April: U.S. Death Rate From All Causes (Including COVID-19) At Multi-Year LOW
The U.S. crude death rate (mortality from all causes) still remains near multi-year lows despite COVID-19. The CDC tracks deaths attributed to the flu, pneumonia, and all causes on this website.
You can download the data as verify for yourself: Just go to this CDC website and click the green ‘downloads’ button. They have complete data for total U.S. deaths up until week 14 (week ending April 5th, 2020).
On April 5th, the U.S. saw 1,344 COVID-19 deaths, as the number of cases in the U.S. accelerated. The overall number of deaths in the U.S., or the crude death rate did not show a correlated rise.
At the very least, this data shows we need to analyze COVID-19 deaths in the context of the broader U.S. mortality rate from all causes. It appears normal deaths are being attributed to COVID-19 if the patient is COVID-19+, even if another underlying chronic cause is responsible.
For The Week Ending April 5, 2020
- – There were 49,292 deaths (all causes) in the week ending April 5th, 2020 – see how this number compares to the weekly number since 2013 in the graph below.
- – 14.92 deaths per 100,000 people in the week ending April 5th, 2020 – see how this number compares to the weekly number since 2013 in the graph below.
In Perspective: The Number of Weekly U.S. Deaths (All Causes) Since 2013
If you are having trouble viewing the graph below on mobile, view the standalone graph.
In Perspective: Weekly U.S. Death Rate (All Causes) Since 2103
We used the CDC weekly death data to calculate the death rate (weekly deaths per 100,000 people using the population of the U.S. for the year in question). If you are having trouble viewing the graph below on mobile, view the standalone graph.
We will update as the CDC updates.
What is the source for the overall mortality data presented in your table? The link that you provide under “The CDC tracks deaths attributed to the flu, pneumonia, and all causes on this website.” takes me to the flu-specific data. Looking for the raw deaths per week for the last several years, regardless of cause.
Yes, use that link and then click the green “download” button. In the download, you’ll get a CSV. Among the many columns, there are three columns that list deaths:
NUM INFLUENZA DEATHS, NUM PNEUMONIA DEATHS, TOTAL DEATHS
-The first one is just flu
-The second one is just pneumonia
-The third one, total deaths, is crude deaths (all causes)
I verified the crude number of deaths by taking the sum of the all cause and matching it to the number of deaths in a year in the us. For example, it’s about 2.7 million total deaths in all of 2015 in USA from all causes. That matches the total deaths this data set has for 2015 (if you add up total deaths per week).
Next, the crude death rate was calculating taking the number of deaths per week divided by the population of the us in that specific year and then multiply that number by 100,000. (Standard crude mortality formula).
I couldn’t find a data set of weekly US deaths anywhere but this site. But the total deaths represent all deaths, and is included with pneumonia and influenza deaths it appears to add perspective.
Do your raw numbers include deaths from homicide, suicide, car crashes and all other non-disease related forms of death? If so, why would you include those in your comparison of disease specific indications?
Why are you disregarding those who have died of comorbid indications due to complications associated with COVID-19 as non-COVID-19 related deaths? Most chronic conditions are well controlled and, If the individual had not caught COVID-19, they might have lived far longer. Disregarding them in the final death count seems disingenuous.
I don’t disregard any deaths. I’m looking at crude mortality (from each and every cause):
CRUDE DEATH RATE is the total number of deaths to residents in a specified geographic area (country, state, county, etc.) divided by the total population for the same geographic area (for a specified time period, usually a calendar year) and multiplied by 100,000.
Be careful. Figures for this year are preliminary, and take at least 8-12 weeks before they before they become *nearly* final, because it takes time for death certificates to trickle in. That flow of information may be further disrupted by health sector layoffs in some areas and health-sector overload in a few of the hard-hit COVID-19 areas.
I think your overall conclusion has a good chance of holding, but it’s way too early to genuinely know that.
I believe this link has the same data you’re referring to:
https://www.cdc.gov/nchs/data/health_policy/Provisional-Death-Counts-COVID-19-Pneumonia-and-Influenza.pdf
Footnote 3 at that link says this:
“Percent of expected deaths is the number of deaths for all causes for this week in 2020 compared to the average number
across the same week in 2017–2019. Previous analyses of 2015–2016 provisional data completeness have found that completeness is lower in the first few weeks following the date of death (<25%), and then increases over time such that data are generally at least 75% complete within 8 weeks of when the death occurred."
