Why We Need to Quantify the Risk of COVID-19
Philadelphia reopened indoor dining in September only to reverse course this November. With policy in flux, it’s no wonder my family does not agree if indoor dining is safe. “I won’t go” my brother protests to the sarcastic reply, “If you want to be perfectly safe, then never leave the house.”
As state and local governments reopen the economy or close business down, they warn that the public is safer at home. Governments have asked individuals to navigate this hazy space between the permissible and the advisable without a clear decision-making framework to guide responsible behavior. As a result, decisions are informed by fear, social pressure, and public health recommendations with a spotty track record. Like so many others, I am left wondering who in my family is right. Is indoor dining safe?
Public health guidance is frustratingly qualitative. The CDC advises, “travel increases your chance of getting and spreading COVID-19. Staying home is the best way to protect yourself and others.” But the CDC does not say by how much travel increase one’s chances. Without this information, one cannot decide if a valuable trip — like Thanksgiving with your family — is worth the risk.
Or take Arizona’s “COVID-19 Risk Index,” one of many similar graphics published by health departments. Working out at an indoor gym is “high risk” while staying at a hotel is “moderate-low” risk. These qualitative statements are not helpful because they compare unfamiliar unquantified risks. If you were dropped into a foreign country and a local merchant quoted the price of a gym visit at 50 Vinders and a hotel stay at 20 Vinders, you would know that the gym is relatively more expensive, but you still need the exchange rate into dollars to know if either were a bargain.
The public needs quantitative estimates of the risk of contracting, spreading, and suffering injury from COVID-19 across a range of daily activities. Rational individuals weigh the costs and benefits of their actions. Given the probability of infection from an average indoor restaurant meal, one can calculate the risk of harm and decide if dinner is worthwhile.
Our public discourse is increasingly broken without these quantitative estimates. Millions of Americans have hardly left their home in months while others attend parties without precautions. Individual behavior may vary based on health profiles, risk tolerances, and other preferences. But, without an estimate of the danger of various activities, it is difficult to gauge who is overly cautious or reckless.
If a friend confided that he was scared of (pre-pandemic) plane travel, I could cite government statistics that show the risk of a plane crash in the U.S. is substantially less than one in a million. The risk is so low that even when multiplied by the value of human life, the result is a very small cost, and this plane crash risk should not dissuade you from any worthwhile trip.
Of course, frequent flyers do not make this calculation when boarding a flight. Flying is a common activity, so we assume that status quo safety precautions reasonably balance costs and benefits. However, when faced with a novel threat like coronavirus, we should not assume society has achieved an equilibrium and we must explicitly model costs and benefits.
To be fair, public health officials may be unable to communicate the probability of contracting and spreading COVID-19 because they lack sufficient data. But without an estimate, one wonders if policies are informed by cost-benefit calculations. As public schools weigh reopening, administrators should weigh the risk of harm to students and teachers against learning benefits and try to clearly explain this balance to the public.
Recently, Arnold Barnett, a professor of statistics at MIT, published an estimate of the probability of contracting and dying from COVID-19 on a flight. The public needs more of these kinds of estimates for a full range of daily activities. Adjustments should be made given one’s age, local community case levels, and mask compliance rates. Even if data is not perfect, scientists should put forward their best estimates to be iteratively refined. A rough estimate of risk is better than no estimate at all. Right now, we are all flying blind.