Stay-at-Home Orders and Movement in U.S. Counties, Spring 2020

In the first months of the COVID-19 pandemic, U.S. states and territories used many kinds of community mitigation to cut down person-to-person contact. One of the most common was an order telling people to stay home. Each jurisdiction sets its own public health laws, so the type and timing of these orders differed widely. CDC and the Georgia Tech Research Institute set out to measure how the orders issued from March 1 to May 31, 2020, by all 50 states, the District of Columbia and five U.S. territories related to how much people moved. They published the results in MMWR.

How the study worked

The team collected each jurisdiction's executive or administrative orders and press releases from official websites and sorted every order into one of five categories: mandatory for everyone; mandatory only in certain areas; mandatory only for people at higher risk; mandatory only for higher-risk people in certain areas; or advisory. A jurisdiction with no order was coded as such. The coding followed each order's legal wording — "shall" or "must" counted as mandatory, "should" or "are encouraged to" as advisory — and did not look at enforcement or public perception. The coded orders were checked a second time and released as a free public data set.

Movement came from anonymized location data from mobile devices such as phones, tablets and watches. For each county and day, the analysts calculated the share of reporting devices that stayed completely at home, meaning within 150 meters of their usual nighttime location, smoothed over 7 days. Counties were grouped by the U.S. Department of Agriculture's rural-urban categories, and the researchers compared the periods before and after four kinds of dates: when each state or territorial order started, when it ended, when the first state order took effect (California's), and when the first state order ended (Alaska's).

What happened to the orders

Between March 1 and May 31, 42 states and territories issued mandatory stay-at-home orders, covering 2,355 (73%) of the country's 3,233 counties. Puerto Rico issued the first territorial order, on March 15, and California the first state order, on March 19. Eight jurisdictions only advised or recommended staying home, and six issued no order at all: American Samoa, Arkansas, Connecticut, Nebraska, North Dakota and Wyoming.

Type and duration of state and territorial stay-at-home orders, March 1–May 31, 2020. CDC figure.

Orders changed often. Over the period, 22 jurisdictions moved from a mandatory order to an advisory one, 11 rescinded their orders or let them lapse, and one had its order struck down by the state supreme court. Alaska, on April 24, was the first state to rescind or let an order expire. Eight jurisdictions still had mandatory orders covering at least part of their population after May 31.

What happened to movement

• When orders began: Among the 2,351 counties with movement data, 2,295 (97.6%) showed less movement — a higher median share of devices at home — in the 28 days after their order started than in the 28 days before. The drop was statistically significant in every rural-urban group.
• When orders ended: 436 (19%) of the 2,355 counties had an order that ended by May 3, the last date that left a full 28 days to observe afterward. Movement rose significantly in every rural-urban group once those orders ended.
• Beyond the state that acted: In the 14 days after California's order, movement fell significantly in every rural-urban group, even in places without an order of their own. After Alaska became the first state to relax its order, movement increased in states and territories that had not relaxed theirs.

County-level mean share of mobile devices at home before and after stay-at-home orders began and ended, by rural-urban category, March 1–May 31, 2020. CDC figure.

What the authors concluded

Less movement means less close contact with people outside the household, which may reduce exposure to the virus. The authors concluded that stay-at-home orders could help limit that exposure and slow community spread.

The drop that followed California's order suggested one jurisdiction's policy can shape behavior elsewhere. But the same weeks brought the pandemic declaration, national and state emergencies, news coverage of deaths and strained hospitals, gathering bans, school and business closures and canceled sporting events, any of which could have changed behavior. Likewise, the rise in movement as orders lifted may have come partly from people returning to old habits, or feeling safer because orders were ending elsewhere, but more study was needed to pin down the causes.

The authors listed five limitations. Some regions or groups may be over- or under-represented in the device data. A person may carry several devices, or leave one such as a tablet at home. The statistical test did not account for neighboring counties influencing one another. The analysis did not look at whether nationwide protests during the period affected movement. And it did not examine whether the orders changed rates of illness or death.

They also noted that stay-at-home orders had public support but disrupt daily life and have caused economic harm, and called for research on when and under what conditions such orders best protect health while limiting harm and ensuring fair enforcement.

Sources

• Moreland A, Herlihy C, Tynan MA, et al. "Timing of State and Territorial COVID-19 Stay-at-Home Orders and Changes in Population Movement — United States, March 1–May 31, 2020." MMWR 2020;69(35).