I believe that the footnote is overly careful, that virtually all data is normally after about 12 weeks, and that about 95% (not 75%) is typically what you see after 8 weeks. But … given the uncertainty of reporting in the current circumstances, we can't be sure.
Yes, agree. To lessen the variance caused by early data not being as accurate, I don’t use the latest data set. I wait two weeks which is why we only show data until April 5th.
By far the biggest errors come from the most recent ‘full’ week of data. That’s why I ignore data from week 15 and week 16; and don’t include it.
A 10% error rate to week 14 in the direction of the most likely revision (upwards) would be 54,000 — which is still lower than all periods of increased influenza deaths. 10% would be an extreme revision considering I’m already using 2 week old data to attempt to keep the data as accurate as possible. In any case, each week the CDC updates, I update the entire data set, and will call out any remarkable revisions.
However, if you look at the most recent full sets (week 9-11), the standard adjustment after labeled 100% complete by the CDC is only 4-6%. So we’re looking at about a 5% upward revision which doesn’t change the conclusions that 1) mortality is very low for a national pandemic emergency and 2) we need further research on what constitutes a covid-19 death.
Furthermore, the Sweden data (they did not lock-down) is starting to challenge more assumptions.
The ultimate conclusion here is we need to take a step back, look at more data points and revisit some of our base assumptions as the entirety of data isn’t matching up with the current mainstream conclusions/assumptions.
You should look at Roy Spencer’s post on his website about the problem with CDC death counts due to delayed reporting. He developed an correction factor. I would like to see how it affects your data
Sam Silver
Atlanta
http://www.drroyspencer.com/2020/04/correcting-recent-u-s-weekly-death-statistics-for-incomplete-reporting/
Roy W. Spencer received his Ph.D. in meteorology at the University of Wisconsin-Madison in 1981. Before becoming a Principal Research Scientist at the University of Alabama in Huntsville in 2001, he was a Senior Scientist for Climate Studies at NASA’s Marshall Space Flight Center, where he and Dr. John Christy received NASA’s Exceptional Scientific Achievement Medal for their global temperature monitoring work with satellites.
I only use complete data. The CDC publishes how complete a set is. Notice I’m only using week 14 in this set. There is actually data for week 16, but week 15/16 are incomplete. I wait until the complete data set filters in and the CDC confirms it’s 100% in from all districts. After the CDC labels data 100%, there are still revisions. But, the recent historical revisions are less than 5% after two weeks.
So, yes, even before I read the post you shared… I identified the same exact issue and accounted for it in the presentation of data.
Dr. Spencer’s data model is a bit flawed. He’s using a 5-year historical average. Whenever you do a data model, you need to weight the model, at a disproportional ratio, to recent-data trends. He doesn’t do that. As it turns out, with all the attention on COVID-19, epidemiological data is being reported at a much quicker rate than the 5-year historical model suggests. We are getting more complete data quicker.
Think about it- if it was your job to provide data that the world needs, wouldn’t you devote more resources to providing it quicker? That’s what’s happening.
I work with trillions of data points, across billions of auctions a day. Identifying trends and patterns is my livelihood.
My only conclusions are this:
1) we need to do a better job at defining certain data points
a) covid -19 deaths: when do we attribute to pre-existing chronic conditions or covid-19 itself? or do we categorically assign death: covid-19 alone or covid 19 w/ pre-existing condition when looking at the death rate? uniformity will help analyze this data in aggregate.
b) death rate to DETECTED covid-19 cases is not really a metric. every state/country detects covid-19 using different methods and thus different rates are attributed to that method rather that qualities of the virus itself.
2) i find it quit surprising that in the midst of a national pandemic emergency, that stops a multi-trillion dollar economy, the chances of dying from any cause are at the lowest levels across multiple years. we deserve proper data as listed in point (1) given how the lock down effects businesses, and the livelihood of billions.
Ultimately, I’m looking for truth and am suggesting areas of further research based on surprising crude mortality rate data.
