Daily Charts
609,025 Views |
2786 Replies |
Last: 2 yr ago by AggieUSMC
That's a very interesting read.agforlife97 said:
https://medium.com/analyticaper/covid-what-the-data-tells-us-about-texas-e319c4406741
Hidalgo county is trying to bring up that death number
A couple notes...
Houston Area Hospitalizations
Updated 7/28 per latest fatality data (current through 7/26)
- Texas started reporting fatalities by the date of death as stated on the death certificate and is only counted as a COVID-19 fatality when the medical certifier attests on the death certificate that COVID-19 is the cause of death.
- By law death certificates must be filed within 10 days, buy may be amended later.
- Data for at least the trailing 10 days is being constantly revised so the shape of the trend line is inaccurate for this period of time. Texas has not re-published the daily counts since
7/24updated to 7/26 overnight. - Texas is reporting incomplete hospitalization numbers due to a transition in reporting to comply with new federal requirements, resulting in a partial data set and the appearance of a drop in hospitalizations back to 7/23. Over this period of time between 82-91% of hospitals have reported data. As of this posting the issue remains unresolved.
- Hurricane Hanna temporarily affected case data collection and reporting in the Rio Grande Valley and south Texas.
Houston Area Hospitalizations
Updated 7/28 per latest fatality data (current through 7/26)
From the notes, we expect a 10 day delay in reporting deaths and since we are backdating to date of death, the end of the curve is inaccurate. Since Texas stopped updating the deaths on 7/24, are we expecting the state to report deaths with a 10 day delay and resume once they establish the 10 day buffer so they don't have to continually revise the tail end of the curve?
We are currently at 7/28 and the peak of the curve is 7/13 (which is about the 10 day delay for 7/24). It would be nice to see what those few days after the peak look like.
BTW, I agree with the idea of backdating to date of death rather than plotting the date deaths are reported. It gives a more accurate picture of where we really are (avoiding noise because a county consolidated a weekend of death reports on Monday, for example). I also think the public isn't really going to take the time to understand that the last part of the curve includes reporting delay and the media will spin that to fit an agenda, either deaths are falling or Texas is under reporting.
We are currently at 7/28 and the peak of the curve is 7/13 (which is about the 10 day delay for 7/24). It would be nice to see what those few days after the peak look like.
BTW, I agree with the idea of backdating to date of death rather than plotting the date deaths are reported. It gives a more accurate picture of where we really are (avoiding noise because a county consolidated a weekend of death reports on Monday, for example). I also think the public isn't really going to take the time to understand that the last part of the curve includes reporting delay and the media will spin that to fit an agenda, either deaths are falling or Texas is under reporting.
Well it will always look like it's getting better. Forever. So we have that going for us.
The new date-of-death epi curve was updated late yesterday afternoon, FWIW. So it seems they are now adding new data on a daily basis?
Well it looks like the state data refreshed overnight and now has data out to 7/26. Above charts have been updated. Not a lot of change at the tail end with most additions being to the body of the curve.
For reference, the data only show an aggregate 5 deaths between 7/23-7/26, so that'll obviously be refined. In my mind that's also a decent way to gauge the lion's share of the reporting lag time.
As Keegan referenced, the state dashboard shows where the new data points are added, in some instances back to late June.
For reference, the data only show an aggregate 5 deaths between 7/23-7/26, so that'll obviously be refined. In my mind that's also a decent way to gauge the lion's share of the reporting lag time.
As Keegan referenced, the state dashboard shows where the new data points are added, in some instances back to late June.
Looking at the TMC data there are a few different ways to present the data that might be interesting to folks.
Everyone hospitalized is part of the cumulative Total Cases. When someone is discharged or dies they become a Resolved Case. That yields the following chart:
The vertical distance between the lines on a given day represents the total current hospitalized Covid patients. TMC has been providing that plot. The horizontal distance between the two curves is something I am calling Latency. You can either cast backwards or forwards. Latency Backwards uses the Total Resolved Cases on a given day and looks back to see when the Total Case curve reached that number (linearly interpolating for fractions of days) and looks like the following:
Something else that seems interesting is to look at is the resolved cases each day and see what percentage are deaths. That looks like this:
These were done with data from the plots TMC releases daily. I do not have discharge vs death data before June 26 when I started recording it. Obviously, YMMV.
