Exoplanets are experiencing a stratospheric rise. In the three decades since the first confirmed planet orbiting another star, scientists have catalogued more than 4,000 of them. As the list grows, so too does the desire to find Earth-like exoplanets — and to determine whether they could be life-sustaining oases like our own globe.

The coming decades should see the launch of new missions that can gather ever-larger amounts of data about exoplanets. Anticipating these future endeavors, a team at the 91̽�� and the University of Bern has computationally simulated more than 200,000 hypothetical Earth-like worlds — planets that have the same size, mass, atmospheric composition and geography as modern Earth — all in orbit of stars like our sun. Their goal was to model what types of environments astronomers can expect to find on real Earth-like exoplanets.

As they report in a paper accepted to the Planetary Science Journal and Dec. 6 to the preprint site arXiv, on these simulated exoplanets, one common feature of present-day Earth was often lacking: partial ice coverage.

“We essentially simulated Earth’s climate on worlds around different types of stars, and we find that in 90% of cases with liquid water on the surface, there are no ice sheets, like polar caps,” said co-author , a 91̽��professor of astronomy and scientist with the ����’s . “When ice is present, we see that ice belts — permanent ice along the equator — are actually more likely than ice caps.”

The findings shed light on the complex interplay between liquid water and ice on Earth-like worlds, according to lead author Caitlyn Wilhelm, who led the study as an undergraduate student in the 91̽��Department of Astronomy.

“Looking at ice coverage on an Earth-like planet can tell you a lot about whether it’s habitable,” said Wilhelm, who is now a research scientist with the Virtual Planetary Laboratory. “We wanted to understand all the parameters — the shape of the orbit, the axial tilt, the type of star — that affect whether you have ice on the surface, and if so, where.”

The team used a 1-D energy balance model, which computationally imitates the energy flow between a planet’s equator and poles, to simulate the climates on thousands of hypothetical exoplanets in various orbital configurations around F-, G- or K-type stars. These classes of stars, which include our own G-type sun, are promising candidates for hosting life-friendly worlds in their , also known as the “Goldilocks” zone. F-type stars are a bit hotter and larger than our sun; K-type stars are slightly cooler and smaller.

In their simulations, the orbits of the exoplanets ranged from circular to a pronounced oval. The team also considered axial tilts ranging from 0 to 90 degrees. Earth’s axial tilt is a moderate 23.5 degrees. A planet with a 90-degree tilt would “sit on its side” and experience extreme seasonal variations in climate, much like the planet Uranus.

According to the simulations, which encompassed a 1-million-year timespan on each world, Earth-like worlds showed climates ranging from planet-wide “” climates — with ice present at all latitudes — to a steaming “moist greenhouse,” which is probably similar to Venus’ climate before a made its surface hot enough to melt lead. But even though most environments in the simulations fell somewhere between those extremes, partial surface ice was present on only about 10% of hypothetical, habitable exoplanets.

The model included natural variations over time in each world’s axial tilt and orbit, which in part explains the general lack of ice on habitable exoplanets, according to co-author , a postdoctoral scientist at the University of Bern and researcher with the Virtual Planetary Laboratory.

“Orbits and axial tilts are always changing,” said Deitrick. “On Earth, these variations are called , and are very small in amplitude. But for exoplanets, these changes can be quite large, which can eliminate ice altogether or trigger ‘snowball’ states.”

When partial ice was present, its distribution varied by star. Around F-type stars, polar ice caps — like what Earth sports currently — were found about three times more often than ice belts, whereas ice belts occurred twice as often as caps for planets around G- and K-type stars. Ice belts were also more common on worlds with extreme axial tilts, likely because seasonal extremes keep the polar climates more volatile than equatorial regions, according to Wilhelm.

The team’s findings about ice on these simulated Earth-like worlds should help in the search for potentially habitable worlds by showing astronomers what they can expect to find, especially regarding ice distribution and the types of climates.

“Surface ice is very reflective, and can shape how an exoplanet ‘looks’ through our instruments,” said Wilhelm. “Whether or not ice is present can also shape how a climate will change over the long term, whether it goes to an extreme — like a ‘snowball Earth’ or a runaway greenhouse — or something more moderate.”

Ice alone, or its absence, does not determine habitability, though.

“Habitability encompasses a lot of moving parts, not just the presence or absence of ice,” said Wilhelm.

Life on Earth has survived snowball periods, as well as hundreds of millions of ice-free years, according to Barnes.

“Our own planet has seen some of these extremes in its own history,” said Barnes. “We hope this study lays the groundwork for upcoming missions to look for habitable signatures in exoplanet atmospheres — and to even image exoplanets directly — by showing what’s possible, what’s common and what’s rare.”

Rachel Mellman, a recent 91̽��graduate in astronomy, is a co-author on the paper. The research was funded by NASA through grants to the Virtual Planetary Laboratory.

For more information, contact Barnes at rkb9@uw.edu and Wilhelm at cwilhelm@uw.edu.

Grant numbers: NNA13AA93A, 80NSSC18K0829.

An analysis of Washington state Department of Health data — — has found that people under 40 years old have continued to become infected by the coronavirus at increasing rates even as the incidence of infection among older populations declines. The publication was available previously in preprint form on medRxiv.

