Some COVID-19 numbers.

On April 22, I began keeping a weekly spreadsheet of COVID-19 deaths in the US and in each of the 50 states and the District of Columbia.  I have kept up with it now for more than three months, and without it, I would not have much of a grasp on the course of the epidemic.  Innumeracy--an inability to make sense of statistics--is rampant in the media. and much of the reporting is about cases, for which our data is lamentably incomplete.  The death count is surely incomplete as well, but not by nearly as much.  Here are some of the most important things that the data now shows.

Nationwide deaths totaled 14,031 for the week ending midnight GMT on April 28.  That figure included more than 2,512 in New York, 1,379 in New Jersey, 821 in Massachusetts, 468 in Connecticut, and 418 in Illinois.  Deaths per million for that week--a key statistic-- were 42 nationwide, 128 in New York, and 155 in New Jersey.  Deaths per million were just 12 in Florida and 5 in Texas.  Nationwide deaths remained at about the same level in the next week, totaling 14,330 for the week ending May 5.  Then they began to drop, to 11,154 on May 12; and they continued to do so pretty steadily for another eight weeks or so, all the way through the month of June.  Deaths for the week ending July 1 totaled just 3,919, less than 1/3 of the week that ended on April 18--just 12 new deaths per million people, rather than 42 on April 28.  Among the states, Rhode Island was the leader for that week with 42 deaths per million, followed by Arizona with 35, Illinois with 30, New Jersey with 26, and Massachusetts with 21.  Connecticut and New York had just 10 new deaths per million each that week,  Both had cut their death rate by 90% or so in nine weeks, a remarkable achievement. 

I do not have this figure at hand, but I saw figures around that time stating that as many as 40% of all deaths had occurred in nursing homes.  The virus had spread through many of them, and through a significant portion (we have no real idea how big) of people in the Northeast, like wildfire, before anyone realized what was happening.  By May, however, if not earlier, nursing homes had become quarantine zones, and that must account for a significant portion of the reduction in deaths.  Meanwhile, the shutdown of the economy and the almost universal adoption of social distancing measures in the large urban areas of the northeast had obviously reduced the infection rate by a very large number as well.

After bottoming out at 3919 on July 1, new deaths nationwide increased to 4,064 on July 8; 5,282 on July 15; 6,039 on July 22; and 7,657 on July 29, last Wednesday. (I should note that all these figures are slightly approximate because they are the ones reported at the time, and states have frequently had to revise their figures upward somewhat later.) Total deaths in that period have been increasing at 5% a week--a rate of increase considerably higher than the long, slow decline from early May through early July.  Meanwhile, the distribution of deaths around the country has completely changed.  The leaders in new deaths per million for the week ending July 29 are Arizona (66), South Carolina (64), Texas (63), Mississippi and Louisiana (47), Florida (46), Alabama (36), Georgia (29), and Nevada (25.) Nationwide we suffered 23 deaths per million.  In the northeast, Massachusetts deaths per million were 16, New Jersey 10, New York 6, and Connecticut 5.  California, interestingly enough, has stayed quite close to the national average throughout most of the epidemic.

The epidemic, then, as been getting worse in terms of deaths since July 1, and therefore, it must have been getting worse in terms of new infections several weeks before that, when the "re-opening" of the country began in early June.  We hear from the news daily that we have set new records for new cases and total active cases.  I am not convinced that that is true.  Deaths for the week ending July 29 were only slightly more than twice as high as deaths in late April.  If we really have more cases now, that means that the death rate per infection must have fallen by 50%.  The early "nursing home effect" might have raised the early death rate that much, but it seems a little doubtful to me.  If the current trend continues, however, we probably will be as badly off in total cases as we were in late April, even though they will be in completely different areas.  It would however take about 50 days of increases of deaths at the current rate to get total deaths per day up to the late April level.  That's plenty of time for the Sunbelt states to get their cases on the decline, if they are willing to do so.

I have not kept week-to-week statistics on other countries, but daily data shows that several of the major European countries have reduced their deaths to very low levels indeed. COVID-19 deaths on July 29 totaled 12 in Canada, 16 in France, 9 in Germany, 3 in Italy, and 2 in Spain.  In the UK they were 38, and Russia 29.  In the United States, the total reported was 1,465, the highest in the world.  Even though we are about six times larger than the major European countries, those are sobering figures.  New England, New York and New Jersey, with about 42 million people total, had 62 deaths on that day, a noticeably higher rate than Germany or Spain.  Brazil is second to the United States in total deaths with 1,189 that day; India, with 3-4 times the population of the US, had 783.

The continental European nations have passed the first big test posed by the pandemic. We have not.