Vaccines and immunization remain one of the most cost-effective ways of preventing disease and it is estimated to avert 2–3 million deaths annually1.Vaccines have resulted in the eradication of deadly illnesses, turning once devastating diseases to distant memories. For vaccination campaigns to be successful from a public health perspective, enough people must take them to achieve herd immunity2,3. Widespread vaccine hesitancy is therefore a public health crisis and has been listed as one of the top ten threats to global health. Defined by the World Health Organization (WHO) as the reluctance or refusal to vaccinate despite the availability of vaccine, vaccine hesitancy is not necessarily new4. The reasons for vaccine hesitancy are complex and not confined to complacency and lack of confidence in vaccines5. However, the COVID-19 vaccine hesitancy demands special attention because this is by far the largest, most destructive, and most sustained public health disaster in the past many decades.

COVID-19 was declared a global pandemic on 11 March 2020, allowing the rapid deployment of resources including manpower and necessary non-pharmacological interventions to prevent further spread. In the US, the urgency of COVID-19 paved way for one of the fastest vaccine development and approval timelines, facilitated in no small part by a dynamic public–private partnership initiated by the US government known as ‘Operation WARP Speed’6. This speedy process, while facilitating access to vaccines to those in critical need, has also raised skepticism in the general population regarding whether due process was followed and whether the vaccine is safe. In an international study of G7 countries, only 12% of Americans strongly agreed that if the public authorities propose a vaccination against COVID-19, they would be confident that the proposed vaccine will not be dangerous. Furthermore, 22% said they were worried about the safety of COVID-19 vaccines because of the speed with which they are being developed and produced7.

Studies have been conducted at the national level to examine prevalence and determinants of vaccine hesitancy. For example, Fisher et al.8 found that vaccine hesitancy was higher among younger adults, females, people with high school education level or less, lower annual household income, people living in rural settings, persons who declined the influenza vaccine in the past year, and those who live in the Southern US8. A Pew survey report released on 3 December 2020, showed that with respect to political affiliation, 50% of republican-leaning persons compared with 69% of democratic-leaning persons would accept the COVID-19 vaccine when it became available9.

Less is, however, known about state-specific variations in vaccine hesitancy; this is important as vaccine rollout in the US is decentralized and at the state level. State-specific data can therefore inform public health planning, programs, and policy. Furthermore, given that wide variations exist in the burden of COVID-19 incidence and mortality rates across states, we were interested in examining whether this variation was associated with vaccine hesitancy. In line with the Health Belief Model which posits that an individual’s course of action with regard to health-related behavior is determined by among other factors, the perceived severity of the condition and their perceived susceptibility to it, not only the perceived benefits or harms.


Data source

Data came from the Household Pulse Survey (HPS), an ongoing weekly, web-based, anonymous, cross-sectional survey conducted by the U.S. Census Bureau and multiple federal agencies. The analyzed wave of the survey was conducted during 6–18 January 2021. The number of respondents was 68348. We obtained state-specific data for the number of COVID-19 cases and deaths from the start of the pandemic up to 5 January 2021 (the period before and just up to the survey), from the COVID Tracking Project website. Due to differences in state population sizes, we standardized the case burden by estimating incidence and mortality rates by state. As this was analysis of secondary, de-identified, and publicly available data, the study was deemed as non-human subject research, IRB therefore was not sought.

Measurements and variables

Sociodemographic characteristics

States were classified into categories based on US census regions as well as by their political leanings (democratic and republican) using the 2020 electoral college votes10. Individual-level sociodemographic characteristics were assessed in the survey including age, gender, race/ethnicity, education level, income, and housing type (important because of concerns about transmission risk in communal dwellings). The latter was assessed as follows: ‘Which best describes your house or apartment building?’. Responses were recoded to include: ‘detached mobile units such as boat, recreational vehicle, mobile home’, ‘detached single house’, ‘attached single house’, ‘up to 4 apartments in the building’, and ‘≥5 apartments in the building’.

Past COVID-19 diagnosis and receipt of COVID-19 vaccine

Ever diagnosis of COVID-19 was defined as a response of ‘Yes’ to the question: ‘Has a doctor or other health care provider ever told you that you have COVID-19?’. Participants were classified as having received a COVID-19 vaccine if they answered ‘Yes’ to the question: ‘Have you received a COVID-19 vaccine?’. Those answering ‘No’ were classified as not having received it. Eligible participants were further asked: ‘Did you receive (or do you plan to receive) all required doses?’. Categorical response options were ‘Yes’ or ‘No’.

