“We have had multiple family members with vague symptoms all of whom tested negative but the father is COVID positive. He then had a follow up test that was negative. Is this sufficient to clear him to go back to school?”

How about this one:

A person had congestion and achiness Monday. The next day she tested negative. She felt a bit better over the next few days but not back to normal (she was also running on very little sleep). A week later her symptoms started up again and were more severe. She is now recovering over the past few days. She has no known exposure. Do you think the negative test on day two clears her and her husband? Technically, she would be 10 days since the start of the second round of symptoms by next weekend but, if this is a second illness and is covid, her husband would not be.
Should she be tested today? Does that then put her and her husband into the questionable awaiting result group?

How would you answer? At best, if you’re wrong and say everyone is clear you could expose an entire school or community and lead to schools shutting down, more people stuck in quarantine, and great financial hardship. At worst, you could be the cause of serious illness or perhaps people dying.

These are not easy questions to answer and there are an infinite number of variations of questions asked to doctors all around the world every day. There is no end to the scenarios of people who were exposed or had symptoms and want to go back to school, work, or just life and are desperate to for a doctor to give freedom with a clear conscience. The answers are as varied as the questions but they often hinge on the outcomes of various forms of COVID-19 testing. This raises a fundamental question, how accurate are these tests?

For all medical testing, it is very difficult to determine the accuracy of a test for the individual in front of you.

It turns out this is a very complicated question and I thought it would be helpful to describe the issue with tests in general focusing on the PCR nasal swab. That is the test where they identify the virus by amplifying its genetic material.

For all medical testing, it is very difficult to determine the accuracy of a test for the individual in front of you. The first problem is you have to know how good the test is in a laboratory environment when tested against samples known to have the virus or in conditions where it was clear that the individual tested had the disease. This tells you the characteristic of the test itself in a way that is important but insufficient to make a decision by itself.

If you have someone who may have coronavirus you want to know the likelihood that the test results accurately reflect the reality of his status as infected. In other words, if the test is positive, what is the chance the person is truly infected? If the test is negative, what it is the chance that they are not? So if you were the doctor and someone has vague symptoms and their test is positive what is the probability that they have COVID-19, and if it is negative what is the probability that they don’t (these are called the positive and negative predictive values if you want to know the jargon)?

These probabilities are very difficult numbers to pin down with certainty. The reason is that they are influenced by how many people in the population have the disease in question. It is a known fact that as disease numbers go up, the probability that a positive test is correct goes up and the probability that a negative test is correct goes down. As disease numbers go down, the probability that a negative test is correct goes up and the probability that a positive test is correct goes down (for the mathematicians in the room you may recognize this as being Bayesian). If you are not a statistician, the important thing is that the numbers of false positive and false negative tests will change if the prevalence of disease changes.

If you’re still with me, you may already be able to tell how this could add to the confusion of a pandemic. The number of infected individuals is constantly rising and falling which will lead to constantly changing rates of false positive and false negative results.

But wait, there’s more. The likelihood that a test is positive varies in one person over the course of their infection. Early in the disease, the virus has not replicated sufficiently to be detected by the test while later in the disease course the immune system has cleared the virus both increasing the numbers of false negative results.

As an example, data from seven studies of PCR tests was combined to see how the likelihood of a false test result changed as time passed from the initial exposure. They found that the false-negative rate was as high as 65% on the day of exposure and still as high as 30% on day 8. It then went up again to as high as 77% on day 21. In other words, on your best day, a negative test could be wrong in as many as 3 out of every 10 tests.

In other words, on your best day, a negative test could be wrong in as many as 3 out of every 10 tests.

Finally, there could be errors in labeling samples, how the samples are collected, stored and transported, errors in the machinery that processes the tests, problems with the reagents, etc.

By the way, this is for the longer and more accurate PCR. The numbers are can be worse for the rapid tests.

If you understand none of this, it’s not the end of the world. If the physician or layperson who is advising you doesn’t understand it we have a bigger problem. What you must understand, however, is that testing for COVID-19 is far more complex than just a positive or negative result would indicate. It’s much more than “I was negative so I’m cleared”.

This may also help you understand a few other aggravations wrought by COVID.

