New York State missed early signs of the COVID-19 outbreak in February because testing was nearly non-existent: it wasn't until February 29th that New York got approval from the federal government to conduct its own tests, and results didn't begin trickling in until March 3rd. From there, testing capacity built up slowly — by March 15th, to about 1000 tests per day, and by April 1st to about 16,000 tests per day. Then, for about three weeks, testing plateaued around 20,000 tests per day, until the 24th, when it began to rise again to hit our current average of 30,000 tests per day.

Governor Cuomo has said that our state's total capacity will continue to rise to 40,000 tests per day in the near future, still far less than the 300,000 we'd need to hit the testing benchmark suggested by some experts, including Harvard's Paul Romer, who have said that to permanently contain the outbreak, we'd need to test the entire population of the state every 14 days. Whether that number is realistic, or even possible, testing will be key to the next phase of the epidemic, as New York aims to shift to a "test, trace, and isolate" strategy to continue to bring down our numbers.

With that in mind, the Gothamist data team has used the testing data provided by New York State to take a closer look at testing from the start of the epidemic. In the graph below you can hover over each series for data by day, and click the menu button at the top right to switch between the subsequent graphs.

First, we'll look at total tests performed in the state by county:

Here we can see that testing was extremely limited well into the beginning of March, with Westchester taking a sizeable amount of the available tests. Testing then begins to ramp up each week, but tends to drop off on weekends, probably because some labs and doctors' offices were closed. Finally, we see a huge spike toward the end of last week, on April 24th, followed by another fall-off. It remains to be seen whether the testing rate will regain this high again, or stay there consistently.

Next, we can look at share of testing by county:

Here you see the massive testing in Westchester after the initial outbreak in New Rochelle — the county continued to take a large share of the tests through the end of March, but by April the other counties had caught up, and share has been stable since then, with occasional spikes in some counties (notably Rockland and Putnam last week).

Another way of looking at testing is by confirmation rate: how many of the tests performed each day have come up positive? Early in the epidemic, we see huge swings as small numbers of tests skew the percentages, and the lines bounce around:

But later, as we get more testing, we see that New York City counties are seeing a higher percentage of tests come back positive, which would seem to confirm that the center of the outbreak was in New York City, and outlying counties had lower true infection rates. Over time, as we performed more tests, the percentage coming back positive decreased. This could have been caused either because less people were actually sick, or more likely, our testing capacity had simply expanded to the point where more people overall were getting tests.

Finally, we can look at testing per capita, which again shows Westchester leading on many days early in the epidemic, eventually eclipsed by Manhattan ("New York County"), with Rockland and Westchester continuing to show up prominently in terms of population sampled each day:

It's worth noting how much more tested Manhattan is than the other four downstate boroughs, as well as noting the spike in Putnam, Rockland, Orange, and Ulster this past week.

We will continue to keep an eye on this testing data set — let us know if you have any specific questions about testing in New York, and we can address them in future posts.