.

PLEASE: Quote the exact words you are discussing in your comment. This avoids endless misunderstandings and problems.

234 thoughts on “#COVID19 Through A Glass, Weekly

    • I have been suggesting for some time that we should be looking at the EU countries which have been serving as crash test dummy for confinement and now relaxation of confinement.

      Two weeks ago I published an analysis of Italian case data which shows a very similar and surprisingly regular weekly cycle. One interesting factor is that the amplitude decreased by at least 50% when initial loosening of confinement came in on 14th April.

      https://climategrog.files.wordpress.com/2020/05/2019-ncov-weekly-filter-italy-1.png

      Since I was looking at the rate of change in daily case numbers ( d/dt of daily new cases ) , I used a light 1-2-1 binomial to remove some daily noise. I then detected the peaks and troughs of the weekly cycle and plotted the midpoint ( a bit climatologists use average of Tmin and Tmax ). This gives a similar extraction of the underlying trend similar to Willis’ CEEMD residual . Full description and code here:
      https://climategrog.wordpress.com/2019-ncov-weekly-projection-italy-2/

      This method turned out to be very sensitive to change and picks up a slight slowdown in the rate of fall due to the initial deconfinement ( the rate of change remained negative but moved a little closer to zero: ie constant daily new cases ). This seems to have settled back down to that same level as under full confinement orders.

      This is rather surprising since all the worlds expert modellers were promising a “second wave” as the spread of infection smothered by restrictions starts to gain ground again. The evidence from Italy shows no measurable difference.

      Italy has gone for a major return to work on 4th May, the effects of which should be visible in the next few days.

      Spanish case data is a little less regular and at least the ECDC version of their data has a stupid break where it shows negative 1400 new cases on one day !! Some clumsy ‘correction’ of totals for some reason.

      However Spain, which put factories and construction back to work on 14th April is still showing firm decline in new cases.

      All this indicated that there are other factors in play that are not included in the models which have so far been allowed to dictate international response to COVID-19. This may mean either wider spread via non symptomatic cases or possibly some inherent herd immunity conferred by exposure to other types of corona virus.

      In any case we should be taking more notice of observational reality and less notice of on validated computer models …. once again.

      It’s interesting that Willis’ totally difference method is exposing a similar pattern.

      • I’m not surprised since in most of those countries, the daily death peak happened one or two weeks before a lockdown could have any effect.

        I even wonder, with regard to some rules that have been applied in France (and in Italy as far as I know), if the effect could have been to increase the infection spread.

        For example, in France, we all had to stay at home (in a confined environment where the virus can infect the most) and all had to go only to a very few opened “essential” stores, again increasing the odds of being infected.

        This remotely reminds me of the weak (or strong) connected graph theory :
        – by closing most of the stores and alternatives ways of shopping or feed ourselves haven’t we just strengthened fewer links between fewer but stronger infection nodes (homes, stores) ?

        By applying a strict lockdown “à la Française”, haven’t we just transformed weak infection connected graphs in stronger ones ?
        Thus how can we make the assumption that the global effect could be even noticeable and if yes, in which direction ?

        Rules that may have increased the infection spread in France by strengthening the infection spread (by suppressing weak links and nodes while strengthening others) :
        – ban of “non essential” shops,
        – ban on walking in (closed) parks in the forest or in the woods,
        – ban on jogging between 10 a.m. and 7 p.m.,
        – ban of cycling for leisure or sport purposes.
        – ban of beaches

        I just wonder :
        – where is the science behind all this madness ?

        • I even wonder, with regard to some rules that have been applied in France (and in Italy as far as I know), if the effect could have been to increase the infection spread.

          I think “the science” was the computer models but in this case you can see a distinctive change in direction in the Italian graph about 10d after confinement. That is about the right timing for attribution to lockdown and certainly they were not past the peak before restrictions came in, they were still firmly in near exponential growth.

          Macron introduced on-size-fits-all restrictions across France which was totally unnecessary, since 3/4 of the country, as in US, is barely affected. If this may not have been evident at first it did not take 2 months of self-destruction to work out.

          He could have got half the country back to work after one month. Much of France is rural and there is nothing short of authoritarian spite preventing citizens from walking in the local woods where they would have been alone and effectively isolated.

          As elsewhere in the West, this seems to have been a bank bailout in disguise. The shutdown businesses now need loans to see them through. The “crisis” means credit rules get changed to allow new loans and banks invent money on to the balance sheet. Help to businesses in France has been done via insurers ( who is auditing how much of that money actually gets to small businesses and how much gets held back ? ).

          Bottom line, loads of tax payers’ money get pumped into banks and insurance. The rest is an exercise is totalitarian population control.

          • In Spain we’re still not allowed to walk in the woods. People were confined to home, work or essential shops until the end of April. Now we’re allowed to go out for a walk or exercise once a day, for up to 1 hour, up to 1 km from home. The police keep enforcing these rules and fining people.
            This means all the runners, cyclists and walkers of each city or town are concentrated on the streets and on a few paths in the outskirts instead of spread out over several square km where they would usually go.

        • the daily death peak happened one or two weeks before a lockdown could have any effect.

          Sorry. Totally incorrect. Deaths peak after cases peak ( 5d later in Italy ). Cases peaked about 10d after confinement.

          I no country I have studied was there anything more than a slowing of the rate of increase before confinement came into force.

          Sweden which took light measures managed to level off new cases some time after the limited voluntary restrictions were introduced.

      • Willis, Greg-
        Nice work trying to analyze the Wuhan death statistics.

        While you’ve found some very interesting cycles I think the main problem is not having reliable data throughout.

        As you point out, the CDC changed the standard a couple of weeks ago which will affect the numbers making it very difficult to interpret the numbers. In addition, the early numbers have to be very poor because it took some time to recognize that a new flu-like disease had arisen making it very hard to distinguish, without rapid, accurate local testing, which disease was causing which death, how infective they were, what were the major types of transmission.

        The early numbers also reflect that the disease started in the US in early January. Later testing appears to have shown that cases were apparently found literally world wide outside of Wuhan months before the epidemic started there.

        I guess the bottom line is that we’ll know when we’ve passed the peak death rate in the US 2-3 months after it occurs. For a multitude of mainly political reasons the CDC was woefully unprepared for the occurrence of a major epidemic and response starting 10-12 years ago.

        • the Chinese have admitted that there were cases in October and November in Wuhan, so what brand of B.S. are you pitching here? They also knew that of the first 41 cases identified, 13 had NO LINK to the Wuhan wet market. Epidemiologically, this made it virtually impossible that it originated in the wet market and they knew that when they claimed it came from there.
          Then, when they knew they had person to person transmission, they stopped allowing Wuhan residents to fly elsewhere in China, but allowed them to fly internationally. This can only be explained a s a deliberate attempt to spread the disease worldwide.
          One thing remains to be done. Examine the records of all the people who flew out of Wuhan in the critical 6 day period. Their reasons for flying and detailed itineraries will show whether they were sent out as a “plague host”.
          If the CCP’s lack of humanity disappoints you, prepare to have your disappointment verified.

  1. Thanks, Willis, for the data analysis.

    And for those who only looked at the graphs, I’ll repeat Willis’s closing statement:
    “Finally, it appears that the US has peaked in terms of daily deaths. Might be another peak to come, might be two more peaks, might be no more peaks, but in any case but it appears we’ve passed the first peak.”

    Stay safe and healthy, all.
    Bob

    • So … a virus basically picks all the low hanging fruit (evidently anyone living in Senior Care) other old people, co-morbid unfortunates, and other compromised humans … first, then runs out of easy victims. I think we’ve seen this graph before. Now let’s go back to work … making a better world for all. Well, for the survivors anyway.

      And if I don’t survive it … enjoy the fruits of my labor.

    • I noticed this a few days ago in the Worldometer data. Sunday and Monday are the lowest days for reported deaths, while Tues. to Friday are much higher. Saturday appears more variable (probably depending on whether folks felt like working on Saturday to enter data). And that pattern goes back 4-5 weeks but weaker in the beginning as you said.
      I also noticed some interesting things from CDC and Sci. American. In some places the CDC says that 1957 and 1968 were very bad pandemics with 115K and 100K deaths in US and 1 million worldwide. But the paper I will link to says the numbers are closer to 30-50K. It also points out how much of flu deaths is modeled and how flu/pneumonia is so hard to disentangle and flu deaths are often not reported so that they have gone to the modeled deaths that flu may have contributed to (as they do now with Covid-19).
      https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4504370/
      And the Sci. Amer. article is from an ER physician who was wondering why he does not remember that many flu deaths, so he checked with other ER colleagues around the country and they don’t either. So, the true number of flu deaths per year from death certificates is closer to 3-16K, not 30-60K. Interesting that even with very broad (2-sigma I imagine) ranges on the flu deaths from models that the actual number. I stopped reading much in SA years ago as they have become very politicized and this article I note has about 4 slams against Trump. Several may be deserved, but they are actually not relevant to the point of the article. Normally, this info might not be published in SA but now that some want to emphasize flu deaths are lower (plus the slams on Trump) it has been. But, before this, my guess is that since we want to encourage people to get flu shot, they would not have published this. (I have had a flu shot for the last 27 years and will continue to get them). Hopefully these two links will make it through the filter.
      https://blogs.scientificamerican.com/observations/comparing-covid-19-deaths-to-flu-deaths-is-like-comparing-apples-to-oranges/

  2. A glimmer of hope.

    Still, many of us live in localities where the death rate is still increasing, so don’t let your guard down–particularly if you’re an old man.

    • the death rate ? daily deaths ? or the total number of dead ? big difference …
      don’t let your guard down only if you are an old man … your guard didn’t need to be up for 95% of the population …

  3. Not only are the weekends lowest, also the big spike appears to be on the Tuesday. Could it be that it’s from delayed reporting? The first reports of weekend diseased coming in on the Mondays, but the bulk on Tuesdays?

    Stay healthy

    • Yes, delayed reporting and the stinking bodies on ventilators are noticed when more people come in to work.

