Suppose that suddenly a new disease X appears to be spreading rapidly and killing people. The symptoms of X are very similar to those of Y and Z, but X may have a far higher fatality rate than Y and Z combined. Thus, we need to be able to distinguish between patients who have X and those who have Y or Z, so we can treat those who have X appropriately.
With the need urgent, multiple labs rapidly develop tests for X, and the FDA cuts red tape so the tests become available quickly.
When we administer a test there are four possible outcomes: true positive, false positive, true negative, and false negative. Ideally the rates of false positives and false negatives are low, but that they occur is normal and not cause for alarm. That our tests are not perfectly accurate is why assessment is more than simply giving a single test. Assessment integrates the available data to reach a conclusion.
Now suppose that we have a large sample of people to whom we give our new test. A meaningful number of people test positive for X but show no symptoms. What should we conclude?
In normal times when a person tested positive for X but had no symptoms, we would conclude we had a false positive. Having some false positives is completely normal and does not mean our test is no good, only that, like other tests, it is not perfectly accurate.
If X is COVID-19 and a person with no symptoms tests positive, we do not conclude that the result is a false positive. We conclude we have an asymptotic carrier who will infect those around him. These are not normal times.
While the possibility of false positives is seemingly ignored, there is a recognition of false negatives. There is a tally of COVID-19 probable cases where a person tests negative, but the evidence from the symptoms strongly suggests that the person does in fact have COVID-19. Some have written that a portion of these probable cases have been classified as positive cases despite testing negative.
From an assessment perspective, we would expect some who test negative to actually have COVID-19, just as we would expect some who test positive not to have COVID-19. Additional data beyond the nasal swab test would be used to determine the appropriate treatment for people whose test results were inconsistent with their symptoms. If it was determined that the test was a false positive, no treatment would be needed.
To make good decisions we must consider the consequences of wrongly classifying a person either as having COVID-19 when he does not have it or of classifying a person as not having COVID-19 when he does in fact have it. In the latter case, we would be denying treatment to a person who needs it and allowing for the further spread of the disease.
But what of those who do not have COVID-19, have no symptoms, but test positive? As the economy is reopened, increased testing of asymptomatic people is expected. I have already written about unintended consequences of the lockdown here.
With the advent of contact tracing, a false positive result could affect the liberty not only of the person who tests positive, but of complete strangers who happened to be nearby. One contract-tracing proposal suggests using cell-phone identifiers. For instance, a person’s cell phone would record identifiers of others’ cell phones when close. The tech companies assure us this would be anonymous and that the information would not be used inappropriately.
Suppose you are at your local coffee shop and two tables across from you is Fred. The next day Fred’s employer administers Fred a nasal swab test. Fred has no symptoms. The testing is part of the business’s plan to keep everyone safe. Fred tests positive, information that he enters in to his phone. You get an alert informing you that you have been in contact with someone who tested positive for COVID-19.
What do you do? What should you be legally required to do?
We need to acknowledge that the tests we use generate false positives. We need to have a good estimate of the false positive rate. Until we do, there is no ethical way to even consider the questions about what actions we expect of people who have been notified that they were close to a person who tested positive.
If in the name of safety we continue to expand testing of those who are not experiencing symptoms, there will be more false positives simply because we are administering more tests. A commitment to science requires that we humbly admit that our tests are not perfect. Only then can we thoughtfully consider how to make use of the information the tests generate.