It’s too early to judge what impact the US’s recent rescinding of stay-at-home and social distancing requirements is having on corona contagion. Newly diagnosed cases have remained flat; however, testing rates have doubled over the past month, while the percentage of test-positives has dropped in half. Does that decrease in test-positives reflect an actual decrease in infections, or merely a widening of the net? Preferably the two go hand in hand: once you reduce the infection rate you can begin testing and treating and isolating people with less severe symptoms, and you can trace and test those who have been in close social contact with already-infected individuals.
Two weeks ago, when I looked at the cumulative state-by-state data since the beginning of the epidemic, there was a strongly positive statistical correlation (r=+0.66) between a state’s testing rate and its test-positive rate. I.e., the stronger the local outbreak of the virus, the more tests were being conducted. That’s not what you want when you’re reopening your state. But when I ran that analysis I was looking at testing data from the beginning of the epidemic. What about recent trends?
Today I analyzed state testing data for the past 20 days. Sure enough, across all states the testing rates are doubled while test-positives are halved. But what about within states: does the prior pattern persist, with hotspot states doing relatively more testing than states with low infection rates? Turns out there’s still a positive correlation between a state’s testing frequency and its rate of test-positives over the past 20 days, but that positive correlation has diminished substantially, from +0.66 to +0.23.
That’s a good sign. Take New York for example. It still has a higher infection rate than the national average, and it does more testing than average. But while the state’s testing rate has gone way up, its test-positive rate has dropped dramatically.
Who’s doing all of this extra testing? Is it doctors diagnosing and treating less severe patients as part of their clinical practice? Or is public health departments doing contact tracing? Almost surely it’s the former. Over the past 20 days the correlation between states’ test-positive rates and death rates is +0.51, suggesting that those being tested are at relatively high medical risk.