Why ARE things so confusing with COVID-19?

All the issues related to the actual number of cases, how many people are dying from COVID-19, how it spreads, where the levels are rising, etc. depends on the reliability of the tests. If a negative test is wrong 1 out 4 or 5 times this could have a huge impact on the epidemiology.

Why do so many doctors seem to be so strict?

There is so much uncertainty and the virus spreads so fast that many public health authorities are justifiably scared that if they rely too much on negative tests there could be disastrous consequences. It is safer from a public health perspective to believe the positive and be skeptical of the negatives. There is no perfect option. Either way there is fallout. You have to choose the best of the bad options. Balancing this with the secondary damage of quarantine and isolation both economically and psychologically is a nearly impossible task.

Why does Dr. Grove always have a pandemic induced stomach ache?

Many people ask if they were exposed and have a negative test, can they end their quarantine. Based on the above you can see why negative tests can be unreliable. People could get a false negative result and think they are cleared and spread the virus further.

Why can’t we get a test to avoid quarantine?

Every question I get involving putting someone in or avoiding quarantine is agonizing. There is so much uncertainty of the meaning of test results. It breaks my heart to tell people they need to quarantine with all the inconvenience and economic hardship that entails but I also am terrified that I could be responsible for giving advice that leads to somebody getting sick or dying.

Every decision we make, especially with COVID-19, is a balance of assessing risk and benefit. Testing helps us assess risk but should not be used alone.

What do we do now?

Beware not to go to extremes. None of this means the tests are useless, it just means they must be interpreted with caution and used in context. Every decision we make, especially with COVID-19, is a balance of assessing risk and benefit. Testing helps with the risk assessment but cannot be used alone.

We can now understand the CDC’s testing recommendations. The CDC recommends testing anyone with symptoms and believing a positive test. They are noticeably silent on what to do with negative test. This is probably because of the inherent difficulty in interpreting a negative test.

So if the test is negative you have to decide what the risk and benefit of believing it is. You have to combine the patients symptoms (how COVID-y are they), their exposure (prolonged close contact with a known positive), and how many cases are in the community they come from. You combine these to estimate the likelihood they have the disease. If it is high then the the likelihood that the negative test is true goes down. In other words, it is more likely to be a false negative. You then have to ask yourself, what if the patient is infected and the negative is a false negative? What could be the result? Are there high risk people that can be exposed? Could this lead to school closure?

The CDC initially recommended that asymptomatic exposures not get tested. I originally thought that this was because of the concerns that people will have a false negative result and perhaps break their quarantine. It turns out on 9/18 they changed the recommendation. Apparently

They recommend testing if you develop symptoms so you can be monitored for severe disease, your close contacts can quarantine to prevents spread, and so you can know when you are no longer at risk of spreading the virus. I was giving them the benefit of the doubt. I was wrong. Apparently there was political pressure from the federal government to make that recommendation. I’m not sure what their motivation was, perhaps it was the same or perhaps there were other nefarious motives. I can’t begin to guess. Now I get to justify the new recommendation which is easier since it is the one recommended by most public health experts. The reason that close contacts should get tested is that, since so many people are asymptomatic or presymptomatic, testing could identify a lot of unwitting spreaders of disease and isolate them. This would be more beneficial than false negatives spreading the disease would be a detrimental.

In medicine things get better with more time and more research but the speed of the pandemic has made perfect (or often even very good) research an impossibility. We cannot wait for more data because questions are in the present and answers need to be given with imperfect information. The CDC and other public health authorities can make general guidelines but the number of specific personal scenarios is infinite.

The bottom line is that we’re all doing our best in a very difficult and uncertain situation. We have no choice but to live with the uncertainty. Anyone who speaks with certainty about this virus or pandemic is almost certainly wrong. Be kind to the doctors who answer the questions; have mercy on the CDC; don’t expect certainty; don’t expect everyone to agree. Also, don’t think that the lack of clear guidelines and occasional contradiction is a sign of incompetence or a conspiracy. Life is messy and complicated.

It is clear that there are higher powers pulling the strings on this one.

More Reading:

4 thoughts on “Is COVID testing reliable?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s