        • It’s on a 24hr schedule. 3rd shift usually depends largely on interns and younger nurses and takes longer to respond to any emergency, and gets overwhelmed more easily with less resources.

    • The data is so corrupted by late reporting one can not draw any collusion from it

    • Using Climate Science ™ logic, it’s obvious that more deaths occur when more health workers are at work during the week.

      The logical conclusion is that health workers are causing the deaths. All you need to do its reduce the number of health workers to reduce the number of deaths! Remember, correlation is causation.

      /sarc (for those who need it)

      Actually it’s probably administrative staff. It makes more sense that they cause excessive deaths in my experience.

  4. Same in UK. Low weekends due to delayed reporting. Best to plot deaths on day they occur. See CEBM website for England showing a single clear peak on 8th April.

      • D., we’re a full-service website, so here’s UK data … I make the peak as being on the 13th. Clearly time for the UK to be getting back to work. End the UK Lockdown Now! … as they say …

        w.

        • Willis,

          The Residual in Fig. 2 runs between 1 and -1. How is it applied to the original data to produce the curve in Fig. 4?

          • The residual is in units of standard deviations. Multiply by the SD, shift vertically to match. Linear.

            w.

          • Exactly.

            The UK has always been showing deaths by date of report, which may be several days after they have occurred. And the lag has been reported on. There is no need for any sort of elaborate statistical treatment, the actual numbers are available.

            The weekend effect is well known and is commented on at the daily press conference. Its purely an effect of reports not coming in as promptly from all reporting bodies at the weekend. Remember there are thousands of care homes and surgeries, and also a lot of hospitals.

            The UK has two systems for recording the deaths. One is pretty rapid,though not instant, and affected by weekends, its from hospitals. The other is the traditional local doctor death certificate route, which is slower. They are trying to get the second to report more promptly. But in the meantime, deaths in hospital come in over a few days, but in other situations over a longer period.

            Anyway, the weekend effect is public and very well known and has a well known cause.

        • W. I took death rates for England not UK and get peak at 8th April. Interestingly lockdown was started on 23rd March so peak is too early to have been the result of lockdown?

    • The CDC website has always carried the disclaimer that their weekend counts were not properly vetted.

      • Kurt Clark
        May 7, 2020 at 9:21 pm

        If that the true effect in consideration of the signal in the data, then the signal Willis has detected and lifted up from the data happens to be the realistic signal of COVID-19 mortality.

        Monitoring, detecting, diagnosing, classifying and recording yearly seasonal influenza mortality is a hard work.
        On top of it, trying to specifically detect-diagnose in “high resolution” the most infectious influenza disease in and during a given season will carry the condition where the further hard working on top of normal hard working will result in revealing the actual signal of the mortality of the infectious disease in question, simply due to the work pattern in the condition of hard working.

        In consideration of death in relation to death certificates for disease cases, that is just a very low validity in consideration of true medium for analyses in consideration of a given condition, as it not really tied to a universal specific.

        In my understanding the signal in the data that Willis has depicted shows the true mortality for COVID-19, which is far far far lower than what we think or believe it really is.

        The dropping of cases either in the consideration of the infections or diseases is simply the closure of the season seen by tracing the most infectious disease there, where actually mortality for the most infectious disease traced is actually very very low,
        but due to what may be called an “umbrella effect” in this case, due to the method of high specification tracing-detection, most of the mortality is listed and therefor dedicated to this disease.

        Kinda of showing a point where hard working consist in a given direction as producing a condition of a strong noise amplification.

        The numbers do not lie.
        In the other hand we do, more often than we realize.

        cheers

        • I will say this thing, maybe not even understood let alone in consideration of it being comprehended.
          it.

          The realization of a click, the click of AK-47 going from safety to rapid firing, heard clearly,
          is a very tormenting schist when happening,
          that I got to have knowing it by real experience…

          If you never touched or carried a fully automatic AK-47, you do not really know of life hazards due to stupidity.

          That is not a joke.

          When the best and the coolest and most rational of all there, cares not any more of the consequences.
          Just ready to go and pull the trigger, when in full rapid firing state, of the automatic AK-47 state of affairs.
          Devastating if it happens but true if still in activated mode, properly with no much regard, to what one thinks as cool or rational… or as a fracking silly stupid malthusian academic claim or proposition.

          I will make this silly again straight forward claim…
          If you never ever touched or handled an AK-47, you definitely know not what you talking about, regardless.
          You do not really know of the flimsiness of the line…when you never subjected to the test of steadiness with the potential of pulling the trigger… and to the consideration of repercussions
          and consequences to you and the rest of the innocents… with no any attachment in reality.

          oh well, silly talking here, please do keep moving along.

          cheers

  5. “Rather than make an explicit distinction between deaths WITH coronavirus and deaths FROM coronavirus, they said to enter COVID-19 on the death certificate if the physician SUSPECTS that the coronavirus MIGHT have CONTRIBUTED to the death.”

    A physician can suspect the coronavirus might have contributed to nearly any death. As already observed by Willis, the criteria is very vague. Makes me wonder how many accidental deaths, suicides, shooting victims, etc will end up being attributed to Covid-19 in the long run.

    Isn’t there Federal Aid being tied to these Covid-19 case numbers? I haven’t been able to keep track.

    Cheers

    Max

    • Apparently they get more money from the medical plan if the death is Covid.

      • Man (85) with heart disease, well overweight, and diabetic goes for a swim, has a heart attack and drowns. The autopsy also shows moderately advanced prostate cancer. Local Health Official asks pathologist “what was the cause of death?” Pathologist replies “what would you like it be?”

        In the context of health politics, cause of death can be important. This man drowned and any honest practitioner would say so, adding ‘heart attack’ as a proximate contributing factor. But depending on where current funding is “directed” that might be the ‘wrong’ answer for the health bureaucracy and after all, the guy’s dead so who really cares how?

        How much of that goes on I wouldn’t know but I’ve not yet met a politician who having started on the current mania for quarantining the whole population doesn’t now need a “reasonable” number of deaths attributable to Covid-19 in order to justify himself. If this can be achieved by a little manipulation (all quite reasonably, of course) in the ‘Cause of Death’ column, well ….it’s win/win, isn’t it?

        • This just popped up on youtube feed. It’s from the end of April and I find a contradiction in what he says but the overall stateent is quite chilling. With the suspicion in most countries about the integrity of the figures surely it renders all models wrong.

          Italian Parliament Leader Slams False COVID-19 Numbers
          https://youtu.be/X9wuFazgpc4

    • IIRC Federal payments to hospitals (e.g. Medicare) have a 20% bonus for Covid-19 cases. I.e. pneumonia unspecified cause = 100, pneumonia Covid-19 = 120.

      • Actually incentivising them to inflate the figures , gotta love the way bureaucrats work.

      • I’d prefer a link showing why a hospital gets paid for death at all.
        I thought hospitals were paid for services delivered, not outcomes. Otherwise why not just pump up deaths, period?

    • I was told that hospitals are paid more if the death is listed as from covId-19.

        • Give ’em more money if they blame COVID-19, then give ’em more money when they use a ventilator, while knowing that using a ventilator appears to cause excessive deaths.

          What could possibly go wrong?

    • If someone is 50 years old and has serious cardio vascular issues and lung issues and then gets covid and dies, what would be the chances he would have survived it he kept healthy during his life ? Assigning death to one thing just doesn’t make any sense.

    • I hear some people believe everything they hear. Doctors, hospitals, and medical examiners are committing widespread fraud for COVID-19 bonus money?

      Of the people who claim the deaths are miscounted, half seem to say they are over-reported and half seem to say they are under-reported.

      • Estimates of annual medicare fraud already range from $90 billion to $300 billion. The incentives for reporting of COVID-19 at the very least legitimizes some amount of fraud.

      • Doctors, hospitals, and medical examiners are committing widespread fraud for COVID-19 bonus money?

        I’d definitely t think that.

        Administrators, on the other hand…

      • Apparently it is OK to use the ensemble mean so applying this to the reporting gives the correct answer.

      • Heard the joke about two linear econometrists out deer hunting? Bang! “You missed, two feet to the left.” Bang! “Missed again, two feet to the right.” “You mean I got him?!”

  6. The next week or two will be very interesting, now that people are starting to get together again. We’ll learn a lot about the virus and what it is going to take to control it enough to carry on our lives.

    • True, Tom. My guess is that we’ll see little in the way of a “second peak”, but hey, I’ve been wrong before.

      w.

    • What it will take is

      A. A cure
      B. An effective vaccine
      C. Herd immunity

      Without one of the those we are spitting into the wind.

      • There are currently no effective vaccines for other CV, and while there have been vaccines developed for cat CV, their efficiency has been questioned. Development of a SARS-CoV-2 vaccine would depend on viral mutation being very low. With reports of at least 2 variants being in the wild, it seems that mutation may well not be low enough. Then there is the problem of production, which needs a large amount of active virus, and level-3 containment in production. And it all depends upon the human body acquiring and retaining the necessary antibodies, noting that other CV infections only confer short-term protection whether by the immune system failing to “remember” the virus or by mutation of the virus.
        Look on the “bright” side….many old people are a drain on the economy.

        • “There are currently no effective vaccines for other CV”

          I heard them talking about a new coronavirus vaccine last night that is in its second phase, and some are predicting it might be ready in the fall. The CNN host was talking to a volunteer that had been injected with the new vaccine (twice) and this volunteer was suggesting that he would also volunteer to be deliberately infected with the Wuhan virus in order to find out if the vaccine is effective and he said this would speed up the third phase of the trial.

          It sounds pretty good. Of course, the proof is in the pudding.

          There are something like 100 different companies working on a Wuhan virus vaccine right now.

      • Or, the virus could just die out after 18 months like the 1957 & 1969 [U.K ] ones did

  7. “the daily record of US deaths”. If you have a reliable source of covid mortality in the US, please share – authorities in the US, UK, Italy and elsewhere are quite honest about the fact that they put every single death they possibly (or impossibly) can down to covid19.

