In Utah, randomized COVID-19 testing provides early estimates of real infection rates
In early May 2020, Utah’s state government, with support from several teams at the University of Utah and other clinical partners, began an innovative testing and surveillance project — named the Utah Health & Economic Recovery Outreach (HERO) Project — to help decision-makers gain a clearer, more accurate picture of the statewide prevalence of Covid-19.
Overseen by a team of statisticians, researchers, and doctors, the HERO project’s random, representative sampling method set it apart from other testing efforts. Instead of relying on data from people who were sick enough to seek out a test, the project sent dozens of field researchers into Utah’s communities to conduct random, representative sampling across four Utah counties.
“When we first met with staff from the Governor’s office, there was a clear desire for the most accurate data possible about the prevalence of the virus in the general population,” said Adam Looney, Executive Director of the Marriner S. Eccles Institute for Economic and Quantitative Analysis at the University of Utah and a lead researcher on the University of Utah team.
“We knew from the start the best way to do that was to conduct randomized testing. We hadn’t seen other states do it, but we knew Utah was in an ideal situation to tackle this problem because it has world-class medical research centers, a national reference laboratory, and a civic-minded population willing to pitch in for their community.”
Benefits of Randomized Testing
Randomized, representative testing can be costly and difficult to implement. But experts say it’s the only way to truly know how widespread the disease is in our communities. There are several reasons why:
- Non-random testing only reveals the tip of the iceberg. Non-random testing—for example, testing available only to those with symptoms or exposures—often excludes individuals who choose not to seek care or face barriers to seeking care, those whose symptoms are too mild to be tested under current protocols, or those who may be asymptomatic but contagious.
When it comes to asymptomatic individuals, we know they may comprise a large share of those who are contagious. Early in the pandemic, estimates of the asymptomatic rate ranged widely from 18% among the elderly/tourist population of the Diamond Princess to 31% among Japanese evacuated from Wuhan to 50% from early testing in Iceland to 86% in updated estimates from China (Qui, J. 2020).
Without random sampling of the population, we can’t know the true prevalence of COVID-19 and how much or how quickly it’s spreading. Even six months into the pandemic in the U.S., most testing is conducted on the sick and vulnerable and reveals only the tip of the iceberg of the infected population. Random sampling allows public health officials to see the entire infected population.
- Representative testing ensures we have data on the full population, not just one demographic sub-set. Even so-called scientific studies that appear random because they use ‘convenience’ samples — for example, visitors to grocery stores or tests of blood samples for non-COVID-related tests—won’t be representative of the population. Using data from a randomized probability-selected sample from the population resolves this uncertainty and ensures officials have data across the full population of their citizens.
- Randomized testing helps policymakers improve public health and economic relief measures. With gold-standard data in-hand, policymakers can initiate appropriate contact tracing, improve estimates of the risks of community-based transmission, and accelerate the treatment and self-isolation of cases at earlier stages of disease progression.
By predicting surges in local demand for hospitalization, high-quality data also helps communities manage situations where healthcare resources are limited. Further, it improves the targeting of testing programs, public health communication, and information campaigns. For example, if local leaders know that children are unlikely to experience symptoms and unlikely to transmit infection, the general population might be more confident in the likely outcomes associated with re-opening schools or children’s activities.
Finally, randomized, representative testing provides policymakers with information not only on the prevalence of the disease, but whether the disease is currently spreading widely within the population without detection and how prevalence rates change over time. This information can shed light on the potential consequences of changing social distancing rules or the likely duration and economic cost of ongoing public health measures.
To see results from the HERO Project’s first phase, including Covid-19 prevalence rates in Utah, download the full Phase One report.
Phase One Field Work and Testing
To conduct random, representative sampling in Utah, project organizers established a “Hope Corps” — a collection of college students that staffed field teams and organized mobile test sites.
In Phase One of the project, Hope Corp volunteers knocked on doors of randomly-selected households to encourage citizen participation. Those who chose to participate, either after talking to a volunteer or by signing up online after receiving information in the mail, were able to. visit a testing bus nearby to receive a PCR (viral) and serology (antibody) test. All participants were compensated with a $10 USD gift card for completing the survey and being subsequently tested.
Ultimately, 10,000 Utahns across four counties underwent testing at one of the HERO project’s mobile testing sites during Phase One of the project.
About the HERO Project
The Utah HERO Project has benefited from the support and guidance of many individuals and organizations dedicated to ensuring the wellbeing of Utahns during the Covid-19 Pandemic. Leading the HERO Project are Stephen C. Alder, PhD, Adam Looney, PhD, and Matt Samore, MD. The project is funded by the State of Utah in coordination with the Governor’s Office of Management and Budget.
Senior advisors to the project are Taylor Randall, MBA, PhD. and Michael Good, MD. The Project team includes Andrew T. Ravia, MD, Julio Delgado, MD, MS, and Tom Green, PhD. The Church of Jesus Christ of Latter Day Saints generously contributed the use of their parking lots and buildings to support mobile testing for this project