The medical experts on President Trump’s coronavirus task force are in an admittedly difficult position. They understand why Trump’s push to scale back social distancing efforts are fraught, but as partners on his efforts, they are forced to calibrate their analyses of the pandemic to find the best possible path forward. For all of the emphasis placed on making decisions based on data, the data aren’t always where the president would like. And so things get a little … gauzy.
On Wednesday, task force member Deborah Birx offered an optimistic assessment of how the pandemic’s course has shifted in the United States.
“Over the last five to six days, we have seen declines in cases across the country. And this has been very reassuring for us,” Birx said. “At the same time, we know that mortality and the fatalities that we’re facing across the United States continue.”
After offering her appreciation for health-care workers, she continued.
“I also wanted to let you know that we do have nine states that have less than 1,000 cases and less than 30 new cases per day,” Birx added. “So, we’re looking at states and metro areas as individual areas.”
This is the sort of argument that Trump has been making: We can rescind containment measures in some places because they’re doing better than the worst-hit states, like New York and New Jersey. The problem is that Birx’s numbers only work as absolutes. As a function of population, those nine states she identified aren’t all exemplars of success in fighting the virus.
According to data from Johns Hopkins University, there are nine states that had fewer than 1,000 cases as of Wednesday: Alaska, Hawaii, Maine, Montana, Nebraska, North Dakota, Vermont, West Virginia and Wyoming.
The metric of “number of new cases a day” is a bit squishy, given daily variance in confirmations. Of those nine states, six have averaged fewer than 30 new cases over the past three days. Let’s set that aside for now.
Let’s instead look at how the number of cases per state changes once we consider population. New York and New Jersey remain the worst-hit, with New York seeing 1,100 cases for every 1 million people. But suddenly, Vermont jumps out — its rate of 122 cases per 1 million residents ranks it not among the lowest nine states but among the 20 states with the biggest per capita outbreaks.
Vermont suffers, in part, due to its proximity to New York and Massachusetts, certainly, but it’s nonetheless hard to depict it as entirely a success story based on these numbers.
In the abstract, it’s certainly the case that the 759 cases in Vermont is better than the 214,000 cases in New York. But having 50 cases in a population of 100 is obviously more problematic than having 500 cases in a population of 10,000, which is the challenge these per capita numbers present.
If we zoom in on the majority of states with per capita infection rates less than 300 per million, we see that the nine states identified by Birx aren’t all sitting at the bottom. In fact, the state with the lowest number of infections per million residents is Minnesota, where only 32 residents per million have been confirmed to have infections. (The actual infection rate, as always, remains unassessed.)
Yes, Minnesota had 1,800 confirmed cases as of Wednesday, but is that worse than Vermont, where the per capita rate is nearly four times as high?
If we consider both metrics offered by Birx on a per capita basis, we see that both total infections and average new infections include several of the nine states in the lower tier — and some much higher up.
What’s important to consider is the metric that Birx started with: how the number of cases is shifting over time in those states. Last week, we created a tool showing how the case totals (and test and death numbers) have changed since March 1. We’ve updated the interactive and the numbers below.
Show data for
Here, we see that states like Idaho and Louisiana (click each state to see a larger version of its graph) have seen a downturn in the number of new cases — the sort of positive sign that isn’t necessarily captured in Birx’s under-1,000 metric.
What happens by focusing on low overall numbers is that we miss states like North Dakota, which is at a peak in its 3-day average of new cases. We miss states like Iowa and South Dakota, where numbers are climbing. All of these states, incidentally, are ones without stay-at-home orders in place.
That’s a lesson worth noting, too — but it’s not the one Trump is interested in.