CDC faces growing criticism for new testing guidance. Revised recommendations suggest those who have been exposed but show no symptoms may not need to be tested. In my view, new guidance makes no sense and smacks of political influence in CDC policy. (CBS News)
A study in India looked at what happened when community health workers started wearing face shields. Before face shields, the rate of COVID-19 infection was 19%. After face shields, no infections were observed. The case for face shields, especially for health care workers as well as barbers, dentists, and others who work in close proximity, is getting stronger (JAMA)
A fishing vessel left Seattle in May. When it returned 18 days later, 104 of 122 people on board had been stricken with COVID-19. Why does this matter? A new study used this as a natural experiment yielding the strongest evidence that neutralizing antibodies protected the three people who had them against the SARS-CoV-2 virus (American Society for Microbiology)
All eyes are on UNC-Chapel Hill as fall approaches, where 26 new cases were reported among students, bringing the total to over 1,000 since the pandemic began. Campus outbreaks there and elsewhere have thrown a wet blanket over plans to reopen campuses across the nation (The News & Observer)
Britain and the U.S. have handled the COVID-19 pandemic worse than other advanced nations, new poll finds (Daily Mail, See Figure A)
Decline in U.S. daily cases hits a road bump, deaths lower languidly Disease detectives take note: we are accustomed to seeing a jagged and zig-zagged appearance in the daily data. There are several overlapping factors that contribute to that appearance. One is the ebbs and flows of the reporting system (offices close down on the weekend, rush to catch up on Tuesday etc.). Some states, most notably Kansas, only report its new cases every 2nd or 3rd day (what’s up with that Kansas?). The testing labs impose their own noise as supplies run out, machines break down and bottle necks appear and are cleared. Then, of course, there is the natural variation that comes from the epidemic itself. For all these reasons, we learn to squint at the data, trying to see bigger patterns and not be fooled by all the noise. Like so much of science, ours is a constant battle to separate signal from noise. Wheat from chaff. On Wednesday, I was seeing a robust signal indicating a decline in daily cases after what looked like a peak in the U.S. around July 18. The last three days of data suggest a slow down in that decline (Figure B). The 7-day average (red line) may be flattening. The data looks more zig-zaggie (to use a technical term) over the last couple of weeks than before. All this makes me want to look at the testing data to see if that helps explain things (see point 3). The story with deaths (Figure C) is somewhat brighter. The 7-day average is still dribbling down. Still, there were 1,260 new COVID-19 fatalities reported on Thursday. Week-over-week, deaths are on the rise in Colorado (+33%), Iowa (+19%), Arkansas (+66%), Georgia (+34%), Mississippi (+17%), Virginia (+35%), and Massachussetts (+33%). Bottom line: As was the case after the previous peak in April and May, the epidemic curves in this pandemic look atypical. We know what shape those curves should have. Like so much of what we see in this disease, SARS-CoV-2 remains inscrutable. The decline in cases may have hit a bump in the road, while deaths droop further. I suspect both are a function of what is happening in testing (keep reading).
Testing Update: Is it live or Memorex? U.S. daily testing mirrors cases, passes peak I last showed you Figure D on August 11. At that time, the pattern wasn’t clear. As August comes to a close, we see more clearly that testing, like daily cases, has peaked and is on the decline. Hmmmmmm! On August 10, COVID Tracking Project says we peaked at over 900,000 daily tests. In recent days, we are averaging just over 700,000 a day. That decline is proportional to the drop we have seen in cases. As always, if testing is declining, but the TPR is below 5%, we are less concerned. That’s because when the test positivity rate is 5% or lower, it means we are casting a wide enough net to reveal the underlying transmission intensity. Here the orange line has been falling gradually, but still remains significantly above 5%. While the decline from 8.5 to 7.8 is welcome, it does not suggest our testing is efficient enough to warrant a sizable drop in the number of tests we are running. At the state level, the variation in testing rates continues to be staggeringly high (Figure E). As I have said before, this is what happens when there is no national policy or leadership. I don’t have the data but I very much doubt that state/provincial variation in testing rates is nearly this high in other nations. Kudos to little Rhode Island! They are kicking butt, having tested 477 of 1,000, nearly half of the state’s population. I was surprised to see Pennsylvania at the bottom, testing only 13 of 100. Louisiana and Tennessee are the two states in the South that are testing near the top of the rankings, six other southern states are below the U.S. median. The bottom line: If you are following the news, you know the White House and the CDC are under severe criticism for the recent release of guidelines that appear to throw a wet blanket on testing. Is that guidance motivated by science and by a desire to improve America’s pandemic response, or is it motivated by politics and a desire to make the numbers look more favorable to the administration as the election inches closer? You, dear reader decide for yourself. I bet you can guess what my answer is.
