Daily COVID-19 deaths in the United States are continuing to rise, reflecting the impact of a surge in newly confirmed cases that began last month. The seven-day nationwide average reported by independent data scientist Youyang Gu, which is based on tallies from the Johns Hopkins Coronavirus Resource Center, rose from a low of 510 on July 4 to 763 yesterday. The Worldometer tally, which is based on deaths reported by local health departments, shows a similar trend, although its numbers are somewhat higher.
Daily deaths are still much lower than they were last spring, when Gu reported a seven-day average of 2,238 on April 18, and the upward trend is much more gradual than it was that month. Furthermore, the crude case fatality rate—deaths as a share of confirmed infections—continues to fall, from more than 6 percent in mid-May to 3.7 percent today. That downward trend likely reflects wider testing, which has added milder cases to the denominator; a younger, healthier mix of patients; and improvements in treatment.
The Worldometer count of daily new cases has risen almost fourfold since Memorial Day, from fewer than 20,000 on May 25 to a record of nearly 75,000 on July 17. But as reflected in the declining case fatality rate, the increase in a daily deaths has been much smaller—about 50 percent, based on the seven-day average.
Some states have seen much larger increases in daily COVID-19 deaths. Since Memorial Day, per Gu's numbers, the seven-day average has risen by 178 percent in Florida and 338 percent in Texas. California, another state that accounts for a disproportionate share of newly identified infections, has fared better by that measure, seeing an increase of just 37 percent. And in Georgia, where the seven-day average of daily new cases has quintupled since May 25, the seven-day average of daily deaths is down a bit since then, although it was rising until mid-June and is once again headed upward. But even in Texas and Florida, the increase in deaths has been much smaller than the increase in cases.
Some of that gap can be explained by the typical lag between laboratory confirmation and death, which the Centers for Disease Control and Prevention (CDC) last month estimated was about two weeks. If expanded testing is catching cases earlier, the average time between a positive virus test and death may be growing, meaning the fatal consequences of the infections recorded in early July might not be apparent until the end of the month.
Even allowing for that lag, however, it is clear that COVID-19 patients, on average, are less likely to die from the disease than they were earlier in the epidemic. That is consistent with data showing that patients in the Sunbelt states that are driving the recent increase in cases tend to be younger than they were in the spring. The median age of people testing positive for the virus in Florida, for example, plummeted from 65 in early March to 35 in mid-June.
Such trends have important implications for the death rate, which is strongly correlated with age. According to the CDC's "best estimate," 1.3 percent of Americans 65 or older who develop COVID-19 symptoms will die from the disease, compared to 0.2 percent of 50-to-64-year-olds and 0.05 percent of people younger than 50. The ultimate death toll will therefore depend heavily on the success of precautions aimed at protecting older and less healthy Americans from infection during the time it takes to deploy an effective vaccine.
While a younger mix of patients implies a lower case fatality rate, we can still expect daily deaths to climb as a result of the recent surge in new infections. Gu, who has a good track record of predicting COVID-19 deaths, has repeatedly increased his projections during the last few weeks. His model still predicts a gradual rise in daily deaths during the next month, followed by a gradual decline through October. But the projected peak has risen from 774 in late August to 918 on August 11, and the estimated death toll as of October 1 has risen from about 186,000, the projection in early July, to nearly 200,000, rising to about 214,000 by November 1.