Friday update - Covid death trend in U.S.

4,170 Views | 20 Replies | Last: 5 yr ago by Rutedown
sleepybeagle
User Profile Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
10:40a, 7/17/20
AG
Here's today's update from today from the CDC website. Great news - Covid deaths are still trending down in the U.S. CDC data link




Many people question this data as it goes seem to go against what is being reported in the news and other web sites. I use the CDC as it is the official and final word on Covid deaths based on coroner reported death certificates with confirmed or presumed COVID-19, coded to ICD10 code U07.1.

It has been rightly noted in past posts, that as new numbers come in on deaths, the graph will change and weekly numbers increase - but overall, the trend has not changed which is a good sign. I wonder if some news agencies and web sites are reporting deaths when they are reported, and not when they occur, skewing the numbers? Also, I use the weekly average for deaths as it gives a better feel for the trend. Covid deaths are going down.

Notes: The data presented here represents when the death occurred, and not when the death was reported. The data for the last couple of weeks is extrapolated based on the number of deaths that have come in vs. past weeks totals. I posted here the "projected" death count, which makes the number higher, for any week less than 100%.



Reply Quote
10
AustinAg2K
User Profile Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
10:58a, 7/17/20
Interesting. If you compare it to the graph you posted yesterday, you can see that the data changed. Yesterday, the graph you posted had approx 1000 deaths, today it's over 2000. 6/27 looks to have gone from about 2k to 2500.
Reply Quote
3
Windy City Ag
User Profile
Private Message
Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
In reply to AustinAg2K • 11:08a, 7/17/20
AG
Not too interesting. The CDC states very clearly that historical data is revised as it is received.

https://www.cdc.gov/nchs/nvss/vsrr/covid19/index.htm

Quote:
The provisional counts for coronavirus disease 2019 (COVID-19) deaths are based on a current flow of mortality data in the National Vital Statistics System. National provisional counts include deaths occurring within the 50 states and the District of Columbia that have been received and coded as of the date specified. It is important to note that it can take several weeks for death records to be submitted to National Center for Health Statistics (NCHS), processed, coded, and tabulated. Therefore, the data shown on this page may be incomplete, and will likely not include all deaths that occurred during a given time period, especially for the more recent time periods. Death counts for earlier weeks are continually revised and may increase or decrease as new and updated death certificate data are received from the states by NCHS. COVID-19 death counts shown here may differ from other published sources, as data currently are lagged by an average of 12 weeks.
Reply Quote
7
Duncan Idaho
User Profile Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
In reply to Windy City Ag • 11:27a, 7/17/20
Wait a second....you mean to tell me that it takes days, weeks even for 3143 unique agencies (counties) that operate on a variety of systems and processes to consolidate and report their data and then even more time to consolidate and report those 3143 data points?

Reply Quote
13
Capitol Ag
User Profile
Private Message
Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
11:43a, 7/17/20
AG
Quote:
as it goes seem to go against what is being reported in the news and other web sites
So I read this as saying it should be VERY reliable then!
Reply Quote
2
sleepybeagle
User Profile Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
11:57a, 7/17/20
AG
Quote:
Death counts for earlier weeks are continually revised and may increase or decrease as new and updated death certificate data are received from the states by NCHS. COVID-19 death counts shown here may differ from other published sources, as data currently are lagged by an average of 12 weeks.
This is true, but in general the data is about 95% in after about 3 weeks. The interesting thing is the CDC calculates a % deaths reported number for each week which has been shown to be very accurate. That's why I plot the data divided by the % reported for the last couple of weeks which creates a higher, but more accurate number. But yes, the numbers do change as new data come in.

The one thing that doesn't change is the trend, which continues to go down.
Reply Quote
5
Windy City Ag
User Profile
Private Message
Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
11:58a, 7/17/20
AG
Quote:
The one thing that doesn't change is the trend, which continues to go down.

Agreed.
Reply Quote
5
Fitch
User Profile
Private Message
Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
12:00p, 7/17/20
AG
OP, appreciate you taking the time to compile the CDC data. I've found their website to be a mess and too hard to work with.

