By pulling anonymized, publicly available location data from mobile devices, researchers have found that the U.S. population generally complied with stay-at-home orders issued by state officials this spring.

The analysis, published in the CDC's Morbidity and Mortality Weekly Report (MMWR), is the latest to use personal devices and the data they generate to describe behaviors relevant to public health.

"These findings can inform public policies to potentially slow the spread of COVID-19 and control other communicable diseases in the future," the CDC and Georgia Tech Research Institute researchers wrote.

TOP-LINE DATA

Between March 1 and May 31, 73% of U.S. counties issued a mandatory stay-at-home order, with several jurisdictions adjusting the restrictiveness of their orders within this window. The researchers observed a decrease in median population movement among 97.6% of counties for which data was available.

A mandatory stay-at-home order was associated with less movement when comparing the 28 days before and after such an order, a trend that persisted regardless of rural-urban strata. Conversely, population movement rose significantly once these orders expired or were lifted.

With that being said, populations across the country reduced their movement during the 14 days immediately following the first stay-at-home order in the U.S. Meanwhile, movement in areas without any stay-at-home orders increased once the country's first order was relaxed.

The researchers noted that the widespread compliance with stay-at-home orders would also have been influenced by other variables, such as the declaration of COVID-19 as a pandemic or increased media attention. They also warned that the reliance on this type of data could underrepresent demographics or regions in which device ownership is less prevalent.

HOW IT WAS DONE

The researchers accessed a bank of public, anonymized location data collected from mobile devices during the study period. The sample included users from all 50 states, the District of Columbia and five U.S. territories, with stay-at-home orders having been enacted in 42 states and territories during this time period.

The researchers codified the stay-at-home orders based on their requirements (e.g., mandatory for all persons or mandatory for those at increased risk in certain areas). Meanwhile, behavior from the location data was estimated at a county level, with movement defined as each day the device had or had not moved beyond 150 meters of its usual resting location. These were combined for a seven-day rolling average, and examined against order dates released by government sources.

WHAT'S THE HISTORY?

The CDC's latest effort follows a similar approach to that of Google's COVID-19 Community Mobility Reports. Released in April, this web tool uses aggregated location data collected by Google to create reports of large-scale behavior trends at the country, state, county or regional levels. Google said that it's putting together the reports to support public health officials as they make decisions to limit COVID-19's spread – for example, changes in foot traffic within certain business types or in transportation hubs.

Still, many individuals and experts in the country object to device-driven location tracking out of concerns for privacy, regardless of whether or not efforts are made to anonymize their data. In the contact-tracing space, this has led to a rise in apps and services that rely on Bluetooth-proximity communications, rather than GPS services. The most prominent of example is the system-level functionalities provided by Apple and Google.