    Here in NZ, the average age of covide-related [sic] deaths is about 80. All bar one of the 20 so far have had serious comorbidities. Many of them were resident in a high-security dementia ward who suffered the trauma of first being moved to another hospital and then being deprived of all contact with their loved ones. Assuming that it was covid19 that killed them rather than this cruel treatment is quite a call.

    The NZ PM and government officers repeatedly tell us that covid19 “can be a very serious disease, particularly for elderly people and also for those with underlying health conditions’. So we are being subjected to a state of emergency for a disease that is serious for those who are already seriously ill.

    NZ figures are about as reliable as everyone else’s. From a typical Press Briefing:

    Sadly today, I have another Covid-19 related death to report. […] The person who passed away was a woman in her 60s; she had underlying health conditions and was considered a probable case of covid-19 due to her clinical presentation and past exposure history despite testing negative. […] a staff member was comforting her when she passed away.’

    “Stay well” is code for “keep swallowing the bs and stay compliant”

    • Makes about as much sense as saying its a smoking related death when you get hit by a bus crossing the road with a cigarette in your mouth.

    • And now it will be interesting to see whether we go down to Level 2 promptly or whether there are more excuses to retain it for another week or so. I’d love to burrow into the situation and find our whose statistics are being used. I suspect that, with our politicians and public servants being risk averse with no business experience, they will make Level 3 last as long as possible.
      I notice the antagonism towards Saint Cindy is growing by the day!

  8. Unfortunately the even the COVID death data is suspect, especially for long term care facilities. The fact is that when someone dies there even without testing positive for the virus they tend to list it as potentially due to it. I miss the days when people were allowed to die of “old age” as opposed to something, anything, that can be turned into a metric.

  9. Not that I want to get the virus – or that I want anybody to get the virus – but there are things worse than getting the virus. The foremost one is to live in a permanent fear of getting the virus. Reading media accounts, I would guess that the virus is going to wipe out 3/4 of us.

  10. Willis,

    Thanks for the analysis. I’ve updated my US deaths/day by manually going through the graph titled “Total Coronavirus Deaths in the United States” presented under the USA data on worldometers.info. I had been downloading the USA data daily from the main world table and observed that the older daily death count in the USA did not match the older data from the “Total Coronavirus Deaths in the United States” graph. I assume the historical data on the graph had been updated due to a change in counting and/or late data that was added a few days after the fact. As such, I updated my table to match the “Total Coronavirus Deaths in the United States” graph data.

    Long story short, here is what I observed:

    7-day moving average new deaths/day in the USA has been very consistent since about April 7th. As we know, the 7-day average includes data 3-days before and 3-days after the fact (April 4th – May 6th in my analysis). The moving average has ranged from 1858 new deaths/day to 2208 new deaths/day. Hardly a peak according to my observations – more like a gradual mound.

    Also of note, the last bar on Figure 1 above does not match my data. The lowest daily death rate I have was 1154 on 5/3 and the death rate on 5/4 was 1324. It appears that Figure 1 above shows less than 1000 rather than 1324 on 5/4. Maybe the figure was taken mid day before all the numbers were final at 0-GMT – just a thought.

  11. @Willis – Are there any charts on deaths because of heart attacks, cancer, traffic accidents, diabetes, suicides etc.?

    I thought you were going to get into that and compare the graphs/statistics, and subtract them…Just sayin…

    – JPP

    • Someone dies in an auto accident – if they had Covid-19 the death is recorded as Covid-19.
      Same with other deaths – the hospitals record them as covid-19 deaths ,,!

      – JPP

      • Substitution effect is rarely mentioned. Per CDC, previous rate that US citizens pass is .04, for all causes, for 65+. 2 mil of the 50 mil per year ( all those other causes will now be lower)

        Why the fascination with “peak death”?
        If we’re only societally 5 to 10% exposed, then 90% of the morbidity is yet to come.

        Long fat tail in the year of the rat.

  12. With the confusion and agendas surrounding this, it’ll be a long time (if ever) before we have reliable information. Until then — figures don’t lie, but liars can figure.

    • You’re not wrong with a blast from the past and are we picking that up in the stats?
      https://www.msn.com/en-au/news/world/its-irresponsible-washington-state-sees-sudden-rise-in-covid-parties/ar-BB13IlkB
      If you catch the Rona you get an expiation notice like the owner of the vehicle caught on camera. The only defense is you have to prove you were either an essential worker or were in lockdown properly social distancing and give us all your contacts. Rona party dudes do not pass muster do not collect $200 and go directly to jail where they will be guinea pigs for testing anything we can possibly come up with. In Oz we’re currently injecting prisoners with Vegemite and the numbers look promising.

  13. The reckoning will come when total deaths by the usual categories (old age, heart, pneumonia, etc.) are tallied for an equivalent period like up until June. I’m guessing #19 has the miraculous side effect of reducing other deaths.

  14. It may be important to discover the process for data acquisition. I’ve heard and read recently, from various sources, that CDC is using an established influenza reporting system to track Chinese virus cases. In the U.S., as I understand it, influenza surveillance reporting data is voluntarily sent to county health offices by doctors, hospitals, and other health care providers, and from county health officials to CDC. How various county data is compiled and sent is up to each state. Territories, the District of Columbia, Military, and Indian Health Services all have similar systems. It is not hard to imagine that there are a dozen or so 36-hour-per-week clerks and administrators serially involved in compilation and transmission of the data before it is received by CDC. I’ve recently heard and read that there is a four-day lag (I assumed average) in data production to recording.

    The influenza surveillance system is designed to capture snapshots on the outbreak and spread of disease, so it has an intentionally low tolerance for accuracy of reporting, rather than pre-filtered, more precise data that has been rigidly verified. The preference is for noisy data to provide a more real-time picture of progression and spread. This is deliberate, but results in a measurable lack of precision.

    To my interpretation, today being Thursday, at the end of the day I can have reasonable confidence that last Monday’s total worldometer report is unlikely to change much. Being a weekday late in the work week, I can be reasonably confident that Tuesday’s posted total is pretty solid, and yesterday’s is in the ball park. I don’t have much confidence that this last minute’s update will stand.

    I have no information on how the data is compiled in other countries, nor how many sets of eyes may have to review it before forwarding to be reported. Just looking at the updates on worldometer, it seems that there are probably similar update and precision issues with UK and Sweden. Moncton writes recently about Italy’s data count being off.

    It took CDC until 2019 for the official report on the 2009-10 U.S. swine flu epidemic to be compiled forensically, and then the rate of infection and deaths were given as the mean of a range of the possible totals. How “good” can or should the first-ever attempt at daily public reporting be?

    • The only number that appears to be reliable is the total weekly mortality. Through week 16 in the US for 2020 we have had 922k deaths. That compares to an average of 899k in the previous 6 years and a high of 946k in 2018 and within 3k of both 2017 and 2019. That is after big weekly spikes in weeks 14 and 15.
      It appears to me we are just trading one kind of death for another so far.

  15. Willis
    I appreciate your efforts, sir. It must be demoralizing for you when the reporting goalposts are changing, and you’re trying to make sense of it all.

  16. So can Willis or anyone else please tell us why they think New York, New Jersey deaths etc are so high per mil, or why Germany fared so much better and why the UK was so much higher than Germany?
    Again I’m trying to compare deaths per mil not total deaths. Here in Australia and NZ we’ve done very well, but I suppose we’ll have a lot more data in a few more months and an even bigger picture by 2021.

    • Some number of NYC deaths are due to incompetence and neglect of hospital staff. I suspect that number is quite large.

    • Different strain? Subways, elevators? If you’ve not been to NYC the density might be hard to get viscerally, just how rare it is to be alone. Doesn’t explain Hong Kong or Singapore’s low numbers though does it?
      I worry NYC highest in the world rate is just advanced in time.

      But again the death cert wont have two causes, it is either covid or heart disease. Substitution has to go into the data analysis.

      • think about the other vectors for the virus, doesn’t have to be passed from human to human, what about rats with fleas, what about bedbugs(new york city is a hot spot for them)

      • “now-new-study-finds …”

        It seems to be the usual fearmongering trick.
        As for the climate scam, my BSmeter is skyrocketing.

      • More likely to be more deadly state policies, like ventilator use, banning HCQ and shoving older patients back into retirement homes where they can infect the most at risk group in society.

        I guess Cuomo reckons older people vote Trump so he can create a edge for the Dems by killing them off.

        • “and shoving older patients back into retirement homes where they can infect the most at risk group in society.”

          An extremely bad decision. It makes no sense.

          • Everybody’s in a state of fear and panic, and trying to offload perceived risk. The well-being of patients and others becomes secondary in many cases.

      • Based on the video you posted of a NYC nurse describing the horrific conditions at her hospital, plus a couple of other videos like it, the deadliest strains appear to be dressed in scrubs.

        • I still can’t get past the stupidity hospital staff as shown on one news cast.

          A group of volunteers gets of a bus and are greeted by clapping hospital staff, in scrubs with masks off. The newbie out of state volunteers have to walk the double lined gauntlet of cheering hospital to get in the door of the hospital … at that point I would have turned an got back on the bus.

    • The difference is in the effort put into contact tracing and testing. When death to resolved ratio is high, say 25% to to 40% like US and Sweden, there is not much effective effort being put into contact tracing. Most cases are found as they present to hospital with about a 60% change of coming out alive.

      In Australia, there was a massive effort put into contact tracing. Australia has identified nearly all the cases in the country; whether presymptomatic, asymptomatic or with symptoms. The death to resolved ratio is 1.6%. South Korea has death to resolved of 2.5%.

      Germany has death to resolved of 5%. US death to resolved is 27%. Sweden is 38%. UK do not even bother to report resolved. Spain is middle of the road with 14%; they had most draconian lockdown and probably better at tracing and was very effective in reducing the death rate.

      US will plod along for many more weeks losing between 1500 and 2000 per day. UK is similar with about 500 to 800 deaths per day locked in for at least a few weeks yet. Sweden is stuck at about 80 deaths per day for the next 5 years or so. Sweden already now has 3k dead and at least another 3k already infected that will not make it out of hospital. So 6k already doomed to die. At overall population death rate of 1.5%, they will have had 400k infected already but only 25k have been identified. So a lot more will need to be infected and some die before they get anywhere near herd immunity. The problem with slowing the rate of infection means that it takes a lot longer to get herd immunity.

