Predictors of willingness to get a COVID-19 vaccine in the U.S

Background As COVID-19 vaccine distribution efforts continue, public health workers can strategize about vaccine promotion in an effort to increase willingness among those who may be hesitant. Methods In April 2020, we surveyed a national probability sample of 2279 U.S. adults using an online panel recruited through address-based sampling. Households received a computer and internet access if needed to participate in the panel. Participants were invited via e-mail and answered online survey questions about their willingness to get a novel coronavirus vaccine when one became available. The survey was completed in English and Spanish. We report weighted percentages. Results Most respondents were willing to get the vaccine for themselves (75%) or their children (73%). Notably, Black respondents were less willing than White respondents (47% vs. 79%, p < 0.001), while Hispanic respondents were more willing than White respondents (80% vs. 75%, p < 0.003). Females were less likely than makes (72% vs. 79%, p < 0.001). Those without insurance were less willing than the insured (47% vs. 78%, p < 0.001). Willingness to vaccinate was higher for those age 65 and older than for some younger age groups (85% for those 65 and older vs. 75% for those 50–64, p < 0.017; 72% for those 35–49, p < 0.002; 70% for those 25–34, p = NS and 75% for ages 18–24, p = NS), but other groups at increased risk because of underlying medical conditions or morbid obesity were not more willing to get vaccinated than their lower risk counterparts. Conclusions Most Americans were willing to get a COVID-19 vaccine, but several vulnerable populations reported low willingness. Public health efforts should address these gaps as national implementation efforts continue.


Introduction
As nations across the globe continue to distribute COVID-19 vaccinations to eligible individuals, some health officials argue that the time is fast approaching when issues with logistics and short supply will be replaced by the need to reach those who are more reluctant to vaccinate [1,2]. A global study found that 72% of people would take a Covid-19 vaccine if proven safe and effective, but willingness varied widely between nations [3]. Studies in Australia and Indonesia have found that willingness varies widely depending on the vaccine's effectiveness and that safety is a concern [4,5].
Given the unique features of this particular virus-its highly contagious nature, high severity for certain groups, and the significant impact it has had on people's daily lives through social distancing and business closures-we sought to better understand the American public's willingness to get a COVID-19 vaccine. We also sought to characterize willingness among groups most at risk for severe COVID-19 (e.g., people age 65 or older, those with specific underlying medical conditions, or with a body mass index (BMI) > 40; and Black and Hispanic Americans) [6]. We examined these questions in a national survey of U.S. adults.

Participants and procedures
We surveyed U.S. households in April 2020 using a prerecruited, address-based web panel consisting of 55,000 adults. The panel is based on probability sampling covering both the online and offline populations in the United States. Households received a computer and internet access if needed to participate in the panel. The resulting panel includes representation from listed and unlisted telephone numbers, telephone and non-telephone households, mobile phone-only households, and households with and without internet access. A random sample of panel members yielded 2279 survey respondents, for a stage completion rate of approximately 34% [7]. Participants were invited via e-mail. The survey was completed online, in English for most participants; 227 respondents completed it in Spanish. RTI International's Institutional Review Board designated the study protocol as exempt from human subjects approval. All methods were carried out in accordance with relevant guidelines and regulations. Informed consent was obtained from all participants.

Measures
The survey assessed willingness to receive a vaccine when one becomes available, beliefs about the virus, and demographic characteristics (Table 1).

Analyses
We report unweighted frequencies and weighted percentages and odds ratios. A post-stratification process was used to adjust for survey nonresponse and any noncoverage, undersampling, or oversampling resulting from the study-specific sample design. The panel provider used data from the U.S. Census Bureau (https://www.census.gov/programs-surveys/cps.html) to weight all respondents to benchmarks for gender, age, race/ethnicity, education level, Census region, household income, metropolitan area, and homeownership status.
Weighted multivariable logistic regression examined predictors of willingness to vaccinate oneself and one's child (dichotomized as willing vs. not willing). Predictor variables were gender, education, race/ethnicity, Hispanic origin, health insurance, income, previous seasonal influenza vaccination, COVID-19 severe disease risk factors (age 65+, BMI > 40, chronic health conditions), perceived threat of the virus, and worry. Critical alpha for statistical significance was set at .05.

Results
Twenty-two percent of participants were age 65 or older (see Table 2). Seventeen percent had a BMI over 40, and 26% had one or more underlying medical conditions that could put them at higher risk of severe illness from the virus. Fifty-four percent had received the seasonal influenza vaccine in the last year, while 18% reported that they had never received the seasonal influenza vaccine. Twenty-seven percent of respondents had children.
Overall, 75% were willing to get a COVID-19 vaccine when it becomes available (37% strongly agreed and 38% agreed). Willingness for their children was similar, at 73% (32% strongly agreed and 41% agreed). Asked about the level of threat from the virus posed to self and family, 8% answered "very high," 15% answered "high," 40% said "moderate," and 33% said "low or very low." 4% did not know. Sixty-eight percent reported being worried about getting the novel coronavirus.
As for others in high-risk groups, willingness was higher for those age 65 or older, and Hispanics compared to other respondents (Table 2). However, those with underlying medical conditions and BMI > 40 were not more willing to get vaccinated than those without these risk factors.
Willingness to get vaccinated was higher among those who received a seasonal influenza vaccine in the past year than those who had not (90% vs. 59%, OR = 4.70, 95% CI = 3.55-6.23). Willingness to vaccinate children showed a similar pattern of results (Table 3).