Intent to decline COVID-19 vaccine and associated reasons

The survey asked participants ‘Once a vaccine to prevent COVID-19 is available to you, would you …’. Participants could select one of the following response options: 1) ‘Definitely get a vaccine’, 2) ‘Probably get a vaccine’, 3) ‘Probably not get a vaccine’, and 4) ‘Definitely not get a vaccine’. We classified response 1 as definite intention to receiving vaccines; responses 2–3 as unsure if to accept or decline the vaccine, and response 4 as definite intention to decline the vaccine. Among the last group, follow-up questions were posed in the survey to assess reasons for showing hesitancy towards vaccines: ‘Which of the following, if any, are reasons that you: only probably will/probably won’t/definitely won’t, get a COVID-19 vaccine/won’t receive all required doses of the COVID-19 vaccine? There were 11 possible responses, and respondents could select all that applied: ‘I'm concerned about possible side effects of a COVID-19 vaccine’; ‘I don't know if the COVID-19 vaccine will work’; ‘I don't believe I need a COVID-19 vaccine’; ‘I don't like vaccines’; ‘My doctor has not recommended it’; ‘I plan to wait and see if it is safe and may get it later’; ‘I think other people need it more than I do right now’; ‘I am concerned about the cost of the COVID-19 vaccine’; ‘I don't trust COVID-19 vaccines’; ‘I don't trust the government’; or ‘Other reasons’. A follow-up question was posed to the group who did not believe in the vaccine: ‘Why do you believe that you don't need a COVID-19 vaccine?’ Categorical response options were: ‘I already had COVID-19’; ‘I am not a member of high-risk group’; ‘I plan to use masks or other precautions instead’; ‘I don't believe COVID-19 is a serious illness’; ‘I don't think vaccines are beneficial’; or ‘Other reasons’.

Statistical analysis

Data processing and analyses were conducted using StataV15.1. Data were weighted to yield representative results at the national and state level. Prevalence of vaccine hesitancy was calculated overall and by state. Of the population who reported vaccine hesitancy, we analyzed the reasons for their hesitancy. Using pooled national data, we examined correlates of vaccine hesitancy using a multivariable Poisson regression model. Probabilistic model selection was done using the Akaike and Bayesian information criterion. Independent variables included in the final regression model were age, gender, race/ethnicity, education level, annual household income, marital status, type of housing, US region, political leaning of state, and previous diagnosis of COVID-19. Statistical significance was assessed at p<0.05, and all tests of significance were two-tailed.


Population characteristics

Within the pooled sample, mean age was 47.2 years (46.9–47.5). By age categories, 10.5%, 35.3%, 34.2%, 2% were 18–24, 25–44, 45–64, and ≥ 65 years, respectively. Overall, 48.4% were male, and 51.6% were female. By race/ethnicity, 62.6%, 11.3%, 5.3%, 17.2%, and 3.8%, were non-Hispanic White, non-Hispanic Black, non-Hispanic Asian, Hispanic, and non-Hispanic other race. Most (55.1%) were married, while 18.3% were either widowed, divorced, or separated, and 26.7% identified as never married. Only 8.5% had less than high school education, 30.6% graduated from high school, 21.3% had some college education, 26.7% had a college degree, and 12.9% had a professional degree. Furthermore, 14.4%, 24.4%, 31.1%, 22.1%, and 8.0%, had annual household income (US$) less than 25000, 25000–50000, 50000–100000, 100000–200000 and >200000, respectively (Table 1).

