One moment please.
Getting fresh, delicious data...
The data feed is dead! Long live the data feed!
Our beloved, The COVID Tracking Project, ceased operations as of March 7th, 2021. This major project, propitiously and generously started by The Atlantic magazine, was the only comprehensive, publicly available source when the world was desperate to understand what was happening. New sources arose over the months, but TCTP was still a primary source for most of the places that provide data online.
Since the bracing news of TCTP's closure, we have been busy scouting out a new a new source of quality data, and adapting the precision tool that is PlagueData.com, to display it to you. Our new source is John Hopkins University - Whiting Engineering School. The integration work was completed on Friday, April 9th. You are looking at it! (Well, if you scroll a little bit.)
PlagueData.com now running on Johns Hopkins University data!
What took so long? Unlike The COVID Tracking Project (TCTP), Johns Hopkins University (JHU) doesn't offer an API. It just provides CSV files, like spreadsheets of data for each day. You aren't expected to understand this, so it won't be on the quiz, but it means that PlagueData.com required a significant rewrite to consume data in this form. In addition, there are significant differences in what actual data are provided. For example, TCTP provided daily changes, but no per capita numbers, while JHU provides provides no daily changes, but does provide per capita numbers. We had to add code to do calculations no longer included in the data.
There will be a few changes in our data:
Hospitalization
TCTP manually aggregated hospitalization numbers, and processed them to provide a workable output. In fact, JHU as well as many other data web sites, used these numbers. Now, Hospitalization numbers are very sketchy and hard to come by, so we don't expect to have current (post 3/7) Hospitalization numbers for a while.
Test Percent Positives
There has been a significant change in this data as testing has become more and more available. TCTP was already struggling with how and what to present, as states have very different methodologies. These different methodologies actually change the meaning of the numbers, so comparing state to state was difficult to make meaningful.
TCTP created a system by which they picked different numbers for each state, so that they could be compared somewhat. JHU doesn't do this, so our Percent positive number will be useful in viewing a states performance over time, but may not be a good way to compare between states. We are still studying this, but for now, this is how they should be viewed. Read more about testing on the TCTP site:
Test Types 101: The Pros and Cons of Different Approaches.
As our users are aware, we've been big advocates of focusing on Percent Positives over "Cases" as the best way to see trends independent of how well each state is doing with testing.
Fortunately, as testing has become institutionalized in every state, testing has become much more consistent, and case numbers have become much more reliable. Hence, Percent positives, are no longer as critical in determining how a state is faring.
Some interesting sets:
This just In - 10/19/2020
The Surge is from Sturgis!
Sturgis Super Spreader Spreads its Wings Across the Upper Midwest
As if science doesn't actually work, people speculated on whether the annual motorcycle event in Sturgis, SD, August 7-16, would cause a surge in C-19 cases. I'm guilty of it, too! I can't believe people are going to do something so misguided, so dangerous, and then wonder if somehow they got away with it, and I'm hibernating in my cave for no good reason.
When Georgia opened early. When other politically right-leaning states followed suit. When the President held his rally in Oklahoma. In every case, due to the initial flatness of the exponential growth curve of contagion, we spend anxious weeks wondering if all our self-denial and sacrifice has been a silly over reaction.
It always happens and Sturgis is probably one of the worst
A half million people traveled to Sturgis, SD, dropping or collecting virus at motels and gas stations along the way. In Stugis, they reveled in the seeming normalcy of being together and celebrating: at concerts, in densely packed bars, in restaurants, in stores, at attractions and at events. Without masks. They danced, sang and shouted to be heard over the crowd. After ten heady days, they headed home, dropping or collecting virus at motels and gas stations along the way. It would be a memorable, bucket list experience - if it was 2019. In 2020, it was the possibly the best way to spread a potentially deadly virus to as many people as possible.
It takes several weeks for an event's seedlings to grow above the general noise.
Without decent contact tracing, something most of the USA has failed at miserably, it is hard to tie specific cases to Sturgis. The first known related death was September 2nd. Subsequently, it was reported that 321 cases resulted from the rally. We'd expect these onesey-twosey occurences three or four weeks out, but it takes severel cycles for the real impact to be seen.
This study concluded, based on tracking attendees via anonymized cellphone data, that the event could account for up to a quarter million additional cases in the subsequent months. The study has been heavily criticized, primarily by people wanting it not to be true. Among those was SD Gov. Kristi noem: "this report is nothing short of an attack on those who exercised their personal freedom to attend Sturgis". But the surge in the upper midwest, may suggest otherwise.
