AZ California Deaths FL New York Predictions Texas update

Leading Causes of Death

There’s no doubt that COVID19 will be a top 10 leading cause of death in 2020. This is quite impressive given that the other diseases on this list typically take YEARS to manifest and are NOT contagious.

The interesting question is… what will COVID19’s exact rank be at the end of the year? It’s been a while since I updated this chart. This graph always stirs up discussion, so I added some more sophisticated analyses to address concerns. Let’s see if I can explain it…

I estimated three COVID19 ranks for the US and 8 other states. I used COVID19 deaths (up to last night) and compared to 2019 causes of death.

Low estimate (green): This estimates COVID19’s rank if the pandemic ended yesterday (i.e. everyone with COVID19 was cured overnight). Unfortunately, we know this isn’t true, but this is the absolute MINIMUM rank COVID19 will be.  

Medium estimate (orange): This estimates COVID19’s rank if we continue on our death trajectory.

High estimate (red): This estimates COVID19’s rank if we continue on our trajectory AND we count ALL excess deaths as COVID19 deaths. I understand that, in reality, all excess deaths are not likely COVID19, but this is the HIGHEST rank COVID19 could be.

So, in the United States, COVID19 will lie between the 3rd (high estimate) and 6th (low estimate) leading cause of death in 2020. In reality, it will be somewhere in the middle. In March, we (epidemiologists) estimated it would be 3rd leading cause of death in 2020. Looks like that’s going to be about right.

In Texas, COVID19 will lie between the 3rd (high estimate), 5th (medium), and 9th (low estimate) leading cause of death. Again, it will likely fall somewhere in between by the end of the year.

For CA, FL, AZ, NY, LA, WA, and IL rankings, see the following graphs. This is all I could get done before my eyes started shutting last night.

Lately, the flu debate has come to surface again. In EVERY state, COVID19 lowest possible rank is still higher than flu. So, I’m not really sure why we are still having this conversation…

Love, your local epidemiologist

Note: Yes, the 2019 numbers will also change this year. But this is the best we got, as CDC only reports these in aggregate form at the end of the year. It will actually be interesting, though, how other ranks change. For example, we know car crashes (unintentional injuries) have decreased while suicide has increased.

Data Sources: 2019 data is from the National Center for Health Statistics at the CDC. COVID19 deaths is from Johns Hopkins (US). Excess deaths is from the Weinburg lab. Graphs/analyses by yours truly.


Excess Deaths

Tracking the mortality impact of COVID19 is difficult. Real-time data is messy and there are vast differences in reporting across states. Critics have raised concerns that deaths not caused by the virus are improperly counted as COVID19. To account for this, epidemiologists started assessing “excess deaths”. In other words, how many “extra” deaths do we have in 2020 compared to the same time period in 2019?

The first “excess death” analysis I saw was in March conducted by epidemiologists in Europe at EuroMono. Following this, a research group in Yale worked on US data and published on July 1 in JAMA. They have continually updated the data since. This post leverages updated US data: excess deaths between March 1-June 27.

The majority of states have excess deaths in 2020 compared to 2019. For example, NJ: there were a total of 40,492 deaths from March 1-June 27, 2020. We expected 58% (23,400) of those deaths according to last year, but 42% (17,200) of the deaths were not expected.

Surprisingly (at least to me), there are 6 states that don’t have excess deaths in 2020: NE, HI, SD, AK, WV, & ND. In fact, they have FEWER deaths in 2020 than in 2019. This could be due to a decline in other categories of deaths, like car crashes.

Figure 2 categorizes excess deaths: COVID19, influenza/pneumonia, and other. So, in NJ, among the 17,200 excess deaths from March 1-June 27, 78% were due to COVID19 (green), 0% were due to influenza/pneumonia (light blue), and 22% were other/unknown (yellow).

The other/unknown (yellow) can be due to: 1) other deaths that have increased because of the pandemic (think suicide, delaying medical care); 2) deaths due to COVID19 but death certificate analyses delayed; or 3) deaths due to COVID19 but were never tested. Nonetheless, the authors of the paper concluded, “Official tallies of deaths due to COVID-19 underestimate the full increase in deaths associated with the pandemic in many states.”

Data is through June 27. As more time passes in the new surge, it will be interesting to see how these numbers continue to change.