Everyone hospitalized is part of the cumulative Total Cases. When someone is discharged or dies they become a Resolved Case. That yields the following chart:
The vertical distance between the lines on a given day represents the total current hospitalized Covid patients. TMC has been providing that plot. The horizontal distance between the two curves is something I am calling Latency. You can either cast backwards or forwards. Latency Backwards uses the Total Resolved Cases on a given day and looks back to see when the Total Case curve reached that number (linearly interpolating for fractions of days) and looks like the following:
Something else that seems interesting is to look at is the resolved cases each day and see what percentage are deaths. That looks like this:
These were done with data from the plots TMC releases daily. I do not have discharge vs death data before June 26 when I started recording it. Obviously, YMMV.
Am I wrong in thinking the back-dated/date-of death reporting is going to make it more difficult to make real-time, society wide changes / suggestions? In the last few weeks the very obvious climb in cases and deaths made it pretty easy to report and show people "wear your masks, stop having gatherings, etc". Now you've got a death chart that a layperson can look at and go "awesome, it's basically over".
But then I guess most people would point to new/active cases which is more of a 'live' look.
But then I guess most people would point to new/active cases which is more of a 'live' look.
you watch the real time metric indicative of severity and volume that we've been watching all along
hospitalizations
hospitalizations
Fair. I'd like all indicators taken as a whole with weighting, agreeing that hospitalizations should be number 1.cone said:
you watch the real time metric indicative of severity and volume that we've been watching all along
hospitalizations
I don't think people would look at the back-dated numbers and says it over. I think it probably gives a better view that it is being managed and not over-running our hospitals.ETFan said:
Am I wrong in thinking the back-dated/date-of death reporting is going to make it more difficult to make real-time, society wide changes / suggestions? In the last few weeks the very obvious climb in cases and deaths made it pretty easy to report and show people "wear your masks, stop having gatherings, etc". Now you've got a death chart that a layperson can look at and go "awesome, it's basically over".
But then I guess most people would point to new/active cases which is more of a 'live' look.
Once this thing was unleashed on the world the fear was not being able to manage the sickness. It appears we are doing that when I look at the numbers from a macro lens. It's unfortunate that people are dying but that has always been the ultimate outcome since it was unleashed on the world. We've just been trying to slow that down IMO.
Cepe - its pronounced "Ceep" and stands for my initials - CP.
I can say our facility is down 15-20% compared to same time last week. TMC hospitals showing a good decrease in number of new hospitalizations this week.
Awesome!
First time all green in a while.
I thought the approach Dallas County used was helpful. They greyed out the recent data but still reported it. Perhaps that along with highlighting the most recent additions, as in the plot above, is a useful technique.
As an aside, I haven't seen an update on Dallas County's Covid site in a week.
This seems to be the latest.
Click here to view the monitoring data.
As an aside, I haven't seen an update on Dallas County's Covid site in a week.
This seems to be the latest.
Click here to view the monitoring data.
Without back dating, the curve is shifted by the reporting delay.ETFan said:
Am I wrong in thinking the back-dated/date-of death reporting is going to make it more difficult to make real-time, society wide changes / suggestions? In the last few weeks the very obvious climb in cases and deaths made it pretty easy to report and show people "wear your masks, stop having gatherings, etc". Now you've got a death chart that a layperson can look at and go "awesome, it's basically over".
But then I guess most people would point to new/active cases which is more of a 'live' look.
Current ICU Cases in TMC
Percentage of Total Current Cases in ICU
Percentage of Total Current Cases in ICU
I hate data without age groups and demographics. If you are under 60, and especially a female, you have hardly a think to worry about. Ironic this is one of the most scared groups.
https://www.cdc.gov/nchs/nvss/vsrr/covid_weekly/index.htm
All USA:
Female only ALL USA:
https://www.cdc.gov/nchs/nvss/vsrr/covid_weekly/index.htm
All USA:
Female only ALL USA:
What happened in Texas yesterday?