Setting aside the study results, its maps tell which states have the most attendees: Arizona, California, Colorado, Idaho, Illinois, Iowa, Minnesota, Montana, Nebraska, Nevada, North Dakota, Washington, Wisconsin, Wyoming. Some of these states have shown big surges. Some have not. The ones that haven't are states that have solid quarantine policies in place: California, Colorado, Illinois, Minnesota, Washington, and the now chastened Arizona.
So how are the remaining states doing? Not well. The substantial increase in cases in these states could be driven by something else, but given the participation in the Sturgis event, and the timing, it is hard to see what else the cause could be.
Take a look for yourself...
---
This just In - 8/26/2029
CDC Pressured to Relax C19 Testing Recommendations
Testing no longer recommended for those exposed if asymptomatic.
Obviously, this is not a good sign. Once COVID data was no longer routed to the CDC, as ordered by the White House in mid July, one expected that agencies closer to the administration, and with less technical expertise, would start making decisions regarding this crisis. This is a scary example.
I leave it to you to read the various articles (links below), but here are a few things to consider. It is well known that 40% or more people capable of spreading C19 are asymptomatic. The only way to know if they are a risk to other people is through testing, or following the future trail of destruction. While it is too much to ask, given our poorly managed testing infrastructure, that everyone get tested, the CDC's (now obsolete) recommendations that those known to have been exposed to the virus, get tested, seems not only smart, but crucial. Then why the change?
Remember: New "Cases" values are easily manipulated by reducing testing!
According to the various reporting, there is no stated basis for this decision. CDC insiders say it was imposed from higher up, and formally refer reporters to HHS for questions. HHS says it was a considered decision, and declines to provide any actual reason.
C19 "Case" numbers will now fall nation-wide while infections increase by untested c19 carriers.
As you've read a million times at plaguedata.com, New "Cases", the primary metric used by the media, is heavily affected by testing rate, and as such is easily manipulated. With an election approaching, and US COVID performance looking terrible, any reduction in testing will be reflected across the media-verse as an improvement in the handling of COVID. The numbers will look better, as the virus spreads more and more quickly.
I have been begging via letters, email, comments, various media outlets to stop using this easily manipulated metric and focus on Percent Positives, which are much harder to fudge. They haven't done so, and as a result, they will be participating in what amounts to propaganda regardless of their political stripe.
Given how badly the no-mask policy went, I'm surprised that the administration would try yet another short-sighted scheme. Perhaps, the hope is that the effects won't be noticed until after the election?
ArsTechnica: In alarming move, CDC says people exposed to COVID-19 do not need testing
NYT: Top U.S. Officials Instructed the C.D.C. to Change Testing Guidelines
CNN: CDC was pressured 'from the top down' to change coronavirus testing guidance, official says
---
8/26/2029
CDC Pressured to Relax C19 Testing Recommendations
Testing no longer recommended for those exposed, but asymptomatic.
Remember: New "Cases" values are easily manipulated by reducing testing!
Obviously, this is not a good sign. Once COVID data was no longer routed to the CDC, as ordered by the White House in mid July, one expected that agencies closer to the administration, and with less technical expertise would start making decisions regarding this crisis. Here is a scary example.
I leave it to you to read the various articles (links below), but here are a few things to consider. It is well known that 40% or more people capable of spreading C19 are asymptomatic, and the only way to know if they are a risk to other people is through testing, or following the future trail of destruction. While it is too much to ask, given our poorly managed testing infrastructure, that everyone get tested, the CDC's (now previous) recommendations that those known to have been exposed to the virus, get tested, seems not only smart, but crucial.
According to the various reporting, there is no stated basis for this decision. CDC insiders say it was imposed from higher up, and formally refer reporters to HHS for questions. HHS says it was a considered decision, and declines to provide any actual reason. Why would this happen?
C19 "Case" numbers will now fall nation-wide while infections increase by untested c19 carriers.
As you've read a million times at plaguedata.com, New "Cases", the primary metric used by the media, is heavily affected by testing rate, and as such is easily manipulated. With an election approaching, and US COVID performance looking terrible, any reduction in testing will be reflected across the media-verse as an improvement in the handling of COVID. The numbers will look better, as the virus spreads more and more quickly.
