Special announcement

After much thought and consideration, I have decided for personal reasons to stop posting updates to this blog. For those of you who have found my work here to be informative, helpful, comforting or disquieting, please know that I greatly appreciate the kindness of your attention, and your feedback. It was my honor to contribute in what ever small way I could to the events of what was a truly extraordinary year.

Please stay safe and feel free to send me your thoughts and experiences.

With regard,
Thomas A. Glass, Ph.D.
Disease detective


Wednesday COVID-19 Briefing

Top news, reports and insights for today:

  1. Curated headline summaries for Mid-week:
  • Top story: U.S. tops 21 million COVID-19 cases with record hospitalizations as states ramp up vaccinations (Reuters).
  • Coronavirus vaccine scams are on the rise in Europe and the U.S.; officials are warning the public against fraudsters who swindle customers out of money and personal data with promises of fake vaccines (Daily Mail).
  • COVID-19 tests are in demand and testing has dropped. You can now get an FDA-approved at-home saliva test at Amazon (CNN Business).
  • UC San Diego installs Covid-19 testing vending machines on campus (CNN).
  • Some 86% of people with mild cases of COVID-19 lose their sense of smell and taste but recover it within six months, according to a new study of over 2,500 patients from 18 European hospitals. Loss of smell was more common in younger patients (CNN).
  1. U.S. testing falls to mid-November levels as cases and deaths continue to surge
      Newly reported cases grew 21 percent this week compared to the holiday week adding 1.4 million cases (See Figure A). That itself is not surprising given what we had anticipated. The more concerning signal comes from the state growth-factor chart (Figure B) showing cases rising in every state and the District of Columbia. The largest increases were in Alaska (+76% increase this week compared to the week previous), Arizona (+51%), Hawaii (+55%), Minnesota (+43%), Florida (+44%), Kentucky (+42%), Louisiana (+64%), North Carolina (+46%), and Vermont (+41%). The daily death data shows a similar pattern (Figure C) as Tuesday’s tally was especially disheartening. The weekend slow-down came to an abrupt end as deaths spiked to over 3,400 on Tuesday, the second highest daily total of the outbreak.
     The thing that makes the current surge of particular importance is that daily U.S. testing has been sagging in recent weeks (Figure D). Over 1.6 million tests were recorded today on COVID Tracking Project, which sounds like a lot, but little noticed has been a marked slowing of the progress that had been made earlier in the Fall. Today’s testing totals put us on par with the 7-day average back in Mid-November, but it’s 300,000 tests fewer than the peak achieved in Mid-December. In other words, the fall off in testing masks the severity of the current surge. Another way to put this is that if we were doing 20% more testing like we were in December, the daily case totals would be breaking records every day. Over the next week or two, the effects of Christmas travel will start showing up in the data.
    Bottom line: January is going to be very rough.
Figure A
Figure B
Figure C
Figure D: COVID Tracking Project screen grab January 5, 2021
  1. How does the U.S. pandemic response stack up to other countries? A novel composite national pandemic response score
      On the weekend briefing, I showed you the ranking of countries in terms of COVID-19 deaths per million (reminder, the U.S. was substantially below average). I’ve been working more on this question this week. That’s partly because we can’t trust the daily numbers yet due to the Christmas reporting slow-down and also because the new year seems like a good time to look big picture. Epidemiologists always struggle with proper ways to compare the performance of people, doctors, hospitals, states or countries when those units differ substantially. One standard approach is to compute something called Z-scores. A Z-score is a statistical tool that expresses how far a unit (here country) is from the average relative to a particular distribution of performances. The units are standard deviations, or the average amount a given unit differs from the mean of that parameter. To compute a composite score that summarizes a country’s performance across different outcomes, I used data from WORLDOMETER as of January 1, 2021 for three different outcomes: 1) COVID-19 cases per million, 2) COVID-19 deaths per million, and 3) tests per million. The direction of each is set so that higher numbers reflect better relative performance. As was true over the weekend, I picked all countries meeting three criteria:
    a) at least 5 million population (to exclude smaller nations that lack an essential health infrastructure);
    b) at least 100,000 total cases (to exclude a small set of countries that are probably pretending not to be impacted by the pandemic);
    c) at least 100,000 tests per million people (to exclude countries whose case and death estimates can’t be trusted because of inadequate testing).
     Among the 222 global nations impacted by COVID-19, that results in a subset of 45 countries that should, in theory, be comparable in terms of their pandemic response. The resulting composite national pandemic response score (NPRS) is created by adding three Z-scores, each of which is based on the entire distribution of those 45 countries (See Figure E). This is one new way to use science and data to better understand how this country stacks up on the world stage. The top 3 scores belong to the United Arab Emirates (UAE), Denmark and Malaysia. Because of the way Z-scores work, countries around zero are (by definition) average, or middle-of-the-pack. Countries in this region include Palestine, Jordan, Oman, South Africa, Ukraine, Slovakia and Serbia. Most are middle income nations that have outperformed a number of high-income countries. The bottom three performing nations include the United States, Belgium and the Czech Republic. The U.S. has the third worst performance among these 45 nations, scoring well-below average on all three components. For those who believe the U.S. should not be compared to Kazakhstan (for example), I also show you just the OECD countries that meet the selection criteria (N=23). The U.S. also was third worst among that set of countries, having been outperformed in 2020 by Russia, Canada, Greece, Germany, Turkey, Israel, Portugal, the U.K., Austria, Chile, Poland, France, Columbia, Sweden, Netherlands, Spain, Hungary, Italy, and Switzerland.
     Bottom Line: After summarizing relative performance of comparable nations based on COVID-19 cases, deaths and testing, the U.S. response in 2020 was the third worst.
Figure D
  1. Quirky Qorner: COVID-19 declares video games the entertainment champ in 2020
     My kids will groan to read this but I don’t own a video game console and I don’t play video games (except for solitaire). Thankfully, they don’t read my blog so I am safe. If they did, they might rub it in my face that as movie theaters and music venues have dried up, Americans have turned to video games as their favorite entertainment option. Geez. An article from the Verge today reports on Nielsen 2020 year end review showing that 55 percent of Americans turned to gaming. Interestingly, 1 in 4 used video games as a way to connect to others. I admit I don’t even know what that means.