The study led by affiliated epidemiologist in the 91̽��School of Public Health, postulates that the increased spread of the virus among young people “creates a possible reservoir of disease with spillover risk to more vulnerable older persons.”

Consequently, Malmgren said, it’s a mistake right now for Washington and other states to relax restrictions meant to contain the pandemic.

“We are going to have a spike in four weeks’ time. But it won’t be as big a spike or as visible a spike as in January because people in older age groups are vaccinated now and so won’t be going to the hospital,” Malmgren said. “So, we’ll be living in a fool’s paradise thinking that we don’t have a lot of COVID-19 out there when we do.”

When the pandemic first erupted, testing found a preponderance of disease among people over 40 who were more likely to show symptoms of the disease and be hospitalized. However, as “hospitalization rates declined without an equivalent rate of decline among confirmed COVID-19 cases,” the, the shift from older to younger populations experiencing disease illustrated “the absence of a true decline in cases.”

Following are quotes by Malmgren related to the results of the study:

“It’s very hard for people to get it out of their heads that the only people affected were old people. That was wrong, right from the beginning. You can see in our study how quickly the cases became predominantly among 20- to 39-year-olds. And that while the cases went down in 40- to 59-year-olds, they went steadily up in 0- to 19-year-olds.

“The number of cases among 60- to 79-year-olds and among 80-plus-year-olds are dropping now that people are getting vaccinated, but the numbers are going up in 0- to 19-year-olds and in 20- to 39-year-olds. So, it’s really a big mistake to go to (the level of reopening designated by Washington).

“It’s going to be really, really hard on the state.

“The elderly population was a harbinger of the disease, because they are more susceptible to poor outcomes. While young people are less likely to be symptomatic and require hospitalization, they spread it like crazy.

“As an epidemiologist, you ask where is the disease? In Washington state, it’s among people under 40. Since submitting the paper for publication, the trend hasn’t changed. And in fact, the number of infections among those younger age groups continues to increase.”

Co-authors are Boya Guo, graduate student in the 91̽��Department of Epidemiology, School of Public Health; and Dr. Henry Kaplan, Swedish Cancer Institute, Seattle.

###

For more information, contact Dr. Malmgren at jmalmgren@seanet.com.

In September, a team led by astronomers in the United Kingdom that they had detected the chemical phosphine in the thick clouds of Venus. The team’s reported detection, based on observations by two Earth-based radio telescopes, surprised many Venus experts. Earth’s atmosphere contains small amounts of phosphine, which may be produced by life. Phosphine on Venus generated buzz that the planet, often succinctly touted as a “,” could somehow harbor life within its acidic clouds.

Since that initial claim, other science teams have on the reliability of the phosphine detection. Now, a team led by researchers at the 91̽�� has used a robust model of the conditions within the atmosphere of Venus to revisit and comprehensively reinterpret the radio telescope observations underlying the initial phosphine claim. As they report in a accepted to the Astrophysical Journal and posted Jan. 25 to the preprint site arXiv, the U.K.-led group likely wasn’t detecting phosphine at all.

“Instead of phosphine in the clouds of Venus, the data are consistent with an alternative hypothesis: They were detecting sulfur dioxide,” said co-author , a 91̽��professor of astronomy. “Sulfur dioxide is the third-most-common chemical compound in Venus’ atmosphere, and it is not considered a sign of life.”

The team behind the new study also includes scientists at NASA’s Caltech-based Jet Propulsion Laboratory, the NASA Goddard Space Flight Center, the Georgia Institute of Technology, the NASA Ames Research Center and the University of California, Riverside.

The UW-led team shows that sulfur dioxide, at levels plausible for Venus, can not only explain the observations but is also more consistent with what astronomers know of the planet’s atmosphere and its punishing chemical environment, which includes clouds of sulfuric acid. In addition, the researchers show that the initial signal originated not in the planet’s cloud layer, but far above it, in an upper layer of Venus’ atmosphere where phosphine molecules would be destroyed within seconds. This lends more support to the hypothesis that sulfur dioxide produced the signal.

Both the purported phosphine signal and this new interpretation of the data center on radio astronomy. Every chemical compound absorbs unique wavelengths of the , which includes radio waves, X-rays and visible light. Astronomers use radio waves, light and other emissions from planets to learn about their chemical composition, among other properties.

In 2017 using the , or JCMT, the U.K.-led team discovered a feature in the radio emissions from Venus at 266.94 gigahertz. Both phosphine and sulfur dioxide absorb radio waves near that frequency. To differentiate between the two, in 2019 the same team obtained follow-up observations of Venus using the , or ALMA. Their analysis of ALMA observations at frequencies where only sulfur dioxide absorbs led the team to conclude that sulfur dioxide levels in Venus were too low to account for the signal at 266.94 gigahertz, and that it must instead be coming from phosphine.

In this new study by the UW-led group, the researchers started by modeling conditions within Venus’ atmosphere, and using that as a basis to comprehensively interpret the features that were seen — and not seen — in the JCMT and ALMA datasets.

“This is what’s known as a radiative transfer model, and it incorporates data from several decades’ worth of observations of Venus from multiple sources, including observatories here on Earth and spacecraft missions like ,” said lead author Andrew Lincowski, a researcher with the 91̽��Department of Astronomy.

The team used that model to simulate signals from phosphine and sulfur dioxide for different levels of Venus’ atmosphere, and how those signals would be picked up by the JCMT and ALMA in their 2017 and 2019 configurations. Based on the shape of the 266.94-gigahertz signal picked up by the JCMT, the absorption was not coming from Venus’ cloud layer, the team reports. Instead, most of the observed signal originated some 50 or more miles above the surface, in Venus’ mesosphere. At that altitude, harsh chemicals and ultraviolet radiation would shred phosphine molecules within seconds.

“Phosphine in the mesosphere is even more fragile than phosphine in Venus’ clouds,” said Meadows. “If the JCMT signal were from phosphine in the mesosphere, then to account for the strength of the signal and the compound’s sub-second lifetime at that altitude, phosphine would have to be delivered to the mesosphere at about 100 times the rate that oxygen is pumped into Earth’s atmosphere by photosynthesis.”

The researchers also discovered that the ALMA data likely significantly underestimated the amount of sulfur dioxide in Venus’ atmosphere, an observation that the U.K.-led team had used to assert that the bulk of the 266.94-gigahertz signal was from phosphine.

“The antenna configuration of ALMA at the time of the 2019 observations has an undesirable side effect: The signals from gases that can be found nearly everywhere in Venus’ atmosphere — like sulfur dioxide — give off weaker signals than gases distributed over a smaller scale,” said co-author Alex Akins, a researcher at the Jet Propulsion Laboratory.

This phenomenon, known as spectral line dilution, would not have affected the JCMT observations, leading to an underestimate of how much sulfur dioxide was being seen by JCMT.

“They inferred a low detection of sulfur dioxide because of that artificially weak signal from ALMA,” said Lincowski. “But our modeling suggests that the line-diluted ALMA data would have still been consistent with typical or even large amounts of Venus sulfur dioxide, which could fully explain the observed JCMT signal.”

“When this new discovery was announced, the reported low sulfur dioxide abundance was at odds with what we already know about Venus and its clouds,” said Meadows. “Our new work provides a complete framework that shows how typical amounts of sulfur dioxide in the Venus mesosphere can explain both the signal detections, and non-detections, in the JCMT and ALMA data, without the need for phosphine.”

With science teams around the world following up with fresh observations of Earth’s cloud-shrouded neighbor, this new study provides an alternative explanation to the claim that something geologically, chemically or biologically must be generating phosphine in the clouds. But though this signal appears to have a more straightforward explanation — with a toxic atmosphere, bone-crushing pressure and some of our solar system’s hottest temperatures outside of the sun — Venus remains a world of mysteries, with much left for us to explore.

Additional co-authors are at the JPL, at UC Riverside, and at the Goddard Space Flight Center, 91̽��researcher , at Georgia Tech and at NASA Ames. The research was funded by the NASA Astrobiology Program and performed at the NExSS Virtual Planetary Laboratory.

For more information, contact Meadows at meadows@uw.edu, Akins at alexander.akins@jpl.nasa.gov and Lincowski at alinc@uw.edu.

As universities and colleges struggle to find the right combination of in-person and online classes combined with protective measures to help prevent the spread of the novel coronavirus, a new study by researchers from four institutions has reached a troubling conclusion.

Reopening university and college campuses with primarily in-person instruction is associated with a significant increase in cases of COVID-19 in the counties where the schools are located.

“Consequently, we are able to predict between 1,000 and 5,000 additional cases per day due to colleges reopening for face-to-face instruction, with our best estimate being somewhere around 3,000 cases per day or around 21,000 cases per week,” said study co-author , professor of health economics and Stergachis Family Endowed Director of the at the 91̽��School of Pharmacy.

More specifically, campuses with mostly in-person instruction contributed to increases in COVID-19 cases in their county by 0.024 cases per 1,000 residents. And, when students come from outside counties with surging cases, an additional 0.0119 per 1,000 residents come down with COVID-19.

“We don’t see similar spikes in cases for counties with colleges that reopened with primarily online instruction. The total spike attributed to face-to-face campus reopenings accounts for nearly 6 percent to 7 percent of all cases in the U.S. during this time,” Basu said.

Researchers from the University of North Carolina at Greensboro, Indiana University Bloomington, 91̽�� and Davidson College conducted the study, which has not yet been peer reviewed. The study was posted Sept. 23 and has been submitted to a journal for peer review.

“Given the timing of the mobility and case spikes,” Basu added, “these results are not likely a manifestation of additional testing or sick cases moving onto college campuses.”

The researchers sampled 1,409 colleges from July 15 to Sept. 13 and out of those classified 886 schools as conducting classes primarily in person, while 483 are teaching primarily online. Out of 1,142 U.S. counties examined in this study, only 779 contained a college in one of these categories, with 15 campuses not open during the sampling period. The researchers then compared these counties to counties without a college and looked at the periods of two weeks before the start of classes and two weeks after instruction began.

One of the signals the researchers used to determine the increase of visitors to a campus, whether for in-person instruction or on campuses with primarily online courses, was the increased presence of cellphones. Regardless of the type of instruction offered, the number of cellphones visiting campus increased significantly, in the week leading up to the start of classes and after classes had begun. However, cellphone traffic was higher on in-person campuses. And, counties with primarily online campuses did not see a statistically significant increase in COVID-19 cases.

“Our main data track cellphone movement on and off campuses and county-level COVID-19 daily reported cases,” Basu explained. “And, all evidence suggests there is a distinct local transmission component, given spikes in cases are happening two weeks after college opens. We also found spikes in cases to be higher for face-to-face colleges that drew students from communities that have seen recent spikes in cases.”

The researchers add that campus administrators and other local authorities should use these findings when considering additional strategies to mitigate COVID-19 outbreaks, and “think carefully” about cases in their counties as well as where students are coming from when planning their spring 2020 semesters.

Co-authors include Martin Andersen, Department of Economics, University of North Carolina at Greensboro; Ana Bento, School of Public Health-Bloomington, and Kosali Simon, O’Neill School of Public and Environmental Affairs, Indiana University; and Chris Marsicano, founding director of The College Crisis Initiative at Davidson College.

###

For more information, contact Basu at basua@uw.edu

Politics, above COVID-19 cases or deaths, determined whether states enacted mask mandates during the first months of the pandemic, a new study finds.

States with Republican governors delayed imposing indoor mask requirements by an average of nearly 30 days, controlling for other factors. The study by researchers at the 91̽�� examined a series of factors surrounding the announcement (or lack thereof) of statewide mask mandates in all 50 states, and found that partisanship, particularly at the state executive level, where such restrictions can be imposed, was the most significant factor in the timing of new rules.

The is posted to the preprint server medRxiv and has not been peer-reviewed.

“Wearing masks in public places is one of the easiest ways to reduce transmission of the coronavirus, and clear, consistent mandates are one of the best tools we have to get everyone to wear masks regularly. Our team has been tracking mask mandates covering indoor public spaces, where the risk of transmission is highest, and we wanted to know whether adoption was really as partisan as it seemed, or if there were other explanations,” said , an associate professor of political science at the 91̽��and lead author of the study. The research is part of the 91̽��COVID-19 State Policy Project, led by Adolph and , political science professor and chair of the department.

Evidence for the effectiveness of mask-wearing in combatting COVID-19 has grown , and scientists in the United States and around the world agree that the virus, commonly spread through the air, can be curtailed . Leaders of some countries have imposed , but such decisions in the United States are currently left to the states.

That’s where politics appears to enter the fray.

Adolph and his team, who earlier this year released a study on the role of politics in imposing statewide social distancing measures, examined statewide mask requirements adopted from early April through mid-August 2020. The researchers grouped mask laws by the breadth of mandates and focused on those that, at minimum, required a mask in all indoor public spaces. States requiring masks indoors and outdoors are considered most restrictive; 25 states now have that requirement.

Currently, 14 states do not have any statewide mask mandate; three states have a limited requirement for masks in some settings.

The team then analyzed the timing of those laws taking into account other state data, such as the number of confirmed COVID-19 cases and deaths, the average partisan ideology of its population, and the governor’s political party (which often, but doesn’t always, match the dominant party affiliation of citizens).

The study found that, after accounting for other factors, states with Democratic governors were seven times as likely as those with Republican governors to impose broad statewide mask mandates. Described another way, if two states were identical except for the party of the governor, the researchers would expect the Republican-governed state to adopt a mask mandate 29.9 days later than its Democrat-led twin.

Adolph argues it’s at least somewhat surprising that Republican governors resisted mask mandates. After all, he points out, requiring masks could help stave off the costly reimposition of social distancing mandates. “President Trump spent crucial months deriding masks and refusing to wear them in public,” Adolph said. “This deepened a partisan divide that few Republican governors have been willing to cross, even as their states’ cases shot up this summer.”

The severity of the pandemic mattered less than governors’ party affiliation, researchers noted. ��The epidemiological indicator that had the largest impact on mask mandates is the rate of COVID-19 deaths, which lags several weeks behind current conditions. Controlling for other factors, states with higher daily death rates were an average of 10.5 days quicker to adopt mask mandates than those with lower rates. But mask mandates did not appear to respond to real-time metrics like new cases per million or the rates of people testing positive, the study points out.

The team analyzed other demographic factors, such as the resident’s education levels, or the percentage of the population over age 70, but those didn’t have any noticeable effect, according to the study.

The paper has been submitted for publication in a peer-reviewed journal. The study was funded by the at the 91̽��and the Benificus Foundation.

In addition to Wilkerson, co-authors were , a doctoral student in health metrics sciences at the UW; and political science doctoral students , , and .

For more information, contact Adolph at cadolph@uw.edu.

Starting in mid-March, many states issued stay-at-home orders or encouraged people to work from home to try to curb the spread of COVID-19. In cities nationwide, .

But did fewer cars on the road lead to cleaner air? Not necessarily, according to a new study led by 91̽�� researchers.

Using air quality data from U.S. Environmental Protection Agency monitors across the U.S., the research team looked for changes in two common pollutants: , and called PM2.5. Compared to the past 10 years, neither pollutant has been consistently lower than expected levels since stay-at-home orders began. But the team found that another pollutant that mainly comes from car exhaust, NO2, was at much lower levels in three cities — Seattle, Los Angeles and New York — (30% lower on average) after stay-at-home orders were implemented.

Update Jan. 27, 2021:��The paper was originally posted on the preprint server , but it has now been peer reviewed. It published online Jan. 2 in .��This release has been edited to reflect that change.

“It’s difficult to determine whether the air really is cleaner now because, in general, there is a lot of variability in weather and emissions patterns,” said senior author , a 91̽��professor of civil and environmental engineering. “You can’t look solely at concentrations today and compare them to the same day a month or a year ago.”

Because pollutant concentrations vary over time and across regions, the researchers developed a method to determine whether pollutant levels in any given week were notably different from normal.

“Let’s say you’re measuring height and you want to know if a difference of 2 inches is a big deal. Well, it depends on how much variability there is,” Marshall said. “If you measured the height of people in a city and then you have someone who’s 2 inches taller than the average height, you’d probably say they’re still about average because people’s height varies by feet. But if you found an NBA hoop rim that was 2 inches different from the average, that’s a big deal. Rim height is so precisely controlled.”

The team’s “robust differences” metric compares a pollutant’s median concentration during a week in 2020 to its median concentration in the same time period over the past 10 years. The metric also adjusts for whether a region has been getting cleaner or dirtier over the past 10 years, a lot like adjusting prices for inflation. This calculation yields a score where negative numbers indicate that a region was cleaner than expected for that week, and positive numbers indicate that a region was dirtier than expected.

The researchers attempted to calculate the robust differences score for the nearly 1,000 PM2.5 and the more than 1,170 ozone monitors across the U.S. But the team excluded monitors that had less than three years of data or were missing data from at least two of the last three years. West Virginia didn’t have any PM2.5 monitors that qualified, and Rhode Island, South Carolina and Hawaii didn’t have ozone monitors that qualified. For all other states, the team used individual monitor data to calculate state-level scores.

​

California was the first state to issue a stay-at-home order, on March 19. That week, average PM2.5 levels across the country were slightly (5%) higher than expected. Since then, PM2.5 levels have continued to be similar to expected values, as of June 17, when this story was initially published. Editor’s note: The maps and graphs in this story will continue to be updated. See them for current information.

PM2.5 levels in individual states also fluctuated over time, which is normal depending on what’s happening with emissions and weather patterns, the researchers said. But in general, the team didn’t see any trends of states being consistently cleaner or dirtier over time.

“Everyone was saying that pollution had to be lower after the stay-at-home orders went into effect,” said lead author , a 91̽��doctoral student in civil and environmental engineering. “But that’s not what we saw across the U.S. It was very surprising.”

​

While the researchers analyzed all states that had enough data, this article highlights three states with stay-at-home orders and large metropolitan areas: Washington, California and New York. As of June 17, Washington had higher PM2.5 levels than normal the week before its stay-at-home order went into effect, and the levels have bounced around since then. Meanwhile, California had lower PM2.5 levels than normal before its stay-at-home order, and that has gradually increased. New York only had a few counties reporting data, but in those counties, PM2.5 levels remained relatively consistent over the course of 2020.

​

Across the country, ozone levels dropped below expected values before stay-at-home orders were issued and have increased since then, reaching expected values by early May. As of June 17, when this story was initially published, ozone levels continued to be similar to expected values. Ozone levels often change over time, the researchers said, depending on what’s happening with emissions and the weather.

​

Similar to the PM2.5 data, ozone levels varied by state. For example, New York’s ozone levels were similar to the national trend. California had lower ozone levels before the stay-at-home order and those levels gradually increased to higher than expected before dropping again the week of May 13. Washington only had one county reporting ozone data – King County – and its ozone levels were within the expected range when the stay-at-home order went into effect. Since then, the levels have fluctuated between lower and higher than expected.

While the team didn’t see any consistent drop in ozone and PM2.5 levels, the researchers wondered if , a pollutant that’s more closely linked to car exhaust, would be affected by fewer people on the road. The EPA has not posted NO2 levels for 2020 yet, so the team used data for Seattle, Los Angeles and New York City from .

NO2 levels for all three cities were lower than expected the week of all three states’ stay-at-home orders, with Seattle 46% lower, Los Angeles 56% lower and New York City 46% lower. These levels generally have remained below normal values, as of June 17. These findings are consistent with , the researchers said.

“Everybody thinks that because so many people are not driving the air is really clean, but it really depends on what type of pollution you’re talking about,” Bekbulat said. “Cars are the main contributor to NO2 levels, which are down, so there’s something to people’s intuition. But we’re still using electricity, and that power plant is generating emissions. When it’s cold, people still burn fires, and that generates emissions. Meteorology plays a role, too: Stagnant air tends to be dirtier. There are all kinds of things happening that could explain why we think we’re all shutting down, but pollution levels don’t show that.”

Other co-authors on this paper are ​ at the University of Texas at Austin, and at the University of Minnesota, and at Carnegie Mellon University. This publication was developed as part of the , which was supported under an Assistance Agreement awarded by the U.S. Environmental Protection Agency.

For more information, contact Marshall at jdmarsh@uw.edu and Bekbulat at bujinb@uw.edu.

Grant number: R835873. This research has not been formally reviewed by EPA. The views expressed in this document are solely those of the authors and do not necessarily reflect those of the Agency. EPA does not endorse any products or commercial services mentioned in this publication.

Update July 9, 2020��— the app mentioned below is now called .

Stay-at-home orders and social distancing have been successful in some areas to help flatten the coronavirus curve. As parts of the world begin to open up again, communities need a reliable way to keep track of the virus and contain its spread.

Contact-tracing apps may provide one option as part of a larger strategy. These apps monitor who has come in contact with whom and can, when appropriate, alert a network of people if someone nearby has been diagnosed with the virus. But many current contact-tracing apps have — for example leaking a user’s location information or taking away people’s control over their own data.

Now researchers from the 91̽�� and 91̽��Medicine, along with volunteers from Microsoft, have developed a new tool, . This contact-tracing app, developed with input from public health officials and contact tracing teams, would alert people about potential exposure to COVID-19 without giving up anyone’s privacy. This app could also help individuals who test positive prepare for a contact tracing interview with a public health official.

CovidSafe is not ready to be downloaded from app stores, but an Android demo version is accessible through the . Users who try the demo version, which doesn’t have full functionality yet, can submit feedback to the team. This app is based off a series of privacy and security guidelines that the team outlined posted earlier this month to the preprint site arXiv.

“Contact tracing is one of the most effective tools that public health officials have to halt a pandemic and prevent future outbreaks,” said author , a 91̽��doctoral student in the Paul G. Allen School of Computer Science & Engineering. “Our contact-tracing app addresses underlying privacy, security and re-identification issues, rather than sweeping them under the rug. With CovidSafe, all information is stored locally on your phone unless you choose to share that you’ve tested positive. Only then is your data sent to a secure server, and the app alerts anyone who has been nearby. After these notifications are sent, all the information is deleted.”

CovidSafe takes several steps to maintain users’ privacy. The app begins by assigning each user a secret code name, which remains private. Then it generates a variation of the code name that changes every 15 minutes and uses Bluetooth to broadcast that to other users nearby. CovidSafe also stores a list of these people’s smartphone signals. With the full version of CovidSafe, if a user tests positive and they choose to share that information with the app, it will alert anyone who has come in contact with them within the past 14 days — the infection window for COVID-19 — without divulging who the person is or where they are.

“Conventional contact tracing already requires a person to give up some measure of personal privacy as well as the privacy of those they came into contact with,” said collaborator , an associate professor in the Allen School. “However, we can make acceptable trade-offs to enable us to use the best tools available to speed up and improve that process, all while ensuring stronger privacy guarantees at the same time.”

Because not everyone will want to use a contact-tracing app, CovidSafe aims to augment — not replace — conventional contact tracing, which public health officials do by interviewing patients who’ve tested positive about where they have been and who they have seen. CovidSafe creates a log of users’ locations over time, so it can help people in these interviews by providing them with the details about where they’ve been lately.

“This is being built first and foremost with contact-tracing teams and public health officials in mind,” said collaborator Dr. , an assistant professor of anesthesiology and pain medicine at the 91̽��School of Medicine. “They are the experts, and much of the functionality has been developed based on direct feedback from teams doing this necessary and difficult work. Combined with extremely thoughtful privacy-preserving designs, this system is built to meet the needs of a privacy-conscious public and to efficiently deliver useful information that can help public health systems and contact tracers work smarter and faster.”

CovidSafe has other features, so that users who have tested positive and are in isolation can track their symptoms, and a messaging system that will eventually allow users to receive tailored health announcements from local public health agencies. The researchers for organizations to customize for their own use.

“Ten years from now, I want to be able to look back and genuinely say, ‘I did something to help in the greatest crisis of my lifetime,'” said collaborator , one of the project volunteers who is also a computer scientist at Microsoft Research. “At this point, dozens of people have contributed hundreds of hours toward making this project happen. We have all the expertise needed to create something genuinely useful, and we are well on the way.”

Other co-authors on the white paper are , , , and in the Allen School, and and at Microsoft.

This research was funded by the Washington Research Foundation, the Office of Naval Research, the National Science Foundation, the National Institutes of Health and the Alfred P. Sloan Foundation.

For more information or to submit feedback about CovidSafe, contact the team at Covidsafe@uw.edu.

Grant numbers: N00014-18-1-2247, #CCF 1637360, #CCF 1740551, K23DA046686, 1914873, 1812559, CNS-1553758 and CNS-1719146

“This indicates an important turning point that each country will reach in the COVID-19 pandemic, and we predict that the United States is on course to reach this point in the coming weeks,” said , a professor of applied mathematics at the 91̽��. “It is a point of maximum strain on a country’s health and medical infrastructure.”

The new model is intended to help health officials and policymakers see at least two weeks in advance how COVID-19 will likely strain medical infrastructure in the U.S. and around the world. It relies on the number of newly diagnosed cases and the number of individuals who have recovered or died in a geographic region — whether an entire country or a subnational level like a state or province.

and , both from the First Institute of Oceanography’s Data Analysis Laboratory in Qingdao, China, created the model with Tung. Their paper describing the analysis is not yet peer reviewed but has been submitted to a journal for consideration. Their study was March 30 to the preprint site .

For the U.S., this model predicts that:

• The rate of daily new COVID-19 cases peaked on April 5-7, a projection that appears to be accurate, according to Tung
• The number of “active” COVID-19 cases, which are individuals who have been diagnosed but haven’t recovered or died, will peak on April 20, plus or minus four days, and will then slowly decline as the number of cases entering the medical system becomes less than the number of cases leaving the medical system
• The U.S. outbreak will taper off in the first week of June with projections of 710,000 total cases, but could be up to 990,000, and 28,000 deaths, but could be up to 39,000, if the current U.S. fatality rate of 4% holds

Their model also predicts that other hard-hit countries, such as Germany and Spain, have either recently peaked in active COVID-19 cases or will do so soon. The United Kingdom will not peak until the latter half of April, according to the analysis.

The model finds that the length of outbreaks will also vary by country. Germany and Italy will take a week longer than the city of Wuhan, China — the earliest epicenter — to reach their turning point in active COVID-19 cases. The United States is projected to take two weeks longer than Wuhan. Wuhan and Hubei Province were placed under strict lockdowns by the Chinese government early in the outbreak, which may explain the shorter course there, said Tung. Italy was slower to roll out lockdowns, first regionally and then nationally. The United States has no national lockdown, though a majority of states have issued stay-at-home orders.

The researchers tested the model’s efficacy using COVID-19 data from China. With an accuracy of a few days, their model predicted key events in the outbreak’s growth, spread and decline of COVID-19 in Wuhan, Hubei and the rest of China — including the peak of new cases, the peak of active cases and the subsidence of the epidemic. Wuhan’s 76-day lockdown April 8.

Scientists at the ����’s Institute for Health Metrics and Evaluation have created , which relies on other pieces of information about the pandemic, such as . In contrast, the model by Huang, Qiao and Tung uses the number of newly diagnosed cases and the number of individuals who have recovered or died. The Institute’s model primarily projects COVID-19 deaths in a region, as well as the demand for hospital resources such as ventilators.

“Our two approaches complement one another, providing the projections that health officials and governments need to understand when the maximum strain on resources is coming, and to show how the course of the pandemic depends heavily on the level of social distancing measures adopted,” said Tung.

Compared to other modeling approaches, such as widely reported from the Imperial College London, the model developed by Tung and his colleagues does not require knowledge of the infection rate or the total number of infected cases — including asymptomatic individuals. These are difficult data to collect or estimate given the relatively sparse testing for COVID-19 in many countries, and the fact that most symptom-free or mildly ill individuals are not entering the medical system for treatment.

As a result of these and other key differences, the predictions by Huang, Qiao and Tung differ significantly from Imperial College London projections of a longer outbreak with 40% to 80% of the U.S. population infected and 1.1 to 2.2 million deaths. The results from the model used by Imperial College London differed significantly because it relied on separate assumptions about COVID-19 and the predictions were generated when key parameters, such as its infection rates, were unknown, according to Tung.

“Those types of models do serve purposes, such as moving policymakers into action,” said Tung. “But once the epidemic begins, it is important to turn to data-driven models that incorporate real-time information such as diagnosed cases, recovered cases and deaths — which reflect the effects of policy decisions and the degree of compliance — so that we can more realistically project the pandemic’s course.”

For more information, contact Tung at ktung@uw.edu.

States led by Republican governors and with a significant share of Trump supporters were an average of nearly three days later than other states to enact social distancing measures related to the COVID-19 outbreak, according to a new study.

The finding is part of new research by the 91̽�� examining factors that contributed to decision-making by governors in all 50 states to combat the novel coronavirus. The explores whether the adoption of state-level social distancing measures depends on the number of coronavirus cases in the state, the affluence of the state, and the partisanship of the state’s governor and voters.

The rapid spread of the novel coronavirus, which as of March 31 has killed more than 3,000 people in the United States and, by many estimates, is projected to cause anywhere from to ��domestically, has prompted public health officials to push social distancing as the key, proactive way of limiting the rise of infection. The World Health Organization more than 750,000 confirmed cases worldwide, and more than 36,000 deaths.

But the response to and attitude toward the virus have been mixed among political leaders. For several weeks at the beginning of the outbreak, President Trump and some right-leaning media outlets .

In the month since the first COVID-19 case was diagnosed – in Washington state – most states have enacted some social distancing restrictions, such as closing schools and businesses, limiting or banning gatherings of people, and advising or ordering residents to shelter in place. Mandates to stay at home are now in effect in 28 states.

The varying restrictions, and the timing of those restrictions, prompted 91̽��researchers to take a closer look.

“We wanted to understand why some American states have been slow to introduce social distancing measures,” said lead author , an associate professor of political science at the UW. “You might expect states to delay if they have fewer confirmed cases — though even that would arguably be a mistake — but we were worried by the appearance of a partisan pattern in responses, both at the state level and in public opinion.”

Update Nov. 19:��The paper was originally posted on the preprint server��but has now been peer reviewed and published in the .��This release has been edited to reflect that change.

Adolph and his team analyzed the measures that states enacted with other data, such as the number of COVID-19 cases in each state, how neighboring states were responding, each governor’s political party and each state’s voter turnout for Trump in 2016.

The team found that partisanship – especially when a state has a Republican governor, as well as the share of the statewide vote for Trump — led to delays in enacting social distancing. That��“combined partisan effect” coincided with a delay of 2.7 days, the team found.��Partisanship had a greater effect than other variables, including the number of confirmed cases in each state, researchers said. The number of confirmed cases, for example, influenced state action by less than half a day.

“ now document that Republican voters in March showed less concern on average about the coronavirus, and were less likely to adopt prudent behavior to reduce their risk of becoming infected,” Adolph said. “If Republican leaders were also systematically slower to act, their reluctance would end up hurting all Americans, but especially their own constituents.”

Under normal political circumstances, governors often make decisions to appease their party and voters, Adolph said. The 91̽��research team wanted to explore how governors adapted to what was essentially an unprecedented threat that emerged at once, nationwide.

The paper is not trying to assign blame, Adolph added. Enacting social distancing measures is difficult for any elected official, because closing schools and businesses has significant economic and personal consequences for a population. But based on public health guidance, until a vaccine is available for widespread use, aggressive social distancing can stem the exponential spread of disease and limit the total number of deaths.

“Fighting COVID-19 shouldn’t be a partisan issue: The virus doesn’t care what party you belong to, and everyone is at risk. There’s still a chance to change this and save lives,” Adolph said.��“The sooner all governors mandate and enforce strict social distancing, and the more they listen to public health experts instead of partisan cues, the more lives we will save, and the sooner we can all recover from this crisis.��Every day matters.”

The paper has been submitted for publication in a peer-reviewed journal. Co-authors are , professor and chair of the political science department at the UW; , a doctoral student in the 91̽��Department of Health Metrics Sciences; and political science doctoral students and .

For more information, contact Adolph at cadolph@uw.edu.

A 91̽�� researcher calculates that 14.4 million workers face exposure to infection once a week and 26.7 million at least once a month in the workplace, pointing to an important population needing protection as the novel coronavirus disease, COVID-19, continues to break out across the U.S.

, an assistant professor in the 91̽��School of Public Health, based her calculations on research she in the American Journal of Industrial Medicine. In that paper, Baker and co-authors calculated that about 8% of workers in Federal Region X — comprised of Alaska, Washington, Oregon and Idaho — work in jobs where exposure to infection or disease occurs at least once a week at work. Those risks include flu-like illnesses, and other respiratory illnesses, like COVID-19, as well as wound infections.

Using federal employment data, and the same analysis method, Baker and her co-authors :

• 10% (14,425,070) of U.S. workers are employed in occupations where��exposure to disease or infection happens at least weekly, based on employee and employer self-report.
• 18.4% (26,669,810) of U.S. workers are employed in occupations where exposure to disease or infection happens at least monthly, based on employee and employer self-report.

Update April 28: The paper was originally posted on the preprint server but has now been peer reviewed and ��This release has been edited to reflect that change.

“While healthcare occupations represent the bulk of workers exposed to infection and disease, other occupations that frequently interface with the public and provide essential services are also at increased risk of exposure. Those include police officers, firefighters, childcare workers, and personal care aides,” said Baker, who is also program director of ����’s program, Department of Environmental & Occupational Health Sciences. “This underscores the importance of all types of occupations developing workplace response plans for infectious disease.”

While those response plans must include how to keep workers safe from exposure at work, Baker added, they must also ensure workers don’t have to show up sick, potentially spreading disease within and outside the workplace.

Some workers who are in higher paying and more secure jobs, often salaried, can work from home or afford more time away from work, but many don’t have these same options, such as workers who participate in the gig economy or are independent contractors and are typically not considered employees. These workers don’t benefit from employee protective policies, such as sick leave, putting them at increased risk of having to work when they or a loved one is sick, despite public health guidance.

Even if a worker does have paid sick leave, and knows how to access it, a variety of other real and perceived pressures — such as an economy that rewards people who are working hard at all times, or pressure to perform work that no one else can perform, encourages workers to come to work sick, a phenomena researchers term “presenteeism.”

“Our findings serve as an important reminder that the workplace should be a focus for public health intervention, especially during disease outbreaks such as COVID-19,” Baker said, adding that researchers produced the new work to help shed light on this fact, and to respond to questions about their 2018 paper and its application to the current, nationwide outbreak. Estimating the burden of U.S. workers exposed to infection or disease is a key factor in containing risk of COVID-19 infection.

“The public health implications from our study,” Baker said, “are that workplaces need to make sure that they are not only protecting their workers at work, but also coming up with contingency plans to make sure that sick workers are not coming to work, and that can be accomplished through training workers to fill in for each other, providing additional paid sick leave during this time and similar measures.”

Co-authors include Trevor Peckham and Noah Seixas, 91̽��Department of Environmental & Occupational Health Sciences, School of Public Health. This research was funded by the National Institute for Occupational Safety and Health (NIOSH).

Learn more about the ����’s Population Health Initiative: a 25-year, interdisciplinary effort to bring understanding and solutions to the biggest challenges facing communities.