Love, your local epidemiologist

Important notes:
-Connecticut and North Carolina were missing mortality data and therefore excluded. I did not forget about them 

Data Sources:
-Data from 2019 was pulled from the CDC National Center for Health Statistics
-Graphs were created by me using data from the Weinberger lab for ease of distribution on social media.
-Here is the EuroMono group website:
-JAMA article:…/jamainternalm…/fullarticle/2767980
-The Weinberger lab:

AZ California Deaths FL GA Texas update

Case Fatality Rates

On July 7, I posted five reasons as to why CFR may be decreasing while cases are increasing. One of which was lag time.

In other words, deaths today aren’t indicative of spread today, but rather a reflection of case severity 20-30 days ago. It’s been 27ish days since exponential growth started across several states. We should start seeing an uptick in CFR if this hypothesis is correct.

And we are. This is obvious in TX and CA. Doesn’t look like there is change in FL, AZ, or GA (yet). Given the spread among the younger population, this lag time may be even more than 30 days.

It’s still too early to see the impact of this recent uptick in TX and CA on cumulative CFR (Figure 2).

So, what’s causing this increase in TX and CA? Either we have reached hospital capacity (which we haven’t). OR COVID19’s reach is so wide it’s starting to reach vulnerable populations. OR we are increasingly testing those that are more sick (indicative of a high test positive rate). It’s likely a combination of the latter two. CFR is a difficult measurement because it’s highly dependent on the number cases we catch. For example, if we are only testing high risk populations (like nursing homes), the CFR will be high. It’s typically missing asymptomatic or mild cases that just never get tested.

Because of this, public health decision makers are starting to use Infection Fatality Rate (IFR). IFR estimates the fatality rate among those infected (detected AND undetected cases).

In the US, the CDC’s best IFR estimate is 0.65%. So, on average, 6.5 people of 1000 infected will die of COVID19. A recent publication pooled global IFR; IFR ranged between 0.53% and 0.82%. IFR is a more direct measure of disease severity, although highly dependent on place.

Understanding the true fatality rate has implications for public health planning. Unfortunately, if you thought the CFR was “low”, you are really not going to worry about 0.65% IFR. Given the reach of COVID19, this is still very much a leading cause of death in the US. The morbidity of COVID19 should still be of great concern too.

Love, your local epidemiologist

Data source: COVID19 tracking project. Graphs by yours truly.
Pooled IRC:…/10.1101/2020.05.03.20089854v4
CDC report:…/…/hcp/planning-scenarios.html

Deaths Texas update

Case Fatality Rates

A hot topic. Specifically, because Case Fatality Rates (CFR) are decreasing, while cases are increasing. Which seems counter intuitive.

I’ve seen several hypotheses floating around explaining this phenomenon:
1. Lag time. Scientists have estimated that CFR lags 14-30 days due to disease manifestation and spread to vulnerable populations. So, for example, the CFR on July 7 isn’t reflective of the cases on July 7 but INSTEAD is reflective of the spread on June 5. Exponential growth started in Texas on June 15ish and 30 days have yet to pass. We may see this start to increase here soon. This delay has a big impact on CFR (see how different the 14-day lag compared to the 30-day lag is in the Figures).

2. Younger populations. Younger people are contracting and spreading COVID19, which mean younger people are going to the hospital more, but once at the hospital they are dying at a slower rate. Keep in mind that “not dying” is not the same thing as “fully recovering”. The younger population also might contribute to #1 (lag time). If there is a lot of spread among younger populations, then the time for COVID19 to get to the more vulnerable will take longer.

3. Treatment. Compared to the beginning of the pandemic, doctors have figured out better ways to treat COVID19. For example, dexamethasone cuts risk of death on a ventilator by 1/3 and those on oxygen by 1/5. Also, plasma transfusions have helped patients recover.

4. Mutation. Since the pandemic started there have been 33 COVID19 mutations. Scientists recently published that the latest strain MAY be more contagious. Could it also be less deadly? There still needs to be a LOT more studies on this, but an important question.

5. Hospital capacity. Texas hospitals have held their own. BUT if cases don’t stop rising, hospitals in some areas will run out of surge units, staff, supplies, etc. If this happens, CFR will increase again regardless of #1-4.

CFR decreasing is fantastic. It’s likely to a combination of the 5 above. However, I’m skeptical that it will stay this way. Unfortunately, only time will tell the bigger picture.

It’s important to note that even if CFR is decreasing, 2% is incredibly devastating. In Texas, COVID19 still is the 7th leading cause of death and 3rd leading cause of death in the US. Compared to other leading causes of death (heart disease, cancer), it’s only been around for FOUR MONTHS and is contagious.

