Public health and medical experts have clamored for increased coronavirus testing to control the coronavirus 2019 (COVID-19) pandemic,1 but how to implement such testing has not been carefully described. We detail a plan for aggressive community screening with the goal of controlling, if not eradicating, local severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) transmission. For at least 2 months, access to reagents, swabs, and tests have been a bottleneck. We believe this bottleneck is abating, as evidenced by recent dramatic increases in testing. The limiting factor, then, is a matter of operations and logistics. In the absence of an effective vaccine, separating persons who are unknowingly infected from individuals with susceptibility is the best way to block transmission. Ongoing community transmission demands an aggressive approach to achieve this goal.
Our proposed screening approach is designed to maximize the number of persons with infections who are detected. Three complementary testing strategies together can achieve this goal. The first is to remove physician referral and cost restrictions on the testing of individuals with symptoms, so anyone with symptoms suggestive of COVID-19 can be screened. Many such persons will have negative test results, but among individuals with symptoms of COVID-19 whose cases have not yet been detected by other means, this option offers the quickest way to isolate them and thereby prevent further transmission to individuals who are susceptible. Existing and additional drive-through testing sites are useful for this purpose but need to be augmented in other ways in poorer communities, where reliance on public transportation in urban areas or a lack of large health centers in rural areas makes testing less accessible.
The second and most intensive approach is to deploy targeted testing in locations where persons more likely to be infected can be found. Such sentinel sites will be initially selected on the basis of population density within dwellings (eg, nursing homes, public or congregate housing, apartment buildings) and by community (population per square mile), as informed by local information on COVID-19 cases, hospitalizations, and deaths. Testing sites would also include locations where residents may congregate, such as grocery stores or pharmacies. Testing teams would be agile and able to move on a daily basis between locations identified for targeted testing.
The third approach is to reserve a small number of tests for random population screening. This approach provides the benchmark for assessing whether sentinel sites are efficiently detecting persons with infections. Sites where the yield is not substantially higher than the estimated population prevalence would be abandoned in favor of other high-risk sentinel sites. Updating screening sites in this way will always direct the program toward areas with greater numbers of persons with infections, thereby maximizing testing efficiency.
Others have proposed massive random screening of the US population with up to 30 million tests per day. Our view is that targeted screening can quickly find sentinel sites where the current prevalence of infection is substantially greater than the mean. For example, a recent investigation2 found that 15% of pregnant women presenting for delivery at a New York hospital tested positive for SARS-CoV-2, and 29 of the 33 with positive test results were initially asymptomatic. Finding other such hot spots could allow for dramatically increased testing efficiency; testing 10% of the population at random each day could be reduced to testing 1% to 2% in a targeted manner.
Our approach still requires collecting millions of samples and corresponding contact information. Those found to have infections would be notified within a day and advised to isolate at home. Recognizing that home isolation is not always feasible, public accommodation of isolation quarters would be offered to those not sufficiently ill to require hospitalization or subacute care but who cannot isolate at home. The outcome of our plan would be substantially diminished if isolation does not follow identification.
Some might worry that imperfect test performance would misdirect our targeting efforts, but such fears are unfounded. For community screening, we focus on polymerase chain reaction testing, because the goal is to find persons with current infections as opposed to signals of past infection indicated by antibody tests. Letting α and β denote the sensitivity and specificity of whatever diagnostic test is used, the observed prevalence pObs is associated with the true prevalence p, in that pObs = p × α + (1 − p) × (1 − β). The corrected prevalence estimate from testing, pCor, is given by pCor = a + b × pObs, where a = (β – 1) / (α + β – 1) and b = 1 / (α + β – 1).
Selecting sentinel sites with the largest observed prevalence (pObs) will yield the same site selections as choosing those with the largest corrected prevalence (pCor), which are the best estimates of the underlying true prevalence at each site. This will be true as long as the test sensitivity and specificity remain consistent.3
We have not emphasized contact tracing in our proposal. Contact tracing is absolutely needed at the beginning or end of an outbreak for containment. South Korea has used contact tracing, including the use of phone apps, to great effect. But even South Korea could not use contact tracing at the peak of their outbreak (which by comparison was a fraction of the population-adjusted outbreak in the US).
Finally, this entire effort (and virtually all community testing) will be wasted if those found to have infections cannot be convinced to isolate in their homes or elsewhere. While the Centers for Disease Control have published guidance on this matter, these rules may not be understood by many members of the public. A well-run public awareness campaign in neighborhoods where infections are most likely is necessary. Perhaps a fund akin to jury duty reimbursement is required to cover the loss of income that could be incurred during home isolation.
We believe our plan provides the best hope of ending this outbreak and escaping from repeated cycles of imposing and releasing stay-at-home restrictions. Earlier modeling4 has shown that effectively isolating persons within 5 days after infection can indeed prevent sufficient transmission to bring the reproductive number of SARS-CoV-2 to less than unity. We need to prepare now, so this plan can be implemented as soon as feasible. Failure to do so may surrender our best hope of ending this outbreak.
Corresponding Author: Howard P. Forman, MD, MBA, Yale School of Management, 165 Whitney Ave, Evans Hall, New Haven, CT 06511 (howard.forman@Yale.edu).
Conflict of Interest Disclosures: None reported.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.