Agree with your overall sentiment that as a final "record of truth" the CDC data is probably the most stringent barometer of overall COVID-related fatalities, but given the lags in time associated with compiling all the agencies nationwide and backdating cases, the data are better used for understanding where we've been rather than where we are.

Houston and Harris County employ a similar methodology of backdating new cases to provide a more accurate picture of the level of community spread and infection rates than what the State (or news) puts out.

The below graph shows the final number of positive lab-confirmed cases as of the date the sample was tested and confirmed positive by the lab (dashed lines), versus the date it was reported by the state (red line).



Because the data are constantly being revised, the more recent data point aren't reliable for understanding where we are in the current trend, but after a few weeks you can start to see trends and tie back the inflection points in the curve to policy changes, holidays, etc. that may have had an impact on transmission.

Certainly better data, but not as timely and by virtue of that aren't as usable by the news agencies who try to constantly report as soon as new information is published.
Reply Quote
8
1 edit
TexAggie5432
User Profile Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
In reply to sleepybeagle • 12:04p, 7/17/20
sleepybeagle said:
Here's today's update from today from the CDC website. Great news - Covid deaths are still trending down in the U.S. CDC data link




Many people question this data as it goes seem to go against what is being reported in the news and other web sites. I use the CDC as it is the official and final word on Covid deaths based on coroner reported death certificates with confirmed or presumed COVID-19, coded to ICD10 code U07.1.

It has been rightly noted in past posts, that as new numbers come in on deaths, the graph will change and weekly numbers increase - but overall, the trend has not changed which is a good sign. I wonder if some news agencies and web sites are reporting deaths when they are reported, and not when they occur, skewing the numbers? Also, I use the weekly average for deaths as it gives a better feel for the trend. Covid deaths are going down.

Notes: The data presented here represents when the death occurred, and not when the death was reported. The data for the last couple of weeks is extrapolated based on the number of deaths that have come in vs. past weeks totals. I posted here the "projected" death count, which makes the number higher, for any week less than 100%.



I agree that deaths decrease significantly from April to June. However, the CDC data is lagged by 2 or 3 weeks. About the time that cases and reported deaths have increased. Based on other sources, I think the last 2 weeks will be revised up to around 6,000 each. Not nearly as high as April and May but certainly not below 2,000.
Reply Quote
2
CompEvoBio94
User Profile Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
In reply to sleepybeagle • 12:09p, 7/17/20
sleepybeagle said:
"The interesting thing is the CDC calculates a % deaths reported number for each week which has been shown to be very accurate"
Where are you getting this % ?
Reply Quote
1
sleepybeagle
User Profile Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
12:16p, 7/17/20
AG
Quote:
the data are better used for understanding where we've been rather than where we are.
I have to disagree. I use the CDC as it is the ONLY data I have found that is 100% accurate. Yes, there site is a mess, which is why I download their data and create this graph myself.

Some data is delayed, but as I said, if you use the % reported to "model up" the last three weeks count, the data is highly accurate overall for showing the trend. The numbers change some as new data comes in, but the curve has not changed.

I've been watching this data for months, and when the trend started to go down in mid April, all I have heard from people is "just wait... there's going to be a second wave and it's going to start going up". Fortunately, no bimodal distribution has occurred.

Where we are is - Covid deaths are trending down.
Reply Quote
1
sleepybeagle
User Profile Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
In reply to CompEvoBio94 • 12:19p, 7/17/20
AG
CompEvoBio94 said:
sleepybeagle said:
"The interesting thing is the CDC calculates a % deaths reported number for each week which has been shown to be very accurate"
Where are you getting this % ?
Good question.

If you go to the CDC link and download the data, you will see a column called "Percent of Expected Deaths". This number has shown to be accurate in projecting the total deaths for that week based on the number that has come in so far for the week. I use this % value to calculate the higher projected value. Generally I use this % for the last 3 or 4 weeks to project 100%. Week by week the % number seems to be fairly accurate.
Reply Quote
CompEvoBio94
User Profile Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
In reply to sleepybeagle • 12:27p, 7/17/20
If you look at the footnote for that column it explains:
"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 (8)."

So when the page reports the count for 7/11 (1,099 deaths right now) is 48% of expected (which is what shows up right now), that is just saying the current count is 48% of the recent years' average. It is not saying that the current count is 48% of the final count.