Table 1

Demographic characteristics of the population

Sociodemographic characteristics%
Age (years)
Non-Hispanic White62.6
Non-Hispanic Black11.3
Non-Hispanic Asian5.3
Non-Hispanic other race3.8
Income (US$)
Education level
Less than high school8.5
High school30.6
Some college21.3
College graduate26.7
Professional degree12.9
Marital status
Ever diagnosed with COVID-19
Plan to get COVID-19 vaccine
Type of housing
Detached mobile units5.1
Detached single house69.1
Attached single house7.9
Up to 4 apartment building6.5
≥5 apartment building11.4
Political Party affiliation

Past COVID-19 diagnosis and receipt of COVID-19 vaccine

Overall, 14.6% reported they had ever been diagnosed with COVID-19 (range: 0.04% in Vermont to 14.8% in California) and 7.7% reported that they had ever received a COVID-19 vaccine (range: 0.2% Wyoming to 11.8% in Texas). Consistent patterns (or reverse patterns as the case might be) were seen in the groups most likely to be diagnosed with COVID-19 versus those most likely to have received a vaccine. The top 10 states in terms of self-reported COVID-19 diagnosis were: California 14.8%, Texas 9.5%, Florida 5.8%, New York 5.7%, Illinois 4.4%, Ohio 4.4%, Georgia 3.8%, Pennsylvania 3.4%, Indiana 2.6%, and Arizona 2.4%. The top 10 states in terms of receipt of a COVID-19 vaccine were: Texas 11.8%, California 8.4%, Florida 6.1%, New York 6.0%, Pennsylvania 3.6%, Illinois 3.4%, Ohio 3.3%, Georgia 3.0%, North Carolina 2.9%, Michigan 2.9% (Table 2).

Table 2

Percentage of the population that reported previous COVID-19 diagnosis, and receiving the COVID-19 vaccine by state

StatePrevious COVID-19 diagnosis %Received COVID-19 vaccine %
District of Columbia0.070.2
New Hampshire0.30.5
New jersey2.42.8
New Mexico0.60.8
New York5.76
North Carolina2.42.9
North Dakota0.30.3
Rhode Island0.50.4
South Carolina1.71.2
South Dakota0.40.5
West Virginia0.40.6

Intent to decline COVID-19 vaccine and associated reasons

Of those who had not received any COVID-19 vaccine, 50.95% would receive the COVID-19 vaccine when it is available, 39.52% said they were unsure if they would or would not get the vaccine, and 9.53% said they would not get the vaccine. By state, the percentage who indicated intent to decline the vaccine was lowest in the following 10 states: Rhode Island 11.7%, Massachusetts 13.9%, California 16.2%, Connecticut 17.0%, District of Columbia 17.3%, Washington 17.5%, Delaware 17.6%, New Hampshire 18.0%, New Jersey 18.1%, and Virginia 18.4%. Conversely, this percentage was highest in the following 10 states: Louisiana 40.2%, Mississippi 35.9%, Idaho 34.8%, Alabama 34.7%, Wyoming 34.1%, Montana 33.5%, South Carolina 32.1%, Arizona 30.8%, Indiana 30.7%, and Oklahoma 30.7% (Table 3).

Table 3

Percentage intent to decline COVID-19 vaccination, political leaning of state, total number of COVID-19 cases, and COVID-19 related mortality, by state in ascending order

