Editor’s Note: Thanks for all those who responded with thoughtful replies and critiques. You are correct about the CDC revisions, which can be large. That’s certainly the case with week 14, which was revised up above 61,000 deaths, which is a lot. That came after the research for the piece was completed last week. Because of the revision, it pushed the death rate per 100,000 up to 18.5. That’s a big number.
More importantly, it pushes the average for the data to the year to date to 16.7, which is slightly above the 16.4 long-term average. Please recall what we wrote in the piece: “As of the first week of April, the last week for which there is solid data, the U.S. overall death rate including COVID-19 fatalities was about 14.9 deaths per 100,000. Since 2013, the U.S. average has been 16.4. During past flu epidemics, the rate got as high as 20.”
So yes, because of those upward revisions, we are now above average in our long-term death rate up to week 14. We stand corrected, not because of error but by dint of new data that superseded the old. Thanks to all of you for calling attention to the changes, and for all your other insights into the data.
But the broader point we tried to make still stands: That is, that the average rate of deaths nationwide has not increased in a way that would suggest a massive pandemic, at least not yet. We’re still about at the average rate. And we used that rate, rather than the raw number of deaths, because raw death numbers mean little given our growing population. It’s the average per unit of population over time that matters.
Along with many of you, we’ll be watching closely in coming weeks to see if the rate does indeed spike as new data come in, or whether it will flatten out. Obviously, we hope we’re right and the death rate stays low, because it would mean we can all go back to our normal lives sooner.
The numbers used in I&I’s piece were the ones we felt most trustworthy at the time. Again, later unforeseen revisions changed the picture somewhat, but not radically. At least not yet. Thanks so much for all your analyses and critiques. We’ll keep them in mind when next we visit this topic.
The story, as printed, follows:
With more than 55,000 deaths attributed to it, COVID-19 would seem to remain a major U.S. health threat, justifying the closure of the U.S. economy. But if so, why is the overall U.S. death rate declining in the midst of a deadly pandemic?
It’s a curious thing. In recent weeks, despite a growing COVID-19 infection rate (due mostly to a jump in testing) and rising fatalities, the overall U.S. crude death rate (total deaths per 100,000 population from all causes) is falling.
That’s right, falling.
“Despite the panic over the Wuhan virus, it now appears that the overall U.S. death rate this winter season is at a multi-year low, no worse than 2014, 2016, and 2019, and far better than 2015, 2017, and 2018 (when we were hit with one of the worst flu seasons in years),” writes blogger Robert Zimmerman.
Let that sink in for a minute.
Centers for Disease Control data indicate the current average U.S. overall death rate, even amid the coronavirus pandemic, is actually lower than it has been for much of the past seven years and well below the long-term average.
As of the first week of April, the last week for which there is solid data, the U.S. overall death rate including COVID-19 fatalities was about 14.9 deaths per 100,000. Since 2013, the U.S. average has been 16.4. During past flu epidemics, the rate got as high as 20. You can find the data here.
There are many possible reasons for this.
For one, the actual number of flu deaths has declined in recent weeks. From the beginning of March, the weekly number fell steadily from 568 to a little more than 200. However, the number of deaths attributed to pneumonia jumped sharply from 3,446 to 6,017, the highest rate since at least 2013.
“The percent of deaths due to pneumonia or influenza (P&I) is high but the increase is due primarily to COVID-19, not influenza,” the CDC noted this week.
Meanwhile, with much of the U.S. in semi-lockdown and not commuting daily to work, the number of traffic fatalities and work-related deaths is surely down.
While reliable recent data aren’t yet available, traffic deaths are in the midst of a long-term decline.
“In 2019, an estimated 38,800 people lost their lives to car crashes – a 2% decline from 2018 (39,404 deaths) and a 4% decline from 2017 (40,231 deaths),” according to the National Safety Council. So it’s reasonable to expect a fairly large drop.
