[OutsiderFan]

So much for giving the HHS Secretary the benefit of the doubt on "streamlining and modernizing the data." I challenge anyone to make heads or tails of the formatting, let alone the actual data.

[tweet]https://twitter.com/Laurie_Garrett/status/1285702131090100232[/tweet]

[OutsiderFan]

So much for giving the HHS Secretary the benefit of the doubt on "streamlining and modernizing the data." I challenge anyone to make heads or tails of the formatting, let alone the actual data.

[tweet]https://twitter.com/Laurie_Garrett/status/1285702131090100232[/tweet]

[lex24]

Here is its quasi mission statement:  https://protect-public.hhs.gov/pages/about

Are you a decision maker?  Don’t think so.  So,  quite honestly it’s irrelevant whether or not you understand this. The question is is it understandable to those who are the decision-makers. If it’s not I’m sure will know it soon. And I suspect changes will be made. If it is, then it’s serving it’s purpose.

As for the rest of us – I’m sure there are some enterprising reporters out there who can explain the data to us.

[lex24]

Here is its quasi mission statement:  https://protect-public.hhs.gov/pages/about

Are you a decision maker?  Don’t think so.  So,  quite honestly it’s irrelevant whether or not you understand this. The question is is it understandable to those who are the decision-makers. If it’s not I’m sure will know it soon. And I suspect changes will be made. If it is, then it’s serving it’s purpose.

As for the rest of us – I’m sure there are some enterprising reporters out there who can explain the data to us.

[OutsiderFan]

Here is its quasi mission statement:  https://protect-public.hhs.gov/pages/about

Are you a decision maker?  Don’t think so.  So,  quite honestly it’s irrelevant whether or not you understand this. The question is is it understandable to those who are the decision-makers. If it’s not I’m sure will know it soon. And I suspect changes will be made. If it is, then it’s serving it’s purpose.

As for the rest of us – I’m sure there are some enterprising reporters out there who can explain the data to us.

https://preventepidemics.org/wp-content/...Report.pdf

"Tremendous effort has been made by state, county and city public health departments to share COVID-19 data. It is crucial that they publicly report essential data on COVID-19 using consistent indicators that can be compared both across regions and over time. This allows individuals and communities to both understand and reduce their COVID-19 risk; transparency about the effectiveness of control measures promotes accountability and motivates continued improvement."

Right now, the only national source of cases, testing, deaths, hospitalizations and ICU beds data, that gives apples/apples view is http://covidusa.net by Johns Hopkins. But it still doesn't go in great depth and is not a government sanctioned data source. The Santa Clara County data is the best I've seen from any government agency in the U.S. That doesn't mean there aren't better ones, but I haven't seen any.

[OutsiderFan]

Here is its quasi mission statement:  https://protect-public.hhs.gov/pages/about

Are you a decision maker?  Don’t think so.  So,  quite honestly it’s irrelevant whether or not you understand this. The question is is it understandable to those who are the decision-makers. If it’s not I’m sure will know it soon. And I suspect changes will be made. If it is, then it’s serving it’s purpose.

As for the rest of us – I’m sure there are some enterprising reporters out there who can explain the data to us.

https://preventepidemics.org/wp-content/...Report.pdf

"Tremendous effort has been made by state, county and city public health departments to share COVID-19 data. It is crucial that they publicly report essential data on COVID-19 using consistent indicators that can be compared both across regions and over time. This allows individuals and communities to both understand and reduce their COVID-19 risk; transparency about the effectiveness of control measures promotes accountability and motivates continued improvement."

Right now, the only national source of cases, testing, deaths, hospitalizations and ICU beds data, that gives apples/apples view is http://covidusa.net by Johns Hopkins. But it still doesn't go in great depth and is not a government sanctioned data source. The Santa Clara County data is the best I've seen from any government agency in the U.S. That doesn't mean there aren't better ones, but I haven't seen any.

[lex24]

Here is its quasi mission statement:  https://protect-public.hhs.gov/pages/about

Are you a decision maker?  Don’t think so.  So,  quite honestly it’s irrelevant whether or not you understand this. The question is is it understandable to those who are the decision-makers. If it’s not I’m sure will know it soon. And I suspect changes will be made. If it is, then it’s serving it’s purpose.

As for the rest of us – I’m sure there are some enterprising reporters out there who can explain the data to us.

https://preventepidemics.org/wp-content/...Report.pdf

"Tremendous effort has been made by state, county and city public health departments to share COVID-19 data. It is crucial that they publicly report essential data on COVID-19 using consistent indicators that can be compared both across regions and over time. This allows individuals and communities to both understand and reduce their COVID-19 risk; transparency about the effectiveness of control measures promotes accountability and motivates continued improvement."

