Diagnostic testing for COVID-19 is highly constrained in the United States relative to other countries, and President Donald Trump has suggested that widespread testing might not even be a necessary step on the way toward opening up the economy. We disagree strongly. Our ability to test individuals will improve drastically in the coming weeks thanks to recent FDA-approved test kits entering large-scale production. Getting the outbreak under control is an absolute necessity before we can move back to a new normal. Now is the time to plan for deploying these tests to gain control of this outbreak.
To date, tests have been used to ensure the appropriate allocation of hospital resources to those critically ill while monitoring health care workers to ensure their safety. Most other tests have been conducted on patients with active and/or progressive symptoms suggestive of COVID-19, or on contact exposures in order to prevent further transmission. These activities must continue for individual isolation and treatment decisions, but now is the time to deploy testing for public health.
The most important goal is to maximize the detection of symptomatic and asymptomatic infected individuals. Asymptomatic (including pre-symptomatic) people have the highest likelihood of further spreading the disease. Once identified, we can isolate them from potential contacts, including the most vulnerable among us.
Screening would pinpoint infections
Doing this requires testing in those locations most likely penetrated by the virus, and we have already learned a lot about these locations. Individuals living in high-density environments — public housing, packed apartment buildings, assisted living facilities — are obvious targets. Screening individuals in such locations is likely to lead to the discovery of more infections than simple random population samples.
Other sentinel testing opportunities are destinations that remain popular, even among those “staying at home” — including grocery stores, gas stations and pharmacies. Targeted screening like this must be agile. There are many more possible testing sites than can be staffed on any given day in this approach, so testing must be highly mobile (for instance, in tents set up in the parking lot by the entrance or exit).
We must make symptomatic testing much more widely available by expanding drive-thru clinics so that anyone with symptoms remotely suggestive of COVID-19 can be tested, rather than the highly restricted symptomatic testing available now. This is consistent with the goal of finding as many previously undetected infections as possible.
A different public health goal is to learn the actual prevalence of infection — the fraction of the population that is infected and where they are — to help steer the targeted screening suggestions above. Learning the prevalence of an infection normally calls for random sampling in a community to ensure representative coverage, and thus seemingly conflicts with the goal of finding as many infected persons as possible. Yet these two approaches need not conflict, and indeed with just a little additional information gathering, targeted screening can complement learning community prevalence.
Take it from a doctor:Fake news about the coronavirus is hazardous to your health. Don't fall for it.
Consider screening in grocery stores or pharmacies. In addition to learning who is being tested and where they live, screeners should ask, “How often do you visit this store?" That will help us correct for oversampling one group (people who go grocery shopping every day versus those who might go once every two weeks). We should also capture information on race, gender and age, so we can more quickly identify communities where hidden outbreaks are emerging before patients present with severe symptoms.
We favor allocating a small proportion of tests for representative screening while reserving the bulk for active targeting: finding as many infected persons as possible. This is the very familiar “exploration/exploitation” trade-off: Some resources must be deployed to explore the community via random sampling, while a much larger share must exploit that knowledge via sentinel screening.
Plan now for critical COVID-19 testing
Simply finding infected persons is of course not the end of the battle. Such persons must be isolated, in most cases at home. This raises the stakes in making sure that people understand just what effective home isolation entails. Detailed public health guidelines from the Centers for Disease Control and Prevention, state and local health departments, medical practices and employers exist, but it is not clear how well these are understood by the population at large. Very clear and engaging messaging is needed for aggressive community screening to succeed.
In some cases, circumstances force many people to share cramped quarters, making home isolation simply infeasible. For these people, we need publicly funded isolation quarters in requisitioned space. Such individuals would not be so ill as to require hospitalization or sub-acute care in field hospitals, but they would need to be removed from their homes to prevent infecting others. This must be done with maximum compassion and respect for individual rights.
Harvard epidemiologist:Coronavirus ‘is the Big One. … I hope never to see bigger’
Finding the majority of infected individuals and isolating them effectively is not merely a Band-Aid. Research shows that detecting and isolating a large majority of infected people within five to 10 days after becoming infected would end this outbreak. Absent a vaccine or reliable therapeutic treatments to lessen the consequences of COVID-19, aggressive community screening and concomitant isolation of those found infected offer our best and perhaps only hope of ending this outbreak without destroying the economy, the health care system, the public’s confidence and/or many thousand lives.
We have only a few weeks until the new tests arrive to plan the logistics and execution of this model. The time to build this system is now, so we can launch it, immediately, once the new tests are in hand. Everything remains at stake until we address this operational issue.
Edward H. Kaplan is the William N. and Marie A. Beach Professor of Operations Research, Professor of Public Health and Professor of Engineering at Yale University. Dr. Howard P. Forman is a Yale professor of management, public health, economics and radiology. Follow him on Twitter: @thehowie