I have been begging via letters, email, comments, various media outlets to stop using this easily manipulated metric and focus on Percent Positives, which are much harder to fudge. They haven't done so, and as a result, they will be participating in what amounts to propaganda regardless of their political stripe.
Given how badly the no-mask policy went, I'm surprised that the administration would try yet another short-sighted scheme. Perhaps, the hope is that the effects won't be noticed until after the election?
ArsTechnica: In alarming move, CDC says people exposed to COVID-19 do not need testing
NYT: Top U.S. Officials Instructed the C.D.C. to Change Testing Guidelines
CNN: CDC was pressured 'from the top down' to change coronavirus testing guidance, official says
---
8/13/2029
The Good News, Bad News
Good: we hit the lowest number of Losers (19) since we started counting!
Good: no states with % Positive Tests > 20% since way back!
BAD: that was a couple of days ago.
Yesterday we had only 19 Losers for % Positive Tests. This metric was at 40 not too long ago and stayed in the upper thirties for a long time. We were excited enough to mention it here, when this dropped to 32. Now, we are at 20, and while it is worth celebrating, we are going the wrong way again. Let's hope it is a blip.
On August 10th, there were no states with % Positive Tests above 20%! This is the dangerous territory recently occupied by states like Arizona and Alabama that really let things get out of hand. Texas and Florida have been flirting with this range for weeks, as well. With some of these states really working to rein things in, we managed to get EVERY STATE out of that zone, this past weekend, for the first time in several weeks.
Unfortunately, on Tuesday, Texas stopped flirting with disaster, and dove into this zone with both feet. The state is now in the mid-low twenties and climbing steeply! If you look at Cases, things appear more optimistic, but that is only appearance - their testing rate is dropping, which, as we've mentioned a MILLION times, gives an artificial reduction in Cases.
Let's hope that the Texans are chastened and take action soon. And, that Florida, which has been just outside this territory for weeks, and mildly descending recently, dosen't do something idiotic, like forbid their sheriffs deputies to wear masks. Oh, wait, that just that just happened in Marion County FL).
---
7/29/2029
Very bracing news: Europe Reignites!
The recent COVID-19 news from Europe and the rest of the globe, begs the question:
IS IT POSSIBLE TO RE_OPEN WITHOUT A VACCINE?
Even countries that patiently waited until infection rates were quite low, and that have responsible citizens and very good compliance with best practices, are seeing significant resurgence in infection after only a few weeks of what is thought to be "responsible" re-opening.
Though admittedly, I was persuaded by experts, that there was some threshold below which we could start interacting again, I never understood how it could be possible. Remember, just ONE infected person got off of a plane from China, and a few weeks of exponential growth later, we were in deep shit.
How, then, are practices designed only to reduce the LIKELIHOOD of contagion supposed to prevent this "patient one" scenario from recurring? They statistically reduce the probability of getting infected, but when it only takes ONE person to start it all up again, this is useless - unless the probability is ZERO.
THESE PRACTICES ARE CRITICAL for keeping the infection rate low enough that we can treat people, that we can save lives, that we can keep healthcare workers from burning out and quitting. THAT is what the practices are designed for.
What does this mean for you? Keep doing all the same responsible things. But, don't get your hopes up, that somehow this is going to be letting up with the waning of summer. Or fall.
And hope to god that the Pollyannas are right, and a vaccine is available within a year.
---
7/27/2020
What's the deal with AZ? Are they cookin' the books?
Arizona's New Cases count is plummeting, because the amount of testing in AZ is plummeting. Is this a cynical attempt to control how they look in the news? ("New Cases" is the most reported metric.) Or, are they having trouble acquiring testing resources? Regardless, less testing is bad. It reduces the accuracy of the numbers needed to make critical policy decisions.
7/27/2020
New Positive Tests Loser count is falling!
New Positive Tests (Percent Positives) Losers are states for which this critical number has risen over the past week. We've watched it grow from low thirties to over 40, and mostly hovering around 37-38. As of yesterday it is down to 32.
The number of Winners for this metric, is up quite a bit to 19. Among these are five of the ten states with the worst infection rates: AZ*, FL, SC, TX, UT.