ETFan said:
What happened in Texas yesterday?
https://www.google.com/amp/s/www.khou.com/amp/article/news/health/coronavirus/texas-covid-19-death-numbers-jump/285-19fc2cc4-a1ab-4d28-88a2-e439a176eb60
What a cluster. We're six months in to this and now they adjust how deaths are reported?Premium said:ETFan said:
What happened in Texas yesterday?
https://www.google.com/amp/s/www.khou.com/amp/article/news/health/coronavirus/texas-covid-19-death-numbers-jump/285-19fc2cc4-a1ab-4d28-88a2-e439a176eb60
I'm not sure how we trust anybody that should be managing this thing if they can't even get the counts right. How can you make decisions based on data this way?
Frustrating.
Change in reporting and catching up. The next few days may also be large numbers as the curve for the last 2+ weeks is filled in using the new aggregation method.
The peak day of death is still 7/12, with a total of 175. That will likely rise and might get pulled forward a bit. But given hospitalization trends and decreasing detected infections, the worst seems to be in our rearview mirror. We'll end up with the usual Gompertz curve over the next month.
The peak day of death is still 7/12, with a total of 175. That will likely rise and might get pulled forward a bit. But given hospitalization trends and decreasing detected infections, the worst seems to be in our rearview mirror. We'll end up with the usual Gompertz curve over the next month.
Any chance you can line with this up with the corresponding number of people actually infected?Premium said:
I hate data without age groups and demographics. If you are under 60, and especially a female, you have hardly a think to worry about. Ironic this is one of the most scared groups.
https://www.cdc.gov/nchs/nvss/vsrr/covid_weekly/index.htm
All USA:
Female only ALL USA:
Oh right, so they're doing a daily reported deaths on their COVID dashboard (shows 313 on DSHS) but also keeping a more accurate death cert chart.Keegan99 said:
Change in reporting and catching up. The next few days may also be large numbers as the curve for the last 2+ weeks is filled in using the new aggregation method.
The peak day of death is still 7/12, with a total of 175. That will likely rise and might get pulled forward a bit. But given hospitalization trends and decreasing detected infections, the worst seems to be in our rearview mirror. We'll end up with the usual Gompertz curve over the next month.
Keegan, is that chart Houston deaths?
Texas
Do you think the old way of reporting was better, given there was a substantially longer delay in reporting some deaths and they were being attributed to the day they were reported rather than the day they died?Cepe said:What a cluster. We're six months in to this and now they adjust how deaths are reported?Premium said:ETFan said:
What happened in Texas yesterday?
https://www.google.com/amp/s/www.khou.com/amp/article/news/health/coronavirus/texas-covid-19-death-numbers-jump/285-19fc2cc4-a1ab-4d28-88a2-e439a176eb60
I'm not sure how we trust anybody that should be managing this thing if they can't even get the counts right. How can you make decisions based on data this way?
Frustrating.
Should they stick with whatever methodology was quickly put together initially or acknowledge that there may be better ways and change methodology if/when a better way is identified?
Part of the issue is they didn't update for several days as the switched the methodology.Keegan99 said:
Change in reporting and catching up. The next few days may also be large numbers as the curve for the last 2+ weeks is filled in using the new aggregation method.
The peak day of death is still 7/12, with a total of 175. That will likely rise and might get pulled forward a bit. But given hospitalization trends and decreasing detected infections, the worst seems to be in our rearview mirror. We'll end up with the usual Gompertz curve over the next month.
Disappointed that Dallas hasn't updated their data. I am really curious about the next update.
After the last couple, I am becoming skeptical everything is on the up and up.
After the last couple, I am becoming skeptical everything is on the up and up.
Really solid. Summarizes a lot of my thoughts nicely, including bewilderment at the collective panic at a very low risk event, especially among the demographics least at risk. Thank you.
Dallas County seems to have gone to a different format. If I am reading this correctly they will put out this summary on Tuesdays and Fridays.
Dallas County 2019 Novel Coronavirus (COVID-19) Summary 07/28/2020
Link found here
2019-novel-coronavirus.php
Dallas County 2019 Novel Coronavirus (COVID-19) Summary 07/28/2020
Link found here
2019-novel-coronavirus.php
Featured Stories
See All
A&M at risk of undercut when undertaking trip to underrated Auburn
by Olin Buchanan
48:02
6h ago
2.3k
5:59
3h ago
770
16:53
7h ago
2.3k