The data is emerging that we did not need to lock down. Sweden did not lock down, and the mortality rate there is lower than france and italy.
https://www.bloomberg.com/news/articles/2020-04-19/sweden-says-controversial-covid-19-strategy-is-proving-effective?srnd=premium
Sweden is currently higher than most Scandinavian nations. But since they did not lock-down, they likely hit herd immunity quicker, and will have lesser deaths down the road. While the other Scandinavian nations, when they come out of lock down, may have an increase in deaths since they are not as far along in establishing herd immunity.
Furthermore, Sweden’s economy not taking a slug to the face with a baseball bat, as other nations.
Thought experiment: If we could reduce automobile accident deaths to virtually 0 by reducing the speed limit to 15 mph, should we do it?
Evan, I’m trying to find your point… that Covid-19 is no big deal?? Much ado about nothing?? What’s your point? Crude oil fell below $20/barrel… what’s the point?? Crude deaths at an all time low… what’s the point??
CRUDE DEATH RATE is the total number of deaths to residents from any cause in a specified geographic area (country, state, county, etc.) divided by the total population for the same geographic area (for a specified time period, usually a calendar year) and multiplied by 100,000. It has nothing to do with crude oil.
you did not respond. What is the point you are trying to make?
I wrote the reply in your other comment. Sorry!
2020 it was the reset of markets. they just used covid19 as smokescreen
It could very well be argued that the drop in numbers of deaths correspond very neatly with the implementation of stay-at-home measures. Are we to conveniently ignore this observation and divert our attention to see only what we are told to see?
Except the stay at home measures started well after the death rates became abnormal and that the impact of the stay at home orders would not be seen in the mortality numbers for 3 weeks or more (so AFTER April 4 for most areas.The first stay at home order was March 19 in CA).
In reality we saw deaths spike during the week of April 4-11 (in other words, 3 weeks + after most of the US was under stay at home orders) and the rate continued to climb for at least 10 days – the period where we should have seen a sharp decline if the lockdowns were effective.
I’m making a narrow point in the article.
1) That the overall death rate for all deaths is one of the lowest it’s ever been and not giving any specific reason.
2) That it’s surprising the death rate is so low being in a national pandemic emergency.
My only recommendation from this, at the very least is this:
At the very least, this data shows we need to analyze COVID-19 deaths in the context of the broader U.S. mortality rate from all causes. It appears normal deaths are being attributed to COVID-19 if the patient is COVID-19+, even if another underlying chronic cause is responsible.
We need to get a clear definition of what constitutes a COVID-19 death. Right now states and countries can really attribute anything if the patient is COVID-19+. But based on the antibodies test from Santa Clara, CA… COVID-19 is even more prevalent that previously thought. So there is ample crossover between what caused the death.
I’m really just asking for a standard definition everyone use. Once we have that and have the data behind that, we can make more conclusions. But with very weak and siloed data, it’s hard to make a lot of conclusions.
You also have to consider with most of the country in lockdown, traffic deaths will be dramatically reduced.
I think I now see what is going to happen. Since they have been adding so broadly to the covid19 numbers, they have been shorting the actual causes just because they have tested positive for the virus despite their state up to that point. And before retesting result because there have been several incorrect tests. The same people who were sadly going to pass anyway will be in their numbers of who died FROM it. That is going to bring the REAL numbers down for all other general causes of death, especially in the elderly. What they will try and feed us is that our staying in and forced distancing have made us healthier people. I know we generally have helped the planet by leaving the animals time and space to live, play and saved some pollution for sure. But I enjoy taking trips and being in nature because it helps our health too. The entitled “govt rulers” will go on MSM soon to get the ball rolling, watch. They will send out scientists with some bs spiel where they picked something that matches the agenda for the day (true or not). Maybe they will determine they need to control us by some new laws since the temporary ones worked so positively for the planet and our health (wink wink) they’ll say. Just wait. Now we find out China bought up all the PPE stuff and continues to hoard it, making the rest of the world be at their feet paying gouging amounts of money. Yet China is getting a pass while it is still going on because they want the orange guy out. Why can’t lawmakers agree that is wrong and act on it? Some are trying very hard and some act like they have no right to try and stop them and stand up for what is really right in this world? Sadly, I think it is because they have ammo against them by either direct outright payoffs or they have some dirt on the officials because that is what they do. (Confirmed Chinese spy act drove Finstien for 20 years. Think about what a driver would be around and see.) It works for China because they have no soul when it comes to other countries. Hence, …Biden son flies out on af2 with dad to China, gets big “Loan” (payment) in real money and a giant diamond from them according to his divorce paperwork) and his dad seems to cover for China and call the people trying to stand up to China the bad ones picking on China as a racist or xenophobic. (Not trying to be political, just using it is an example because the info is verified, but not considered to be ILLEGAL in the eyes of the laws then. And I am not implying it was illegal. Slimy yes, illegal no. This is an election year something goes wrong very often in those years for some reason. Perfect storm, really.