Right now, the only national source of cases, testing, deaths, hospitalizations and ICU beds data, that gives apples/apples view is http://covidusa.net by Johns Hopkins. But it still doesn't go in great depth and is not a government sanctioned data source. The Santa Clara County data is the best I've seen from any government agency in the U.S. That doesn't mean there aren't better ones, but I haven't seen any.

 I don’t think HHS -  at least the information from the link provided - is designed for you and me to interpret. I think it’s designed to provide information to the decision makers. At least that’s what it says on it’s homepage. So the fact that you have difficulty interpreting the data (not a criticism I certainly cant interpret it)  is not determinative of the effectiveness of the site.  What is important is whether the decision makers can interpret the data. And certainly there are those out there who can interpret the data and provide information to the general public as to what it’s setting forth. Right?

 If I am incorrect and this site is supposed to be something that informs the general public – then I will stand corrected and agree with you. It’s does not appear comprehensible to the layperson. (Or at least to me).

[lex24]

Here is its quasi mission statement:  https://protect-public.hhs.gov/pages/about

Are you a decision maker?  Don’t think so.  So,  quite honestly it’s irrelevant whether or not you understand this. The question is is it understandable to those who are the decision-makers. If it’s not I’m sure will know it soon. And I suspect changes will be made. If it is, then it’s serving it’s purpose.

As for the rest of us – I’m sure there are some enterprising reporters out there who can explain the data to us.

https://preventepidemics.org/wp-content/...Report.pdf

"Tremendous effort has been made by state, county and city public health departments to share COVID-19 data. It is crucial that they publicly report essential data on COVID-19 using consistent indicators that can be compared both across regions and over time. This allows individuals and communities to both understand and reduce their COVID-19 risk; transparency about the effectiveness of control measures promotes accountability and motivates continued improvement."

Right now, the only national source of cases, testing, deaths, hospitalizations and ICU beds data, that gives apples/apples view is http://covidusa.net by Johns Hopkins. But it still doesn't go in great depth and is not a government sanctioned data source. The Santa Clara County data is the best I've seen from any government agency in the U.S. That doesn't mean there aren't better ones, but I haven't seen any.

 I don’t think HHS -  at least the information from the link provided - is designed for you and me to interpret. I think it’s designed to provide information to the decision makers. At least that’s what it says on it’s homepage. So the fact that you have difficulty interpreting the data (not a criticism I certainly cant interpret it)  is not determinative of the effectiveness of the site.  What is important is whether the decision makers can interpret the data. And certainly there are those out there who can interpret the data and provide information to the general public as to what it’s setting forth. Right?

 If I am incorrect and this site is supposed to be something that informs the general public – then I will stand corrected and agree with you. It’s does not appear comprehensible to the layperson. (Or at least to me).

[M T]

Doesn't seem that unreasonable to me as a data source.  It isn't good for visualization.
I don't know what qualifications your favorite pundit has, but if she doesn't seem able to understand something like this, I sure wouldn't trust her to understand economy or anything complex!


First thing I noticed (on a Wednesday) is that the data hasn't been updated for 2 DAYS!!  I would expect this to be updated continuously as the data comes in.  Are they fat-fingering the data in order to introduce a delay??

Another thing I notice is that history is not seemingly available.  You can't see how things have changed in the last day (uh, in a whole week or 3, depending upon how often they update it) or in the last month.

Another thing I notice is that the data rows do not indicate date of the data.  You can't know how stale the data is.

(I'm not sure the above are really any different than the CDC.  So much of the old infrastructure seems built not thinking about fast-changing epidemics.)

Another thing I don't see is any clear metadata info, explaining the rows and columns.  Yeah, they do say that certain columns are numbers or text (duh, but you do need that) but not any metadata indicating whether numbers are integers, ratios, percentages, etc. or how they are calculated.
There's nothing in the data table to indicate how much data is new or old or missing.

There is an important statistic NOT being reported:  The number  or percentage of ICU patients with COVID.

First, the rows are the states or other regions.