      Countries with effective border control; effective interpersonal quarantine and effective contact tracing have eliminated the virus and freedom of movement and association can occur inside those borders and between like countries without risk of CV19.

      Taiwan set the gold standard – warned WHO of risk in December 2019, closed border to Chinese in January and implemented contact tracing through every available means, electronically and manually. So far 439 cases and 6 deaths.

      UK is still letting 10,000 potential virus carriers through Heathrow every day to board public transport and roam through the streets and hotels spreading CV19.

        • Survivors is such a relative term.
          Those that “survive” often have continuing health problems needing a continuation of healthcare. The damage done to lungs needs, in many cases, a lengthy stay in hospital after recovery from the initial infection. You also forget that a large number of admin staff are now working from home, with rather patchy IT, and some have removed themselves from the chance of infection by leaving the job. Others booked holidays and just haven’t come back!!
          You should also consider that many ordinary staff have been infected, and many others will be, because PPE is pretty basic in the non-frontline staff (it is not exactly good in the frontline staff, with masks and gowns being worn past the recommended period)

      • Nice to see someone give Taiwan a pat on the back – they certainly deserve it! The media always praise S Korea, Hong Kong, etc for doing well, but leave Taiwan out as they are afraid of a slap from the ChiComms.

  17. Some humble observations.
    First, the data you are brilliantly analyzing are even more problematic than surface temp recs.
    Second, we a dealing with al less previously understood niological agent—no physics equations need apply (altho I did as an undergrad show the equivalency of Probabalistic Markov chains to partial differential calculus equation solutions for the classic rabbit/fox predator/prey equation.

    • I must be missing something. In urban industrialized areas a death should be noted in pretty close to 100% of the cases, with a time of death accurate to the day in at least 90% of the cases. People who live alone, die at home and are not discovered for long enough to make the day of death uncertain by more than a day or two must be pretty rare.

      So while I can understand uncertainty over the cause of death, I don’t understand why the date of death wouldn’t be reliable in the vast majority of cases. So it seems unlikely the periodic spikes are the result of delayed reporting.

      • You aren’t getting how the reporting works … to understand just do the following

        So goto this site and lets enter some data for US
        https://www.worldometers.info/coronavirus/country/us/
        Now click on the “Report coronavirus cases” and you can add numbers yourself just by putting in a link
        They then man handle checks an viola numbers appear they don’t know when the deaths occurred beyond the day they are reported.

        In many cases the numbers being reported are just hospitals reporting, death certificates are a whole other kettle of fish as they have legal/admin processes they pass thru in most countries.

    • The article said that hydroxychloroquine was given to sicker patients.

      Since the drug’s effectiveness is apparently best when given early in the disease (and with Zn), it does not seem surprising that no benefit was observed.

      • Exactly. There is no indication of whether is was given with the antibiotic azithromycin.

        In short they seem intent on doing the damnedest to ensure it does not work.

        When your article title shouts about Trump “touting” the drug , you can guess how objectively they are reporting the facts. They also report gleefully that remdesivir is now approved but fail to mention that has also been found to be ineffective.

        But as long as the political bias fits your worldview and attacks Trump that’s all you need to know. Media outlets are just for bias confirmation now, not news or facts.

        • Actually, there is also mention of azithromycin being given. However, as with ALL the “rigorous” studies I’ve seen or heard about, Zinc was NOT added to the cocktail.

          Given the putative mechanism of action of hydroxychloroquine as a zinc ionophore, leaving out the zinc would appear to be ridiculous. It’s supposedly the zinc that screws up the viral RNA replication so the treatment would be less effective if zinc was in short supply in cells.

          Many older people are zinc deficient and a supplement of 20mg/day would seem to be a cheap, harmless top-up to provide yo patients.

  18. As of right now, .126% of North Carolinians have tested positive for CV-19 and .0048% have died from it.

    We go to phase 1 of our get out of jail program at 5pm tomorrow. Really hoping to keep the death percentage below .01%.

  19. Maybe its flattening because fewer flu shots are being given with the end of flu season?

    I don’t get flu shots so I’m just finding out that seniors over 65 get their own flu vaccine in US , at least since 2016, and its used in only 37 other countries, including many in Europe (UK, Italy, etc), Canada, Australia. It has a powerful adjuvant called squalene (MF59). The vaccines called Fluad in US. China makes its own vaccine so I don’t know if they use the same super adjuvant. Its meant to stimulate the immune system and is 4 times more powerful than aluminum adjuvants. Elderly have poor immune systems so they need the boost. Squalene was also allegedly behind the Gulf War Syndrome thought to be caused by vaccines , hence the delay being approved in US (Italy was first to sign up in 1997, Uk approved in 2018) . Italy actually suspended Fluads use for a short period following 19 deaths in 2014 blamed on the vaccine. US fast tracked it in 2015 skipping important trials. At one time it was banned in Germany but I cant find out if that was lifted.

    Anyways, we have a new “virus” that from a large amount of data affects the elderly more severely than the young , primarily because their immune system goes into overdrive despite have poorer functioning immune systems. You think someone might look into an association between vaccines given with a super immune stimulant and severe covid infection due to immune over reaction. In the US 60% of elderly get the flu vaccine , many getting the super adjuvant, so there are at least 40% out there who don’t , so that would make a nice control group for a retrospective study.

    If such a study is not proceeding that is very disturbing, and quite telling in itself

    I am not saying the vaccine is the definite reason for increased disease severity, just that its something which should logically be looked into. Absence of evidence is not proof of absence. If you don’t look for evidence you don’t find evidence.

    • Good comment. There are lots of questions to be answered.

      I’m no antivaxxer, but there is a fair amount of “witchcraft” involved in all of this. Squalene is from shark liver or possibly whale oil, if natural. It’s possible that the Chinese cheat and adulterate it with synthetics.

    • Thanks for this, very interesting. I didn’t know about squalene, but I haven’t seen it listed in our (NZ) flu vaccines – possibly because I have taken more interest in what goes into babies. I’ll check our schedule. I myself have never had a flu shot, more from lack of interest than for ideological reason, at least until a couple of years ago when I started looking into the ingredients …

      Fyi, two major vaccination campaigns against influenza and meningococcus were carried out in Lombardy in the months immediately preceding the outbreak of Covid19, notably in the later hotspots of Bergamo and Brescia. https://www.bergamonews.it/2019/10/21/vaccinazione-antinfluenzale-a-bergamo-ordinate-185-000-dosi-di-vaccino/332164/

    • “Elderly have poor immune systems so they need the boost.”
      ————————
      Actually the elderly have a very highly experienced immune system, due to life long experience, the compensation for overall ageing of the body.
      Good long history,
      oh well, till they start getting vaccinated for seasonal flu, and get boosted,
      or more accurately busted.

      cheers

  20. “Next, about a week ago the CDC changed its guidance on the reporting of deaths involving the COVID virus. Rather than make an explicit distinction between deaths WITH coronavirus and deaths FROM coronavirus, they said to enter COVID-19 on the death certificate if the physician SUSPECTS that the coronavirus MIGHT have CONTRIBUTED to the death … “suspects the virus might have contributed” to the death??? Could they possibly be more vague?”

    It might just not be easy to discriminate from other causes of death as cytokine storm, heart attack, kidney failure and stroke can all be caused by an infection of SARS-CoV-2 and pneumonia is just one part of a broader picture.

    Which would partly explain why ventilators are not such an efficient treatment.

    Evidence goes into the direction of cardiavascular malfunction:

    http://joannenova.com.au/2020/05/the-stroke-virus-covid-causes-hundreds-of-microclots-throughout-the-lungs-and-everywhere-else/

    That would also partly explain why there is such a high hospitalization rate in younger people:

    https://www.sciencenews.org/article/coronavirus-covid19-young-adults-can-face-severe-cases

    This disease exposes just any weakness of the cardiovascular system. Would also partly explain the difference in susceptibility of different ethnicities:

    https://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1538-7836.2011.04443.x

    • JoNova’s continued use of Nick Cordero’s blood clot in an attempt to link it to covid is either deceptive or ignorant. The ECMO machine that Nick was on caused the blood clot.

        • Whatever. You lose credibility when you justify deceptive/ignorant behavior like that.

          • You can’t basically discriminate the clotting from ECMO or COVID-19 without looking at coagulation factor data from before, during and after treatment. Without knowing this the hypothesis COVID-19 was responsible for the amputation can’t be disproofed.

          • The doctor(s) told Nick’s wife that the ECMO machine caused his clot. Those machines are prone to do that.

            “They put the ECMO machine in him to save his life,” Kloots said. “It was literally to save his life, and it did, thank God. And sometimes the repercussion of putting that machine on can cause some blood issues, and it did with his leg.”

            https://www.nbcnews.com/pop-culture/pop-culture-news/broadway-star-nick-cordero-s-wife-his-coronavirus-leg-amputation-n1187706

          • This is no proof. No proof without data. Could be one. Could be the other. Could be both together. Nobody knows in the context of blood clots in COVID-19.

          • It’s not proof and it’s not science either; it’s medicine. The chances of getting a large blood clot from an ECMO machine are enormously larger than getting a large clot from corona-chan. Most of the ones she helps cause are micro-thrombi. Therefore, ascribing Nick Cordero’s massive blood clot to corona-chan is simply deceitful alarmism.

    • Oh, and it could explain the described blood type correlation as well as the higher death rates in the African-American population:

      “Elevated factor VIII (FVIII) increases the risk of thrombosis. African-Americans appear to have higher levels whereas individuals with blood group O tend to have lower levels of FVIII.”

      https://www.ncbi.nlm.nih.gov/books/NBK538251/

    • Autopsy finding:

      “Deep venous thrombosis in 7 of 12 patients (58%) in whom venous thromboembolism was not suspected before death; pulmonary embolism was the direct cause of death in 4 patients.”

      https://annals.org/aim/fullarticle/2765934/autopsy-findings-venous-thromboembolism-patients-covid-19-prospective-cohort-study

      Take heparin. Maybe harder stuff like phenprocoumon if you are at risk and don’t have contraindications.

