COVID-19 Fake versus Real Death Count

(Last updated on: September 9, 2020)

What is the Truth of Coronavirus? 

COVID-19 fake versus real

COVID-19: Fake or Real? I surprise myself by posing this question. But I saw a TV interview that left me in shock.

In the interview, a reporter asks a woman a question about the coronavirus deaths reported in her city. And she replied something like this: “It’s all fake. Somebody dies, so they call it coronavirus just to frighten us.” If she had continued, I think she might have said something like, “they just want to control us, tell us what to do” without specifying who “they” were.

I was shocked because I want the facts, the truth. But, as Pontius Pilate said, “what is truth?”

To the woman, I suspect that the truth of “COVID-19 Fake” came from whichever source she uses for news. And for many people, that source is social media or a news supplier that suits their politics.

However, to me and to other scientists, truth is only provisional. In fact, science itself is a search for truth.

Even when science confirms a hypothesis, a possible “truth,” it is subject to revision following a more accurate, more thoroughly proven alternative. For example, CDC initially advised not to wear masks, to save masks for healthcare workers, but later advised that everyone should wear them. And the change in advice does not make science untrustworthy: on the contrary, the change affirms that science works. The conclusions of science change when understanding advances.

            Contents of Today’s Blog

Media Boost COVID-19 Fake Claims

Today, you can find public media that deny the very existence of coronavirus as a threat. And some people – too many people – believe that despite careful rebuttal by Reuters, Axios, Rolling Stone and many others.

But there is a grain of validity in the woman’s televised comment calling COVID-19 fake.

Many people have died during the last few months. And most of them were not actually tested for coronavirus although they may have had some symptoms. How did the physician or coroner decide what to certify as the cause of death? Moreover, if they list multiple causes, how do the health agencies decide which deaths count as coronavirus fatalities?

Science Rides to the Rescue with Excess Death Analysis

Fortunately, the Centers for Disease Control (CDC), using science, have a way to tell how many deaths are due to the presence of COVID-19. And that is to count the total deaths each week in 2020, compared with the number of deaths reported during the corresponding week of the previous three years.

Both statisticians and epidemiologists have endorsed the validity of such an Excess Death analysis to distinguish COVID-19 fake versus real data.

            Excess Death Measurement Has Important Advantages

1. The Excess Death measurement approach is objective. It does not depend upon what is written on a death certificate. And it does not depend upon any counting or data rules that may be biased by judgment, politics, hidden agendas or differences of opinion.

2. In addition, it provides information that’s valuable for public health experts in judging the penetration of COVID-19 and its advancement or its retreat. The data can be used to guide policy and public health rules at any geographical level (national, state, metro, etc.) where there are sufficient deaths for statistical validity. (CDC only treats areas and time intervals where there are 10 or more deaths.)

            Excess Death is a Lagging Indicator

Although excess deaths are a valuable objective measure to distinguish COVID-19 fake versus real data, as an indicator they are delayed in time. They explain the past but don’t try to predict the future. After all, as my figure on Disease Progression showed, it takes 3 to 4 weeks for infection to lead to hospitalization to lead (in many cases) to death.

To determine whether we are controlling coronavirus, earlier measures such as infections and hospital use are more valuable, as given by IHME. And the earliest part, tracking infections, requires a broad program of testing, including testing people who don’t appear to be infected, as some other countries are doing.

In addition, death data is reported in an uneven, spotty manner. As a result, the CDC data contains this comment next to excess death data from February 1 on: Data in recent weeks are incomplete. Only 60% of death records are submitted to NCHS within 10 days of the date of death, and completeness varies by jurisdiction. Thus any conclusions we draw may need adjustment when additional data arrives.

            Excess Death Measurement Has Data Limitations

The Excess Death analysis does have limitations:

1. The excess death measurement tells absolutely nothing about the cause of death of any individual. The data only applies to a large group.

2. Reports of deaths have different time delays in different jurisdictions. Therefore, the analysis has to make good faith estimates where firm data is not available.

3. Moreover, we can never know what would have happened in 2020 without the coronavirus. All we can do is to estimate what would have happened based on recent years. To do this, CDC computes two estimates:

  • The average number of deaths expected based on previous years; and
  • A larger number, the Upper Bound Threshold (UBT), such that there is less than a 5% probability that deaths would exceed that number without coronavirus. That is, there is uncertainty in death rates, which randomly vary above and below the average. If deaths exceed the UBT there is a very high likelihood that they are due to some factor present in 2020 but not in preceding years. The most obvious candidate for that factor is the coronavirus. 

                        All Causes of Death Are Lumped Together

4. Here is yet another point: Excess Deaths measures the additional deaths due to the presence of the coronavirus. But there are many ways that the COVID-19 pandemic might cause death, and this technique does not distinguish between them:

  • Death by the virus infection, acting all by itself.
  • Death by the virus infection, made extra deadly by the presence of other conditions such as heart trouble, diabetes, obesity, pneumonia, advanced age and other factors not well understood.
  • Or death by other diseases, caused because the patient either did not seek or was not able to obtain medical care. Perhaps the healthcare system was overloaded, or perhaps the patient feared infection risk if he visited an emergency room.
  • Death by other causes, such as alcoholism or suicide triggered by life changes caused by the pandemic: isolation at home, domestic violence, depression.
  • Also, death by indirect causes triggered by job loss or income loss.
  • The presence of COVID-19 may also have prevented some deaths. For example: fewer concussions due to the suspension of team sports; fewer drunken brawls due to the closure of bars; fewer auto accidents due to less commuting to work. (COVID deniers sometimes quote such examples as “proof” that COVID isn’t so bad. However, the fact that there are significant total excess deaths shows that any offsetting negatives are less than the positive factors.)

            The Bottom-Line Limitation of Excess Deaths

The most important drawback to measuring Excess Deaths is that it measures both the direct and indirect fatal effects of the pandemic. Some people will use this duality as an excuse to deny the validity of Excess Deaths as a measure. They may do this so that they can advance a political agenda that asserts either a larger or smaller death count than data would otherwise show. However, such denials do not change the fact that people have died, many of them while suffering from the complications of coronavirus.

            Excess Death Measurement Also Has Logical Limitations

1. In a sense, there are no “Excess Deaths” because every one of the deceased people would have died eventually anyway. Excess Deaths only makes sense when measured for a particular time period (one week in the case of the CDC data). Perhaps a better term would be Accelerated Deaths, or Premature Deaths, or Lives Shortened.

2. Moreover, if victims suffer premature deaths due to COVID-19, then there should be fewer deaths in the following weeks, months and years. However, the death statistics from some jurisdictions are not final for up to one year even in normal times, and during a pandemic data may be delayed or absent for various reasons. CDC does not attempt to compute death rates below normal, because they can’t tell whether the rate is truly lower or whether events have interrupted the reporting of deaths. In addition, the decrease in death rates will be spread over a number of years, during which other changes may occur (A cure for cancer? Prevention of dementia? Better treatment of heart disease?) that confuse the analysis by reducing deaths due to non-COVID causes.

The CDC Excess Deaths Dashboards: How to Distinguish COVID-19 Fake versus Real

I highly recommend the CDC Excess Deaths page for your attention. It has thirteen dashboards with amazing layers of detail. And the charts can be downloaded in many formats, including Tableau Workbooks which can be studied using a free reader app.

I will show a few samples of the data available from this important resource.

Total Predicted Excess Deaths Due to COVID-19

Dashboard #4, “Number of Excess Deaths,” provides a chart like the following. CDC updates its data weekly; this version was complete through September 2.

COVID-19 Fake versus Real

You can choose which states to include in the display, and they are displayed in order of their totals. Here is what the headline summary numbers mean:

  • There were 253,841 deaths more than the average expected based on the previous three years. Some part of this total might be a normal fluctuation that occurs from year to year.
  • There is at least a 95% chance that there were 190,912 deaths more than would have occurred in a “normal” year, without coronavirus. Therefore, at least this many deaths can be confidently attributed to COVID-19.

            What Does This Death Count Signify?

Let’s gain some perspective. Dashboard #1 (of which the first image shows part) shows that actual US deaths from all causes during 2019 averaged 54,852 per week. From February 1 through September 2, the time period covered in this chart, there were 30 weeks, during which one would expect 30 X 54,852 = 1,645,560 deaths. And during an entire year, one would expect 52 X 54,852 = 2,852,304 deaths.

  • Some folks look at these numbers and say, oh, 200,000 extra deaths? That’s only 7% of last year’s deaths (thus far), that’s not much.
  • Others say, from March 28 when excess deaths became significant until September 2 there are 22 weeks, when we would expect 22 X 54,852 = 1,206,744 deaths. And 254,000 is 21% of that number, which is an alarming increase.
  • Even worse, during weeks 14 to 19 of this year, when deaths peaked, there were 34% more deaths than in the corresponding weeks of 2019!
  • Still others say, 675,000 died from the H1N1 flu virus in 1918-1919 in the US. And the US has three times as many people today. Thus although coronavirus is still killing people, so far its percentage of deadliness is only one-tenth that of the 1918 pandemic.
  • Other people point to the 61,000 deaths from the 2017-2018 flu season and say, coronavirus has already killed three times that many people. It’s really dangerous!

So, how serious is COVID-19, anyway? It depends on your point of view.

My advice is, take precautions and try not to catch it! 

COVID-19 might turn out to be minor in the broad expanse of history. However, if you acquire a coronavirus infection, it may disrupt or even end 100% of the life of you and others around you. And that is not minor for you. Please, stay safe!

CDC Dashboard #6: Deaths by Age Group

The CDC dashboards present lots of other information, which you can display for all the US or by state. I will show you several that I found especially helpful.

Dashboard #6 shows “Weekly Number of Deaths by Age”:

            Art’s Analysis Approach

To analyze the data behind this chart I needed a systematic way to flag the weeks in which something was noticeably out of kilter compared with pre-COVID dates.

I didn’t want to treat every increase in 2020 as COVID-related, because that would mix up COVID cases with the normal year-to-year fluctuation that occurs. Therefore, I decided to calculate the average deaths in the first 4 weeks of 2020 to establish a norm prior to the known presence of coronavirus. I then designated each week with at least 10% more than that many deaths as a “significant week” in which there are unusually more deaths than the year’s norm.

I’ve looked at several different threshold numbers, and I like 10%. Besides being a round number, I found that 10% is large enough to weed out most of the spurious fluctuations. In addition, I tried 11%, 12% and 13% and found very little difference in the results.

For the significant weeks I then averaged the 2020 actual deaths and the 2015-2019 average deaths, and calculated the percentage increase: (In the CDC data, weeks are numbered consecutively beginning with the first Saturday of the calendar year.)

            Art’s Data Summary – Age Groups

Age GroupSignificant Week Numbers2020 Actual Deaths, Avg Sig Wks2015-19 Actual Deaths, Avg Sig WksPercent Increase
Under 25 years23, 24, 281,3201,2654.4%
25 – 44 years14 to 303,4952,62833.0%
45 – 64 years14 to 1913,36510,29929.8%
65 – 74 years14 to 1914,43110,09642.9%
75 – 84 years14 to 1917,94912,61742.3%
85 years & older14 to 1922,95316,56738.5%

You can see that the significant weeks are mostly weeks 14 to 19 of the year, as in the previous figure. And as a resident of the penultimate category, I found these numbers to be pretty interesting.

Yes, it’s true that coronavirus is relatively more deadly for people older than 65. However, the difference is not immense. The most obvious distinction is that older people have a higher expected death rate even without coronavirus. Being older only adds about 10 points to the relative virus risk compared with people under 65. In round numbers, you could say that if you’re over 65, the presence of coronavirus increases your risk of dying by about 40%. And if you’re 25 to 64 years, the increased risk is more like 30%. Whether or not that difference is significant depends on your perspective.

What about those under 25, who are also those most likely to socialize unsafely and acquire infections? Their risk of dying is already quite low, and COVID-19 only adds 4.4% to that risk. That is, if you’re under 25 you are much less likely to die from coronavirus than anyone a few years older. COVID-19 may make you very sick, and may leave you with long-term impairments, about which we don’t yet have good data. However, if you’re young, usually the virus does not immediately kill you.

CDC Dashboard #7: Deaths by Race / Ethnicity

CDC dashboard #7 shows “Weekly Number of Deaths by Race / Ethnicity”:

Once again, for each ethnic group I calculated the average deaths in the first 4 weeks of 2020 to establish a norm. I then designated each week with at least 10% more than that many deaths as a “significant week” in which there are unusually more deaths than the year’s norm. For those significant weeks I then averaged the 2020 actual deaths and the 2015-2019 average deaths, and calculated the percentage increase:

            Art’s Data Summary – Race / Ethnicity

Ethnic GroupSignificant Week Numbers2020 Actual Deaths, Avg Sig Wks2015-19 Actual Deaths, Avg Sig WksPercent Increase
White Non-Hisp14 to 1752,31841,74125.3%
Black Non-Hisp13 to 23, 28, 309,9136,21259.6%
Hispanic13 to 326, 2043,67768.7%
Asian Non-Hisp13 to 212,1661,24773.7%
Native Non-Hisp16 to 3045232339.7%
Other13 to 19, 3071846654.9%

The significant week numbers are somewhat more scattered than was the case with various age groups. However, the peak is mostly in the range 13 to 19.

As has been often pointed out, the COVID-19 death rate is relatively higher for African-Americans than for whites. Moreover, it’s larger for all the non-white groups than for whites. In fact, the highest coronavirus death ratio based on excess deaths is for non-Hispanic Asians! However, we don’t hear as much about this group.

The Native groups see somewhat less effect from coronavirus than the other non-white groups. How can this be? Well, native communities may have limited access to healthcare, which would increase their mortality. However, they may also be protected by being physically isolated from infectious groups.

There has been lots of discussion about why whites appear to be especially favored in COVID-19 mortality. Most explanations have centered on socio-economic status, access to health care, exposure at work and cultural factors.