So, here's what I'd put as the metadata for the columns

ObjectID: Index Number; consecutive, beginning at 1, and unique across all rows.
State Name: Text.  Full name of the state (unique)
State Abbreviation: Text.  2 letter abbreviation of the state (per some standard I'm not going to chase down) (unique)
State FIPS: Numeric.  2 digit code (per some  standard)
State HHS Region: Natural Number.  (per some standard)
Inpatient Bed Utilization (Estimate):  Percentage (0.0 - 100.0)  The percentage of occupied inpatient beds to the total number of all available beds (ICU & otherwise, I think)
Percent Inpatients with COVID (Estimate):  Percentage (0.0 - 100.0) The percentage of occupied inpatient beds assigned to patients with confirmed COVID-19 (? plus suspected COVID-19 ??) to the total number of all available beds.
ICU Beds Used (Estimate): Number.  The number of occupied inpatient beds for all ICU patients.
Inpatient Beds Used (Estimate): Number.  The number of occupied inpatient beds (ICU & otherwise, I think)
Inpatient Beds Used COVID (Estimate): Number.  The number of occupied inpatient beds (ICU & otherwise, I think) assigned to patients with confirmed COVID-19 (? plus suspected COVID-19 ??)
Shape_Area: Number.  This is some number telling you the area of the state.  I don't know if  these are hectacres or what.
Shape_Length: Number.  I suspect this is some number telling you the longest e/w or n/s dimension of the state, so you can know how to scale the image of the state when you display it.

If you click on the Overview tab and then click on one of the attributes (columns), you get a graph.  Clicking on FIPS doesn't tell you much, except that none appear to be missing.  If you click on "Inpatient Bed Utilization (Estimate)", you get a STATE-LEVEL representation of how full hospitals are.   One state is 90-100% full.  Of course, when you look at California, half the hospitals could be keeping patients in parking lots while the other half were 20% full and it would look like the state is doing fine.

[M T]

Doesn't seem that unreasonable to me as a data source.  It isn't good for visualization.
I don't know what qualifications your favorite pundit has, but if she doesn't seem able to understand something like this, I sure wouldn't trust her to understand economy or anything complex!


First thing I noticed (on a Wednesday) is that the data hasn't been updated for 2 DAYS!!  I would expect this to be updated continuously as the data comes in.  Are they fat-fingering the data in order to introduce a delay??

Another thing I notice is that history is not seemingly available.  You can't see how things have changed in the last day (uh, in a whole week or 3, depending upon how often they update it) or in the last month.

Another thing I notice is that the data rows do not indicate date of the data.  You can't know how stale the data is.

(I'm not sure the above are really any different than the CDC.  So much of the old infrastructure seems built not thinking about fast-changing epidemics.)

Another thing I don't see is any clear metadata info, explaining the rows and columns.  Yeah, they do say that certain columns are numbers or text (duh, but you do need that) but not any metadata indicating whether numbers are integers, ratios, percentages, etc. or how they are calculated.
There's nothing in the data table to indicate how much data is new or old or missing.

There is an important statistic NOT being reported:  The number  or percentage of ICU patients with COVID.

First, the rows are the states or other regions.

So, here's what I'd put as the metadata for the columns

ObjectID: Index Number; consecutive, beginning at 1, and unique across all rows.
State Name: Text.  Full name of the state (unique)
State Abbreviation: Text.  2 letter abbreviation of the state (per some standard I'm not going to chase down) (unique)
State FIPS: Numeric.  2 digit code (per some  standard)
State HHS Region: Natural Number.  (per some standard)
Inpatient Bed Utilization (Estimate):  Percentage (0.0 - 100.0)  The percentage of occupied inpatient beds to the total number of all available beds (ICU & otherwise, I think)
Percent Inpatients with COVID (Estimate):  Percentage (0.0 - 100.0) The percentage of occupied inpatient beds assigned to patients with confirmed COVID-19 (? plus suspected COVID-19 ??) to the total number of all available beds.
ICU Beds Used (Estimate): Number.  The number of occupied inpatient beds for all ICU patients.
Inpatient Beds Used (Estimate): Number.  The number of occupied inpatient beds (ICU & otherwise, I think)
Inpatient Beds Used COVID (Estimate): Number.  The number of occupied inpatient beds (ICU & otherwise, I think) assigned to patients with confirmed COVID-19 (? plus suspected COVID-19 ??)
Shape_Area: Number.  This is some number telling you the area of the state.  I don't know if  these are hectacres or what.
Shape_Length: Number.  I suspect this is some number telling you the longest e/w or n/s dimension of the state, so you can know how to scale the image of the state when you display it.

If you click on the Overview tab and then click on one of the attributes (columns), you get a graph.  Clicking on FIPS doesn't tell you much, except that none appear to be missing.  If you click on "Inpatient Bed Utilization (Estimate)", you get a STATE-LEVEL representation of how full hospitals are.   One state is 90-100% full.  Of course, when you look at California, half the hospitals could be keeping patients in parking lots while the other half were 20% full and it would look like the state is doing fine.