Weekend Covid-19 Briefing

Top news, reports and insights for today:

  1. Curated headline summaries for Saturday/Sunday:
  • Top Story: The coronavirus variant first identified in the U.K. has been confirmed in the U.S. with cases reported in Colorado, California and now Florida. Dr. Fauci says he is “not surprised”, and I agree. The new variant (named B.1.1.7.) appears to be more contagious although this has not been confirmed. Whether it causes more severe disease remains unknown (CBS News).
  • December was deadliest month of the pandemic, with over 77,000 deaths compared with 58,000 in April as the U.S. passes 20 million total cases (NBC News).
  • L.A. County hospitals turn away ambulances, put patients in gift shop: ‘I’ve never seen anything like this’ (Los Angeles Times).
  • Concern increases over slow rollout of COVID-19 vaccines. President Trump had promised 100 million Americans would be vaccinated by the end of the year. That number was lowered to 20 million shortly after approval. As of last Thursday, less than 3 million have been vaccinated (Slate).
  • At least two California hospitals have diverted COVID-19 vaccines from front-line health workers to family members. This signals that drug diversion is most likely a common and unanticipated part of why vaccine rollout has been slow (Daily Mail).
  • Up to half of health care workers are refusing the vaccine. While not true everywhere, 60 percent of nursing home staff in Ohio, between 20-40 percent of health workers in Los Angeles and about half of health workers in Riverside County California are saying no to a vaccine (Daily Mail).
  • In just a few weeks, Israel has managed to vaccinate more than 10 percent of its population becoming the world leader. This shows what can be done with the right public health infrastructure and political leadership (Axios, See Figure A).
Figure A: https://ourworldindata.org/grapher/covid-vaccination-doses-per-capita?tab=chart&stackMode=absolute&time=latest&country=CAN~CHN~DNK~EST~FIN~FRA~DEU~ISL~ISR~ITA~MEX~PRT~RUS~GBR~USA~OWID_WRL&region=World Note this figure is updated by me compared to what appears in the paper.
  1. New record high daily totals set for cases yesterday and deaths on Wednesday. The Christmas slow-down is over.
     The Christmas reporting slow-down is over; daily cases spiked to their highest level yet yesterday soaring to over 250,000 (See Figure B). Weekly cases rose by 1.3 million, rising by 8.3 percent compared to the week before. Cases are rising broadly in 34 states (Figure C), led by surges in Connecticut (+74%), Rhode Island (+38%), Texas (+36%) and Washington DC (+32%). Cases also rose substantially in Kansas but the numbers are unreliable because that state is having data problems and consistently fails to report its numbers 3-4 days a week. The South and Northeast are again the regions to watch with rising states outnumbering falling states 10 to 2 in both.
     After 5 consecutive days of fewer than 1,500 daily deaths during the Christmas break, daily fatalities spiked to over 3,250 a day on Tuesday, then leaping again to a new record high of 3,572 on Wednesday (Figure D). This sent the 7-day moving average line back in an upward climb signaling the end of the Christmas reporting slow-down. Overall, deaths rose 19 percent last week compared to the previous week. The state pattern was similar to cases with weekly deaths headed higher in 31 states and in all four regions (Figure E). The largest jumps in weekly deaths were in Wyoming (+195%), Virginia (+169%) and Kansas (+124%). California saw deaths jump 53 percent recording more than 2,300 last week with a record 585 on New Years day. Similar to cases, deaths also increased strongly in Connecticut (+47%), and Texas (+21%).
      Bottom line: I had hoped the data were telling us that the virus may have finally peaked, while suspecting that the numbers were being driven by a slow-down in reporting rather than the virus. That now looks highly probable. Coronavirus keeps coming. Those currently exposed, infected, sick and dying won’t be helped by the vaccine.
Figure B
Figure C
Figure D
Figure E
  1. 2020 is finally over. How did the U.S. do compared to other countries?
     Way back on November 20, I read an article in Mother Jones by the political blogger Kevin Drum. His piece was about how the U.S. stacked up to other countries in terms of COVID-19 mortality rates. He showed a graph plotting cumulative deaths per 1 million people across 15 counties (reproduced below as Figure F). He described the U.S. rates as about the middle of the pack. Sure enough, it looks like the U.S. was in the middle of the lines on the graph. I’ve been pondering this graph and his summary ever since. As 2020 is finally over, I thought I would take this opportunity to look back over the first 10 months of the U.S. response to the coronavirus pandemic and assess where we are and whether Mr. Drum is right that we are about “average” in our handling of the epidemic.
