Top news, reports and insights for today:
- Crude death ratios widen, little progress toward the true CFR
I thought I would revisit a graph I last posted several ‘years’ ago (um, actually April 5) showing variation across nations in the crude death ratio (or CDR). As a reminder, the Holy grail is the actual Case fatality rate (CFR), or the risk of death among all infections. If we knew that number, we could work out lots of useful stuff. We would be able to pretty accurately predict how many deaths we will have and where in the epidemic cycle we are. Right now, we don’t have the data to calculate the CFR because we don’t know the prevalence of infection in the whole population. We have used our scarce testing resources to test only (at first) people who had been to an effected country, and later just those who are sick. In my view, this has been an unwise use of testing resources for an epidemic like this one. We believe that half of those infected have no symptoms or mild symptoms. We also think that pre-symptmatic and asymptomatic transmission is occurring. The epidemic is being driven not by people sick in hospitals or in their beds, but by secondary transmissions emanating from the thousands or millions of people who are infected and don’t know it. Currently, we are entirely in the dark about how many people are actually infected or who they are. Today’s top pick of the day tells us that if our positive testing rate is 20%, we are failing at testing the people who matter most. That’s a big part of why we don’t yet have a solid answer to the CFR question. All we have now is a the crude death ratio (the number of dead/confirmed cases). I won’t bore you with the technical debates about what you call this number, but I’ll call it the CDR just to keep it clear that it’s a very poor substitute for the CFR. To get to the Holy grail, we need to be able to count all cases of infection to get our denominator. If we had that, the CFR should be roughly similar across different countries. Looking at the graph, the CDRs are staggeringly variable, ranging from over 15% in Algeria, to over 13% in Belgium, Italy and the UK, to less than 1% in Russia. Based on data from WORLDOMETER on April 15, the global CDR (deaths divided by confirmed cases) is now 6.5% (in purple). If that were the true CFR, we would predict 4.6 billion infections and 300 million deaths globally, assuming a 60% attack rate. That’s no where near where we are based on the the reality on the ground and our data. The true CFR must be significantly lower than that. On March 4, when there were only 93,000 global cases, the WHO announced an “estimated” CFR of 3.4%. More than 2 million cases later, that number hasn’t been updated. What these numbers do show is that the range of estimates is getting bigger, not converging toward consensus. When I posted this 2 weeks ago, virtually all of the CDRs were lower than they are today. All of the hardest hit countries have seen rates slide upward. Some countries rose more than others. The two nations that were the focus of global attention 2 weeks ago, Italy and Spain, both rose a modest 0.6%. Belgium’s CDR spiked 5.4% higher, while Algeria, the United Kingdom, France and Sweden, all saw hikes of around 3%.
The bottom line: A country’s CDR is a better indicator of the limits of the testing regime than the impact of the virus. Because of increasingly selective testing, we are no closer today to determining the real CFR than we were weeks ago. It it is vitally important that nations conduct comprehensive, population-wide viral and antibody testing on a random sample to determine true population prevalence of COVID-19 so that we can get to the Holy grail: CFR.
- U.S. Deaths dip, cases rise on Thursday, state variation in crude infection rates widen
After two days of increasing deaths, Thursday saw a 7% decline compared to Wednesday as 2,140 Americans died of COVID-19. The cumulative total death toll now stands at 30,356. The true death toll is certainly higher due to undercounting of deaths outside of hospitals, particularly in nursing homes, and deaths in patients without test results or where a negative test was false. Despite the overall decline, 8 states matched or set new record high deaths including New Mexico (8), Oregon (6), Illinois (124), Nebraska (3), Arkansas (3), New Jersey (362), Rhode Island (18) and Vermont (5). Three day growth in deaths over 50% (or doubling in 6 days) were seen in Wyoming, North Carolina, Connecticut, Washington DC, and Maryland. U.S. cases rose by 5% as 32,067 new confirmed cases were reported, bringing the total to over 664,000. As the graph below shows, New York again leads the nation with 1,142 confirmed cases per 100,000 population. Another 16 states have rates over 100 per 100,000 including: Colorado (151), Washington (145), Illinois (203), Indiana (142), Michigan (293), South Dakota (148), Louisiana (485), Connecticut 446), DC (333), Delaware (213), Massachusetts (467), Maryland (178), New Jersey (848), Pennsylvania (217), Rhode Island (362) and Vermont (123).
The bottom line: Cases and deaths continue to increase. Daily fluctuations both up and down are to be expected. Overall national trends hide large variations by state and region. This is not peak. Rescinding social distancing is likely to be very costly in terms of preventable deaths.