It’s a simple law of clinical laboratory testing that everyone wants a clinical lab test to be cheap, fast, and accurate — but you can only ever have two of those traits. Pick the two you want, but the third is out of reach. A corollary to this law: Lab people like me are nearly always stalwarts for accuracy, and those who don’t work in a lab are often willing to sacrifice accuracy for speed and cost.
This is the context in which I’ve been reading the increasingly strident calls for increased COVID-19 testing. Lately, these have emphasized fast and cheap tests that are not particularly accurate.
I’m no better at predicting the future than anyone else, so I can’t say exactly how things might change if we go this route. But I will offer some perspective gained over 2 decades in the clinical laboratory and especially over the last 6 months, overseeing a massive COVID-19 testing response in an academic medical system.
Doing a bad test daily is not the same as doing a good test once. Repetitive testing can indeed increase the sensitivity of a test, meaning the ability to detect COVID-19 if you have it, but always at the expense of specificity — which means more false-positive results. This is how the math of testing works.
Compounding this problem is that test manufacturers that have acquired “emergency use authorization” by demonstrating acceptable performance of rapid, convenient, or cheap diagnostics (antigen, saliva) with data comprised mostly of selected patient samples that contained high viral loads. Thus they are proving that their tests work on the easy cases and performance will likely disappoint when they fail to detect patients’ coronavirus at lower levels.
In that way, a “rapid” COVID-19 antigen test functions like a drugstore pregnancy test that requires a billion molecules of human chorionic gonadotropin, the pregnancy hormone, to be present for the stripe to turn blue, indicating positive. We have technology to amplify the sensitivity of antigen tests, but even so, they can never match the performance of gold-standard polymerase chain reaction (PCR) tests, which detect a positive viral result from a little over 100 SARS-CoV-2 particles.
Also worth noting: The emergency use authorization for antigen tests allows testing in symptomatic people only.
The calls for ubiquitous implementation of cheap and fast tests stretch the imagination even of people like me, who have more than a passing interest in the creation of cheap tests.
There is much more to a clinical lab test than a chemist making a white spot turn red. The path from where we are now to $1 paper-based tests, performed at home with a drop of saliva, is long and expensive at best, and maybe imaginary: Abbott’s BinaxNOW is supposedly available for $5, but it still doesn’t work on saliva, and I don’t know anyone who has it in stock.
One other salient bit of math: BinaxNOW, by my calculation, is about 100,000 times less sensitive than PCR tests.
Consider the fact that the rationale for cheap, inaccurate testing comes almost entirely from mathematical modeling studies, not real-world experience. There is an old story about a dairy farmer who goes to the local university complaining that his cows’ milk production has decreased. Several university departments work on the problem, to no avail, until the chair of the physics department announces, “I have solved the milk production problem. However, it applies only to spherical cows in a vacuum.”
The current rash of recommendations for cheap testing may similarly apply only to robots in the dreams of mathematicians, and not to our real families and friends.
The COVID-19 testing process has slowed appreciably. We are experiencing an imbalance between testing supplies and demand, which can be resolved by increasing supplies or decreasing demand. The latter is hard to imagine, with so many individuals, populations, and politicians willing to ignore public health guidance for masking and social distancing that would slow the pandemic.
Unfortunately, the rush to increase supplies has fueled an emergence of bad tests. This leads me to ask: Do you think it is a good idea to tell people infected with COVID-19 that they don’t have it? Telling people that they are virus-negative almost certainly emboldens behavior that undermines public health.
In any event, boosting production of test supplies at a linear rate gives us little hope of matching the demand of unchecked virus spread. The pandemic will not be ended simply by producing more test technology.
And to those who still believe that daily testing is the Holy Grail? Ask the Miami Marlins how that worked out. Positive coronavirus tests among players haven’t shut down the baseball season, but consider that these are healthy young athletes, not nursing home residents.
COVID-19’s spread is due to human behavior. Short of a vaccine, it will be stopped only by changing human behavior and addressing the demand side of our national imbalance. Introducing a bunch of subpar tests to the market might enrich several diagnostics manufacturers, but the rest of us will be left, as they say, to rearrange the deck chairs on the Titanic.
Geoffrey Baird, MD, PhD, is interim chair of Laboratory Medicine and Pathology at the University of Washington School of Medicine.