Category Archives: Short Tips

Short Management Tips

Covid-19 Testing Scarcity: A Self-Inflicted Wound

It seems obvious that the smart way to control of an epidemic is to prevent infected patients from triggering exponential chains of secondary infections. It also seems obvious that to do this we must first find infected patients. So, this is how we fight the war against Covid-19 today. We try to optimize the use of a single test to find a single infected patient. We focus on improving the accuracy, cost, and speed of this individual  test. And, we’ve done a great job; we test with a precision, efficiency, and speed that our predecessors could never dream of. Yet, there is a dark side. By focusing on the one test/one person problem, we have neglected another critical problem. This is what we might call the many tests/many people problem, a problem that is vitally important during an epidemic. What may not be obvious, is that the many people problem is very different from the single patient problem, and it has a very different solution.

Optimize a Single Test or A System of Many Tests?

Let’s use a simple analogy. We know that a chain is as strong as its weakest link. One way to make a strong chain is to separately test each individual link in the chain. We reason that if all the links are strong, then the chain will be strong. This frames the testing problem as a one link/one test problem.  A different way to view this problem is to see the chain as a system composed of many links. Viewed this way we would realize that we could attach 100 links together and test them in a single test. This allows us to use a single test to establish 100 links are good. The many links/many tests problem has a different solution than the one link/one test problem.

What does this have to do with Covid-19 testing? By now we have done about 20 Million Covid-19 tests worldwide. Most of the time, our test will establish that 1 patient is negative. Less frequently, it establishes that 1 patient is positive. What is certain is that we never identify more than one negative per test. What would happen if there was a way to identify as many as 10 to 30 negative patients in a single test? We would increase our testing capacity by a factor of 10 to 30. The 20 million tests we’ve already done could have done the work of 200 to 600 million individual tests.

The Magic of Sample Pooling/Block Testing

The method for producing more results per test is already in use. It is called sample pooling or block testing. It has been described in the Journal of American Medical Association. It has been reported on in the New York Times. It is used in Germany and Korea. It works by combining samples from 10 patients in a single batch and testing this batch in a single test. If the test is negative, which happens most of the time, it has identified that 10 patients are negative in a single test. If the test is positive, which happens less frequently, we need additional tests to determine how many patients were positive. The crucial difference is that we need much less individual testing. If a disease is only present in 1 percent of the population, 90 percent of the time a pool of ten tests will be negative, necessitating no additional tests. Sample pooling is perfectly suited for Covid-19 PCR testing because any dilution caused by combining 10 samples is trivial in the face of  the power of a PCR test to amplify the presence of DNA by a factor of a million.

What’s Stopping Us?

A key question is, why don’t we use this higher productivity approach to testing in America today? Quite simply, because our clinical laboratories are required to follow testing procedures mandated by the FDA and CDC. A clinical laboratory can lose its certification if it does not follow mandated procedures. Unfortunately, the current test prescribed by the CDC, , dated 3/30/2020, CDC 2019 Novel Coronavirus (2019-nCov) Real-Time RT-PCR Diagnostic Panel has no procedure for pooled samples, it only permits testing individual samples for individual patients. In other words, inefficient Covid testing is mandated by the US government.

We pay a high price for this inefficiency. Because we require 1 test per patient, we have a scarcity of tests. Because we have scarcity of tests, we focus these tests on people who already show Covid-19 symptoms. Since about 20 percent of Covid-19 infections are asymptomatic, this leaves thousands of untested people spreading Covid-19 through our communities. And, since scarce testing prevents us from locating the sources of infections in our community, we resort to brute force approaches like locking down our entire economy.

Even worse, in addition to forcing lockdowns, inefficient testing cripples our ability to reopen the economy. Reopening restarts the free movement of asymptomatic or pre-symptomatic people in a large pool of the uninfected people. This is only workable if we can find new infections quickly, trace their contacts, and quickly isolate them. While we can’t prevent all chains of infections, we can keep these chains short by shutting them down quickly. And, what does it take to find chain quickly? Frequent testing. Frequent testing is the key to preventing new waves of infection, and testing efficiency is what makes frequent testing cost-effective.

We Can Do Something

In fairness, the choice of the CDC and FDA to mandate inefficient testing is not motivated by malice. They are keenly aware of the danger of trying to fight an epidemic with unreliable testing and they are simply trying to do their job and promote public welfare. Test scarcity is simply an unintended and perhaps unexamined consequence of their choices. Now is a great time to change this and save lives.

Don Reinertsen

Boiled Frogs and Thought Experiments

Aristotle understood the value of thought experiments. They can be done quickly without get your hands dirty. Measurements are always precise and consistent. Results are perfectly repeatable. So, why bother with a real experiment when you can do a thought experiment?

Silly Aristotle. We are much smarter than that now. Perhaps. Management thinkers constantly quote the example of the boiled frog. Put a frog in boiling water and it will jump out. Slowly raise the temperature of the water and it will fail to notice the gradual change and boil to death. Since most of us lack both the frog and the insensitivity needed to do this experiment, like Aristotle, we accept the results of the mental experiment.

What would a real experiment reveal? The frog dies if placed in boiling water. The frog jumps out when the water is heated slowly. What happens if the frog is placed in freezing water? It will die. What happens if the temperature of the water drops slowly, such as from summer to winter? The frog will activate genes that produce metabolic enzymes that function at low temperatures.

As any dinosaur could tell you, slow change gives us time to react, rapid change can lead to extinction.


Fast Company: “Next Time, What Say We Boil a Consultant.” November 1995 (p. 20) The “boiled frog” story.

Bad Outcomes vs. Bad Decisions

Only a fool assumes that bad outcomes are always a sign of bad decisions. If you play Russian Roulette with 5 of 6 chambers full, and don’t blow your head off, you still made a bad choice.

When we get a bad outcome based on a good choice, we often incorrectly “learn” that we made a bad choice. It is even worse when we get good outcomes from bad choices — we “learn” that we made a good choice. Real learning requires that we clearly distinguish the decision from its outcome, and such learning is critical in a stochastic world.