Testing Targets: October 1, 2020
From Diagnostics to Screening for COVID-19: Introducing the U.S. Testing Calculator
Brown University School of Public Health and HGHI update state testing targets with NPR | New screening approach to test more asymptomatic people
How many daily tests are needed to frequently screen for COVID-19 and keep people safe in schools, prisons, nursing homes, among grocery store and restaurant workers? How can we democratize testing? With our new tool, you can find out. We have developed the COVID-19 testing calculator, which estimates the number of tests needed based on a chosen screening strategy, to support decision makers in refining strategies for this type of testing.
In the coming months, the U.S. needs to be performing millions of tests each day to open and keep open the economy, send children back to school safely and finally suppress the coronavirus, The arrival of millions of new antigen tests – such as the 150 million tests for nursing home testing the administration announced on Monday – and other new testing modalities will allow cities and states to gradually increase their testing and expand from efforts to test mostly people with symptoms to regularly and strategically test more and more asymptomatic people in critical settings. This screening approach to testing – for example, let’s test all school children in Colorado once a week – is supported by our COVID-19 testing calculator to inform decisionmakers on the number of tests needed.
The testing calculator is an update to our previous testing target estimates, as published with NPR and visualized with the support of the Microsoft AI for Health team via our testing target dashboards (see top of page). It provides a more granular way to illustrate how millions of tests can be applied to suppress the virus.
“As the pandemic evolves, we have seen that the science evolves. It is crucial that we continue to adapt to this change and have a full understanding of the COVID-19 pandemic,” says John Kahan, VP, Microsoft Chief Data Analytics Officer and global AI for Health lead. “We have partnered with the Harvard Global Health Institute and now Brown University School of Public Health on this journey to provide visualizations on their new mitigation targets and surveillance approach of high risk populations as well as a calculator to allow US states to adapt their local strategies to the changes in the future.”
In a new story out today, NPR reports on the calculator and our updated recommendations for a basic scenario and an ideal scenario for this screening testing approach.
We built the calculator and these recommendations to help illustrate the trade-offs decision makers face as they consider how to distribute increasing testing capacity. Too often in the current environment of market-driven testing distribution, those who have the means (sports teams, elite universities, companies on Wall Street) can afford this screening testing approach, while cities and states struggle to build the infrastructure and regularly screen in critical contexts as they lack support and funding from the federal government and congress.
Such support could also address some of the key challenges we currently see as the new tests become available, such as the need for better coordination between federal and state efforts, for consistent reporting of results, and for studies that can help us assess how well different antigen tests are performing in a screening environment.
The testing calculator is currently available for download as an Excel file. A more accessible version will be added to our testing dashboard in the coming weeks.
Methodology
The calculator is built on the latest available data on how many people currently are members of each of the population groups that should be considered in this strategic screening approach, such as school children, healthcare workers, prison inmates, nursing home residents, essential workers, frontline workers, and others. We selected these populations based on the latest evidence about where and how the virus spreads easily, and current best practices for coronavirus screening.
The calculator allows exploration of U.S. total numbers as well as specific numbers for each state. For example, the calculator includes the number of school children per grade in any given state, as well as teachers and staff; and the information can be accessed on the state level or on the national level.
For each population group included in our screening tool – such as school children, university students or essential workers – the calculator allows users to pick from three different strategies:
- no testing
- moderate screening (one test per person per week)
- active screening (two tests per person per week)
Wherever relevant, the calculator also allows for another level of stratification: K-12 students for example, are sorted into grade groups and listed based on our risk levels map. Risk levels help assess the level of community spread, which impacts recommendations. In general, the higher the community spread, the more screening testing is recommended. To build an effective and granular strategy, officials can pick moderate screening for students in green or yellow risk levels, for example, and active screening for students in orange (schools should move to remote learning when their communities have reached the red risk level and therefore weekly testing of students is not necessary.)
The calculator also allows for making proportional adjustments to the number of university students and retail/restaurant workers who need testing, with the understanding that not all restaurants have reopened and not all students have returned to college campuses. In our baseline model of minimum screening needed we assume that about 25% of college students have returned to campuses and that half of retail and restaurants have reopened, though these assumptions can be modified when using the calculator. These adjustments simply deflate the population sizes of these groups based on the selected proportion.
In adding up the numbers, the calculator assumes that testing will be spread evenly across each week, thus daily testing can be obtained by multiplying the size of a group by the number of weekly tests for that group and then dividing by seven.
Once decisions about screening tests are made and numbers of needed tests per population group are displayed, the calculator also adds in the estimated need for diagnostic tests (tests of people who have symptoms) based on the projected number of infections in each state, using our previously established methodology.
Five Scenarios
The calculator allows for the exploration of a wide number of strategies. As funding for this type of testing remains in question, we have outlined five scenarios that illustrate trade-offs and decision-making points as jurisdictions decide how to apply a limited number of tests for maximum suppression and economic relief.
Our base scenario tests school teachers, inmates, university students, and first responders once a week, and nursing home residents and staff twice a week. We prioritize teachers over students given the relatively lower risk of serious illness in children and include inmates, nursing home residents, and emergency service workers given the significantly elevated risk of infection that they face.
While this baseline scenario allows for minimizing the number of daily tests needed, it does not include all of the populations we believe may require regular testing for at least two months to suppress viral spread. Our scenarios 2-4 build upon our baseline scenario by adding in specific other groups that may benefit from screening.
Scenario 2 prioritizes education by adding in testing for K-12 and university students. Scenario 3 excludes students but adds testing for restaurant and retail workers on top of our baseline testing. Scenario 4 excludes students and restaurant and retail workers but adds testing for healthcare workers.
Our final scenario is an ideal scenario in which we test all of the groups included in the calculator at levels that could effectively suppress viral spread. This is the scenario that would allow us to find, control and suppress the virus while opening the economy and getting our lives and work back on track. For a limited time – until the virus is under control in the entire nation — it will require a substantial expansion of testing capacity and infrastructure. We know many states are trying to build up this infrastructure and capacity but need more coordinated support from government agencies and Congress.
Delaware leads the way
The state of Delaware is leading the way by making free testing available for all in the state, regardless of symptoms. An Interstate Compact project involving 10 states and supported by the Rockefeller Foundation is another example of states working hard – and in this case, together – to create the capacity needed to start this screening testing approach in schools and other critical contexts.
Data Sources
K-12 Students and Teachers (Note: Data is not divided by K-8 and 9-12, so we divided the total K-12 enrollment in each state based on the proportion of each state’s population in the corresponding age ranges)
Nursing Home Residents and Staff (Note: Nursing home staff estimates are challenging to compile as they encompass nursing, medical, therapeutic, administrative, and other staff. Total nursing home population was estimated here by multiplying nursing home residents by 1.9, given the finding that on average nursing homes employ 90 full time employees for every 100 residents)
Healthcare Workers (Note: Not all healthcare workers are patient facing and this source makes a very inclusive estimate of healthcare workers. We divided the number of total workers in healthcare settings by 4 to bring them more in line with our estimates of patient-facing healthcare workers on a national level while still offering state-level variance)
Prison Employees, Firefighters, Police, EMTs, Food Service Workers, Cashiers, and Salespeople
