Top news, reports and insights for today:
- COVID-19 deaths decline over the last two days.
For the third straight day, absolute numbers of U.S. deaths confirmed COVID-19 deaths have dropped. On Monday 1,492 new deaths were reported, a 7 percent rise in the cumulative total. The U.S. continues to have the highest death toll in the World, although the crude death rate (deaths per 1 million population) at 71 ranks 15th behind Ireland (74 per million) and ahead of Iran (55 per million) according to Worldometer. New deaths dropped by just 1 since Sunday, suggesting the possibility that reports were lagged over Easter weekend. I am adding a twist to my daily graph of deaths. Below, take note of the “whiskers” above and below each bar. Those are called error bars and they convey information about how much uncertainty there is in our estimates. Any daily differences not larger than this uncertainty zone, has to be taken with a grain of salt. Rises or falls in deaths smaller than the width of these bars could be due to random error in how we count and report. Based on that, we can see that the number of deaths from April 7 through 11 was about the same. The drop we see on Sunday and Monday is more significant because it’s bigger than our uncertainty. The difference between Sunday and Monday is too small to be considered more than just random wobble.
The bottom line: The number of new deaths on Sunday and Monday was importantly lower. We must be patient to see what happens the rest of the week before we conclude that we are nearing peak deaths. It’s possible that New York and other states will start reporting “probable” or “suspected” COVID-19 deaths in addition to deaths among those with a positive test. This could cause numbers to go up again for a while. The data in this graph don’t include “suspected” deaths from New York (see point 2).
- New York City adds 3,700 “probable” COVID-19 deaths, raising the City’s death toll to over 10,000
Today, the New York Times is reporting that New York City began reporting probable COVID-19 deaths separately from confirmed deaths. Previously, only those who died and had a positive lab test were counted. This has been a major problem because many people who are hospitalized die before the test comes back from the lab, or die in nursing homes or at home without a test. The new numbers are only available on the NYC Health website and are not yet folded into the numbers on popular dashboards. The inclusion of probable or suspected deaths adds more than 3,700 to the total in that ravaged city. Using these new numbers, New York city now has more than 10,000 deaths, making the overall crude death rate higher than Italy.
Why this matters? Getting the right death toll is vitally important, not just as an academic exercise, but also because we can’t monitor and track the course of the epidemic unless we know the true numbers of deaths. If we miss half the deaths during an outbreak because we don’t have adequate testing, we won’t know when peak occurs so we won’t know when it’s safe to send people back to work.
- It’s time to start looking at RATES of COVID-19 by state
For the first time, I am posting a new kind of graph showing not numbers of deaths and cases, but rates of COVID-19 cases by state. Lets start with big caveats: testing has not been adequate and is very uneven across states. So, rates are a long way from being perfectly comparable. Second, ideally we would calculate true incidence rates (the number of infections divided by the total number in the population). But, we know that a quarter to a third of infections have no symptoms or mild symptoms; those people are almost entirely untested. The numerator for this first set of rates is therefore not infections but the subset of infections that got tested because they were sick enough to come to the attention of the health care system. Saying all that, the COVID- Tracking Project now tells us we have done about 3.1 million tests, with another 17,000 that are still pending. As testing spreads out, our confidence in rate calculations will grow. And when comprehensive antibody testing kicks in this summer, we will start to have very solid numbers. So, consider this a baseline preview of what things look like now.
The first thing to notice is that the scale is truncated to make comparisons visible. New York and New Jersey are not fully shown and are literally off the charts at 1,003 and 727 per 100,000 respectively. These differences are no doubt even bigger because of large numbers of probable but not confirmed cases. The red dashed line is the overall crude case rate in the U.S. (currently 175). States above this line have higher crude rates than the Nation. They include no western states, Illinois and Michigan in the midwest, only Louisiana in the south, and 8 or 12 Northeastern states: Connecticut, Washington DC, Delaware, Massachussetts, New Jersey, New York, Pennsylvania, and Rhode Island.
What does this mean? We need to squint our eyes in looking at these numbers. Comparing rates across states is most meaningful when there is comprehensive testing everywhere. We are a long way from that. However, sticking to big patterns, if looks like California’s containment efforts are succeeding. The crude rate of COVID-19 cases is 68% lower than the U.S.; Colorado, Idaho, Nevada, Utah, and Washington are all higher. In the midwest, Illinois, Indiana, and Michigan all have more than 100 cases per 100,000, but it is Michigan that really stands out at 46% higher than the U.S. rate. Southern states are similar and below the nation with the exception of Louisiana which currently has the third highest rate. The eastern seaboard clearly remains the epicenter of the outbreak in the U.S. Looking at rates rather than total cases reveals surprisingly high concentrations of disease in smaller states like Connecticut, Delaware, DC, Massachussetts and Rhode Island, all of which are now comparable to Michigan. Maryland now has twice the rate of COVID-19 cases as California per capita.