[allrightynow]

Right now, the only national source of cases, testing, deaths, hospitalizations and ICU beds data, that gives apples/apples view is http://covidusa.net by Johns Hopkins. But it still doesn't go in great depth and is not a government sanctioned data source. The Santa Clara County data is the best I've seen from any government agency in the U.S. That doesn't mean there aren't better ones, but I haven't seen any.

Correct me if I'm wrong, but I don't think covidusa.net actually tracks any hospitalization or ICU beds data, it only does a rough estimate based on an assumed percentage of new cases and is pretty much way off of actual reported numbers. The best "national" site I've found is https://covidtracking.com/data/charts which tracks what you list except for ICU beds data.

[allrightynow]

Right now, the only national source of cases, testing, deaths, hospitalizations and ICU beds data, that gives apples/apples view is http://covidusa.net by Johns Hopkins. But it still doesn't go in great depth and is not a government sanctioned data source. The Santa Clara County data is the best I've seen from any government agency in the U.S. That doesn't mean there aren't better ones, but I haven't seen any.

Correct me if I'm wrong, but I don't think covidusa.net actually tracks any hospitalization or ICU beds data, it only does a rough estimate based on an assumed percentage of new cases and is pretty much way off of actual reported numbers. The best "national" site I've found is https://covidtracking.com/data/charts which tracks what you list except for ICU beds data.

[OutsiderFan]

Right now, the only national source of cases, testing, deaths, hospitalizations and ICU beds data, that gives apples/apples view is http://covidusa.net by Johns Hopkins. But it still doesn't go in great depth and is not a government sanctioned data source. The Santa Clara County data is the best I've seen from any government agency in the U.S. That doesn't mean there aren't better ones, but I haven't seen any.

Correct me if I'm wrong, but I don't think covidusa.net actually tracks any hospitalization or ICU beds data, it only does a rough estimate based on an assumed percentage of new cases and is pretty much way off of actual reported numbers. The best "national" site I've found is https://covidtracking.com/data/charts which tracks what you list except for ICU beds data.

I don't know for sure, but I haven't ever really put much stock in Covidusa.net's hospital numbers, especially because the percentages never seemed believable to me.  I really only follow the case numbers, deaths, test positivity rates and projections on that site. I've even started a table that records the 7/14/30 day predictions to see if they line up with reality when those days arrive. 

The 7-day case numbers projected last Friday appear to be right in line with what is actually going to be reported.  Deaths seem to be much more difficult to project accurately. Last Friday's 7-day death projection was 160k by 7/24. That's a lot to be off by, but it's also possible the death numbers aren't being accurately reported. I mean it's hard to believe that after this much data, the projections for cases would be pretty accurate and deaths would not be. I get there are possible issues related to the age of patients now being younger and less likely to die than previous periods that had more older people sick, plus possible improved treatments, but the data and projections are updated daily, so these big discrepancies in death projection accuracy versus case prediction accuracy leave me skeptical. In any case, it's way easier to undercount actual infections than it is deaths, so it's puzzling for sure.

[OutsiderFan]

Right now, the only national source of cases, testing, deaths, hospitalizations and ICU beds data, that gives apples/apples view is http://covidusa.net by Johns Hopkins. But it still doesn't go in great depth and is not a government sanctioned data source. The Santa Clara County data is the best I've seen from any government agency in the U.S. That doesn't mean there aren't better ones, but I haven't seen any.

Correct me if I'm wrong, but I don't think covidusa.net actually tracks any hospitalization or ICU beds data, it only does a rough estimate based on an assumed percentage of new cases and is pretty much way off of actual reported numbers. The best "national" site I've found is https://covidtracking.com/data/charts which tracks what you list except for ICU beds data.

I don't know for sure, but I haven't ever really put much stock in Covidusa.net's hospital numbers, especially because the percentages never seemed believable to me.  I really only follow the case numbers, deaths, test positivity rates and projections on that site. I've even started a table that records the 7/14/30 day predictions to see if they line up with reality when those days arrive. 

The 7-day case numbers projected last Friday appear to be right in line with what is actually going to be reported.  Deaths seem to be much more difficult to project accurately. Last Friday's 7-day death projection was 160k by 7/24. That's a lot to be off by, but it's also possible the death numbers aren't being accurately reported. I mean it's hard to believe that after this much data, the projections for cases would be pretty accurate and deaths would not be. I get there are possible issues related to the age of patients now being younger and less likely to die than previous periods that had more older people sick, plus possible improved treatments, but the data and projections are updated daily, so these big discrepancies in death projection accuracy versus case prediction accuracy leave me skeptical. In any case, it's way easier to undercount actual infections than it is deaths, so it's puzzling for sure.