Statement in Response to U.S. Department of Health and Human Services’ REPORT TO CONGRESS on its COVID-19 Strategic Testing Plan

May 26, 2020
By Danielle Allen and authors of the “Testing Millions” white paper, E. Glen Weyl and Divya Siddarth; all are co-authors on the Edmond J. Safra Center’s “Roadmap to Pandemic Resilience” and the Center’s supplementary report, “Pandemic Resilience: Getting It Done.”

The U.S. Department of Health and Human Services has issued a REPORT TO CONGRESS on its COVID-19 Strategic Testing Plan dated May 24, 2020. That report cites our work, writing:

Estimates about how many tests are needed vary widely based on assumptions such as the prevalence of active cases, the number of contacts per case, the effectiveness of mitigation in the community, the sensitivity of the assays to detect cases, and the overall level of immunity within the community. Estimates in the literature vary from a few hundred thousand per day to twenty million per day. For example, the simulation model developed by the Safra Center at Harvard (Edmond J. Safra Center for Ethics. COVID-19 Rapid Response Impact Initiative - White Paper 6: Why We Must Test Millions a Day. Retrieved from: ) assumes: tests are 80% sensitive (actual average sensitivity of laboratory based nucleic acid detection test is 95%); hospitalization rate is 20% (actual hospitalization rate is <5%); no effect of mitigation (actual at least 35%); days to recovery is 15 days (actual is 11 days or less). If testing can be targeted to likely positive individuals (f = 10), the actual number of tests needed per day is reduced from Safra’s estimate of more than three million to just over 300,000 if the correct assumptions are used. This number is already being achieved through the current testing regimen, and will be far exceeded by mid-summer.

We’re encouraged to learn that the White House has engaged with our Roadmap to Pandemic Resilience (released April 20, 2020) and our accompanying whitepapers.

That said, we stand by our assumptions and our models.

The Department’s Report does not provide an accurate summary of the modeling supporting our recommendations. In both the cited paper, and in a more recent paper, we ran multiple models, with different methodologies, all of which pointed to the same order of magnitude of levels of diagnostic testing and contact tracing needed for disease suppression, defined as returning disease prevalence close to zero. That order of magnitude is 2-6 million diagnostic tests per day (targeted by effective manual contact tracing), with the range dependent on the synchronicity of suppression efforts across various locales. In the Department’s report, the authors have selected for review one non-primary model that appears in an appendix of our first report, and the Report’s authors have selectively adjusted assumptions used in that model to generate a different number. They have not offered a review of the models in the main body of the paper.

Also, the number of 300,000 tests per day that the Department claims is generated from our model when used with revised assumptions is incorrect. Using the Department’s assumptions with the equilibrium model drawn from our appendix yields a result of more than 1 million tests per day. Using the Department’s assumptions with the main model in the body of our paper results in a recommendation of at least 1.5 million tests per day.

Our most updated policy recommendations (released May 12, 2020) are available here; and the modeling supporting them is available here.

We have used the South Korea model to generate our numbers; that model applied to the U.S. generates a higher per capita number than in South Korea itself. This is because required tests scale (roughly) with disease prevalence and prevalence in the U.S. is higher than in South Korea. Also, we have begun broader testing at a much further developed point in the trajectory of community spread. We will continue to advocate for that higher per capita rate of testing that is necessary for disease control in our context given our late start. We are not aware of a single country that has achieved disease suppression with a positivity rate of 10 percent.

With regard to the specifications in our primary model, hospitalization rates are irrelevant, as our focus is on the control of disease spread, not control of rate of hospitalization. We assume a 20% false negative rate (rather than the 5% assumption made by the administration). Given the spread of testing devices in use and significant error in the administration of tests, this has been borne out as a reasonable average. Our R0 is 2.5, which doesn't take into account severe social distancing but 40% of people are essential workers who likely have more contacts than are modeled into a 2.5 R0, and states are already reopening, which makes ours a reasonable estimate. Our supplement available here addresses these disparities in spread, prevalence, and contact rate, and breaks down our national numbers into local guidance based on region-specific estimates.

At the end of the day, the debate about the necessary level of diagnostic testing and manual contact tracing comes down to a choice of strategic objective. Is our goal mitigation or suppression?

The core issue is whether the country is pursuing a mitigation strategy (some reduction in the rate of R through diagnostic testing and contact tracing) or whether the country is pursuing a suppression strategy (seeking to get to zero prevalence). We have been arguing for and modeling a suppression strategy, which our numbers reflect. The administration appears to have embraced a mitigation strategy. We continue to encourage the administration to aim high. Mitigation should be but a stepping stone to suppression.

We are recommending a suppression strategy which other countries have successfully implemented that enabled them to fully reopen without fear of further lockdowns or disease spread.

This is a can-do America and we can build a diagnostic testing and manual contact tracing infrastructure as a powerful tool of disease control to fend off the need for further closures and restore a healthy, vibrant economy unhampered by shutdowns and fear of catching a virus. We have routinely communicated to the White House Task Force our willingness to provide support and assistance to the country’s efforts. We look forward to continuing to work with the White House, Congress, states, and tribal and local governments to help build the diagnostic testing and manual contact tracing infrastructure that will restore our country and make our economy vibrant again.