      Increasing your vitamin K level might actually be detrimental.

  21. There’s so much crap in corona-virus catastrophism that no wonder the rush on toilet paper happened when it did. What could possibly be more poetic for the times?

    How can anybody have any confidence that the numbers mean what we think?

    Stump your toe, … it gets infected, … blood poisoning, … you die, … test for the virus is positive. We have a winner! Enter on line whatever of the death certificate. Adjust mortality count accordingly. It’s the fashionable thing to do.

    • Maybe we need to be including venereal diseases in the testing along with COVID-19. If the younger population has a higher incidence of VD and the older population have a lower incidence then maybe the older geezers need to break loose! Just think of it — VD protects you from CORONA !!!!! List VD on the death certificate. Pandemic over!

    • I stopped by a liquor store the other day and they were selling toilet paper. I think they missed that fad.

  22. Dear Mr. Eschenbach,

    With your data set, is it possible to sort out all deaths of individuals over 70? And examine graphically the daily death rate of only those under 70?

    Without being too cold blooded about it and just for statistical purposes, let us assume those over 70 with comorbidities were going to die anyway, covid or no covid.

    When did the under 70 daily deaths peak? What does the CEEMD of that time series look like? Just curious…

  23. https://www.washingtonpost.com/health/2020/05/07/blood-thinners-coronavirus-clots/

    https://www.medrxiv.org/content/10.1101/2020.03.28.20046144v3

    https://onlinelibrary.wiley.com/doi/pdf/10.1002/rth2.12353

    I would take low molecular weight heparin for starters (and aspirin probably).

    Heparin is safe, anti-inflammatory, prevents blood clotting and inhibits cellular entry of many different viruses.

    And if I would be in the risk group or end up in the hospital I would ask my doctor what else can we do without risking bleeding.

  24. https://www.washingtonpost.com/health/2020/05/07/blood-thinners-coronavirus-clots/

    https://www.medrxiv.org/content/10.1101/2020.03.28.20046144v3

    https://onlinelibrary.wiley.com/doi/pdf/10.1002/rth2.12353

    https://annals.org/aim/fullarticle/2765934/autopsy-findings-venous-thromboembolism-patients-covid-19-prospective-cohort-study

    I would take low molecular weight heparin for starters (and aspirin probably).

    Heparin is safe, anti-inflammatory, prevents blood clotting and inhibits cellular entry of many different viruses.

    And if I would be in the risk group or end up in the hospital I would ask my doctor what else can we do to thin my blood without risking excessive bleeding.

  25. I do not trust the “CovID-19” attributed deaths, so I am rather inclined to look at the total deaths. Now a lot of deaths have been averted due to the lock-down (at least in theory), but the total deaths is capturing a lot of CovID-19 deaths that are not counted otherwise. I cannot say for certain that Flu did not cause these extra deaths, but I am a fan of comparing year to year and Flu deaths are likely to follow similar curves.

    A CovID-19 death is in the eye of the doctor, but a death is a death and always counted.

    There is another possibility…although I kind of hate to share it out loud. Wives may be killing their husbands at an unprecedented rate since they are locked up with them…I know mine is thinking about taking me out! [ 🙂 ]

  26. “Next, about a week ago the CDC changed its guidance on the reporting of deaths involving the COVID virus. Rather than make an explicit distinction between deaths WITH coronavirus and deaths FROM coronavirus, they said to enter COVID-19 on the death certificate if the physician SUSPECTS that the coronavirus MIGHT have CONTRIBUTED to the death … “suspects the virus might have contributed” to the death??? Could they possibly be more vague?

    actual guidance

    https://www.cdc.gov/nchs/data/nvss/vsrg/vsrg03-508.pdf

    https://emergency.cdc.gov/coca/ppt/2020/04-16-20-transcript.pdf

    https://emergency.cdc.gov/coca/calls/2020/callinfo_041620.asp

    https://emergency.cdc.gov/coca/ppt/2020/Final_COCA_Call_Slides_04_16_2020.pdf

    https://www.cdc.gov/nchs/nvss/covid-19.htm

    hmm?

    ‘”Next, about a week ago the CDC changed its guidance on the reporting of deaths involving the COVID virus. Rather than make an explicit distinction between deaths WITH coronavirus and deaths FROM coronavirus, they said to enter COVID-19 on the death certificate if the physician SUSPECTS that the coronavirus MIGHT have CONTRIBUTED to the death … “suspects the virus might have contributed” to the death??? Could they possibly be more vague?”

    hmm citations needed

    ya know when people cite “Ar5” and you get pissed because they dont use page numbers?

    • Sorry, Steven. I’d posted the link to this guidance so many times that I figured even folks like you would have heard about it. My bad, though, should have re-posted the link. Here’s the actual quote from the CDC:

      Should “COVID-19” be reported on the death certificate only with a confirmed test?
       
      COVID-19 should be reported on the death certificate for all decedents where the disease caused or is assumed to have caused or contributed to death.

      Which is … well … pretty exactly what I’d said from memory. From the CDC here … sorry that you wasted all that snark, maybe you could collect it up and direct it at someone who actually deserves it.

      w.

      • Sorry.

        1. Thats march 24th guidance
        2. It does not say this

        ‘‘”Next, about a week ago the CDC changed its guidance on the reporting of deaths involving the COVID virus. Rather than make an explicit distinction between deaths WITH coronavirus and deaths FROM coronavirus, they said to enter COVID-19 on the death certificate if the physician SUSPECTS that the coronavirus MIGHT have CONTRIBUTED to the death … “suspects the virus might have contributed” to the death??? Could they possibly be more vague?”

        A) you said a week ago. the document you refer to now is stated March 24th.
        B) here is what I found for latest guidance
        https://www.cdc.gov/nchs/data/nvss/vsrg/vsrg03-508.pdf, April guidance
        C) Maybe you did not go through the latest training materials
        D) The codes are U07.2 and , U07.1
        U07.2, when there is no test, , U07.1 when there is a test

        here is the actual words

        “Should “COVID-19” be reported on the death certificate only with a confirmed test?
        COVID-19 should be reported on the death certificate for all decedents where the disease caused or is
        assumed to have caused or contributed to death. Certifiers should include as much detail as possible based
        on their knowledge of the case, medical records, laboratory testing, etc. If the decedent had other chronic
        conditions such as COPD or asthma that may have also contributed, these conditions can be reported in Part
        II. (See attached Guidance for Certifying COVID-19 Deaths)”

        there are 2 codes. One code for when there is a test. One code for when there is no test.
        That way you keep them separate and can count on, deaths with tests, deaths with no tests

        “Sorry, Steven. I’d posted the link to this guidance so many times that I figured even folks like you would have heard about it.”

        it is weird. I remember this kind of argument from the climate wars. With respect to hide the decline. Briffa and CRU arguing that the decline had previously been disclosed in the literature .

        • B) here is what I found for latest guidance
          https://www.cdc.gov/nchs/data/nvss/vsrg/vsrg03-508.pdf, April guidance

          Maybe you could clarify further? For example:

          “here is the actual words “Should “COVID-19” be reported on the death certificate only with a confirmed test?”

          I don’t find those words in the “latest guidance” doc you cited here: https://www.cdc.gov/nchs/data/nvss/vsrg/vsrg03-508.pdf

          The actual words in *that* document are (emphasis added):

          An accurate count of the number of deaths due to COVID–19 infection, which depends in part on proper death certification, is critical to ongoing public health surveillance and response. When a death is due to COVID–19, it is likely the UCOD and thus, it should be reported on the lowest line used in Part I of the death certificate. Ideally, testing for COVID–19 should be conducted, but it is acceptable to report COVID–19 on a death certificate without this confirmation if the circumstances are compelling within a reasonable degree of certainty.

          This appears to blatantly contradict your claim:

          “there are 2 codes. One code for when there is a test. One code for when there is no test.
          That way you keep them separate and can count on, deaths with tests, deaths with no tests”

          How does one “keep them separate and … count on, deaths with tests, deaths with no tests,” when “it is acceptable to report COVID–19 on a death certificate without this [i.e., testing] confirmation if the circumstances are compelling within a reasonable degree of certainty”?

          Not that you didn’t find those words you quoted somewhere else in CDC documentation. I’m sure you did.

          But which words should we follow?
          Those before or those after (depending on which came first and when)?
          Who can know?
          How can they know?
          Where is the final word of wisdom?

          Oh dear, “it’s a ‘puzzle'” . . . said a wise man once, somewhere . . .

          🙂

        • Huh? What’s this? Surely not crickets right?

          I mean you were JUST here bouncin’ all over W. E. for quoting an old document when your own up-to-date citation says exactly what his old one did?

          Sup dood? Come on “techno” bro step up and be the man you want W. E. to be!

          “it is weird. I remember this kind of argument from the climate wars. With respect to hide the decline. Briffa and CRU arguing that the decline had previously been disclosed in the literature .”

          Yeah, “it is weird,” right? Somebody said: “hmm citations needed,” but when he got ’em he cricketed the field? Surely not the accuser? He wouldn’t do that would he?

          🙂

  27. “Why these numbers are different
    Provisional death counts may not match counts from other sources, such as media reports or numbers from county health departments. Our counts often track 1–2 weeks behind other data for a number of reasons: Death certificates take time to be completed. There are many steps involved in completing and submitting a death certificate. Waiting for test results can create additional delays. States report at different rates. Currently, 63% of all U.S. deaths are reported within 10 days of the date of death, but there is significant variation among jurisdictions. It takes extra time to code COVID-19 deaths. While 80% of deaths are electronically processed and coded by NCHS within minutes, most deaths from COVID-19 must be coded manually, which takes an average of 7 days. Other reporting systems use different definitions or methods for counting deaths.

    Things to know about the data
    Provisional counts are not final and are subject to change. Counts from previous weeks are continually revised as additional records are received and processed. Provisional data are not yet complete. Counts will not include all deaths that occurred during a given time period, especially for more recent periods. However, we can estimate how complete our numbers are by looking at the average number of deaths reported in previous years. Death counts should not be compared across jurisdictions. Some jurisdictions report deaths on a daily basis, while others report deaths weekly or monthly. In addition, vital record reporting may also be affected or delayed by COVID-19 related response activities.