      In the field of epidemiology, we constantly run into what I call the “denominator problem”. The claim that the U.S. is middle of the pack depends on getting two things right: 1) the pack and 2) the middle. Both are quite problematic in this graph, making the overall conclusion shaky at best. I thought I would try to do better. Below is my attempt to address this question (Figure G). The first puzzle is to get the pack right. That means avoiding the tendency to select (or exclude) countries that don’t properly describe the whole pack (aka population of countries we want to compare). The Drum figure includes just 15 countries. To get the pack right, I chose all countries in the WORLDOMETER set of 222 impacted nations meeting 3 criteria: 1) greater than 5 million total population (to exclude smaller countries with non-comparable health sectors), b) countries with more than total 100,000 cases (to rule out some countries that don’t have the capacity to accurately track the epidemic), and c) countries that have done at least 100,000 total tests per million residents (to exclude countries that really have no way of knowing how much coronavirus they even have). Email me if you want details on how I chose these criteria. Figure G shows current cumulative COVID-19 deaths per million people ranked from worst to best for all 45 countries that met these criteria. The range of COVID-19 death rates is nothing short of staggering, from a high of 1,686 per million in Belgium to a low of 15 per million in Malaysia. That represents a 100-fold difference! We don’t see that wide a range for other diseases. The United Arab Emirates (UAE) is a particularly interesting case. With a population of just under 10 million, they have run 2 tests for every citizen for a testing rate of 2.1 million per 1 million people, just about tops in the world. Yet, they have only about 211,000 total cases and only 674 cumulative deaths. Given that level of testing, it is virtually impossible that their success is due to severe undercounting. More likely, aggressive testing has meant they have managed to control secondary transmission to achieve such a low death rate. The U.S. has the 8th worst COVID-19 death rate at 1,080 per million. Of these 45 countries, 37 have lower death rates than the world’s sole remaining “superpower”. That doesn’t sound like middle of that pack. The countries worse than us include less developed nations (Peru, Bulgaria and the Czech Republic) as well as countries hit early and hard when the case-fatality rates were much higher than they are now (Spain and Italy). I’m not sure anyone knows what exactly is going on in Belgium (topic of a future entry). The UK is also a strange case in many ways. The U.S. looks “average” in the Drum graph because most of the countries with low mortality rates aren’t included in the pack. That’s the denominator problem in a nutshell.
    Now let’s tackle the question of the middle. For that we need a number (not just our eyeballs) to describe the central tendency of this distribution. There are two options: the mean and median. Both are averages but any good disease detective knows to be cautious, especially when there is a big outlier value like Belgium in this case. I won’t bore you with more technical issues, but the best number to use here is the median, which for all 45 countries is 473 per 1 million (purple dashed line). The U.S. now looks very far from middle of the pack indeed, with a cumulative mortality that is more than double the “average”. Let’s say you don’t think some of the countries on this list should be compared to us. Maybe we can’t trust the death tally in countries like Jordan, Iraq, Morocco or Belarus. Ok fine, lets make another cut, looking this time at just OECD countries that met the first three criteria (blue bars). The U.S. has the 6th worst OECD mortality rate. That is more than double the death rate of Canada, Germany and Slovakia. Look at Israel, which vaccinated more than 10% of its populace in about 2 weeks and has a COVID-19 death rate that is a third of ours. Among OECD countries, the median jumps to 856 (blue dashed arrow), still considerably less than the U.S. rate.
    Bottom Line: I respectfully disagree with the conclusions reached by Mr. Drum in Mother Jones. It all depends on getting the right middle of the right pack. Now consider this: if the U.S. response to the COVID-19 pandemic had been average, fewer than half the total cumulative deaths (over 350,000 from lab-confirmed infections) would have occurred and nearly 200,000 dead Americans would now be alive.
Figure F From November 20 issue of Mother Jones (https://www.motherjones.com/kevin-drum/2020/11/heres-how-the-us-is-doing-on-covid-19/)
Figure G
  1. Quirky Qorner: French snails farmers lament sluggish year as COVID crisis dents sales
      Yep, I copied the headline right from the article in yesterday’s Guardian (including the word ‘sluggish’). In France, snail farmers (aka heliciculteurs) depend on Christmas for up to 70 percent of their annual business, thanks to the fondness the French possess for the curly beasts as a holiday hors d’oeuvre. Seems that with COVID-19, business has slowed to, um, a snail’s pace.