Discussion
In a large nationally representative survey conducted in April 2020, the willingness of U.S. adults to receive a novel coronavirus vaccine was high. Most U.S. adults were willing to get the vaccine for themselves and for their children, which is encouraging in a climate in which vaccine hesitancy around childhood immunizations has received global attention. Those most willing to get a COVID-19 vaccine were over age 65, had a bachelor's degree or higher, were worried about the novel coronavirus, and had received an influenza vaccine in the previous year.
Our estimate of the percentage of people who indicated that they were not willing to get vaccinated is similar to other polls during the same time period (e.g., [8]), but our estimate of those willing to vaccinate was somewhat higher. One explanation is that our survey forced respondents to report willingness or unwillingness to get the vaccine because the item did not include "don't know" or "not sure" response options. Thus, our findings suggest that those who report they are unsure in some polls may be willing to do so when pressed to decide. More research is needed to determine the conditions under which that willingness will translate into action when the time comes to vaccinate.
Studies during previous infectious disease outbreaks suggest somewhat lower willingness to get a new vaccine during a pandemic of a novel pathogen. During the 2009 influenza A (H1N1) pandemic, 64% of adults in two North Carolina counties intended to receive a vaccine when available [9]. In that study, seasonal influenza vaccination was associated with intentions to get the H1N1 vaccine, but other factors (H1N1 vaccine knowledge, age, gender, race/ethnicity, work status, and having children under age 18 living at home) were not. During the 2016 Zika virus pandemic, 56% of respondents in a national study were willing to get the hypothetical vaccine [10]. The higher willingness to obtain a COVID-19 vaccine than those for these two previous disease outbreaks may reflect higher media coverage and higher perceived threat of COVID-19.
Black respondents were markedly less willing than White respondents to get a COVID-19 vaccine, a finding that is consistent with adult vaccination trends of, for example, lower seasonal influence vaccine coverage in the same group [11]. Given that Black people have experienced disproportionately high rates of hospitalization and death from COVID-19 [12], the disparity in COVID-19 vaccine willingness raises the possibility that vaccination could amplify existing disparities. Respondents of Hispanic origin expressed higher willingness than their non-Hispanic counterparts to get the vaccine. Although adult vaccine coverage has generally been lower for Hispanics than non-Hispanic Whites in recent years [13], their greater willingness to get a COVID-19 vaccine may reflect the virus's disproportionate impact on Hispanic communities (e.g., [14]). It will be imperative to ensure equitable access to the vaccine even as efforts address vaccine hesitancy among diverse communities. The upcoming report by the National Academies of Sciences, Engineering, and Medicine on this question will provide critical guidance for the nation on addressing disparities in vaccine distribution and uptake.
People with lower socioeconomic status were generally less willing to get vaccinated. Lower income and lower education respondents had less willingness to get the vaccine. Those without insurance were nearly 30% less willing to get the vaccine than insured respondents. Interventions aimed at reducing financial barriers to vaccination and increasing vaccine availability are among the most successful [15]. Equity-oriented policy   initiatives focused on cost and access barriers will need to be coupled with communication efforts to make sure those who need them are aware of the options available to them.

Implications for health communication
Our findings have implications for public health communication interventions. Perceived and actual risk showed different patterns of association in our study. Individuals with higher risk appraisals-perceived threat and greater worry-were more willing to get vaccinated, a finding that is consistent with previous research [16]. However, willingness to get a vaccine was not always associated with being at objectively high risk for severe COVID-19. Individuals over age 65 were more willing than some younger age groups; however, individuals with underlying medical conditions and those with BMI > 40 were not more willing. Healthcare providers treating individuals with these conditions may have an opportunity to help them better understand their risk.
The public has been inundated with information about COVID-19. The question is no longer whether to communicate, but how to effectively provide accurate information to meet the vaccination needs of U.S. adults. The Extended Parallel Process Model [17] suggests that health messages are more likely to result in health behavior when they are threatening and lead people to believe the recommended behavior is effective at reducing the threat. Messages that emphasize both the significant threat of this virus to people most at risk and the effectiveness of a recommended vaccine (once that is known) could help persuade those who are on the fence about vaccination [18]. That message strategy paired with an audience segmentation approach that addresses those less willing or unsure about vaccination (e.g., those with lower education, those of Black race, and those without insurance) would be a good starting point for health communication campaigns.

Strengths and limitations
Our study relied on a representative national sample of U.S. adults to quickly gather data on a pressing public health problem. We recognize that we were asking about a hypothetical behavior. Nonetheless, understanding how perceptions are changing can provide helpful information for public health planning and communication. Prior research has shown that behavioral willingness is highly correlated with actual behavior and may even predict behavior independent of intentions [19].

Conclusions
The findings from this study provide direction about areas to focus public health planning and communication efforts as vaccine efforts continue.
Future research should focus on reasons for some of the differences we found as well as explore how communication about the virus and vaccination has changed perceptions since this survey was conducted in April 2020. These results offer a snapshot of public perceptions early in U.S. COVID-19 pandemic experience. As public perceptions change over time, we can expect communication to continue to play a critical role in generating confidence in the programs and policies that allow nations to achieve high vaccine uptake [15]. Social processes including provider recommendations, social norms, and social media sharing also likely play an important role in vaccine uptake. Our study shows great willingness among Americans to vaccinate and yet we also can see possibilities for notable differences between some groups that warrant efforts to mitigate health inequity.

Acknowledgements NA
Authors' contributions BJK contributed to conception of the paper; designed the analyses; interpreted the data; developed the first draft and revised the paper. LBS and LAM designed the survey and data collection; and substantively revised the paper. BGS helped conceive of the paper and analyses; interpreted the data and revised the draft. NTB helped to develop the plan for analyses and revised the draft. CB and AF helped with design of the survey and collection of the data and also revised the draft. PDDM helped with design of the