Quirky Qorner: Why isn’t contact tracing working well? Look to strip clubs for an answer The Toronto Star reports that public health officials this week said are talking about conduct contact tracing in a prominent strip club in that city. Surprisingly, they found that two-thirds of visitors provided false or incomplete contact information when approached. While provincial infection rates have dropped in Ontario, the stage was set for contact tracing to help keep the outbreak at bay. Contact tracing is only as good as the information people provide about their social network interactions and patterns. While strip club patrons may seem like a special case, trends across the nations point to a wide-spread reluctance on the part of citizens to trust public health authorities. Few users are downloading and using phone-based tracking apps. As Figure A shows, Canadians generally like how their government has handled coronavirus, so we won’t be surprised to learn that American’s are even more wary about giving authorities personal information on their social connections.
Ditch the valve masks; they are selfish. The whole point of wearing a mask is to protect others from being infected by you. If your mask has a valve, it doesn’t do that. That valve just spews your respiratory secretions straight out potentially infecting others if you are sick and don’t know it (ABC News)
COVID-19 deaths in Los Angeles County pass 1,500 with 1,800 currently hospitalized as trails, golf courses and some businesses reopen (Los Angeles Times)
U.S. FDA has granted emergency use authorization to a more rapid kind of COVID-19 test called an antigen test. It’s basically similar to the rapid flu test done in doctor’s offices. They give quick results but are less accurate than the PCR test we all love to hate. In my opinion, this is a bad move because these tests have a higher false-negative rate than the current test (which already has an unacceptably high false-negative rate). Use of this test will mean more infections and deaths because people who actually have COVID-19 will falsely believe they don’t and are likely to behave less carefully (The Verge)
U.S. deaths fell yesterday, but grew by 21% last week, surpassing 70,000 On Friday, an additional 1,695 Americans lost their lives to COVID-19, a rise of 2% and a third straight day of declines. Arizona had the highest increase in deaths last week, rising 62%, roughly doubling every 10 days. As I reported yesterday, Arizona has tested fewer of its residents than any other state. New record highs were set in Arizona (67) and Washington DC (19) yesterday. In addition to Arizona out west, there are 4 midwestern states reporting more than 40% growth in deaths last week (Iowa, Minnesota, North Dakota and South Dakota), along with 1 southern state (Mississippi) and 3 northeastern states (New Hampshire, Pennsylvania and Rhode Island). Overall, there were 4,369 COVID-19 deaths last week, an increase of 21% in the cumulative total. Several midwestern states that had been slowing have surged in the last 3 days including Illinois and Ohio. Pennsylvania has had a very rough week, adding more than 1,000 deaths in the last 4 days. What this means: Deaths have plateaued in the U.S. over the recent few weeks averaging 1,780 a day. That tells us the epidemic remains in a steady state of linear growth. This means the epidemic is not waning, it is not ending soon and the worst is not yet behind us.
Global cases pass 4 million, how epidemiologists see the big picture Most major data sites suggest that more than 4 million global cases have occurred. The graphic below is a screen capture of the WORLDOMETER dashboard at this moment. This is a useful big picture summary of the pandemic. How do epidemiologists look at this information?
First, what is the ratio of active to closed cases. Here, it’s considerably greater than 1, indicating the global pandemic is still growing significantly. There are almost half a million more estimated active cases than closed cases now.
Next, we look the breakdown of cases by severity. In this case, we can see 3 categories: mild, serious/critical and death. The vast majority of active cases are mild (98%), a proportion that has been rising around the globe as more nations are testing people other than just those who are sick. That sounds like good news to most people, but not to us. The very high proportion of mild cases is bad news because it tells us why this pandemic will be remembered for generations. Paradoxically, it is much easier to stop an epidemic that kills or sickens a far higher fraction of its victims.
