Friday COVID-19 Briefing

---

Top news, reports and insights for today:

1. Daily headline summaries for Friday:

• Data shows signs the epicenter of the coronavirus epidemic may be shifting to the Midwest on Thursday while Sunbelt states saw new infections begin to decline while cases rise in Ohio, Kentucky, Tennessee, Missouri, Kansas and Nebraska (Reuters)
• The Trump administration abruptly required hospitals to stop reporting COVID-19 data to CDC and use a new reporting system set up by a contractor. That system remains plagued with delays and inaccuracies. Hearings and investigations now underway (NPR)
• Scammers are making millions selling bogus dietary supplements to treat and prevent COVID-19. NPR found over 100 supplements listed for sale on Amazon that make unsubstantiated and potentially illegal claims that they fight viruses. These include products sold by a company recently sued by the Department of Justice over fraud allegations related to COVID-19 (NPR)
• African-Americans are known to suffer disproportionately from COVID-19 in many ways. A recent study in 5 hospitals in Baltimore/Washington from March to May shows that Latinos had test positivity rates 5-times higher than whites and double that of blacks. The Latino community needs greater attention (JAMA)

2. New daily cases remain flat but “bouncy”, 3 states pass New York for cumulative rates of COVID-19; new case growth pops up in Northeast and Mid-west
    On Tuesday I noted that new daily cases seemed to have flattened. That trend has largely continued this week, although numbers have fluctuated substantially (See Figure A). The U.S. added 442,658 new cases last week, a rise of 11% in total. Over the last 3 days, new record high daily cases were set in Hawaii (124), Missouri (2,084) and Mississippi (1,775). Figure B shows the 1-week growth factors for each state indicating the ratio of new cases this week to the week before. They show something important and unexpected. Our attention has been on the Sunbelt states where the epidemic has been surging. For the first time in weeks, we see bigger hikes in new cases in the Northeast and Mid-west while cases declined in 6 Southern states. Connecticut added a startling 1,438 cases, tripling the previous week’s total. New Jersey also saw new cases more than double, prompting Governor Murphy to update quarantine advisories issued for travelers from 3 new states, Puerto Rico and Washington DC. CBS News in New York is reporting that New Jersey hospitals are bracing for a surge of new cases. New cases also rose by 20% or more in Massachussetts (+22%), Maryland (+23%), New Hampshire (+26%), and Rhode Island (+84). In the Midwest, Missouri (+53%) and Oklahoma (+72%) saw the largest spikes. Michigan (+20%), Minnesota (+14%), Nebraska (+16%) and South Dakota (+21%) all saw new cases rising. In the West, big increases were seen in small states as new cases surged in Alaska (+37%), Hawaii (+280%) and Montana (+33%).
    Figure C updates us on the overall rates of COVID-19 cases at least for that subset captured in our testing. A couple of months ago, it looked like no state could possibly catch up to New York. Now three states have a higher overall rate of cases per 100,000: Louisiana now has the highest (2,463), followed by Arizona (2,347) and Florida (2,148). As a region, the South now nearly matches the Northeast at 1,560 per 100,000 (compared to 1,582). In the South, only Kentucky and West Virginia have kept rates below 1,000 per 100,000, a feat also matched by 7 of 13 Western states and 10 of 13 Midwest states.
    The bottom line: While the flattening of new cases is good news, the whack-a-mole continues as falling new cases are offset by brush fires of transmission intensity breaking out in previous hot spots. Key factors appear to be lax social distancing associated with vacation travel and people staying indoors to avoid the heat.