A smaller impact could be felt by the many people working at home, thereby reducing workplace fatalities. Some 5,250 people died in 2019 from workplace accidents, a record. So any decline in that would show up in lower overall death rates.
So why are death rates falling while COVID-19 deaths continue to grow? It may be that many deaths that aren’t really COVID-19 related are counted as that anyway. The net effect is to exaggerate COVID-19 deaths, while underestimating others.
Employing simple arithmetic, in a pandemic you would expect a surge in overall deaths, all things being equal. Could it be that the number of COVID-19 deaths has been inflated? There’s strong evidence for this.
Pennsylvania, for instance, recently removed some 200 deaths previously recorded as COVID-19 fatalities from its data after coroners complained that state numbers and methods for tallying virus deaths didn’t match their own.
“Pennsylvania Health Department officials had included in their count ‘probable’ coronavirus deaths in cases where they believed the virus was the cause of death but did not have confirmation from a positive test result,” according to the Philadelphia Inquirer.
Similarly, in California, a 37-year-old man recently overdosed on non-prescribed drugs. But because he tested positive for COVID-19, it was declared a COVID-19 death.
But New York’s COVID-19 death reporting is even worse, if that’s possible.
The Empire State has fewer than 6% of the nation’s population but nearly 30% of all U.S, COVID-19 deaths, or nearly 17,000 deaths out of a national total of 55,383.
On April 13, New York mysteriously added 3,700 new COVID-19 deaths to its tally, jacking it even higher. Did it miss a bunch of deaths? No. Officials just started counting as COVID-19 fatalities the deaths of some who died at home or in nursing homes and hospitals, even if they didn’t get tested.
This exaggeration of death counts has added to the sense of panic that led to the lengthy national shutdown of our economy. There is no clear national standard. Indeed, a recent Washington Post investigation found that COVID-19 death count methodology varies wildly from state to state.
So do we really have 55,000 COVID-19 deaths? Not likely. Meanwhile, evidence is growing that the total number of people exposed to the virus is much greater than thought. That means the fatality rate is far lower than first estimated.
It’s important to note that the wild inaccuracy of supposed experts led to a U.S. overreaction and the disastrous shutdown of our economy, which could end up costing us $4 trillion in lost output, by some estimates. That includes former President Barack Obama and current Joe Biden health care adviser Dr. Zeke Emanuel, who predicted as many as 100 million infected and a million deaths.
A recent Stanford University study suggests that the true death rate from COVID-19, rather than being 3% or higher, as the World Health Organization predicted, is really about 0.1% to 0.2%. That’s about the same as the flu.
Studies elsewhere, including in Los Angeles and New York, have found the same. COVID-19 exposure is much more widespread than thought, so reported death rates are likely much lower.
“‘Coronavirus deaths’ soared by assuming that people dying with the virus were dying from it and then by ascribing to the coronavirus other deaths among people with symptoms of pulmonary distress, even without being tested,” wrote American Enterprise Institute economics fellow and futurist George Gilder in RealClear Markets.
Two California doctors recently did a deep dive into the data and concluded: “Lockdown versus non-lockdown did not produce a statistically different number of deaths. That is the bottom line.” Other looks at the data have found the same thing.
The point is, Americans got stampeded into closing the economy for a pandemic that may end up no worse than a very bad flu season. And for what? States now are moving to end their lockdowns. It’s about time.
— Written by Terry Jones
And YT pulled those Doctor’s video last night…less than an hour after they were interviewed on Laura Ingraham!
You could not put out a video of this on YouTube. It turns out that if you contradict the “never wrong” WHO, you are violating Google’s opaque “Community Standards”. That is why the doctor’s video was taken down.
Actual statistics shows that 95% of the COVID-19 fatalities are from underlying illnesses. The overwhelming majority of people DO NOT have any significant risk of dying from COVID-19. Why not protect JUST the older, at-risk?
and the obese?
Fake Pandemic.. Just a Bad Flu Season hyped up to get power back.. Democrats Run Media is working for China..We are at War and nobody wants to except it.
