National changes Testing Texas update

National Update

Looks like we’ve hit our peak in the U.S.? Unfortunately not yet. We won’t start seeing the “true” holiday impact until mid- January. 

You may remember my post (back in April, I think) showing that test and case numbers are dependent on human behavior. In Texas, for example, daily case counts are much lower on days with a thunderstorm or tornado warning because testing sites close due to inclement weather. Case counts are highest on Tuesdays and lowest on Saturday/Sunday/Mondays. Reporting 7- or 14-day averages takes care of these day-to-day fluctuations. 

The same is true for the holiday season; it’s basically a really long weekend in terms of case and testing numbers. Facilities and labs have been inconsistently open, reporting is delayed, places are catching up, etc. You can see a dip in the graphs, but this is due to human behavior (rather than true reduced transmission). We won’t see the holiday impact until reporting has stabilized. It’s irresponsible to make conclusions from case, test, and death data right now. 

Right now, a much more reliable metric is hospitalizations. Hospitals are open 24/7 and, by now, have rigorous, systematic, and completely separate reporting systems than testing/case data. Because hospitalizations lag cases, we have a pretty good idea of whats been happening with cases behind closed doors. 

Unfortunately, hospitalizations are not looking good. Across the U.S., there are 123,614 COVID19 patients hospitalized right now. This is the highest it’s ever been. 

Not all places are equal though.

The Midwest continues their downward trend. Not one Midwest state reported an increase in hospitalizations since Christmas. 

The South, on the other hand, is responsible for 50% of the country’s hospitalizations. This is mainly driven by the high Texas hospitalizations. Also, the South just has more people than the Midwest or Northeast. 

Texas Hospitalizations by Trauma Service Area (TSA)

After we adjust for population, though, the West and Northeast is in worse shape than the South (but not by much) and closing in on Midwest’s peak. The West’s numbers are mainly due to Southern CA where some hospitals literally (not figuratively) have no more ICU beds.

The way we “define” hospitalizations is also changing compared to the beginning of the pandemic. The threshold for sending someone home is much lower than in the past. For example, some hospitals will send someone home with a 90% O2 level, which wouldn’t have happened prior to the pandemic or in the past few months when hospitals had beds. So, in reality, the hospitalization numbers are underestimated compared to hospitalization rates in the past. 

Nonetheless, hospitalization is the best metric we have right now. Continue to keep an eye on this number until other metrics stabilize.

Love, YLE

Data Source: Covid Tracking Project, which never ceases to amaze me with their clean, state-level data. And our dashboard at

Deaths FL Hospitalizations National changes Social Distancing Testing Texas update

Thanksgiving Surge?

It’s been 2 weeks since Thanksgiving and I was curious if we had a “surge upon a surge”.

In other words, did our acceleration (the rate of cases) change after Thanksgiving? Unfortunately, it’s a very simple question with a very complicated answer. If this blog were my day job, I could statistically figure this out. But it’s not, so I triangulated a few other data sources instead. This is what I found…


TPR is now 20.3% in the United States. It increased 15% since Thanksgiving. TPR is particularly concerning in the Southeast, where it’s increasing at higher rates since Thanksgiving than the rest of the country. While the Midwest finally seems to be moving past their peaks, their TPR’s are increasing again (likely related to Thanksgiving), which will slow their decline. The WHO has stated that countries need a TPR below 5%. While testing does not have a direct benefit because there is no cure, there are a number of indirect benefits: 1) public health officials know the “true” rate of infection and can deploy resources to the right areas to stop spread; 2) psychologically if someone tests positive then they are more likely to quarantine (hopefully).


Cases have increased 21% since Thanksgiving. Cases increased 22% two weeks before Thanksgiving. New hot spots have popped up since Thanksgiving, particularly along the Sun Belt (southern CA, AZ, TX) and the Northeast. Boston, in particular, has surpassed 100 daily cases per 100,000. And while Vermont and Maine have been more than impressive this entire pandemic, they too are seeing doubling rates.


Not enough time has passed since Thanksgiving to see the impact on hospitalizations/deaths. But there is no reason to believe they will not continue to mirror case trends. Fatality rate (number dead out of the number with positive COVID19 tests) continues to remain steady in the United States at ~1.9%. We should continue to see this, unless out health systems are strained too much. Then hospitals will have to start making hard decisions on who to save and who not to save. In April, Italy had to make these decisions (they decided not to treat those 80+ years) and fatality rate increased.