    For more detailed technical information, visit the Provisional Death Counts for Coronavirus Disease (COVID-19) Technical Notes page.”

    warning to all data monkeys

    ‘Death counts should not be compared across jurisdictions.”

    here is a clue. you will not see this data monkey plotting any death data.
    and never compare death data across jurisdictions.

    When data providers give you fair warning ,you precede at your own risk.

    But the nice thing about looking at provisional data is that when you make a mistake
    and the data changes under your feet like all provisional data does, you can always blame the supplier.

    don’t be bad data monkey.
    It is ok to look at the data and start to understand the issues,
    but if you publish results based on provisional data,
    expect to have the carpet pulled from under your feet.

    • I’m wondering about classifications of murders by spouses locked in together due to Covid?

      Clearly seems caused by Covid.

      • If you use a death certificate not possible they would die from blunt force trauma, gun shot etc. The problem is the death certificate data can take weeks to get. If you use hospital or media company data maybe, that is up to what data source manager decides.

      • Depends on coroner, I guess. Two cases:

        Twenty-something showed up at UK hospital, had a heart attack, then died. Coroner listed death as covid because he heard she had a cough. Hospital staff disagreed saying she didn’t test positive.

        US infant died of drowning and tested positive postmortem. Governor of state publicly declared infant’s death was linked to covid, and was then publicly exposed to be a liar. Coroner refused to register death as covid.

        • The first case you gave was a massive reporter error the BBC and British newspapers had to offer very public apologies as they cause distress to the family and were looking at damages case.

    • Mosh,
      Just asking.

      Are you are you not a techno guy????!!!

      I know that not to be the proper question, but never the less.

      cheers

  28. a week ago the CDC changed its guidance on the reporting of deaths
    ============
    politics as usual. TDS. Changing the accounting method midstream to panic the country and justify their jobs.

  29. I would like to point out that the CDC says hospitals, doctors, and nurses kill 100,000 patients PER YEAR just from preventable hospital infections.

  30. Figures for England and Wales
    from…https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/datasets/weeklyprovisionalfiguresondeathsregisteredinenglandandwales

    All deaths week 17 2019 10,059
    5 year average deaths week 17 10,317
    All deaths week 17 2020 21,997

    and for Scotland from…https://www.nrscotland.gov.uk/statistics-and-data/statistics/statistics-by-theme/vital-events/general-publications/weekly-and-monthly-data-on-births-and-deaths/weekly-data-on-births-and-deaths
    All deaths week 14 2019 1032
    All deaths week 14 2020 1741

    So co-morbidities or not people are being pushed off the branch earlier than previously.
    After the virus has passed the death rate may, perhaps, possibly drop as people who were likely to die “soon” have already died.
    That won’t be clear until the data are in.

  31. I think this analysis was a nice waste of time. One major problem with looking at deaths by day is that they tend to be *reported* date, not *death* date. For example, there have been several days when a batch of nursing home deaths were added to the NYC statistics, spiking the deaths on those days. But, those deaths happened over a period of time and should be understood to be spread out over multiple days. The weekend spikes could be spikes in *reporting* not *deaths*.

    • Trivial.
      The statistics must week-integrated.
      Even this does not help as many report deaths at the end of month.
      It’s bureaucracy, not epidemic.

    • Ann, all your points show why the filtering Willis did was necessary, not why it was a “waste of time”. The

      Alex: “statistics must week-integrated.”

      No, you do not need to integrate, you can filter. That is Willis did and what I did in by a different means. Both mean you get more than one point per week, and see the trend underlying the weekly variation.

      • “you do not need to integrate, you can filter”

        One can, but one should not do it like Willis did.

        The Gaussian average is stable. You cannot apply it at the end of the series of course.
        What Willis did, he applied his – whatever – procedure towards the end of the series.
        This is a kind of extrapolation, even if it is still within the available data.
        Extrapolations are always unstable.

        Whatever he got towards the end of the series – a bump or a decay – is meaningless.

    • “I think this analysis was a nice waste of time. One major problem with looking at deaths by day is that they tend to be *reported* date, not *death* date. ”

      yes this has been pointed out many times but guys still continue to plot daily death data.

      Now what is worst is that some politicians are also looking at daily death data.

      At this stage it is largely pointless to look at time series of death data.
      But folks will continue to push agendas by plotting up data before its been through a proper
      cleaning analysis.

      Again, people did the same thing with early DImand Princess data, same thing with early Korea
      death data.

      Their is no stopping people from monkeying around with data that hasn’t been properly vetted

  32. Hi Willis, – An old geezer & I were passing the time of day quite a long time ago. We got around to our respective ages after I admired his daily activities. Upon hearing I was in my late 50s he deadpanned: “I got shoes older than you.”

  33. Well, let us watch those, who started earlier.
    Europe is by far not the early bird.
    Rather, check Iran.
    I do not like what I see there.

  34. For a long time, I’ve known and commented about the dip in the death counts on weekends and the subsequent rise on Tuesday. A less obvious aspect of the counts is that just because a death is counted on a certain day doesn’t mean it occurred on that day or the day before. Indiana has one of the best run COVID websites. One of their graphs shows when the newly-counted deaths occurred. In some cases, some of the deaths happened weeks before.

    On April 29th I notice a very large jump in the Covid Tracking Project’s U.S. count, and decided to see which states it was coming from. I noticed 164 of the deaths came from Indiana. However, when I looked at the Indiana website, it only showed 63 deaths. When I questioned the Covid Tracking Project, they replied that they were, per CDC guidance, also including suspected deaths. However, the 29th was the first time they included suspected deaths in Indiana’s count; the previous day’s count matched Indiana’s confirmed death count. So all of Indiana’s suspected deaths, which happened throughout the epidemic, were added to the death total in a single day; and 101 deaths in the jump in U.S. deaths on April 29th were attributable to those suspected deaths.

    • I’ll also add that many of the 63 Indiana confirmed deaths reported on April 29th didn’t occur on that day. Currently, Indiana shows 38 deaths on the 29th, and 33 on the 28th. So the report of 164 deaths greatly distorted the actual trend’

      I haven’t looked to see if there are similar examples for other states, w]here suspected deaths were added to a total the previously included only confirmed deaths, but I suspect there are.

  35. It appears that the most of European countries that introduced universal BCG vaccination of young children in the 1950s have been spared from the most severe Covid-19 impact.
    http://www.vukcevic.co.uk/EuropeCV.htm
    Currently only hypothesis, but future research will clear matter one way or the other.

    • Abstract
      “The reasons for a wide variation in severity of coronavirus disease 2019 (COVID-19) across
      the affected countries of the world are not known. Two recent studies have suggested a link
      between the BCG vaccination policy and the morbidity and mortality due to COVID-19. In
      the present study we compared the impact of COVID-19 in terms of case fatality rates (CFR)
      between countries with high disease burden and those with BCG revaccination policies
      presuming that revaccination practices would have provided added protection to the
      population against severe COVID-19. We found a significant difference in the CFR between
      the two groups of countries. Our data further supports the view that universal BCG
      vaccination has a protective effect on the course of COVID-19 probably preventing
      progression to severe disease and death. Clinical trials of BCG vaccine are urgently needed to
      establish its beneficial role in COVID-19 as suggested by the epidemiological data, especially
      in countries without a universal BCG vaccination policy. ”
      https://www.medrxiv.org/content/10.1101/2020.04.07.20053272v1.full.pdf
      (p.s. In the ethnically east European compact countries with very little or no ex-european immigration the BCG effect appears to be further reinforced.)

      • Vuk

        According to your effort;

        Do actually BCG vaccination or BCG vaccines consist as a stupid beyond stupid flu vaccination?

        just asking Vuk!
        You tell me please!

        • BCG is one of the oldest vaccines, if administered at an early age, preferably to preschool children it gives lifetime protection from tuberculosis, one of the better known lung disease. Wikipedia has all about it.

      • My advice: vaccinate medical staff with a proven and safe tuberculosis vaccine. The autumn and winter wave may be worse than the spring one. Even more so when combined with seasonal flu.

        • Adults aged 16 to 35 who should have the BCG vaccine

          BCG vaccination is recommended for people aged 16 to 35 who are at occupational risk of TB exposure, including:

          laboratory staff who are in contact with blood, urine and tissue samples
          veterinary staff and other animal workers, such as abattoir workers, who work with animals that are susceptible to TB, such as cattle or monkeys
          prison staff who work directly with prisoners
          staff of hostels for homeless people
          staff who work in facilities for refugees and asylum seekers
          healthcare workers with an increased risk of exposure to TB
          https://www.nhs.uk/conditions/vaccinations/when-is-bcg-tb-vaccine-needed/

    • Easy to test: take blood samples from vaccinated vs non-vaccinated, run T cell, B cell and neutralizing experiments.

      If you don’t find any mechanism it is just a correlation like storks and babies.

      Ireland clearly argues against the BCG hypothesis. One of the worse countries by infected/M (4,533) and deaths/M (284) in Europe.

      • The Republic of Ireland population is just under 5 million
        The largest immigrant groups, with over 10,000 people, are the British, Croats, Poles, Americans, Lithuanians, Latvians, Germans, Nigerians, Indians, Pakistanis and Chinese.
        Also there has been large exchange between North and South in the last two decades since the GF treaty. Ireland is far more ethnically diversed than the East European countries.
        Both tuberculosis and Covid are lungs diseases, so it shouldn’t be a huge surprise that BCG reduces effects of the Covid lungs infection.
        Explanation looks plausible and the medical science needs to find answers urgently, one way or the other.

        • “The largest immigrant groups, with over 10,000 people, are the British, Croats, Poles, Americans, Lithuanians, Latvians, Germans, Nigerians, Indians, Pakistanis and Chinese.”
          Most of those countries have BCG vaccination programs still active. No big impact to be expected. Immigrants might also be vaccinated. That is what health care systems do.