‘accept it’
Our death rate is down because Covid-19 had already run havoc through our assisted living homes! There are quite a few less individuals that are already near death to be assisted by Covid-19.
I think it would be very informative, and extremely good journalism, if an organization could actually list the number of deaths per state where victims were already in an assisted living facility. In our county the majority of deaths occurred in that population. From the number of assisted living homes that have been reported infected it seems they are the majority of victims.
So please do some good reporting. That just may put quite a few million people more at ease and assist in opening up our country!
The 2020 death rates are incomplete, and 3/4 of Covid deaths have occurred after April 7.
I agree. The linked post, mentions week 14 having 49,292 deaths, but it’s now up to 60,195 from the same CDC data source. Judging from that fact alone, the analysis didn’t fully account for the time for the CDC to receive latest death totals. If you re-run the figures as of today, there’s a dramatic upswing in total deaths (all causes) across weeks 13-14.
It appears that the title and overall premise of the linked article and this article are incorrect now. The US started on a multi-year low pre-COVID-19, but has taken a significant turn for the worse over the last two to three weeks.
Please read the top of the article – it acknowledges the updated data but you will find the rest of the post’s premise remains valuable.
I posted my earlier comment prior to the editor’s note. Really appreciate the editor’s update to clarify for future readers. I do disagree with the reaffirmation of the broader point that “the average rate of deaths nationwide has not increased in a way that would suggest a massive pandemic, at least not yet.”
Plotting the CDC data points and controlling for population increase still indicates a significant spike in total US deaths, which is cause for continued concern. Please stay safe and continue to focus on the data over political and media viewpoints.
One last CDC data split by year and state for reference: https://i.imgur.com/yHIYhos.png (CDC data source: https://www.cdc.gov/nchs/nvss/vsrr/covid_weekly/) Continue to be vigilant, especially if you live in these hardest-hit states!
If you look at total mortality for the nation (https://gis.cdc.gov/grasp/fluview/mortality.html then hit the green download button for weekly deaths back for five years) you will see that for the first 15 weeks (the furthest week into 2020 with up to date info) total deaths is as follows:
2016: 825,308
2017: 866,478
2018: 892,029
2019: 865,242
2020: 869,881
So, YTD is very similar to 2 recent previous years (without factoring in population growth) and still 22,148 behind 2018. After I ran these numbers, I recalled that 17/18 was a harder than normal flu season, so I ran 17/18 vs. 19/20 adding the last 4 weeks of the preceding year. The 17/18 total is 25,976 higher (1,125,589 minus 1,099,613). There has been about 2% growth in the US population since 2017, so there would need to be about 1,045,000 deaths in 19/20 to be the same percentage as 17/18 – meaning that the current death tolls, as influenced by Covid 19 deaths, would have to be about 46,000 higher to match the mortality of the flu season 2 years ago.
The next month will probably bring the current toll to an equivalent of the 17/18 season but the death tolls are declining. In addition, there are two additional factors to consider. When Dr. Birx announced that the public totals would include deaths where the virus was a “contributing” factor (versus cause) as well as “probable” we were assured of an overcount. How much is yet to be seen but there will have to be some adjustment. Additionally, in order to truly gauge the ‘value” of the lockdowns, we will have to track the excess mortality that is being caused by the lockdowns. Empty ERs and delayed testing and treatments indicate that at least some of the excess mortality is the result of these draconian measures.
Ultimately, the underlying point – that the outbreak’s impact is not at a “massive pandemic” level but more at a bad flu season level remains valid.
Also, people are avoiding the hospital so there are many less medical mistakes made as well as hospital caused infections.
Unless those pushing this contention have better data than CDC, I believe their conclusion is premature. There is a 1-8 week and sometimes longer delay in reporting deaths to CDC, meaning that the only weeks of this year which are truly current only go back to end of Feb./early March.
