What now for COVID-19? Here is one idea. It may be right, but untenable?

As a researcher, I had two habits. First, rethink everything, and, second, the best research ideas are 180 degrees opposite the prevailing. So, a good place to start to think about what to do next about COVID-19 is to think differently. This blog is about one alternative, I will propose others. 

But thinking differently must aim at a core problem. It does no good to rethink something that did not need thinking about in the first place. In my view, there are two main problems with COVID-19 activities that need rethinking. First, how to let science inform us, and how do we foster confidence in people who must decide when to return to life in full. 

Counter idea one. Reduce testing to 0% in the general population, and the ill. (The present science about the consequences of COVID-19 is bad, so, let’s stop reporting consequences until the science improves. Stopping testing and reporting will reduce the pain of hearing about new cases that may not even be cases. The reduced reporting will slowly build confidence in the public). 

There are three reasons to consider changing the prevailing idea of more testing to less. 

First, the science of prevalence/mortality must, before reporting numbers, carefully consider who to test, when to test and, where to test. There needs to be a systematic plan, and we have none. It does not help to report prevalence or mortality data if one area of the country tests only sick people in ICUs and another tests only in the work force. Haphazard testing patterns lead to incomparable numbers of outcomes for both those who die and those who do not.

To illustrate, suppose in my town, we test every person, 30,000, and find that 1000 have a positive test. The prevalence of COVID-19 in my town, then, is 1000/30,000 or 3.3%. However, one town over decides to test only people with symptoms, and assume 1000 had symptoms and 100 of those had COVID-19. The prevalence in my neighbor’s town is, then, 100/1000 or 10%. Comparing these numbers is meaningless. Publishing numbers without a systematic plan may falsely alarm, or sooth, but both emotions are unfounded as the numbers are confounded. 

Second, testing should stop because the test is not good. A “good” test will be positive in those with disease, and not in those without. A perfect test, then, finds 100% of diseased patients and falsely detects 0% of non-diseased people. But, tests for COVID-19 are imperfect. The rapid tests were approved under an emergency provision at the FDA. This means they did not have robust clinical evaluation to determine how well the tests perform. The estimates of their diagnostic abilities have been examined only in laboratory situations. Hence, the accuracy of the tests are likely worse than reported. The estimated false negative rate is 1%, but the false positive percent (perhaps due to cross reactions), is 10%. (https://www.medicaldevice-network.com/news/fda-bodysphere-covid-19-test/) 

Why are false positive tests a problem? An example in another epidemic informs. In the 1980s people applying for marriage in Illinois were tested for HIV. The prevalence of HIV was about 4/100,000. The test for HIV detected all 4 cases of true HIV, but nearly 40 normal people were told the test was positive (false positive percent was 4/10,000, much better than the false positive for the COVID-19 test). For every 44 positive tests, 4 had HIV, 40 did not. The consequences of this policy, ordered by edict, were horrible, and the policy was rescinded.

What did this epidemic teach us? When the prevalence of a disease is low, there are lots of non-diseased people susceptible to a false positive test. We are unsure of the prevalence of COVID-19 due to the haphazard plans for testing noted above, but in Iceland about 5-6000 well people were tested and 48 had positive tests, about 1%. Assume the prevalence is 1% (1 with, 99 without). The COVID-19 test will pick up the 1 in 100 with COVID-19, but 10% of the other 99 people without will have a positive test, or about 10 people. So, for every 11 people with a positive test,1 will have COVID-19 and 10 will not. This is a best case scenario for the test, also. Some of the 48 who tested positive in Iceland were false positives so the ratio of false to true positives is likely worse than 10 to 1. 

Imagine the consequences if we base a public policy on a 90+% error rate? Just think how nice it would be to not wake up to, “1000 more cases of COVID-19 last night” especially if most of the positive tests are wrong. 

For sure, like any epidemic, seasonal flu, or H1N1, COVID-19 is dangerous. We can’t deny the fact that people are very ill. But, our ideas about who should be tested, or who should not be allowed to come into contact with others will be partially based on the truth of the prevalence in people who are well. Consider a workforce, if all workers were tested, about 10% could be falsely positive making it hard to imagine how a company might want employees to return to work, or how often they would have to keep applying the test?

Third, the reason to test for a disease is to treat if disease is present. In this epidemic the treatment is isolation, and we are already treating everyone as infected. We ask people to stay at home, wash hands, don’t touch your face, wear a mask, if sick, stay away from people and, if well, don’t approach sick people. These are universal precautions for all infections, not just COVID-19, and if we are already treating everyone as if they were positive, we don’t need a test to tell us otherwise. (I will not address another important issue about screening; we must have a useful treatment when disease is found, and it is unsure if we have a useful treatment). 

Now, what about testing in people who are ill? People with symptoms are more likely to have COVID-19, but other diseases as well. In other words, the prevalence of COVID-19 is greater. Now, the testing issue is not the false positive, but the false negative. When the prevalence of a disease is high, there are lots of people susceptible to the false negative, and how many missed, false negative people does it take for a contagion? The description of the index case in Washington State is illustrative. The person, coming from Wuhan, knew they were ill, knew of the illness in Wuhan, got off the plane, masked, went directly to care, was isolated, and contacts, as many as 60, were identified, tracked down, tested, and isolated. This is classic epidemic treatment, isolate the case and contacts. But, the contagion still spread. In other words, maybe it only takes one to slip through the cracks. 

So, any non-critical illness should be isolated and assumed to have COVID-19. All critically ill people, alternatively, will be cared for just like anyone who is critically ill with a possible transmittable disease, under hospital infection policies, including isolation. In other words, if we are ill, critical or not, let’s assume we have something transmittable and act accordingly. 

Also, if we are going to test for COVID-19 in the ill, we should also test for other infectious agents, like flu. Co-infection is likely. If we only report on COVID-19 tests in the ill, we inflate COVID-19 as a cause of death and doom. If we don’t test for competing diseases, like flu, COVID-19 information is false information. 

So, if we can’t stop the contagion completely, and if we are acting on inflated, unscientific numbers, let’s stop testing.

Why I might be wrong. 

The mathematical basis for the counter idea is well known. Hence, we can’t progress in our understanding of what to do with this epidemic with poor tests, poor test strategies, and haphazard interventions.The truth is that we have an imperfect test, comparisons, and interventions because we have an imperfect prevailing idea of how to measure the burden of COVID-19.

But, even if this counter idea is correct, it won’t be followed. A dear friend of mine said, nice idea, Bob, but it won’t be followed, so it is just a useless idea for helping people understand the epidemic. He may be correct. While I wrote this, I listened to news reports and websites all saying, more tests are needed. That idea is so prevalent that it dwarfs the prevalence of COVID-19. However, I think the testing plan presently in place feeds fear with inflated numbers that get lots of press, but little accuracy. Accuracy is what we need, and if this haphazard testing and reporting goes on, I sometimes wonder what it will take for the public to feel like they can reenter the world without flinching at every passerby.  Businesses will ultimately open, jobs will return, but when will we return to them?

So, in the next blog, I will float another counter idea. Maybe somewhere along the line something will make sense to all of us. We need new thinking, our present approach is not sustainable, if COVID-19 is.