Figure A

Figure B

Figure C

3. U.S. COVID-19 deaths pass 140,000; daily totals continue to rise
    On Thursday, 1,231 Americans were reported dead from the SARS-CoV-2 virus. This caps a four-day run of more than 1,000 daily deaths and continues a trend of rising mortality that started around the 4th of July (see Figure D). The U.S. reported over 7,500 deaths last week, a cumulative rise of 6%, which is double the 3% rise seen the week prior to July 26. In all major data tracking sites, the U.S. has now exceeded 140,000 deaths. The U.S. remains the nation with the largest death total with more than twice that of the next two highest countries (Brazil and the UK) combined. Figure E shows the states that rose the most last week compared to the week before. As always, states with fewer than 25 weekly deaths are shown with a patterned bar since the ratios of two small numbers tend to be unstable. As expected, the most consistent region is the South, where despite an apparent peak in cases, deaths continue to surge. Weekly deaths rose by 20% or more in Arkansas (+24%), Florida (+27%), Georgia (+21%), Kentucky (+38%), Mississippi (+37%), North Carolina (+28%), South Carolina (+27%), Texas (+183%) and Virginia (+175%). Deaths spiked particularly in Kansas (+456%) and Delaware (+500%) although the absolute numbers are smaller. Deaths were on the rise significantly in Ohio (+22%) and Oklahoma (+51%) but fell in Illinois, Michigan, Missouri, Minnesota and Wisconsin. In the West, deaths rose in California (+29%), Idaho (+63%), Oregon (+79% and Washington (+49%) while holding steady in Arizona for the first time in several weeks.
    The bottom line: Consistent with the lag between infections, testing, and mortality, the summer surge in cases is now resulting in a rise of COVID-19 deaths. Importantly, we are still significantly below the peak of 2500+ deaths a day seen in April in May despite vastly more cases. This is because of the doubling of testing. While we were capturing approximately 10% of the total number of actual infections occurring in the population in May, we have now roughly doubled the number of daily tests from 40,000 a day to around 80,000 (see Tuesday’s blog). This expansion of testing as lowered the TPR from 10% to 8% allowing us to capture closer to 20+% of true infections. All this suggests that the overall mortality rate is likely fairly stable. The reality is that we had far more cases of active infection occurring in April and May than we knew – like 10-times more. The best performing model by Youyang Gu and colleagues, suggests that new infections in the U.S. peaked around July 16 and that deaths are projected to peak at just over 1,100 a day around August 13. We will see.

Figure D

Figure E

4. Two timely and important studies released about the role of schools: reopening this fall carries unavoidable risks
    Here we are in the middle of the strangest summer of our lives. Now is the time people are thinking about the Fall and what it will mean for schools. Should they reopen? Should they stay closed? Is online training even worth it? Why not let the kids go back to school given that they don’t seem to be getting very sick. I want to make you aware of two important and timely studies that have come out that bear on this issue and should be kept in mind as we lurch toward the Fall.
    The first study was published July 29 in JAMA by Katherine Auger and colleagues from Cincinnati. They looked at whether there was a relationship between the timing of state-wide school closures and COVID-19 infection and death rates across all 50 states in the period from March 9 to May 7. The analysis they did was a bit complicated but their approach was very solid. They found that state-wide school closures were associated with a 62% reduction in weekly cases and a 58% lower weekly rate of deaths. This doesn’t prove causation, but it suggests that despite the fact that school-age children are not the ones at highest risk of sickness and death, shutting down schools may have a substantial impact on the overall risk in the community. Children can be conveyors of disease directly by transmitting to more at risk groups. In addition, through the patterns of interaction and mixing that school requires, transmission intensity can increase among all age groups because of the increase in overall interactions in and out of school.
    The second relevant study came out in Eurosurveillance on July 21 and was done by researchers affiliated with the Jerusalem District Health Office in Israel. (Note: I apologize if either of these links are behind a firewall. I can’t tell. Let me know if they are and I can post the article). This study looked at what happened when schools were reopened on May 17 after being closed across the country in Mid-March. Considerable planning went in to the reopening plan; daily health reports, hygiene procedures, facemarks, social distancing and minimal interactions were all required and put into place. Disease detectives were put on stand-bye to monitor the situation. Despite these efforts, ten days later, the first significant outbreak occurred in an Israeli High School. After an official outbreak declaration, mass testing and contact tracing was undertaken. Overall, 153 students and 25 staff members were confirmed to be COVID-19 positive, yielding an attack rate of 13% in students and 17% in staff. This outbreak appears to have started from 2 initial index cases not related to each other that sparked the larger outbreak.
    What do these studies mean? In thinking about what we should do as a nation, these studies suggest two tentative conclusions, recognizing that we are only looking at 2 imperfect studies when we would like to have 200. The first is that open schools can increase the risk for entire communities, not just the students in those schools. We have good evidence that school closures were a vital tool for controlling the spread of influenza in 1918-1919. Cities that reopened schools suffered a worse fate than those that closed schools early and kept them closed. But coronavirus is not influenza and we are tempted by the fact that children seem to be at lower risk. But the JAMA study tells us that what happens in schools, doesn’t stay in schools. Returning to classrooms invites greater chance of community-wide transmission, leading potentially to higher rates of hospitalization and death. The second study warns us that even with the right planning and the best available epidemic control measures in place, outbreaks can happen in school settings and happen fast. Testing, monitoring and social distancing do matter and they do help, but the school environment is to some extent an unavoidable petri dish for infectious disease transmission.