I believe we’ll ultimately find that total US deaths went up slightly compared to prior years, and that almost all if not all of the increase will be traceable to the COVID-19 trouble spots (NYC, Long Island, and north suburban counties; Metro Detroit; Louisiana; Massachusetts; most of New Jersey; Rhode Island; DC; Cook County IL; and parts of PA) and nowhere else.
No, this is not correct! The linked article makes a fundamental error in its interpretation of the data. The 2020 CDC data on all-cause mortality show an INCREASE of roughly 10,000 deaths per week for weeks 14 and 15 of 2020 over the same period in 2019.
If you look back at previous CDC updates, for example https://www.cdc.gov/flu/weekly/weeklyarchives2018-2019/data/NCHSData17.csv you will see that there is always a precipitous “drop” in mortality over the last month or so of data. I presume this is due to slow reporting and collection of deaths, but I don’t know. I do not have an opinion on the reliability of the CDC data; I can only tell you what it says. Regardless of the cause, the only way to make useful year-on-year comparisons for recent weeks is to compare against the data that was being reported at the same time last year. If you look, you will see that the number of deaths for week 14 of 2019 reported at the time was 49054, but if you look at that same week as reported in the latest data, it is 56594. Obviously 7500 people didn’t retroactively die in April 2019. Rather the CDC is retroactively updating their numbers as new data come in.
I have a spreadsheet that compiles some representative CDC data and shows graphically what is happening. I will gladly share it.
The virus is here and like other viruses will continue to infect until herd immunity is achieved. THEREFORE:
* Lockdowns merely slow the spread of infection without significantly reducing the final number of cases, and make sense only if medical facilities are being overwhelmed, which they’re not.
* The number of deaths will depend on the efficacy of treatment.
* Testing provides little benefit over the long run except accurate statistical analysis.
The President and his task force are aware of all these realities and have acted accordingly. His critics are TDS asses, peddling disinformation.
The CDC total death data shows a significant spike over the last month.
Pulling the CDC data does show a spike in overall deaths for the 14th week of the year, which corresponds to around April 8th:
2019 – 2020 week 14 = 60,195
2018 – 2019 week 14 = 56,596
2017 – 2018 week 14 = 55,294
2016 – 2017 week 14 = 55,160
2015 – 2016 week 14 = 54,066
2014 – 2015 week 14 = 52,696
2013 – 2014 week 14 = 50,749
The average for previous years = 54093.5, so this season is about 11% above the average of the previous 6 years, and about 6% higher than the 2018-2019 season.
Other data sources often mention that the most-recent data is usually preliminary and partial as more data comes in over time.
This is the CDC’s comment on the timeliness of their data:
“NCHS Mortality Surveillance System data are presented by the week the death occurred at the national, state, and HHS Region levels. Data on the percentage of deaths due to P&I on a national level are released two weeks after the week of death to allow for collection of enough data to produce a stable percentage. States and HHS regions with less than 20% of the expected total deaths (average number of total deaths reported by week during 2008-2012) will be marked as insufficient data. Collection of complete data is not expected at the time of initial report, and a reliable percentage of deaths due to P&I is not anticipated at the U.S. Department of Health and Human Services region or state level within this two week period. The data for earlier weeks are continually revised and the proportion of deaths due to P&I may increase or decrease as new and updated death certificate data are received by NCHS.”
The premise of the article is incorrect. The linked article makes a fundamental error in its interpretation of the data. The 2020 CDC data on all-cause mortality show an *increase* of roughly 10,000 deaths per week for weeks 14 and 15 of 2020 over the same period in 2019.
If you look back at previous CDC updates, you will see that there is always a precipitous “drop” in mortality over the last month or so of data. I presume this is due to slow reporting and collection of deaths. I do not have an opinion on the reliability of the CDC data; I can only tell you what it says. Regardless of the cause, the only way to make useful year-on-year comparisons for recent weeks is to compare against the data that was being reported at the same time last year. If you look, you will see that the number of deaths for week 14 of 2019 reported at the time was 49054, but if you look at that same week as reported in the latest data, it is 56594. Obviously 7500 people didn’t retroactively die in April 2019. Rather the CDC is retroactively updating their numbers as new data come in.