State% (95% CI)Electoral voteTotal casesTotal deaths
Rhode Island11.7 (7.7–15.7)aDEM938521870
Massachusetts13.9 (10.9–16.9)DEM39720212734
California16.2 (13.5–19.0)DEM2452334b27003
Connecticut17.0 (13.4–20.6)DEM1969686192
District of Columbia17.3 (12.0–22.6)DEM30166801
Washington17.5 (14.8–20.1)DEM2564353482
Delaware17.6 (13.2–21.9)DEM61100947
New Hampshire18.0 (12.7–23.4)DEM47992792
New jersey18.1 (14.3–21.8)DEM55141919382
Virginia18.4 (14.5–22.2)DEM3719135191
New Mexico18.8 (15.0–22.6)DEM1484992594
Colorado18.9 (15.0–22.8)DEM3468934991
Hawaii19.8 (13.6–26.0)DEM22650289
Maine20.6 (14.9–26.3)DEM26565369
Illinois21.0 (17.5–24.4)DEM99171918562
Utah21.1 (18.1 – 24.2)REP2889511312
Vermont21.2 (14.6–27.7)DEM8038149
Nebraska21.7 (17.1–26.3)REP1695851682
New York21.8 (17.3–26.3)DEM104102830802c
Maryland21.9 (18.0–25.7)DEM2897586082
Minnesota22.9 (19.2–26.7)DEM4252615461
Texas22.9 (19.5–26.2)REP184315328219
Pennsylvania23.2 (19.3–27.0)DEM67391516546
Nevada23.4 (19.3–27.4)DEM2354553235
Oregon23.4 (20.1–26.7)DEM1184531506
Iowa23.6 (19.3–27.9)REP2438293999
Wisconsin23.6 (19.4–27.8)DEM5318905366
Kansas24.5 (20.6–28.5)REP2313172897
Florida25.0 (21.1–28.9)REP136777822515
Michigan25.6 (22.1–29.1)DEM54664213608
Arizona25.7 (22.4–29.1)DEM5674749317
Kentucky26.6 (22.0–31.2)REP2808362772
South Dakota27.0 (21.4–32.6)REP1010761513
Tennessee27.3 (23.0–31.6)REP6176497267
Georgia28.0 (23.3–32.6)DEM70615411072
Ohio29.0 (24.5–33.4)REP7350039247
West Virginia29.3 (23.6–34.9)REP931621442
North Carolina29.4 (24.5–34.3)REP5753966996
Alaska30.1 (25.6–34.6)REP47006218
North Dakota30.3 (24.4–36.1)REP934941336
Missouri30.4 (25.3–35.4)REP4055895825
Oklahoma30.7 (26.0–35.4)REP3082682571
Indiana30.7 (26.5–34.9)REP5330838663
Arkansas30.8 (25.5–36.1)REP2388883836
South Carolina32.1 (27.3–36.8)REP3280735498
Montana33.5 (27.7–39.3)REP833781005
Wyoming34.1 (26.5–41.7)REP45569464
Alabama34.7 (28.5–40.9)REP3795934886
Idaho34.8 (30.2–39.4)REP1433051459
Mississippi35.9 (29.5–42.3)REP2254444975
Louisiana40.2 (34.1–46.2)dREP3266487635

a Rhode Island has lowest mean intent to decline COVID-19 vaccine.

b California has the highest COVID-19 cases.

c New York has the highest COVID-19 related mortality.

d Louisiana has the highest mean intent to decline the COVID-19 vaccine.

DEM: Democratic-leaning states. REP: Republican-leaning states.

Within the pooled national sample, the 3 most common reasons for vaccine hesitancy were: ‘I plan to wait and see. If it is safe, I will get it later’ (22.1%); ‘I think others need it more than me’ (17.8%), and ‘I do not trust the government’ (18.0%). Only about 9.3% in the pooled national were concerned about the side effect of the vaccine, and 2.4% did not believe they needed a vaccine at all. The most common reasons cited for not believing the vaccine was needed were: ‘I plan to use masks and other precautions’ (25.9%), ‘COVID-19 is not a serious illness’ (25.4%), and ‘I am not a member of high-risk group’ (23.8%). Within republican and democratic-leaning states, the top 3 reasons cited for vaccine hesitancy were identical, and reflective of the general population: ‘I plan to wait and see. If it is safe, I will get it later’ (22.8% democratic vs 21.4% republican), ‘I think others need it more than me’ (18.0% democratic vs 17.5% republican), and ‘I do not trust the government’ (16.8% democratic vs 17.1% republican) (Table 4).

Table 4

Reasons for COVID-19 vaccine hesitancy

I do not plan to get the COVID-19 vaccine because%
I plan to wait and see. If it is safe, I will get it later22.14
I think others need it more than me17.76
I do not trust the government16.96
Other reasons10.24
I do not trust the COVID-19 vaccine9.37
Side effects9.31
I am concerned about cost5.77
I do not believe I need a vaccine2.42
I do not like vaccines2.34
I do not know if a COVID-19 vaccine will work1.99
My doctor has not recommended it1.69
I do not believe I need the COVID-19 vaccine because
I plan to use masks and NPI25.85
COVID-19 is not a serious illness25.35
I am not a member of high-risk group23.75
Other reasons9.05
I already had COVID-19 infection8.48
Vaccines are not beneficial7.51