Dividing by 0.48 (as you said you were doing yesterday), just takes the data from a week and rescales it so that it is the average for that week in 2017-2019.

Fitch's response above is a nice explanation of why the CDC provisional data count data is not appropriate for the conclusion that you are trying to make.


Edited to correct em-dashes being dropped when copying from the CDC site.
Reply Quote
4
1 edit
Fitch
User Profile
Private Message
Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
In reply to sleepybeagle • 12:30p, 7/17/20
AG
I certainly hope you're correct, but candidly my expectation is the last week or two will be upwardly revised to account for the recent uptick in fatalities in the southern states given that trend has appeared in other sources which are faster to aggregate and publish data from the states.
Reply Quote
1
CompEvoBio94
User Profile Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
In reply to CompEvoBio94 • 12:34p, 7/17/20
Note: I should have said "rescales it based on the average for that week in 2017-2019." instead of "rescales it so that it is the average for that week in 2017-2019."

It's not going to make the COVID # equal to the all-causes deaths. But the point remains, this is not a valid way to rescale the data when we don't know what % of the all-causes deaths will be for a week in this year. If the final number is above 100%, then your correction will be dampening the count unjustifiably.
Reply Quote
1
sleepybeagle
User Profile Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
In reply to CompEvoBio94 • 1:31p, 7/17/20
AG
CompEvoBio94 said:
Note: I should have said "rescales it based on the average for that week in 2017-2019." instead of "rescales it so that it is the average for that week in 2017-2019."

It's not going to make the COVID # equal to the all-causes deaths. But the point remains, this is not a valid way to rescale the data when we don't know what % of the all-causes deaths will be for a week in this year. If the final number is above 100%, then your correction will be dampening the count unjustifiably.
Nope... It is valid.

The % is not used to dampen any weeks number, only to increase the numbers for the last couple of weeks where the % reported is not yet 100% This method of estimating the total deaths for the week has held very accurate over the last few months.

Overall - the trend has held accurate. Week by week covid deaths are trending down. The good news is there is no sign of the bi-model second wave that many have feared.
Reply Quote
1 edit
reb,
User Profile Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
In reply to Fitch • 1:35p, 7/17/20
AG
hello good sir, might i ask where you found this chart
Reply Quote
Fitch
User Profile
Private Message
Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
In reply to reb, • 1:47p, 7/17/20
AG
Self made from data from the Texas DSHS and Harris County dashboard's epi curve:
Reply Quote
CompEvoBio94
User Profile Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
In reply to sleepybeagle • 1:48p, 7/17/20
Yesterday (at https://texags.com/forums/84/topics/3125107/replies/56966298 ) you said:
"7/11 - 272 counted with 30% reported = 907"

Today the numbers on the provisional site for 7/11 are:
a count of 1,099 and 48% for "percent expected deaths"
Using your method that would predict 2,290.

So the corrected number increased by a factor of 2.5 in one day. That is not what most people would call "accurate".


Reply Quote
3
bay fan
User Profile
Private Message
Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
In reply to CompEvoBio94 • 2:24p, 7/17/20
S
It illustrates why making an argument that there is no threat from this virus is impossible to make at this time.
Reply Quote
2
Marcus Aurelius
User Profile
Private Message
Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
2:27p, 7/17/20
AG

Hospitalization rates climbing steadily.
Reply Quote
3
1 edit
Rutedown
User Profile Ignore User Stop Ignoring
How long do you want to ignore this user?
24 hours One week Permanently Cancel
In reply to Marcus Aurelius • 2:47p, 7/17/20
Marcus Aurelius said:

Hospitalization rates climbing steadily.
Those are number we should be most worried about at this time. We know that doctors have gotten better at treating COVID since April, but if there is not enough hospital beds or staff in these hotspots, people with and without COVID will be collateral damage. How do we count a death of someone who had an appendix burst, but there was no one to treat them?
Reply Quote
1 edit
Refresh
Page 1 of 1
 
×
subscribe Verify your student status
See Subscription Benefits
Trial only available to users who have never subscribed or participated in a previous trial.