The analysis of this data is wrong. The CDC data used is noted very clearly to be by flu season (2019-2020 wk1 = 1st wk Oct’19) and not by calendar week (2020 wk1 = 1st wk of Jan’20). The data being used here ends in January, before COVID. This article should be taken down.
I think it’s important for all of us to stop soft-pedaling the issue of the number of deaths attributed to COVID-19 and how those deaths are determined. The medical officials are lying (pretend that’s in all caps). They. Are. Lying. And those lies are parroted throughout society, serving as justification for crushing people’s livelihoods, trashing the constitution, and inserting fascism into nearly every nook and cranny of government — all while shaming anyone who has the temerity to question this headlong dive into madness and doling out praise to whoever exhibits the most sheep-like behavior. If we simply and relentlessly tell the truth about what’s going on, we’ll be on our way to the solution.
They aren’t lying so much as they are utilizing differing categorization criteria for COVID than we have used for the annual flu. This was directly advised by CDC and has been advocated for ad nauseum by Dr. Brix (sp?). There has never been a completely accurate template for reporting cause of death, it varies not by just each state but by each doctor. The assumption being that over the entire data set those doctors who are more liberal with their findings will be counteracted by those who are more conservative. In this current COVID panic, the entire dataset has been shifted to more liberal by policy.
Well, and of course Medicare having tweaked payment terms for cash strapped hospitals such that the COVID death designation is worth money, that is also fueling COVID death designations.
I am estimating roughly 20-30% overcounting of COVID deaths vs the same parameters for flu reporting. It is still fair amount of sadness for those thousands of families for sure, but it is still being counted inaccurately.
The simplest explanation for the lack of excess deaths: CDC is way behind in its reporting. It is a part of the guvment after all. In addition, Dr. Erickson made a fundamental mistake in extrapolating COVID cases that is so sophomoric as to call into question many of his other assertions. He extrapolated from tests, not from a random sample of the general population. It gives you a much larger number of infections. I’m in full agreement that these lockdowns should be lifted in almost all areas, but we have to get our facts and methods of interpretations correct!
what frightens me the most…how many people were so willing to give up all their rights simple because they were frightened. these same people were all too willing to turn on their friends and neighbors also
One minor note. Statistically you are much more likely to die in home accidents than work ones. OSHA has actually ended up making the workplace safer than your house. While I doubt it will move the needle overall enough to notice, the stay in place orders will actually increase accident deaths.
Interesting response to Dr. Erickson
While the author’s candor in updating the article based on updated data is to be commended, do not think that this reduces the value of the original article and its key take-aways. I reviewed the CDC data just minutes ago and confirmed that the raw death totals for the first 14 weeks of 2020 remains significantly lower than the previous 2 years (about 6000 and 9000 lower, before accounting for population growth).
This is important for two reasons – first, Worldometer indicates there were roughly 23,000 Covid 19 deaths by then. This confirms that there has to be a significant reduction in deaths by other causes for this period – at least on paper.
As Carl Heneghan, an epidemiologist and director of the Centre for Evidence-Based Medicine at the University of Oxford indicated in the April 1 edition of BBC Future; “During an epidemic, doctors are more likely to attribute a death with complex causes as being caused by the disease in question – a trait known as ascertainment bias. We know, during an epidemic, people will call every death as though it’s related to Covid-19. But that is not the case,” says Heneghan. “Always, when people look back at the case notes and assign causation, they realise they will have overestimated the case fatality in relation to the disease.” The reason for the bias is that “there’s a tendency to focus on the worst-case scenario”, says Heneghan. “That’s the only message that gets out there.”
In other words, especially given the extremely high ratio of elderly patients with serious comorbidities, many of these deaths are likely “with” not “from” the virus.
