Garbage-In, Garbage-Out

Here’s a logic puzzle for you. If testing for COVID-19 only if another test for COVID-19 is positive, when do you ever test?

Answer: you don’t.

Luckily, humans aren’t computers, though rules can make them behave, sometimes, like they are.

The question above was prompted by personal concern. It’s that time of the year when folks, including me, tend to get sick. I’ve had symptoms of the common cold or allergies for a week or so: a dry cough, runny nose. Number One Son presented symptoms of a viral infection on Monday evening: A fever, a dry cough, aches and pains. He did not have trouble breathing. Could be an influenza.

We can test for the flu. The test results came back negative. It’s not the flu. Given the current situation, it could be SARS-CoV-2. We could test for that, if test kits were available. But test kits are unavailable, for a variety of reasons including but not limited to a fragile supply chain. Therefore the Centers for Disease Control has certain recommendations on when to test, in order that the available tests not be used up in a panic.

Decisions on which patients receive testing should be based on the local epidemiology of COVID-19, as well as the clinical course of illness. Most patients with confirmed COVID-19 have developed fever1 and/or symptoms of acute respiratory illness (e.g., cough, difficulty breathing). Clinicians are strongly encouraged to test for other causes of respiratory illness, including infections such as influenza.

Epidemiologic factors that may help guide decisions on whether to test include: any persons, including healthcare workers2, who have had close contact3 with a laboratory-confirmed4 COVID-19 patient within 14 days of symptom onset, or a history of travel from affected geographic areas5 (see below) within 14 days of symptom onset.

Evaluating and Reporting Persons Under Investigation, Coronavirus Disease 2019 (COVID-19), Centers for Disease Control and Prevention
(updated March 4, 2020; retrieved March 11)

Practically speaking, however, there’s no testing because we can’t. This limits our knowledge of the spread of the illness and whether or not we should test.

And that’s not even getting to sins of omission. The cases reported to the World Health Organization are an approximation due to input error. We can somewhat extrapolate the scope of this problem with math–The Economist mapped typical tourism flows to elucidate where COVID-19 is being under-reported, for example–but small errors in variables cause very different results. How many people have been exposed? How many people have been infected? How long is the incubation period? When does an infected person start spreading the disease? How long does an infected person spread the disease? How much can you trust people who habitually lie?

All of which means that the very nice map from Johns Hopkins is a picture of the past, and not a clear guide to what’s happening now, in your neighborhood. The same goes for any news reporting. Still, decisions have to be made with incomplete or inaccurate data. And they can be, from knowledge of how diseases spread in general, and boil down to two things: reduce contact and reduce travel. The decisions by many colleges to hold classes online and to send students home is then, in this light, only half right.

We know that we don’t know if we’re infected.

We know that it spreads before symptoms appear.

We know that the symptoms are deceiving.

We know that we aren’t testing.

We know that the virus cannot spread without hosts.

Waiting until we have a confirmed case before acting seems the height of silliness, closing the barn doors after the horse has bolted.

But how easy is it to ask people to stay home for a fortnight? Martial law shouldn’t be necessary: ask folks to stay home and make it possible for them to do so. That’s the difficult part.

One thing that can be done, relatively easily, is projecting the result of doing nothing. [edit to add link that starts from a different assumption]

First, assume everyone is infected. 80% will be fine. 20% will develop complications. 1% will die. In the United States, that’s ≈3,500,000 people dead. Luckily, we aren’t doing nothing. People do wash their hands and stay apart from each other. We’re just doing practically nothing.

When this is all over, we shall see.