Sociodemographic characteristics associated with vaccine hesitancy

Within the pooled national sample, factors associated with increased likelihood of being vaccine hesitant included: being Black than White (APR=1.63; 95% CI: 1.53–1.73); having already tested positive for COVID-19 versus never having tested positive (APR=1.18; 95% CI: 1.12–1.25); being female versus male (APR=1.15; 95% CI: 1.11–1.20); and living in a republican versus democratic-leaning state (APR=1.26; 95% CI: 1.20–1.32). Conversely, the likelihood of vaccine hesitancy was lower among people aged 45–65 years (APR=0.85; 95% CI: 0.76–0.95), and ≥65 years (APR=0.35; 95% CI: 0.31–0.40), compared to those 18–24 years; living in a multi-unit dwelling ≥5 apartment building (APR=0.63; 95% CI: 0.57–0.69), compared to those in detached mobile unit (APR=0.78; 95% CI: 0.72–0.84); those with a professional degree (APR=0.51; 95% CI: 0.45–0.58) versus less than a high school education; and annual household income more than $200000 (APR=0.44; 95% CI: 0.39–0.50) compared with less than $25000. Persons living in the South (APR=1.12; 95% CI: 1.04–1.21), Midwest (APR=1.16; 95% CI: 1.07–1.25), and West (APR=1.18; 95% CI: 1.09–1.26), compared to the Northeast (all p<0.05) (Table 5). For details on vaccine hesitancy among people who have not received any COVID-19 vaccine by age, gender, race/ethnicity, see Supplementary file.

Table 5

Poisson regression for vaccine hesitancy

Intent to vaccinateAPR (95% CI)
Age (years)
18–24 (Ref.)1
25–441.24 (1.12–1.38)
45–640.85 (0.76–0.95)
≥650.35 (0.31–0.40)
Male (Ref.)1
Female1.15 (1.11–1.20)
Non-Hispanic White (Ref.)1
Non-Hispanic Black1.63 (1.53–1.73)
Non-Hispanic Asian0.49 (0.41–0.58)
Hispanic0.85 (0.79–0.91)
Non-Hispanic other race1.33 (1.22–1.44)
Education level
Less than Highschool (Ref.)1
Highschool graduate1.10 (0.98–1.24)
Some college0.96 (0.86–1.08)
College graduate0.74 (0.66–0.83)
Professional degree0.51 (0.45–0.58)
Income (US$)
<25000 (Ref.)1
25000–500000.90 (0.85–0.96)
50000–1000000.80 (0.75–0.85)
100000–2000000.61 (0.57–0.66)
> 2000000.44 (0.39–0.50)
Marital status
Single (Ref.)1
Married1.14 (1.09–1.20)
Divorced/widowed/separated0.84 (0.80–0.89)
Type of housing
Detached mobile unit (Ref.)1
Detached single house0.78 (0.72–0.84)
Attached single house0.67 (0.60–0.74)
Up to 4 apartment building0.71 (0.64–0.79)
≥5 apartment building0.63 (0.57–0.69)
Northeast (Ref.)1
South1.12 (1.04–1.21)
Midwest1.16 (1.07–1.25)
West1.18 (1.09–1.26)
Political party
Democratic (Ref.)1
Republican1.26 (1.20–1.32)
Ever diagnosed with COVID
No (Ref.)1
Yes1.18 (1.12–1.25)

[i] APR: adjusted prevalence ratio.


This study examined variations across states in the percentage showing COVID-19 vaccine hesitancy. We found that vaccine hesitancy varied widely with a difference of close to 30 percentage points between the state with lowest prevalence (Rhode Island, 11.7%), and that with the highest prevalence (Louisiana, 40.2%). This state-specific variation is a novel finding from our study that state programs can use to plan for their vaccine campaigns. Our study also confirmed findings from other studies, including key sociodemographic factors that directly impact intent to decline the COVID-19 vaccination. For example, Black race and female gender were strongly associated with vaccine hesitancy, consistent with other studies11,12. A possible explanation for the higher likelihood of vaccine hesitancy among females is perception of reduced risk, since some studies have shown that females have less risk of acquiring the infection and experiencing COVID-19 related mortality13. Another factor that may be contributing to vaccine hesitancy among females is the unknown effects of COVID-19 vaccine on reproductive capability, fetus in pregnant females, and breastfeeding children in lactating women. However, it must be noted that despite female gender being a seemingly protective factor, individual-based risk factors such as the presence of lower socioeconomic status, inability to physical distance due to housing conditions, chronic medical conditions including obesity, and asthma, are associated with increased risk of acquiring COVID-19 infection and developing severe disease. Identifying as Black, was the strongest determinant of vaccine hesitancy, a finding that aligns with well documented mistrust in government and research14-16. That one in four of the general population reported hesitancy is likely attributable to the widespread climate of mistrust about the origins of COVID-19, mistrust in government, and for some, mistrust in vaccines in general, all of which have dovetailed to engender conspiracy theories with resultant higher levels of vaccine skepticism, and hesitancy especially among racial minorities. Research studies have also shown that people who believe conspiracies about COVID-19 report that they will be less likely to access a COVID-19 vaccine once one becomes available, they are also more likely to indicate less support for COVID-19 public health policies17,18. Data show that Blacks are more likely to get infected and experience severe COVID-19 disease and mortality when compared with other races19,20. However, our study shows that they are also the group with the highest odds of declining the vaccine. Interventions to address this multifaceted distrust must first acknowledge and address issues of social inequity and promote transparent partnerships to vaccine confidence and acceptance.