We note with sadness that the global death toll is now above 280,000 human beings. That represents 16% of the closed cases. Epidemiologists don’t confuse this with the case-fatality rate because we know that the denominator (total number of infections) is no where near accurately estimated by this case total. Also, because the duration of illness is long for a virus, closed cases are not a good metric even for the crude death ratio, especially when there are more active cases than closed cases. Finally, we can’t forget that deaths are being undercounted probably by half.
The bottom two curves show us global cases and deaths on the log scale. We look at the data this way because we want to tell the difference between exponential and linear growth. When the curves rise on the log scale, that’s exponential growth. Flat sections indicate a transition to linear growth. Everyone is talking about a “second wave” in the fall. Based on this graph, I would say there is partial evidence that on a global scale, we are in the second wave now, having seen a plateau in cases and deaths in early March. More importantly, the trend in both the blue and orange is for continued non-linear growth if not acceleration in the pandemic in the last 2 weeks. Bottom Line: There is no evidence here that the global pandemic is weakening.
Scientists at Mount Sinai in New York show that COVID-19 patients treated with blood thinners had better survival adjusting for other factors. This was not a randomized clinical trial but strengthens the case for using established treatments to improve outcomes in very sick patients (Journal of the Am College of Cardiology)
48 hours after pledging to have a vaccine by the end of the year, President Trump backpedals (Politico)
FDA revised its policy regarding antibody testing, now requires companies to seek Emergency Use Authorizations (EUAs) including submitting validation data (FDA website)
SARS-CoV-2 found to have infected at least one person in France in December, a month earlier than first official case (Intl. J Antimicrobial Agents)
Anyone who thinks this is just like the flu hasn’t seen this graph! I continue to hear people justify their resistance to social distancing measures by comparing COVID-19 to seasonal flu. A couple of weeks ago, I posted a graph comparing the U.S. COVID-19 death toll to other major mass casualty events in history. Below is an update of that graph. A month ago (April 5) the U.S. had reported just over 9,500 deaths. As of Tuesday, 55,000 deaths were added in that month. That is nearly as many as all American deaths during the 20 years of the Vietnam war. The U.S. has now eclipsed 65,000 deaths in a period of just over two months (if Puerto Rico and Guam are included). To put this in perspective (see graph below), as of yesterday, more Americans have died of COVID-19 than died from Ebola, SARS, The Las Vegas shootings, the Gulf war (Operations Desert Shield & Desert Storm), Hurricanes Andrew and Katrina, average deaths from the flu in March and April, H1N1 Swine flu and the Korean War. Combined. Bottom line: This is not the usual flu.
U.S. records fourth highest death toll on Tuesday Yesterday, 2,435 Americans were reported to have died of a confirmed COVID-19 case, the fourth highest daily total thus far, an increase of 4%. U.S. deaths rose by 24% in the last week. Yesterday, 7 states met or exceeded the highest daily death toll including Arizona (87), Utah (6), Iowa (19), Illinois (176), Mississippi (32), Maryland (74), and Pennsylvania (554). Pennsylvania had a particularly sharp increase. The PA department of public health posted this in their press release: “As a result of our continued work to reconcile data from various sources, the state is reporting an increase of 554 new deaths today bringing the statewide total to 3,012 deaths in Pennsylvania. These deaths have occurred over the past two weeks.” Several of the states that set record deaths are nevertheless moving towards reopening including Pennsylvania. The lower figure shows rate of change in deaths over the last week. Seven states saw deaths near or above 50% last week. Midwest hot spots continue in Iowa, Montana and South Dakota. To that, additional states with high growth rates emerged in the south (Arkansas, South Carolina and Virginia). According to the CDC, 35 U.S. states plus Guam are reporting widespread community transmission. That this means? Weekend lag in reporting continues as deaths spiked on Tuesday. Hopes that the worst is over appear to be premature as the U.S. remains just below 2,000 daily deaths. The 7-day moving average is flat over the last week. This is especially troubling given that we have not yet confronted the effect of states that reopen due to the lag between increased behavioral mixing, infections, illness and deaths. Those effects won’t be seen for another 2-3 weeks.