  1. Airports. The CHOP Policy Lab found circumstantial evidence that the most concerning areas of the country post-Thanksgiving are adjacent to our busiest airports: Los Angeles, Boston, DC, Atlanta, and Dallas. In other words, Thanksgiving air travel led to increased local transmission. For example, in Clayton County, Georgia (home to the Delta Hub and Atlanta airport), cases are doubling compared to surrounding counties.
  2. Distance Traveled. Interestingly, distance traveled did not change, on average, by much. We see an increase right before Thanksgiving, but honestly not as high as I would have expected. This only means people, on average, didn’t travel far. This doesn’t mean that family wasn’t close by and people didn’t get together. It was also very obvious that distance traveled varied by states too (see Figures). Wish I had more time to look into this. But still adds a little piece to the puzzle.

Conclusion: Right before Thanksgiving we were starting to see a plateau in cases. Then, after data reporting caught up, our cases continued exponentially increasing after Thanksgiving. So, I don’t think we saw a surge upon a surge. But we definitely didn’t stop our original surge. The pandemic continues to ravage our communities across the United States.

Love, YLE

Data Sources: I triangulated many sources of data for this report. I couldn’t have done it without the beautifully clean and workable data and graphs from the following sites:

COVID19 Tracking Project:

CHOP Policy Lab:



COVID19 Testing

As cases skyrocket, you should know the different types of tests out there. Which is confusing.

There are 2 types of diagnostic tests. Diagnostic tests tell you if you CURRENTLY have COVID19. If these come back positive, you NEED to quarantine or isolate yourself from others.

1. Molecular test. (Also called RT-PCR, viral test, molecular test, NAAT, LAMP test). This detects the genetic material of a virus. This test can be a nasopharyngeal (part of the throat behind the nose), nasal or throat swab or saliva test. You should get results the same day in some locations or up to a week at other locations. It is highly accurate (almost 100% in detecting infected people) and highly sensitive (so it will detect minuscule amounts of virus).

2. Antigen test. This detects specific parts of the virus (called proteins). It is a nasal or nasopharyngeal swab. You will get results very quickly (within 15 minutes). Positive results are usually highly accurate, but false positive can happen. Antigen tests are more likely to miss an active COVID19 infection because you need more of the virus particles present for detection (compared to PCRs). Negative results may need to be confirmed with a PCR test if you have symptoms (check with your doctor).

The other type of test is an antibody test. (also known as serological test, serology, blood test). This test detects whether you were ever infected with COVID19 in the PAST. This will not tell you if you are currently infected. This is a finger stick or blood draw. You should know results 1-3 days after. Sometimes you need a second antibody test for accurate results. It sometimes takes weeks after infection to develop enough antibodies to be detected in a test. You do not need to quarantine if this test is positive.

Check out the figure for the most ideal time for each test.

The best way to get a COVID19 test is to contact your doctor. Your state and local health departments also typically have testing information on their website.

Love, YLE

Data Sources: Nature article:… ;…/coronavirus-disease-2019-testing…

California Deaths National changes Testing

Early Spread of COVID19

Many tested negative for the flu in January and February. Could this have been COVID19?

Short answer: Yes.
Long answer…

We have to outline the first “original” cases:

• Jan 21-Feb 23: U.S. detected 14 COVID19 cases all related to travel from China

• Feb 26: 1st non-travel US case confirmed in CA (patient was ill starting Feb 13)

• Feb 28: 2nd non-travel US case popped up in WA

So, from this timeline, COVID19 started spreading in the United States on Feb 26 right? Nope.

CDC found that it was spreading earlier in the US by looking at four things:

1. Seattle Flu Study: Some scientists, during this time, just happened to be conducting a Seattle Flu Study. They were basically monitoring the flu from Nov 2018- March 2020 by testing people randomly. After the pandemic started, they went back to test these samples for COVID19. From Jan 1-Feb 20, none of their tests were positive. Their first sample was positive on Feb 21; the following week there were 8 positives; the following week there were 29 positives.

2. Gene analyses: Genes from early cases suggest that a virus imported directly or indirectly from China began circulating in the US between January 18 and February 9, followed by a COVID19 strain from Europe.

