RALEIGH – There has been a lot of emphasis on testing lately. An apparent epidemic with an uncertain trajectory tests to create a demand for testing, but how we are testing and what we are doing with the information is important. Though total confirmed cases of SARS-CoV2 infections are advertised are advertised daily, the testing pool tells us very little about the full extent of the pathogen’s spread. In lieu of testing everyone, which isn’t plausible, an alternative would be the random testing of people through out the state, and extrapolating more telling information about where the virus is spreading, among whom, and what the varying rate of effects are.
Andy Jackson of the Civitas Institute is suggesting exactly that, because, despite the headline numbers, we don’t really know how many people have the virus. He argues the current ‘paucity’ of reliable data is leading to an over reliance on models that lead to more drastic fears and actions.
“[…] To be sure, there are more than 2,870 actual coronavirus cases in North Carolina. But how many?
Once again, we don’t know.
Give the paucity of actual data on coronavirus, decision-makers have depended on models that predict a range of possibilities based on how they assume the virus will spread and how it will affect people. But models are not reality. Erica Thompson of the London School of Economics and the London Mathematical Laboratory is aware of the danger of over-reliance on models, nothing that they can create a “fairy-tale state of mind.” Public officials can forget that they are “only models” and place too much confidence in them, forgetting that their underlying assumptions can fall apart when you “move back to reality.”
Humans also have a well-documented negativity bias; our minds tend to gravitate towards the grim. That explains why so many paid such close attention to an “entirely fanciful” March 16 Imperial College model projection of 2.2 million deaths in the USA.
The net effect of all this is that, right now, decisions are being made at all levels of government in North Carolina based on assumptions and fear as much as data.
While modeling has an important place in decision-making, it is not a substitute for real-world data. That is why we need to start random testing of the North Carolina population for coronavirus.
Aside from telling us the prevalence of coronavirus in North Carolina, random testing could also tell us if it has spread across the population or is concentrated in clusters within the population, information that health officials and policymakers can use to craft the best response to the disease.
Iceland offers an example of the importance of random testing. Official government data, which includes tests of suspected Coronavirus cases at the National University Hospital of Iceland, found that seven percent of those with coronavirus are asymptomatic. However, random (if imperfect) testing by the private biotech company DeCODE has found that the actual number of asymptomatic coronavirus carriers in the general population could be as high as fifty percent.
We cannot continue shutting down large segments of the economy without causing catastrophic damage to people’s lives and inflicting misery. We will have to allow businesses to reopen and relax social distancing at some point in the coming weeks. But exactly when should we do that? To know the answer to that question, decision-makers need data on the general population. That is why countries such as Norway and states such as Ohio are starting random testing.
It is long past the time to start random testing for coronavirus so that public officials can use actual data from the general population to make informed decisions. An added benefit to random testing is that the data it would provide would improve the models decision-makers are currently using.
While statistical analysis is not perfect, it would let officials make decisions based on real-world data. So, what’s to fear?”
Fear itself? The fear of this virus, the fear that has led us to sabotage our own economy and livelihoods, is driven by uncertainty and pessimistic guesstimates. The only thing to address this is with reliable information.