Advertisement

Daily COVID-19 Briefing: 3/31/20

Top news, reports and insights for today:

1. The White House predicts 200,000 US deaths even in the best-case scenario.
   The Washington Post and others are reporting on the latest warnings from the White House coronavirus coordinator suggesting that up to 200,000 deaths are projected in the US. This is described as a ‘best-case scenario’. These projections appear to be behind the President’s recent decision to extend social distancing guidance till the end of April. The US now reports 162,000+ cases and 2,958 deaths (with Washington State not reporting). Most experts believe the peak in US cases is still weeks away. On Monday, 517 new deaths were reported, a jump of 21% with notable rises in Maryland (+48%), Tennessee (+80%), Ohio (+34%), and Michigan (+39%).
   What does it mean? It is not clear where the White House projections are coming from. To forecast the number of US deaths, we need to know 2 critical pieces of information: the attack rate (what percentage of American’s will be infected over the next two years) and the case-fatality rate (the percentage of infections that will end in death). Right now, we don’t know those numbers with any accuracy. Our confidence is starting to improve however. I review the models and projections from the top experts on a regular basis. The trend in projections suggests that the number of deaths will unfortunately be considerably higher than the Whitehouse numbers. The estimates I trust the most right now suggest between 450,000 and 1,500,000 deaths. The President’s numbers indeed represent best-case expectations. It is still too early to have confidence in any single number. What is clear is that the impact of the COVID-19 pandemic will be substantial and unprecidented in our lifetimes.
   
2. So what the heck is the mortality rate for COVID-19?
   Last night I looked carefully at mortality rates across 2 dozen countries where the pandemic is established. If testing was far enough along that we were capturing 90% or more of all cases, and if we were able to identify everyone who dies of COVID-19, we would expect the case-fatality rate (CFR) to be roughly similar among at least developed nations with good health care. On March 3, the WHO announced that their estimate of the case-fatality rate is 3.4%. Most experts believe this estimate is too high. So, until we know what the CFR really is, we can estimate by computing a simpler number: the crude fatality ratio (deaths divided by confirmed cases). I made the graph below to summarize what we now see. It’s a very confusing picture. The good news is that the average of these mortality rates is 2.7%, which is not that far from the WHO estimate. The surprise here is that the estimates are so very different and inconsistent. They range from a stratospheric 11.4% in Italy, to Iceland at only 0.2%. The standard deviation (for you nerds out there) is a whopping 5%.
   What does this mean? Basically this tells us that we don’t yet know the true CFR. Looking at the extremes, the 11% death ratio in Italy is most likely due to the fact that only severe cases are being tested due to stress on the health sector. When a health system is overwhelmed keeping people alive, surveillance goes to the back burner. On the other hand, Iceland’s number (0.2%) is probably the estimate that is closest to the real CFR. That’s because Iceland has tested more of their citizens than any other country by far (4,800 per 100,000). Compare that to the US where only 400 per 100,000 have been tested. Because they are testing virtually everyone in Iceland, they capture the highest proportion of all cases, making their fatality ratio most likely to be accurate. Based on all of this, my own estimate (shared by others) is that the CFR will end up being somewhere between 0.5% and 1.5% depending on where you are.

3. Workers strike at Amazon, Instacart and Whole Foods over safety concerns
   As Americans shift their shopping and purchasing online, the web-based retail giants are struggling to keep up with the demand. As reported today by USA TODAY, stay-at-home measures have created a frenzied work pace at shopping sites. This week, workers have said they have had enough of long hours and unsafe conditions. Amazon warehouse employees on Staten Island New York walked out at lunch yesterday over concerns about workplace safety. The grocery delivery service Instacart is facing a nationwide strike over similar concerns. Amazon workers across the country are warning of a nationwide “sick out” over conditions and pay. Some workers have been fired for making waves. Warehouse workers argue that they are not being given adequate protective equipment to protect themselves and their families. Sound familiar? Business Insider reports that Amazon prime deliveres are delayed for up to a month as the company struggles with the crush of online orders and problems keeping their pipeline adequately staffed.
   
4. Experts question China’s numbers, asymptomatic cases not counted
   The world has looked to China’s experience with the COVID-19 outbreak for benchmarks and insight. After spending the first weeks of the pandemic on center-stage, the Chinese have claimed success in controlling the epidemic through drastic outbreak control measures. The Wall Street Journal (article not freely available) reports on new concerns about the reliability of Chinese numbers. The graph below shows the daily tally of new cases from China since the epidemic started in January. It shows that new cases slowed to a trickle by February 14 and have remained very low ever since. Only 76 new cases were recorded on March 29. The WSJ reports that more than 1,500 infections that showed no symptoms were not included in their count. At very least, caution is warranted in reading the data coming from China.

Image captured from Johns Hopkins CSSE website on March 31, 2020. It shows the epidemic curve for confirmed COVID-19 cases since late January