Love, your local epidemiologist

Data source: DSHS. Graphs by yours truly.
Mutation study:…/10.1…/2020.06.12.148726v1.full.pdf
Dexamethasone study:…/10.1101/2020.06.22.20137273v1

Deaths Leading Cause of Death Texas update

Leading Cause of Death

I’ve seen lots of comments about the low COVID19 fatality rate in Texas.

Although 2% seems low, it’s relatively high compared to other causes of death in Texas. If we stay on our trajectory, COVID19 will be the 7th leading cause of death in Texas this year. COVID19 would average 20 deaths per day compared to, for example, stroke that caused 30 deaths per day in Texas in 2019. This is impressive considering stroke and diabetes and heart disease take YEARS to develop. COVID19 has only been around for 109 days and is contagious.

There are a few things that can/will change Texas’ COVID19 rank this year:

1) Overflow hospitals. Many Texas metroplex hospitals are opening up “surge” units and scrambling for qualified staff. Less people will survive due to less quality of care. This will INCREASE COVID19’s rank.

2) Spread among populations. As I said in my previous post, CFR is decreasing in TX because of spread among younger populations. However, the far reach of this disease may start impacting the circles of younger populations, including those at high risk and kids. This will INCREASE or DECREASE COVID19’s rank. We just don’t know yet.

3) Other deaths will shift. The figure is comparing COVID19 deaths to 2019 deaths rates for other diseases, which isn’t necessarily an accurate comparison. A global pandemic WILL have impact on how we die other ways too. For example, some of the chronic lung disease deaths that WOULD have been in this category are now in the COVID19 category. On the other hand, unintentional injury (like motor vehicle crashes) have decreased. These shifts will INCREASE or DECREASE COVID19’s ranking, we won’t know HOW until the end of the year.

4) Counting “deaths”. This is very complicated. In short, the number of “COVID19 deaths” could (and are likely) off. The only way to “truly” know is through examination of death certificates. This is done each year by national organizations, but certainly not at a pace that the public wants. The problem is that states and hospitals and counties count covid19 deaths differently. Also, in the beginning of the pandemic, no one was getting tested for COVID19 but still dying, so we think the deaths could be underestimated. Again, this will INCREASE or DECREASE COVID19’s ranking, we won’t know HOW until more time passes.

If we stay on our trajectory in the US, COVID19 will be the 3rd leading cause of death. In other words, COVID19 would average 1118 deaths per day compared to heart disease (which killed 1,774 people per day in 2019) and cancer (which killed 1,641 people per day in 2019).

Translation: As for now, Texas’ COVID19 rank in terms of leading causes of death IS lower than the US as a whole. We should keep it this way by taking the Texas epidemic seriously.

Love, your local epidemiologist

Data source: 2019 data is from the National Center for Health Statistics at the CDC. COVID19 data is from DSHS (Texas) and Johns Hopkins (US). Graphs/analysis by yours truly.

Deaths Predictions Texas update

Texas Today

And this is my best shot at giving your the most comprehensive picture of COVID19 in Texas today.

Cases: While Harris County is getting a lot of attention due to an increase in cases (and we should be worried about this), it looks like after adjusting for population, Dallas is in even worse shape. Another worrisome county is Bexar (San Antonio) given their recent steep incline. In order compare the severity of COVID across Texas counties, we NEED to adjust for population. I noticed this is missing in much of the news.

Testing: TPR has always been a problem in Texas. Our goal is 5%, and we are over 10% in most of Texas. We need to allocate resources to increase testing. Like NOW. We are not testing enough to get ahead of the epidemic.

Mortality: CFR looks steady in Texas. In fact, it’s decreasing. This is driven by COVID spread among younger Texans, who have the ability to fight off the disease compared to older populations. HOWEVER, this JUST measures mortality. This does NOT account for lifelong complications we are seeing among the young patients that survive. Which is a whole other story.

Future cases: I included classic Epi curves for each county (with 7-day average- blue line). I also included our June 13 projections calculated based on R(t), which takes into account testing, temperature/humidity, social distancing, and population density. Projections in Bexar, Travis, and Harris county are grossly UNDERestimated. We NEED to especially improve social distancing here. Projections for Dallas and Tarrant look accurate so far (but still need to improve social distancing so this projection goes flat).

Translation: We need to increase testing. We need to improve our social distancing. We need to be worried for all of Texas, but particularly the rate in Dallas and Bexar right now.

Love, your local epidemiologist

Data: DSHS. Graphs and analyses by yours truly.