The second most important aspect of these numbers is that it logically eliminates the claims that Covid deaths were undercounted because we now know it appeared earlier than originally reported. With mortality below average for these 14 weeks, the number of missed cases is highly unlikely to be significant. This should have been obvious based on historical patterns for disease outbreaks (essentially and inverted pyramid) but there are those determined to make this situation look worse than it is.
Does it seem likely that the death rate will actually rise for the rest of April? Probably. There have been about 40,000 deaths added in the 3 weeks since then. 13-14,000 per week will have a lot of impact on a number that runs around 53,000 for a median. However, the highest weekly total for the years included in the CDC download is over 67,000, so probably within or near the historical norms. Since daily deaths appears to have peaked about 10-14 days ago, it appears that the rate will decline, again likely keeping mortality in the normal range.
Finally, there is another factor people are just beginning to track that may have to be accounted for in the death totals. Several European countries – particularly in the British Isles – are reporting higher overall mortality than average since the lockdowns, which public health officials believe cannot be explained by the outbreak alone. Growing evidence (especially in terms of dramatic reduction in ER visits) indicates that people are dying at home either fearing to go to the hospital or having been shut out for “non-essential” treatments.
Its bad enough that the overreaction is killing the economy (with all the long term mortality problems inherent in that), If it also turns out that we have just started killing one at-risk population in lieu of another, there is going to have to be serious ramifications for the public health officials who made these decisions.
I commend you for printing a correction; however, I’m puzzled as to why the original article is still online, including its inaccurate headline “…Why’s U.S. Overall Death Rate Down.” The fact is, the overall US death rate is substantially increased during the relevant weeks, and the earlier post relied on CDC data that were not complete. As of today, the CDC website reports that during the week ending April 4, 2020, 64,313 Americans died (115% of expected deaths), and during the week ending April 11, 66,577 Americans died (120% of expected deaths). In NY City, total reported deaths since early February are more than double (219%) the expected amount. And, those figures will only continue to rise as CDC receives more death certificate records from the states (which can lag behind by as much as 8 weeks or more).
See: https://www.cdc.gov/nchs/nvss/vsrr/covid19/index.htm
The entire premise of the original article was wrong, including its headline. Why not just remove the article, and print only the retraction?
Because the premise is not wrong – just that the degree is not as great as they indicated in the original. Please note my post above for a more detailed explanation but the point you are missing is that mortality for the season (through the date used in the original) remains low compared to previous years.
Moreover, as bad as the 2 weeks you cite were, they are comparable to the highest 3 weeks shown in the CDC download (actually a bit below the peak – and I have not corrected for population growth – a minor but not irrelevant factor).
The numbers are high right now but not unprecedented. The mortality numbers are already dropping, so we are going to see a very rough April but probably not one that goes significantly beyond historical ranges. Worldometer indicates that the two weeks following the 2 you cited experienced about 15 and 16 thousand deaths related to the virus, respectively.. Adding to the median number of deaths per week this would represent about 68 and 69 thousand deaths (versus a historical peak of 67.5 in the CDC data). When accounting for about a 2% growth in US population since that peak week, the numbers look like they will come out statistically even.
This is not to make light of those deaths but to add perspective. Previous flu seasons have been in the same range, there remain serious questions about the number of deaths caused by the virus (vs. with), a growing set of questions about excess mortality caused by the lockdown, and there is no scientifically valid way to quantify the alleged benefits of the lockdowns. However, we can absolutely compute the cost – at trillions of dollars and going up. There is a point where society has to make decisions about cost/benefits. As we find out that the virus is more prevalent than supposed and less lethal, it becomes harder and harder to justify the extreme measures and costs of the response.
The CDC data has again been revised, with the totals reduced significantly. As of May 4, total deaths reported and attributed to COVID-19 are 38,576. https://www.cdc.gov/nchs/nvss/vsrr/covid19/index.htm