And CDC Dashboard #11: Deaths by Cause

CDC dashboard #11 shows “Change in Number of Deaths by Cause.” It classifies deaths by the “underlying cause of death” as reported:

This chart is not very informative until you look at the underlying data. With that data, I performed the same analysis as in the other charts:

Underlying Cause of DeathSignificant Week Numbers2020 Actual Deaths, Avg Sig Wks2015-19 Actual Deaths, Avg Sig WksPercent Increase
Hypertension (blood pressure)14 to 172644170654.9%
Ischemic disease (coronary artery disease)159173712428.8%
Alzheimers & dementia14 to 176650494434.5%
Diabetes14 to 172250159541.0%

What can we learn from this data?

  • Hypertension: If you have high blood pressure, the presence of coronavirus greatly increases your chance of dying. That may be due to synergy between the diseases, or because during the pandemic people with heart problems are less likely to obtain treatment.
  • Coronary artery disease: This heart condition somewhat increases your virus risk, but not nearly as much as does high blood pressure.
  • Alzheimers & dementia: These conditions increase your risk of dying if you acquire a COVID-19 infection. That may be because many demented patients are elderly, and housed in nursing homes where the chance of catching coronavirus is particularly high.
  • Diabetes: Coronavirus greatly increases your risk of dying if you have diabetes.

The message seems to be that the mortality risk from each these conditions is magnified, in some cases substantially, when coronavirus is circulating. However, the mortality from other diseases listed in the figure (COPD, other circulatory diseases, cancer, renal failure, sepsis) is not magnified by the presence of COVID-19.

Summary: What Have We Learned About COVID-19 Fake versus Real Data?

Let’s summarize what the above discussion seems to show:

  • In 2020, there are undeniably many more deaths from all causes than in previous years. At the same time, hospital intensive care units and ventilators have been oversubscribed treating very ill patients. The simplest conclusion from these facts is that many of the deaths are directly or indirectly caused by coronavirus, which was not a factor before 2020 but which is a factor this year. COVID-19 fake? I don’t think so!
  • Coronavirus appears to be more deadly than the 2017-2018 flu, but less deadly than the 1918-1919 H1N1 flu pandemic.
  • When people under 25 catch coronavirus it usually does not kill them (although they may experience aftereffects). Those over 25 suffer about a 30% increased mortality rate. For people over 65, that goes up to 40%.
  • When white Americans catch the coronavirus their mortality rate goes up noticeably, about 25%. However, non-white Americans experience an even bigger boost in mortality, anywhere from 40% to 74%. The higher impact of COVID-19 on all ethnic groups other than whites suggests that the difference might arise from socio-economic and cultural factors that give whites less virus exposure and/or better access to medical care.
  • Many of us live with chronic disease conditions which we can usually manage reasonably well. However, there are specific conditions whose normal probability of death is increased if coronavirus is also present. The riskiest conditions appear to be high blood pressure, coronary artery disease, dementia and diabetes. In contrast, if you have other circulatory diseases, COPD, cancer, liver disease or sepsis (a serious infection), the mortality associated with those conditions is not increased very much by COVID-19.

I hope that this discussion helps you distinguish COVID-19 fake versus real facts and information. And I hope that this introduction to the CDC Excess Deaths dashboards may inspire you to browse the data base for information that applies to you. I wish you the best of health!

Image Credits:
– Cartoon viruses by liftarn on openclipart.org
– All other images in public domain, adapted by Art Chester from the CDC Excess Deaths page

 


Comments

COVID-19 Fake versus Real Death Count — 6 Comments

  1. The most detailed analysis of this deadly disease that I have seen so far. Masterful explanation and interpretation of the complicated and confusing statistics.
    Great job Art.

  2. Thank you for this. Another thing I like to point out is that this is more contagious than many other things that put people in hospitals, and so we’re putting first responders and healthcare workers at more risk than they signed on for.

    • Thanks Cheryl, that’s a very appropriate observation. Healthcare jobs are already risky but this is a whole new level of risk. And even jobs such as teaching and bartending have much more risk than ever thought possible. The folks who take risks to serve certainly deserve our sympathy and praise!

  3. As we know Pilate did not want the truth and Jesus was vague about “his” truth. Do not ask the question if you do not care to know the answer.

    If you do find out how reliable the death count in the US is let us know. Note that now the number of deaths is approaching 200,000. Total US deaths in all wars since 1945 were 102683 https://en.wikipedia.org/wiki/United_States_military_casualties_of_war

    • Thanks, Linos! Unfortunately, when there are multiple causes of death, which deaths should be attributed to coronavirus has become a political issue. For that reason, statistics may be being tallied differently from one state to another, so computing a US total might be iffy. However, taking data from one state and comparing it with past data from that same state might be reasonably reliable.
      Your comparison with wartime deaths is unfortunately relevant, in underlining the seriousness of this disease.

Leave a Reply

Your email address will not be published. Required fields are marked *