Education level can influence the ability of an individual to comprehend scientific information21, and this plays a pivotal role in vaccine acceptance and uptake. In our study, having a high school education or less was also associated with vaccine hesitancy. The ability to comprehend scientific information such as what vaccines are, how they work, and what efficacy of the vaccine means, for example, is directly impacted by literacy. Lack of access to internet and internet connectible devices can also negatively impact vaccine attitudes. While information about the vaccines is present on the websites of public health agencies such as the Centers for Disease Control and Prevention, and the Food and Drug Administration, the style and language of information can also be a barrier if it is difficult to understand. Anti-vaccination contents are readily available and accessible on social media platforms22,23, to increase vaccine acceptance, information about vaccines need to be just as easily accessible and understandable by lay audience.

Our results showed a bipartisan distinction in attitudes towards vaccines. Republican-leaning states had higher likelihood of vaccine hesitancy, while states with democratic-leaning showed more vaccine receptivity. This can be partly explained by the politicization of public health measures including facemask wearing, policies on state shutdowns, and physical distancing measures. These attitudes have carried over into public sentiments about the COVID-19 vaccine. Our study did not highlight systematic differences between democratic-leaning versus republic-leaning states in reasons for vaccine hesitancy. Our study, like other, has shown that vaccine acceptance is higher in populations who perceive that they are at higher risk of contracting the infection and having more severe disease24. Our study also highlighted that person who reside in larger apartment buildings are less vaccine hesitant; one possible explanation may be perception of higher risk due to less ability to physical distance since they are constantly exposed to their co-tenants while sharing common high-contact areas such as mail, laundry, and exercise rooms. We also noted that vaccine hesitancy was higher among those with a previous diagnosis of COVID-19. Vaccine hesitancy in this subpopulation may be due to perception of reduced risk of reinfection. However, this may be an incorrect assumption since case reports have highlighted serologically confirmed reinfection in an individual with previous diagnosis of COVID-1925,26. Knowledge about the longevity of natural COVID-19 immunity is still evolving. The perception of reduced risk due to innate immunity also has public health significance since some individuals may become less compliant with non-pharmacological interventions such as facemask wearing and physical distancing.


This study has several limitations. First, the nature of the survey is designed to be a short-turnaround instrument that provides valuable data to aid in the pandemic recovery, as such data products may not meet some of the Census Bureau’s statistical quality standards. Data are subject to suppression based on overall response and disclosure avoidance thresholds. Second, data were self-reported and subject to bias. Third, at the time of data collection vaccines were available to only persons at high-risk such as healthcare workers and elderly persons, and this may have negatively skewed vaccine acceptance responses. Despite these limitations, this study fills an important knowledge gap in state variations in vaccine hesitancy. These data are important for public health practice and has potential to inform vaccine campaign efforts.


One in four Americans indicated vaccine hesitancy, especially women, Blacks and those living in republican-leaning states. Enhanced and sustained efforts are needed to boost trust and confidence in the COVID-19 vaccines. Knowledge of state-specific information can have significant impact on how clinicians, frontline workers, public health professionals, state departments of health, and overseeing agencies tailor specific campaigns to promote vaccine confidence and uptake, while also encouraging non-pharmacological interventions27.