This makes sense given that the governments in these states, which at first rejected the recommendations of health experts, reversed course and embraced the precautions taken by states which have better managed the spread. For example, on July 2nd, Texas announced an appropriately strong face covering mandate, backed with fines. It is no surprise that now, 2-3 week later, percent New Positives have plateaued and are starting to decline.
Some of these states are improving, without policy changes, as their citizens take precautions of their own in response to reports of rapidly increasing infection in their state.
*But, what's with AZ? Their New Cases count is also falling, and even faster, because the amount of testing AZ is doing is falling sharply. Are they having trouble acquiring testing resources? Or, could this be a cynical attempt to control how they look in the news? ("New Cases" is the most reported metric.) Regardless, less testing is bad. It reduces the accuracy of the numbers needed to make critical policy decisions.
---
7/20/2020
PlagueData.com minor update grab bag!
We've implemented a few new interface enhancements that don't quite merit their own block in the sidebar, adorned with an exciting red headline. You will find them useful, nonetheless:
- Settable graph starting date - even with normalization, the very high values during the initial C-19 spike, make it hard to see trends in the relatively low values of states that are currently doing well. By, changing the starting date, you can start the graph after that initial spike, making it much easier to see what's going on down in the weeds.
- Smart(er) graph scaling - when toggling between performance sets (Win/Lose, Red/Blue etc.), the graphs now retain the same, optimized scale. Previously, the graphs automatically re-scaled based on the data. This can be confusing, because it isn't obvious that one graph is 0 - 80, and the next is 0 - 30. An update last week improved things by fixing the normalized graphs to a range of 0 - 100. Less confusing, but pretty lame if the data points are all under ten, for example. Now the max Y-value is calculated and fixed, so if the highest value for a metric is 29, the graph goes to 30, and stays there when change between various sets.
These changes support whole point of plaguedata.com, which is to provide key metrics that are as important as they are hard to find, to be able to see trends, despite the disparity in state population and case loads, and to be able to compare states' performance quickly and easily.
---
7/17/2020
PlagueData.com has a new graph layout!
For a small subset of users, the supporting graphs (the tiny ones) were useful, but not for most. Now, you will have two graphs per section, both time averaged (smoothed). One will have have the normalized data, you are familiar with, which makes it easy to see trends. The other will not be normalized, so you can easily see the actual values with out referring to the tabular data.
Clicking the ↔ symbol near the smaller graph will make it full size. As such you can toggle between the two!
Note: this will affect the winner/looser calculations slightly, possibly pushing states that are very close to the threshold out of a given category.
---
7/16/2020
Bad News for USA: White House orders hospitals to bypass CDC with data reporting!
Good News: This should have no impact on PlagueData.com.
Per "The COVID Tracking Project", our source data source, "Every day, our volunteers compile the latest numbers on tests, cases, hospitalizations, and patient outcomes from every US state and territory."
In other words, the data is being manually gathered directly from various state sources, and should not be subject to manipulation at the federal level, as I understand it. This doesn't mean that states can't and won't manipulate data for political advantage, as has always been the case. However, it makes it very difficult for anyone to manipulate the numbers across the board.
---
7/10/2020
New Cases numbers are meaningless without accounting for testing increases!
Why can't the media and state agencies grasp this simple fact that plaguedata.com users have know for weeks?! Case in point...
Yesterday, friends in Illinois (which is doing a fantastic job, BTW), were seriously bummed because of a jump in New Cases following some thoughtful re-opening. My first thought, "Really, they were doing so well!" Of course, my next thought: "check the % New Positives before you get your skivvies in a bunch" (see Hot TIP, above right, for the easiest way to do this.).
Lo and behold, Illinois is still a PlagueData.com New Positives winner. A little more digging shows that their new tests went from 21k to 36k in the previous 4 days. Even with declining % Positives, the New Cases jumped 70% in four days. Terrifying! And, absolutely meaningless.
PLEASE! If your news source is citing New Cases spiralling out of control, go to plaguedata.com and check your state's New Positive Tests % before defenestrating yourself.
---
7/9/2020
I have to hand it to Utah.
A mere week ago, they were the WORST of the WORST, with a New Positive rate sky-rocketing into the ether. Had Elohim forsaken the desert nation? Evidently, not.
They have sharply turned things around, and now appear in the best category for that, most important, metric. Their absolute numbers still suck, but the trend is strongly in the right direction.
I'm giving an honorable mention to Wyoming and South Carolina, which have also corrected sharply to the good.
It can be done, people!