3. CDC has found other cases before Feb 26

• Jan 31: CA women became ill. Died 6 days later. She did not travel internationally. Postmortem, COVID19 positive.

• Feb 11: An infected passenger boarded the Grand Princess in Seattle leading to two outbreaks.

• Feb 13: CA man died at home. He did not travel internationally. Postmortem, COVID19 positive

4. Surveillance: ER records did NOT show an increase in visits for COVID19–like illness until February 28. CDC thinks this is because there were too few people with the disease to see an increase in ER visits in a meaningful way.

Translation? Community spread of the Chinese COVID19 strain likely started in January. Community spread of the European COVID19 strain started in Feb.

Why do we care? There are many reasons. One being that we can better estimate how many people truly died from COVID19 that were missed. We know from excess death analyses that we missed a lot in the beginning, meaning our current numbers are underreporting.

Love, your local epidemiologist

Data Source: Jorden MA, Rudman SL, et al. Evidence for Limited Early Spread of COVID-19 Within the United States, January–February 2020. MMWR Morb Mortal Wkly Rep 2020;69:680–684. DOI: icon

Deaths Hospitalizations National changes Testing


Well, July wasn’t pretty…either.

In order to get the best comprehensive picture, I triangulated several constructs:

Cases (i.e. incidence): In the month of July, 23 states jumped to a higher CDC COVID19 risk category. For example, OK jumped from orange (13.7 daily cases per 100,000) to red (26.8 daily cases per 100,000). Shout out to VT…the only state that got better (jumped from yellow to green). VT is the first state to make it to the green risk category.

Deaths: Because of increased incidence, cumulative and daily deaths have increased in July for the majority of states. The figure includes daily deaths July 31 compared to July 1. NJ is looking good! CFR or IFR are incredibly difficult to estimate (and take a lot of time), so I didn’t include. See earlier posts:;

Testing (i.e. test positivity rate [TPR]): 41 states look like they have testing under control (under 10% TPR). Although we really need to get this to below 5%. There are 10 states that need some serious help (over 10% TPR). We are looking at you AL, AZ, FL, GA, ID, LA, NE, SC, TX, and VA.

Hospitalizations: Not even going to try to compare July 31 to July 1 because of the data reporting switch. But hospitalizations did increase in July. We know this because incidence and deaths increase.

The good news is that it LOOKS like some states may have recently (like past two days) reached their peak. If you live in one of these states, DO NOT CHANGE A DARN THING. We need to be well down the curve to start opening strategically and changing individual behaviors.

Maybe August will be our month?? Here’s the BEST peer-reviewed scientific article of how to get out of this mess by October.

Love, your local epidemiologist

Data sources: Analysis and graphs (except the first one) by yours truly. Data came from many sources: COVID19 tracking project, Harvard , and CDC.


Testing disparities

Your data-driven update…

An article from was released yesterday regarding COVID19 and health disparities. I’m proud to have contributed. Here is a high-level synopsis:

Across the country, Hispanics and Blacks continue to be hit hardest by the COVID19 pandemic. This is due to many reasons. They are more likely to: be an essential worker, not have sick leave, live in densely populated areas, rely on public transportation, live multigenerational households, have preexisting chronic conditions, be under- or uninsured, have language barriers, etc. etc. etc.

Another reason is access to care, and more specifically, access to affordable COVID19 testing sites. It’s incredibly obviously when we look at a map.

I’m going to pick on Dallas, Texas. For those of you not familiar with Dallas, it’s greatly segregated by one freeway. South of the freeway is majority low income and minority households. North of the freeway is majority high income and white households.

As of this morning, there are 163 COVID19 testing sites in Dallas-Fort Worth. Only 7 are south of the freeway. We can see this in the figures with raw number of testing sites (Figure 1) and the NEED for testing sites (Figure 2).

This article also highlights other cities across the United States. I highly recommend you read it.

You may think this doesn’t apply to you. However, the virus doesn’t care about freeway lines. We need to improve access to affordable testing for the hardest hit populations so we stop playing whack-a-mole in this nation. Without access to testing, people won’t know they are positive, so they won’t quarantine (or don’t have the choice to quarantine), and the disease will continue to spread. We are all in this together, now let’s start acting like it.

Love, your local epidemiologist

Data Sources: