key: cord-0877065-5mnj3qr7
authors: Clements, John M.
title: Knowledge and behaviors toward COVID-19 among U.S. residents during the early days of the pandemic
date: 2020-04-02
journal: nan
DOI: 10.1101/2020.03.31.20048967
sha: 16ffae6b21685dfcd71c62398ba65adefc9b5d44
doc_id: 877065
cord_uid: 5mnj3qr7

Abstract Objective: To test the hypothesis that knowledge of COVID-19 influences participation in different behaviors including self-reports of purchasing more goods than usual, attending large gatherings, and using medical masks. Methods: Cross-sectional online survey of 1,034 U.S. residents age 18+ conducted on March 17, 2020. Results: For every point increase in knowledge, the odds of participation in purchasing more goods (OR=0.88, 95% CI:0.81-0.95), attending large gatherings (OR=0.87, 95%CI: 0.81-0.93), and using medical masks (OR=0.56, 95% CI:0.50-0.62) decreased by 12%, 13%, and 44%, respectively. Gen X and Millennial participants had 56% to 76% higher odds, respectively, of increased purchasing behavior, compared to Baby Boomers. Results suggest politicization of response recommendations. Democrats had 30% lower odds of attending large gatherings (OR=0.70, 95% CI:0.50-0.97), and 48% lower odds of using medical masks (OR=0.52, 95% CI:0.34-0.78), compared to Republicans. Conclusions: This survey is one of the first attempts to study determinants of knowledge and behaviors in response to the COVID-19 pandemic in the U.S. A national, coordinated effort at pandemic response may ensure better compliance with behavioral recommendations to address this public health emergency.

Some of the most important problems in the world require an understanding and acceptance of science by the general public, including addressing health problems such as the emergence of the novel coronavirus (SARS-Cov-2) and subsequent disease (COVID- 19) transmission. SARS-Cov-2 first emerged in December 2019 in Hubei Province, near Wuhan, China 1 . By mid-January 2020, Thailand and Japan were the first countries outside of China to report COVID-19 cases 1 .

The Chinese government subsequently quarantined the greater Wuhan area on January 23, 2020 to prevent COVID-19 spread 2 .

On January 21, 2020, the first COVID-19 case in the United States was reported in Washington State 3 and it was later reported that public health officials thought the virus was prevalent in the community for at least several weeks prior 4 . In the United States, the federal government ordered certain flights from China halted, and screening of passengers from other locations at different ports of arrival 5 . The Centers for Disease Control and Prevention (CDC), and the National Institutes of Health (NIH), began making recommendations, based on the scientific knowledge of the situation, to limit social contacts, encourage wise use of medical supplies including masks, and assure the public about the reliability of the food and consumable goods supplies 6 . However, even after these recommendations, there were reports of college students waiting in long lines at bars to celebrate their campuses closing, 7 people buying medical grade masks, 8 and hoarding everything from toilet paper to eggs and milk, 9 even as the President sought to reassure the public that the supply of food and supplies was secure 10 .

Scholarship on public understanding of science (PUS) aims to explain public understanding of, involvement in, and trust in science. In the face of the current pandemic, this requires the public to understand and trust those who are making recommendations to limit exposure and spread of illness. The deficit model of PUS posits that a lack of support for science (and a subsequent rejection of recommendations) is due to a lack of understanding about science, and if scientists can find a way to fill this knowledge deficit, then support for science will increase. A more contemporary view of PUS is that the public's knowledge is not deficient, but rather there is a deficit in trust in science, and in scientific experts specifically. Because of an increasing lack of trust in these institutions, Solomon 11 observed that there is an increased personal rejection of science which then leads to lower levels of scientific literacy and understanding of science. Low literacy and understanding may influence people to not follow recommendations for addressing science-based problems, as evident with the current pandemic.

Much of the PUS literature examines trends in scientific knowledge (albeit self-reported knowledge, for the most part) and attitudes about science. Results are mixed as to whether increased knowledge leads to positive attitudes (variously described as trust, support, confidence, and support for funding) about science. Allum et al 12 observed a small positive correlation between knowledge about science and positive attitudes about science and Miller 13 reports that there is public support for science even in the face of a scientific literacy rate of 20%. The public's support for science is necessary when addressing many important social issues, including an immediate need for the public to understand, and trust, the science about the novel Coronavirus pandemic currently plaguing the world. If the public does not trust the underlying science about these issues and does not trust in institutions that are tasked with managing this threat, it will be difficult to count on public support for policies to address these issues. This paper describes a cross-sectional online survey designed to gauge public knowledge and behaviors about COVID-19 in the United States. Zhong et al 14 

China, approximately one week after the Hubei Province was put on lockdown (approximately . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.03.31.20048967 doi: medRxiv preprint eight weeks after the first case emerged), to determine the level of knowledge and public sentiment about the emerging pandemic in China. This study essentially replicates questions about knowledge from that study, while asking about more specific behvaiors, with a sample drawn from an online work platform (Amazon's Mechanical Turk) to determine level of knowledge about COVID-19 and characteristics that influence knowledge and behaviors toward COVID-19. This is among one of the first attempts to investigate determinants of knowledge and behaviors in the public related to COVID-19 in the United States.

The general hypothesis guiding this research is that lower levels of knowledge about the coronavirus pandemic are associated with behaviors that are contrary to current guidelines that suggest that panic buying is not necessary, to avoid large gatherings, and to avoid using medical masks. Further, there are differences in knowledge and behaviors in different age groups, by sex, education level, race, income, and political party identification.

This cross-sectional study recruited respondents from Amazon Mechanical Turk (MTurk).

MTurk is an online platform for recruiting remote workers to complete small tasks for small amounts of money. Some studies report that MTurk sample demographics are closer to the U.S. general public than typical university samples, 15, 16 and tend to be more diverse than other Internet samples 17 . MTurk provides a quick, inexpensive method to collect data from a wide cross-section of the general public.

The MTurk interface allows Requestors (myself) to advertise Human Intelligence Tasks (HITsthe survey in this case) to Workers (survey participants). I advertised for workers age 18 and older who resided in the U.S. and offered to pay them $1.00 to complete the survey. The . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.03.31.20048967 doi: medRxiv preprint Institutional Review Board at Michigan State University determined that this research was exempt from full board review. Participants provided consent by answering a yes-no question at the start of the survey before they could move to the first question.

The survey was administered in two parts. The first part asked participants basic demographic characteristics including year of birth, which was used to determine age and generational membership, i.e. Baby Boomers, Gen X, 18 education, sex, income, race, political party affiliation, and place of residence (U.S. State). Age was included to determine differences in knowledge and behavioral patterns based on age. Some reports in the U.S. essentially call-out different age groups for ignoring public health recommendations 7, 19 . In addition, there are well described patterns of health literacy based on education level, 20 and race 21 which may not be present in a homogeneous society such as China. Political party identification is associated with many attitudes and behaviors in the U.S. related to science and science-based recommendations 22, 23 . Leaders from both major parties in the U.S. have reacted differently to the COVID-19 pandemic, likely influencing those who follow them 24, 25 .

The second part of the survey included 12 questions that were adapted from Zhong et al 14 to measure knowledge about COVID-19, including clinical characteristics, transmission, and prevention and control. The knowledge questions were scored with one point for each correct question and an aggregate score calculated (range 0 to 12), with higher scores indicating more knowledge about COVID-19. Three additional questions were asked to determine participation in specific behaviors related to recommendations from CDC and/or NIH including whether participants had spent more money than usual in the last two weeks on cleaning supplies, personal hygiene products, and food (a proxy measure of hoarding), whether they had gone to . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.03.31.20048967 doi: medRxiv preprint any place in the last five days where there were more than 50 people present (contradicting CDC recommendations to avoid such gatherings), and if they had worn a mask when leaving the home in the last 5 days (contradicting CDC, NIH, and healthcare official guidance).

Sample characteristics were generated using frequency analysis and other descriptive statistics as appropriate (Table 1) . Knowledge scores were compared using independent sample t-test for differences in mean score between males and females, as well as groups based on whether people had engaged in hoarding activity or not, had attended large gatherings or not, and had worn masks or not. In addition, independent sample t-tests were used to determine differences in mean age between people who had engaged in these activities or not. Analysis of Variance (ANOVA) was used to determine differences in mean knowledge scores among groups based on education, race, income, political party, and generational age groups (i.e. Baby Boomers, Gen X, etc.) ( Table 2 ). Multivariable linear regression was used to determine which demographic characteristics influence knowledge scores, while binomial logistic regression was used to determine which characteristics influence participation in hoarding behavior, attending large group events, and using masks ( Table 3 ). All analyses were conducted using SPSS (v.25) 26 .

A total of 1,070 participants completed the survey. The use of VPN networks allows people from all over the world to mimic U.S. IP addresses and so I asked for the U.S. state of residence for each participant and matched those to each IP address location to determine matches. I then excluded responses from participants whose IP address location did not match their given State.

Thirty-six participants were excluded for a final sample size of 1,034. The survey was offered to MTurk workers on March 17, 2020 at 4:05 PM Eastern time, and all 1,070 responses were . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

is the (which was not peer-reviewed) The copyright holder for this preprint Knowledge scores were significantly different between groups based on sex, generational ages, education, race, income, and political party identification. In general, Baby Boomers, females, those with some college, and higher incomes were more knowledgeable about COVID-19, while Black/African Americans, and Republicans were less knowledgeable ( Table 2) .

Regarding behaviors, participants who reported spending more money in the last two weeks, or going to gatherings with more than 50 people, or wearing masks outside the home, were less knowledgeable about COVID-19 compared to participants who did not report these activities. In . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.03.31.20048967 doi: medRxiv preprint addition, participants who reported the above behaviors were also significantly younger, except for increased spending where there was no significant difference in age (Table 2) . 

The PUS literature posits that an increase in knowledge leads people to understand science, and trust in the institution of science. Extending this to the current COVID-19 pandemic, I hypothesized that increased knowledge should lead to willingness to follow public health recommendations. In this sample, lower knowledge is associated with self-reports of engaging in purchasing more goods than necessary, attending gatherings of more than 50 people, and wearing medical masks outside the house. In addition, there were differences in knowledge about COVID-19 based on age group. In fact, contrary to recent U.S. media, Baby Boomers in this . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.03.31.20048967 doi: medRxiv preprint sample are more knowledgeable about COVID-19 than all other age groups and are less likely to engage in purchasing behavior that could be considered hoarding. In general, people who do not engage in these behaviors had significantly higher knowledge scores. Finally, people who reported attending large gatherings and wearing masks in public were younger on average.

The average knowledge score for this entire sample is about 9.72 out of 12 total points (approximately 80%) at a time about eight weeks after the first case was diagnosed in the U.S.

Approximately eight weeks after the first diagnosis in China, the mean knowledge score for a sample of Chinese citizens was 10.8/12 (approximately 90%) 14 and it was suggested that the knowledge of Chinese citizens is high because of their experiences with the SARS outbreak in the early 2000s, and the observation that this sample was relatively affluent and highly educated.

In this sample nearly 30% of people reported attending gatherings or going to places with more than 50 people in the previous five days, contrary to advice from the CDC since March 12, 2020 (survey conducted on March 17, 2020). In China, only 3.6% of people reported going to crowded places in the previous two weeks 14 . It is possible that the coordinated effort from, as well as the unchecked authority by, the Chinese government, to lockdown provinces provided most of the motivation for Chinese citizens to obey these mandates. To date, there has not been a coordinated effort by the U.S. government to lockdown the nation. There is some debate whether the federal government even the Constitutional authority and so individual states are left to make decisions about "shelter at home" policies and similar efforts. As of this writing, California, Illinois, New York, Washington, Michigan, Massachusetts, Indiana, Oregon, and West Virginia have issued stay-at-home orders. With about 1 in 3 U.S. citizens ordered to stay home it is likely in the coming weeks that fewer people will report attending large gatherings.

. CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.03.31.20048967 doi: medRxiv preprint Use of masks is an evolving, and cultural phenomenon. In Asia, people are encouraged, and even mandated to wear masks outside the house. In China, only 2.0% of people reported not wearing masks outside the home 14 . In this sample, approximately 76% of people did not wear masks outside the home in the last five days, which is perhaps reflective of the CDC and NIH recommendations that the general public not use masks so that they are saved for front-line healthcare workers 28 . However, it is probably more likely that masks could not be found in the U.S. because of lack of supply combined with hoarding behavior 10 . Still, 24% of people reported using masks, indicating that a large section of the U.S. public chose to ignore recommendations.