          “Both tuberculosis and Covid are lungs diseases, so it shouldn’t be a huge surprise that BCG reduces effects of the Covid lungs infection.”
          The misconception COVID-19 to be a lung disease might have killed a lot of people on ventilators. It is not at all primarily a lung disease.

          Different mechanisms can lead to the same symptoms. Just looking at symptoms might put you on the wrong track. The evidence is accumulating that this has happened with COVID-19.

          • BCG efficacy depends on the genetic variation in populations.
            “Trials conducted in the UK have consistently shown a protective effect of 60 to 80%, but those conducted elsewhere have shown no protective effect, and efficacy appears to fall the closer one gets to the equator.
            ….. Native Americans immunized in the 1930s found evidence of protection even 60 years after immunization, with only a slight waning in efficacy.” -wikipedia
            It looks that the Covid-19 may have some genetic factors attached to intensity of infection and consequently mortality.
            Out of season flu or devious killer, the Covid-19 science is far from being settled. It would be good to know.

    • Vuk, the UK had universal BCG vaccination from the 50s until 2005 as I recall. I remember we were tested/vaccinated at high school. It’s hard to believe that the UK has been spared.

      • BCG vaccination apparently is most effective it administered at a very young age, however the high school age vaccination may lose effectiveness some 15 to 20 years later, it could be something to do with the body mass at the time of the vaccination. I am told that was vaccinated at age of four.
        Apparently differences in genetic make-up of different populations may explain the difference in efficacy, some studies carried in India show low vaccine’s efficacy in the indigenous population.
        Currently in the UK there are reports that people of Asian and African genetic make up have four times mortality to that of the indigenous British (apparently mostly those of the high age, those with serious underlining medical issues and the overweight not to say obese).
        Boris Johnson may have or have not been vaccinated as a young child it has mixed Turkish/white European ethnicity and was rather overweight, while dozen of his parliamentary colleagues including two or three ministers were infected at more or less same time but none (as far as I know) needed hospitalisation.

        • That’s not what the paper you cited said Vuk. It referred specifically to revaccination policies not vaccination of young children. Also one of the countries supposedly benefiting from this practice is Russia, the table cites Russia as having 1534 cases and 8 deaths, latest data is 187,859 cases and 1,723 deaths!

          • Revaccination is adults not very effective against tb and it is unlikely to be much effective against Covid-19. Only difference between west and east European countries is that east is genetically more compact and all adults were BCG vaccinated as small children. CV arrived there late and they are already lifting lockdowns.
            Hopefully science might be able to find out reasons for the large disparity in death rate.

          • Vuk May 9, 2020 at 12:45 am
            Revaccination is adults not very effective against tb and it is unlikely to be much effective against Covid-19.

            But that’s exactly what the paper you quoted is about.

            Only difference between west and east European countries is that east is genetically more compact and all adults were BCG vaccinated as small children.

            But the difference that was cited in that paper was whether they revaccinated or not.

  36. Finally, I prefer the CEEMD residual method over say a Gaussian smooth because it goes all of the way out to both the start and finish of the data. Not only that, but the information out near the ends is meaningful. Here’s a comparison of the CEEMD residual with a Gaussian filter.

    Willis, I’ve always been suspicious of the technically invalid extensions of spreading convolution or iterative filters out to the end of data, though I’ve never wanted to spend the time to find a case where it does not work. You have just provided such an example.

    The gaussian filter removes all visible trace of the cycle up to the last week where you need to start padding and recycling. There we see that there is a fair amount of the weekly variability which makes it through the filter, though it is strongly attenuated. This clearly leaves a misleading visual impression as to where the underlying trend is going.

    • “CV19 is just a bad seasonal flu that selectively kills oldies with existing health issues”
      Well, yes, it does. The list shows it clearly.
      This virus will limit our life expectancy.

      The medics are exposed to huge viral loads and one has to deconvolute the age of the medics and the probability to die.

    • Same old, same old-

      1957 flu killed over 1 million of all ages. These things happen .

  37. The weekly cycle appears to me an artifact of the reporting process, bunching counts near particular days of the week. Otherwise one has to explain a little conundrum: how does the virus know it is Sunday?

    • Just working hours pattern statistic in the testing labs, hospital emergency units staffing and ‘working from home’ of the data compilers. “Friday afternoon car” syndrome.

    • The weekend effect has been noted by many.
      FFS Cuomo talks about it constantly.

      basically you should be looking at 3 day or 7 day averages

  38. I would like to know how data on deaths is parsed and resolved due to the decision by many hospitals and facilities to include the death of the patient by the presence of the Covid virus regardless of the actual cause of death. A statistical nightmare of resolution.

    • Jeff Smathers,

      I’ve been asking the same questions. I’ve come to find out that a counted COVID-19 death is an opinion. Some further questions are: does a positive COVID-19 death mean no other tests were administered or that a whole range of other tests were administered and they all came back negative? Or just some other tests? What other viruses were present? Does anyone know?

      Andrew

  39. I suspect that a large part of the weekly cycle is simply a failure to report deaths until the next working day.

  40. comes down to one thing , are the numbers correct.

    Withe youtube and Google removing any voices of skepticism we can see that the numbers are not to be believed. It has become political.

  41. Illinois counts all deaths “with” Corona as death “from” corona. “All” not just suspected. It’s not vague in the least. It is over counting. This method is in widespread use across the USA.

    The death counts are NOT to be trusted. I think the best way to measure the increase or decrease of the virus is Covid admissions to hospitals. But try to find that by state on a daily basis. Good luck.

  42. It seems to me that one of the errors of our country is not having carried out a total quarantine from the first moment that it had the first case of coronavirus in the country ….. those sifras had to be less and more controllable … which was the main mistake of the country? underestimate the virus.

  43. Swedish Covid-19 death rate has overtaken Ireland and now is just behind Holland, the next in line is France, which was badly affected and had very strict lock-down.
    http://www.vukcevic.co.uk/EuropeCV.htm
    Is it now question of time when Sweden catches up with France or it may have to introduce a lock-down to arrest rise in the death rate of its population. About 14% of Swedish population was non Swedish born, hence it could be very vulnerable to more serious CV infection/death rate.

      • Hi Richard, Mosher is your expert on Korea.
        Swedish model is in more trouble, in the last 12 hours their death rate has overtaken both Ireland and Netherlands and it is racing ahead, might catch up with France soon. Introducing strict lock down now might take a week or two to slow down the death rate.
        I’ve just updated the link, so have another look.
        What any individual thinks at the moment, be it Donald Trump, Boris Johnson, Vladimir Putin, Kim Jong Un, Xi Jinping, Mosher, Richard, Vuk or anyone else for that matter, at this stage is not much better than a guess. Only thing we can do is follow the data, good or bad, see what happens and hope for the best. Keep safe, just in case.

      • South Korea closed schools, canceled all big events including holy mass, it is mandatory to wear masks if you want to go into a bar, restaurant or club you have to put your name on a list to allow tracking. Tracking App is mandatory as well and the app is using credit card information.

  44. Willis, what you have done is definitely more sophisticated than what I have with UK deaths. But my method is on eline of R code:

    dow=rep(0,7); for(i in 1:floor(length(dM)/7)) dow=dow+dM[(7*i-6):(7*i)]; print(dow)

    And here’s the current result, for Saturdays through Fridays:
    4774 4512 2769 2990 5323 4662 4847

    I don’t need a chi-squared test to see that the sub-3000’s are significant. In a recent email on the subject I said: “Either divine intervention is reducing mortality on a Sunday or less reporting is done on a Sunday.”

    I think we can agree there is a pervasive weekly effect arising from something.

    Rich.

    • Thanks, Rich. Seems to me that it is a reporting issue, not a variation in the actual rates. Viruses don’t have calendars.

      Your way works to determine if the irregularity is there. However, I wanted to be able to remove the fluctuation so that I could see if deaths in a given location have peaked or not.

      w.

    • Check-out the even later reporting on bank-holiday weekends, such as this weekend….todays (Sat) figures are 346 deaths…Friday was a holiday. By Tuesday the figures will have increased, as they do after every weekend.

    • When is lock down ever right-

      The rise in the suicide rate caused by lockdowns in Australia is predicted to exceed deaths from the Wuhan coronavirus by a factor of ten, the Australian reported Thursday.

    • You may be right there Steven.
      Swedish model is in a bit of a trouble, in the last 48 hours their death rate has overtaken both Ireland and Netherlands
      http://www.vukcevic.co.uk/EuropeCV.htm
      and it is racing ahead, might catch up with France soon. Introducing strict lock down now might take a week or two to slow down the death rate.

  45. I am seeing the analogy of bloodletting a patient and the economy…. by allowing the patient to bleed out , the fractional ratio of virus in their blood can almost go near zero. Unfortunately the patient void of the virus is dead…much like the economy.

  46. However the increase in the rate of developing antibodies means that the much sought-after group immunity – where there are so many people immune to the virus that it has little or no opportunity of spreading – is still a long way off.

    Experts generally reckon that it would take a minimum of 70% of people with antibodies before group immunity is present. At a rate of 3% every three weeks, that target would only be attained in late July 2021.
    https://www.brusselstimes.com/all-news/belgium-all-news/110496/national-security-council-used-phone-data-to-help-inform-decisions/

  47. The aftermath of lockdowns will be felt by families for many years to come. There is no model that predicts when the economy will recover. Locking down the healthy made no sense. An attempt to protect the most vulnerable would have made some sense. Of course a focus on treatment is of the highest importance. Statements of the obvious. Here in California our lockdown has been proposed to extend until 4 weeks after there are no virus deaths, at least that is what I gleaned from listening to his newscast. Economy be damned.

  48. Interesting update of euromomo data, more age groups:

    https://www.euromomo.eu/graphs-and-maps

    COVID-19 kills unprecedented number of people in the age 65-74y than the flu, more people <65y, more people from 75-84y and more 85y+ with that age having the smallest difference in relative increase.

