State-by-State Rankings of Coronavirus Testing. Is Your State Testing Enough?

Is your state doing enough testing for coronavirus? Which states are doing the best and worst jobs?

Here's what the data says.

The Right Metric: Tests Completed per Infected Resident

Comparing states on total tests completed is obviously not an apples-to-apples comparison. Not only do states have different populations, they also have different timelines of the outbreak, with some states like New York heavily affected early on, and others only beginning to see cases rise.

For example, New York has a much higher population than Alaska. More testing in NY doesn't necessarily mean NY is doing a better job testing than AK.

Percentage of the state's population tested is a better metric... but it too is imperfect. Namely, it ignores the timeline and current severity of outbreaks in each state.

Returning to our example, if NY and AK each tested 10% of their residents, that wouldn't mean they were doing equally adequate testing. NY currently has many, many more confirmed cases than AK. NY's testing would be much worse than AK's relative to the severity of the outbreak and the need for testing. Thus comparing states on percentage of their population tested is also not apples-to-apples.

There is one metric that accounts for both differences in population and differences in outbreak timeline, allowing us to make an apples-to-apples comparison: Number of Test Completed per Infected Resident.

"Number of Tests Completed per Infected Resident" is the metric that matters

For example, if NY had tested 8,000 total residents and had 1,000 total confirmed cases of infected residents, their value would stand at 8. Said in reverse, a value of 8 means that 8 residents in a given state have been tested for the virus for every 1 resident of that state confirmed to be infected.

Because confirmed cases of infection is a function of both population and timeline/severity, it allows us to compare apples-to-apples across each state, even though those states have drastically different populations and different outbreak severity.

I will refer to this metric in shorthand as "Tests per Infection."

A higher "Test per Infection" score means that a state is testing more people beyond just those who are likely sick, meaning more of their population is prepared and informed to counter the virus. A "Test per Infection" value near 1 means that only 1 resident has been tested per every 1 confirmed case, i.e. the state is not testing anyone other than residents they highly suspect to be infected.

National Benchmark

Before comparing individual states, let's look at the entire nation to get a benchmark. As of the latest available data on April 15th, 2020, across the entire United States, "Tests per Infection" sits at 5.1.

In other words, 5.1 US residents have been tested for every 1 infected resident.

Let's compare that to South Korea, which has been widely praised for its containment of the virus despite its vicinity to China and its short window of time to prepare. South Korea's recent "Tests per Infection" value is over 30.

In other words, more than 30 residents of South Korea have been tested for the virus for every 1 infected resident.

The United States has done a 6x worse job of testing their population than South Korea.

Plenty of people are quick to point out that South Korea is much smaller and much more densely populated than the United States. Which is a fair point, South Korea is just 7% larger than the state of Indiana, and has an 8% higher population density than America's most dense state, New Jersey.

So let's see how individual states, such as Indiana and New Jersey, are faring...

Best States vs Worst States

Top 3 States (of those w/ at least 1,000 cases)

Though these states should be commended, it's important not to lose sight of the wider context.

Of states with at least 1,000 cases, not a single state in the US has done a better job at testing than South Korea.

1. Minnesota: 23.3 Tests per Infection
2. New Mexico: 22.7 Tests per Infection
3. Oregon: 19.6 Tests per Infection

Of the national leaders, New Mexico has been sharply trending down over the past several weeks, possibly reflecting that testing has not been keeping up with the spread of the virus in that state.

Bottom 3 States

Of the worst performing states, I was surprised to see that Michigan has squandered a great start to testing in mid-March. They have since fallen to one of the very worst states in the nation.

1. New Jersey: 2.0 Tests per Infection
2. New York: 2.5 Tests per Infection
3. Michigan: 3.2 Tests per Infection

Those that say South Korea can't be compared to the United States should pay these numbers attention.

America's most densely populated state, New Jersey, which is only 8% less dense than South Korea, is performing 15x worse than South Korea at testing.

Full State-by-State Rankings

This below displays the current Tests per Infection figures for every state. Note that some states such as Hawaii, North Dakota, and Alaska currently have less than 1,000 cases so were excluded from top performers listed above.

Note here that America's most similarly sized state to South Korea, Indiana (just 7% smaller than South Korea) is doing a 6x worse job than South Korea at testing.

This chart is useful as a snapshot of where states currently stand, but what if we want to understand how the situation has been evolving over the past month?

Visualizing Daily Changes in State Testing

The below visualization displays a state's number of confirmed cases on the x-axis, and that state's Tests per Infection over time.

• moving to the right represents a rise in confirmed infections
• moving upwards represents an increase in tests relative to infections
• moving downwards represents a decrease in tests relative to infections (i.e. infection cases growing faster than testing efforts)

Worryingly, most states seem to be continuing to fall and move to the right (more infections, and relatively fewer tests to keep up). Though New York, the nation's hardest hit state, has begun to inch up in recent days, suggesting larger testing efforts may (finally) be rolling out.

The data is painstakingly clear: Amercia's coronavirus testing efforts continue to be severely lacking. If we want to save lives and "return to normal" soon, the quickest way to do both is to drastically increase testing.

What do you make of the data? I'd love to hear your thoughts.

Data Sources

The COVID Tracking Project: https://covidtracking.com/

Data from The New York Times, based on reports from state and local health agencies: https://github.com/nytimes/covid-19-data