Social distancing it going to cause all other contagious diseases to drop. It just means COVID deaths are replacing the other deaths that aren’t as prevalent due to social distancing.
Perhaps. We know that flu death rates vary widely from year to year without draconian social distancing as a factor. Flu deaths this year (per the CDC) are well within the normal rage through April 4 (and the flu season is normally considered to be October – March https://www.cdc.gov/flu/about/season/flu-season.htm) Widespread social distancing started late in this period.
The underlying problem that anyone faces when claiming that the current high levels of social distancing work is that there is no historical comparison or true control group to compare with. As the evidence mounts that COVID 19 was present sooner than previously believed and that higher percentages of the population have been infected/exposed, the more likely it is that the lockdowns exacerbated the problem by exposing the high risk populations to higher concentrations over longer periods of time by confining them with family and caregivers who were already exposed. If there were any benefit to the lockdowns, there is no empirical basis for claiming these benefits are as good as those that would have been achieved by traditional “vertical” approaches that would focus on the sick and known exposures.
Where’s the facebook share option?
Thank you Evan for this data. With public policy decisions allegedly science-based it’s good to see some comparative research. The govt-induced shutdown may have more harmful consequences than we can yet predict; it might have been better for people at risk to be permitted to self-shield rather than to give in to mass hysteria. But, I remind myself, this is an election year. What’s next? Here are Dr. Rima’s thoughts: http://www.opensourcetruth.com/covid-second-wave/
Hi Evan,
Thanks for the info and thoughtful presentation…
Have you had a chance to sum the flu deaths over the year? As a reality check, I summed the various weekly deaths (from flu and pneumonia) and found a discrepancy. For example, the final data the CDC summarized for 2017 said “55,672 deaths from both flu AND pneumonia”. However, the data that is referenced in the article shows the “pneumonia” deaths alone for the 2016-2017 or 2017-2018 seasons to be around 180,000! Do you have any ideas about this major discrepancy? I’d like to be able to cross reference CDC sources and have consistency between them…
Thanks in advance for your continued diligence.
Kind regards, Fred Lutfy
The analysis of the CDC is entirely wrong. A quick download of the CDC data referenced in this article shows that it is clearly noted as reported by flu season (2019-2020 week #1 = 1st week Oct’19) and not by calendar year (2020 week #1 = 1st week Jan’20). The CDC data referenced is only thru 1st week of Jan’20, before COVID. This article should be taken down.
The notation used by the CDC is confusing. I believe the week number *is* for the calendar year, not the “season”. For example, total deaths in New York City peak at 7632 for week 15. If “week 15” was week 15 of the flu season, it would mean everyone failed to notice the death rate in NYC being at 7X normal in the last week of December.
Just a quick comment on the reference to traffic fatalities as an influence on the mortality rate. In 2016 (the most recent data I could find) there were about 170,000 accidental deaths of which 40,000 were traffic related. The weekly average then for accidental deaths per 100,000 would be 1 and cutting traffic fatalities in half would only reduce that by .125 (.875 or realistically, stil 1) https://www.cdc.gov/nchs/fastats/accidental-injury.htm
Moreover, the other largest causes of accidental death are falls and poisonings which are unlikely to decline while on lockdown and I think there is a pretty logical case for say they might increase.
It looks like the actual number of deaths started to drop dramatically well before any “stay at home” measures were put in place. Those didn’t really start until the last half on March.