  49. “Fighting the inflammatory process
    He also explains that the cytokine storm is the body’s immune state in which it begins to produce various types of substances that on the one hand are designed to fight the inflammatory process, but on the other can also intensify the inflammatory response. So, as a consequence, the patient’s condition can get much worse.
     – Knowing about the occurrence of such a situation, we reach for Tocilizumab – a drug that counteracts the inflammatory storm arising as a result of virus infection – says prof. Życińska. He points out that Tocilizumab is an antibody that blocks an important substance that enhances the development of inflammation – interleukin 6.”
    The Australasian Society for Clinical Immunology and Allergy recommend tocilizumab be considered as an off-label treatment for those with COVID-19 related acute respiratory distress syndrome. It states this because of its known benefit in cytokine storms caused by a specific cancer treatment, and that the cytokine storm may be a contributor to mortality in severe COVID-19.[36]

    On 11 March 2020, Italian physician Paolo Ascierto reported that tocilizumab appeared to be effective in three severe cases of COVID-19 in Italy.[37] On 14 March 2020, three of the six treated patients in Naples had shown signs of improvement prompting the Italian Pharmacological Agency (AIFA) to expand testing in five other hospitals.[38] Roche and the WHO are each launching separate trials for its use in severe COVID-19 cases.[39]

    In March 2020 a randomized study, at 11 locations in China, which should conclude by 31 May 2020, started to compare favipiravir versus tocilizumab versus both.[40]
    https://en.wikipedia.org/wiki/Tocilizumab

    • The Central Clinical Hospital of the Ministry of Interior in Warsaw uses a modern monoclonal antibody to fight COVID-19. – The effects are spectacular – says prof. Katarzyna Życińska from the CSK MSWiA. – Patients who received the medicine after a few days were disconnected from the respirator – he emphasizes.
      So far, 20 patients have been given the drug, all responded “spectacularly”

  50. Willis, good effort with the data that are available to you, and us. You can’t always despair of imperfect data and shy away from them. In this case, the liabilities in case definition, error rates in the tests, and vicissitudes of reporting are severe.

    Thank you for the essay.

  51. I do wonder as to the bone fide deaths FROM covid19 as opposed to WITH!

    I cannot find any site/publication that can,with honesty, supply this information.

    Figures used are as supplied!!!

  52. Weekly reports less? You don’t say Willis.
    Look at this saw tooth!

    https://depts.washington.edu/labmed/covid19/

    The real artifacts affecting the infection rate curves I believe were the publicity of the ‘Plannedemic’ and the availability of tests and testing facilities. There is nothing like looking in the storeroom and finding all the test kits have been used up to crush a curve. The other curve crusher may be medical staff at the coal face realizing they have been had.

    The Mortality curves are affected by the actual Lockdown, Cold weather and the CDC directive to add anything that looks like Covid-19 to the Schedule 1 of the Death Certificate. NVSS directive 24th March.

    https://www.cdc.gov/nchs/data/nvss/coronavirus/Alert-2-New-ICD-code-introduced-for-COVID-19-deaths.pdf

    What I find interesting from the UW Virology dashboard is the virus appears to run at about %10 of those presenting with symptoms in the early period and now at %5 after lock down or warmer weather. Why?

    If the usual suspects were going to choose a virus for a ‘Plannedemic’ initiating the ‘Slump to get Trump’, this was a beauty as it is prevalent everywhere not like say Ebola.

  53. Yitzak Ben Israel hypothesis is the virus peaks on day 42 and declines rapidly after day 56 regardless of country or policy.

    He looked at USA, Israel, Sweden, Germany, Italy, S Korea, etc (20 countries total).

    I find his hypothesis most convincing.

    Him and Michael Levitt

    also: zero lupus or rheumatoid arthritis patients have tested positive for covid

  54. Late to the party on this one…
    I appreciate the analysis of the data Willis, but unfortunately the data is crap. Too many hidden variables, too much incentive to lie, and far too much that’s hidden. For example, here in NJ there is virtually no risk to otherwise healthy people. About 50% of the deaths have been patients in long term care facilities. Once that is understood, the real question is why the rest of us had to go into lockdown? 4,000 excess deaths in a state with 10 million people? That’s basically a rounding error.

    Heartless? Perhaps, but the numbers are still inflated. Every COVID case is a COVID death, regardless of comorbidity? You’ll never get the answer to the question of who many everyday people developed COVID and died. Never. As soon as I see hundreds (literally!) of MSNBC videos attacking the president over the virus, I know that the “science” has been coopted by people who will say anything to make this Trump’s fault.

    Also, their fawning praise of Dr. Fauci was reminiscent of their love for DeGrasse Tyson. Just another media hyped celebrity who happened to believe their narrative. Well, good for him. In the meantime, let the infirm and the elderly continue the lockdown, put them in Hazmat suits if you like, but let the rest of us get back to work. I want the restaurants, sports venues, and bars opened and I hope I never see another mask.

  55. Willis
    Thanks – the advantages of the CEEMD decomposition-smoothing seem very compelling. The performance at the end is particularly impressive. According to Luukko et al., the Finnish authors of the “libeemd” paper you referred to (and posted on ResearchGate),

    https://www.researchgate.net/publication/280114554_Introducing_libeemd_A_program_package_for_performing_the_ensemble_empirical_mode_decomposition

    this is due to a clever approach to the problem of artificial twists at the end due to the truncation of the series:

    Several ways have been proposed to mitigate the end effects by adding artificial extrema to the ends of the data, such as simple wave forms defined by the extrema near the end (Huang et al, 1998). We have adopted the method described by Wu and Huang (2009), where additional extrema are added to the ends of the data by linear extrapolation of the previous two extrema. However, if the extrapolated extremum is less extremal than the last data point, the value of the last data point is used as an additional extremum instead. This method successfully reduces the end effects while avoiding the possible complications of more complex data extrapolation.

    This extrapolation if additional extrema beyond the ends of the data series allowed your fit for instance to “predict” the uptick after the end of the series that happened to finish in a weekend (lower reported deaths).

    Did your method incorporate the “CEEMDAN” modification which improved removal of noise by separately averaging for each IMF (intrinsic mode function)?

    They also discuss the stopping problem which I guess always afflicts iterative methods. The approach they used was again a nice one, to quote:

    Therefore Huang et al (1999) proposed a simpler stopping criteria, in which iteration is stopped when the number of zero crossings and extrema differ at most by one and that these numbers stay the same for S consecutive iterations. This criterion was extensively studied by Huang et al (2003) and the optimal range for the S-number was found to be from 3 to 8. Our code … used S=4.

    Did you have to choose an S value?
    I would be interested to try out the CEEMD method although my software skills are rather limited.

    • Thanks, Phil. I don’t know the answer to the question about CEEDAN. I’m using the function “CEEMD” from the R package “hht”. It has the following form:

      CEEMD(sig, tt, noise.amp, trials, verbose = TRUE,
      spectral.method = “arctan”, diff.lag = 1, tol = 5, max.sift = 200,
      stop.rule = “type5”, boundary = “wave”, sm = “none”,
      smlevels = c(1), spar = NULL, max.imf = 100, interm = NULL,
      noise.type = “gaussian”, noise.array = NULL)

      I’m generally using the defaults. The question of stopping criteria is handled by the “stop.rule” type variable.

      stoprule
       
      stopping rule of sifting. The type1 stopping rule indicates that absolute values of envelope mean must be less than the user-specified tolerance level in the sense that the local average of upper and lower envelope is zero. The stopping rules type2, type3, type4 and type5 are the stopping rules given by equation (5.5) of Huang et al. (1998), equation (11a), equation (11b) and S stoppage of Huang and Wu (2008), respectively.

      I haven’t messed with the stopping rule at all.

      The boundary conditions controlling what happens at the beginning and end of the data are described as “wave”, but I can’t find anything describing the possible alternate conditions … hang on, just thought of another way … ok, tracked it down.

      boundary
       
      specifies boundary condition from “none”, “wave”, “symmetric”, “periodic” or “evenodd”. See Zeng and He (2004) for evenodd boundary condition.

      Hmmm … I suspect that “wave” would be my choice just from the names.

      If you’re not using the R computer language, I can’t recommend it enough. Steve McIntyre practically had to beat me up to convince me to learn it, and he was 100% correct. It’s free, cross platform, and has free “packages” for just about anything you might want to do.

      Onwards, stay well,

      w.

    • Looked good, but their splash screen stopped me dead in my tracks:

      We’re tracking racial and ethnic data from every state that reports it—and pushing those that don’t to start. Together with the Antiracist Research & Policy Center, we’re analyzing this data to uncover the true impact of the outbreak on vulnerable communities.

      Sigh.

  56. Thanks for this week end analysis,
    There are many other aspects, I believe:
    – bank holidays (at Oster, the previous Friday, or the following Monday, May 1st, May 8th, etc.)
    – the way fatalities from nursing homes were added to the deaths from hospitals, e g in France the catch up from nursing homes took place in several days split over approximately 2 week before the reporting became regular
    – at some moment deaths at home will probably be added one way or another
    For this ex appreciating Gauss curves, a Singapore based institute publishes curves for each country including an « end of pandemics date » per country. If trust is to be granted, it would mean for the US an end by mid October, vs end of Sept for UK, mid or end August in Western continental Europe.
    https://ddi.sutd.edu.sg/covid-19
    https://user-images.strikinglycdn.com/res/hrscywv4p/image/upload/c_limit,fl_lossy,h_1000,w_500,f_auto,q_auto/679545/
    594764_580902.jpeg
    https://user-images.strikinglycdn.com/res/hrscywv4p/image/upload/c_limit,fl_lossy,h_1000,w_500,f_auto,q_auto/679545/185197_241486.jpeg
    https://user-images.strikinglycdn.com/res/hrscywv4p/image/upload/c_limit,fl_lossy,h_1000,w_500,f_auto,q_auto/679545/346964_831872.jpeg

  57. Hi Willis,

    What is not sufficiently commented is the curious 7 days periodicity. Why should people, on average, die less on weekends and peak around Wednesdays?

    Best,

    Robert

    • My theory is that it is delayed reporting from outlying stations over the weekend. Virii are sadly lacking in the calendar department.

      w.

Comments are closed.