It is important to note that the debate on masks has changed even since this survey was conducted only six days ago. The FDA and CDC are currently debating the merits of wearing masks in public because of the understanding that many people with mild symptoms may not even know they are infected with COVID-19. Mask use could prevent infecting others by asymptomatic carriers. . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.03.31.20048967 doi: medRxiv preprint

Conversely, Republicans believe scientists should stay out of policy debates. These attitudes may be reflected in these results that Republicans have lower knowledge about COVID-19 and have higher odds of participating in behaviors that are not recommended by authorities to stem the tide of the current pandemic.

There are some limitations to this research. First, knowledge questions are not validated and scientific knowledge is currently a moving target. For example, while the current consensus is that eating wild animals will not transmit the disease, living and working in close proximity to animals clearly influenced this outbreak and could influence future outbreaks. As such, the argument for banning wet markets in China is gaining momentum, but knowledge about proximity to animals, as opposed to using them as a food source, might be conflated. In addition, knowledge regarding who is most at risk for COVID-19 may change as the pandemic proceeds, as well as with experiences in different countries. For instance, fewer younger people in China were infected, while in the U.S. a different pattern appears to be emerging 29 . Next, this was a convenience sample of U.S. residents from every state in the country, but people were able to self-select based on their interest and experience with the topic. It is possible that sample demographics may not completely represent the U.S. public. Finally, although the survey questions were not able to be validated given the fast-moving nature of the pandemic response in the U.S., the questions do have face value in the context of the situation at the time the survey was conducted.

This survey is one of the first attempts to describe determinants of U.S. public knowledge and behavioral response to the emerging COVID-19 pandemic in the United States. While knowledge about COVID-19 is generally high, there are differences in knowledge based on age, . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.03.31.20048967 doi: medRxiv preprint sex, education, income, race, and political party identification. These differences appear to prevent a coordinated effort at slowing the spread of the pandemic in the U.S. in these early days.

Ignoring official recommendations for crowd avoidance, use of medical supplies, and purchasing behaviors that signal hoarding of goods, does not bode well for efforts to contain the spread of the virus and limit exposures to vulnerable populations. Without a coordinated national response, it is likely the U.S. will experience a longer, more drawn out battle than if such coordination would occur. The author reports no conflicts of interest.

. CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.03.31.20048967 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

is the (which was not peer-reviewed) The copyright holder for this preprint . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.03.31.20048967 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.03.31.20048967 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.03.31.20048967 doi: medRxiv preprint 

World Health Organization. Novel coronavirus (2019-nCoV) situation report -1

Lockdowns rise as China tries to control virus

Novel coronavirus outbreak 2020

The New York Times. Coronavirus may have Spread in U.S. for Weeks

Trump declares coronavirus outbreak a public emergency, will ban foreign travel from China and quarantine US citizens

What coronavirus? College students flood Myrtle Beach to party as COVID-19 spreads. The Charlotte Observer

Americans hoarding hand sanitizer, face masks and oat milk amid coronavirus fears

Hoarding in the USA? Coronavirus sparks consumer concerns

Trump calls on Americans to cease hoarding food

Reception and rejection of scientific knowledge: choice, style and home culture

Science knowledge and attitudes across cultures: a meta-analysis

Public understanding of, and attitudes toward, scientific research: what we know and what we need to know

Knowledge, attitudes, and practices toward COVID-19 among Chinese residents during rapid rise period of COVID-19 outbreak: a quick online cross-sectional survey

Conducting behavioral research on Amazon's Mechanical Turk

Amazon's Mechanical Turk: A new source of inexpensive, yet high-quality, data?

Convincing Boomer parents to take the coronavirus seriously

The New Yorker. Published

Determinants of health literacy and health behavior regarding infectious respiratory diseases: a pathway model

Racial disparities in health literacy and access to care among patients with heart failure

Trends-Americans' trust in science and scientists

Trust and mistrust in Americans' views of scientific experts

Negative partisanship: Why Americans dislike parties but behave like Rabid Partisans

Age (Reference: Baby Boomers)