Abstract
This study uses data from a nationally representative longitudinal study of adults in the US to examine individual-level change in vaccine intent and uptake between April and July 2021.
Since May 2021, the US has offered COVID-19 vaccines to all adults,1 yet only 66% of adults were fully vaccinated by September 25, 2021. The Delta variant surge heightens the importance of vaccination.
To optimize outreach and education, understanding the degree to which an individual’s intent to vaccinate changes over time and assessing factors that relate to rising vaccine likelihood are critical.2 For example, whether individuals who are initially “unsure” or “unlikely” will eventually be vaccinated is unknown. Most studies of vaccine intent are cross-sectional3 and cannot assess these changes.
Using data from a nationally representative longitudinal study of adults in the US,4 we assessed individual-level change in vaccine intent and uptake between April 2021 and July 2021 and characteristics of individuals who reported an increase in vaccine likelihood or uptake.
Methods
The Understanding America Study (UAS) repeatedly surveyed a probability-based internet panel of approximately 9000 US adults (aged ≥18 years).5 Panelists were recruited using address-based sampling, allowing for valid statistical inferences and avoiding coverage problems from convenience web-based panels. Internet-enabled tablets were provided if needed. Respondents received $20 per 30-minute survey time. Panelists were surveyed beginning March 10, 2020, initially biweekly and monthly after February 17, 2021, in English or Spanish about COVID-19. Overall, 89% of panelists consented to participate in the longitudinal survey. With each survey wave, about one-fourteenth (1/28 since February 17, 2021) of these consenting panelists were invited daily on a rolling basis to complete surveys over 2 weeks.
We analyzed 2 UAS waves, April 14, 2021, to May 25, 2021 (70% completion rate), and June 9, 2021, to July 20, 2021 (67% completion rate), focusing on respondents who were unvaccinated in April/May. For each wave, we asked panelists whether they received a COVID-19 vaccine; if they had not, they were asked “How likely are you to get vaccinated?” (response options: “very likely,” “somewhat likely,” “unsure,” “somewhat unlikely,” or “very unlikely”). We compared intentions in April/May with vaccination uptake or intentions in June/July and used robust Poisson regression to assess demographic predictors of vaccine intentions in June/July among respondents who were unsure, somewhat unlikely, or very unlikely in April/May. Analyses accounted for survey sampling weights, using 2-sided significance levels of .05, with significance defined as 95% CIs not containing 1 (SAS, version 9.4 [SAS Institute Inc]).
Participants provided written informed consent. The University of Southern California’s institutional review board approved the study.
Results
The April/May survey included 6052 respondents (including 2039 who were unvaccinated) and the June/July survey included 5839 respondents (5747 with vaccination/likelihood responses), and 1683 of these respondents (weighted N = 1967) also met the April/May inclusion criteria (reported being unvaccinated). The mean time between surveys was 56.5 (SD, 4.6) days. The analytic sample was 55.6% women (mean age, 44.9 [SD, 16.1] years).
The likelihood of vaccination among unvaccinated respondents remained the same or changed only somewhat for the majority of individuals between April/May and June/July (Table 1). Of the 564 participants who were somewhat or very likely to get vaccinated in April/May, 257 (45.6%) reported being vaccinated by June/July, 211 (37.3%) remained somewhat/very likely, and 96 (17.0%) became unsure/somewhat/very unlikely in June/July. Of the 1403 of 1967 respondents (71%) in April/May who were very or somewhat unlikely or unsure about getting a vaccine, 1199 (85.5%) remained so in June/July, 102 (7.3%) reported being vaccinated by June/July, and 101 (7.2%) became somewhat/very likely in June/July. Results for individual likelihood response categories are shown in Table 1. Factors significantly related to rising likelihood included being aged 50 to 64 years, being in an urban/suburban location, being Asian, and having a Democratic party affiliation (Table 2).
Table 1. Vaccinated or Intent to Get a Vaccine in June/July 2021 Among Unvaccinated Respondents in April/May 2021 (N = 1967).
Likelihood of vaccination in April/May 2021 | Likelihood of vaccination in June/July 2021, No. (%) | ||||||
---|---|---|---|---|---|---|---|
Very unlikely | Somewhat unlikely | Unsure | Somewhat likely | Very likely | Vaccinated | Total | |
Very unlikely | 720 (83) | 41 (5) | 26 (3) | 9 (1) | 30 (3) | 38 (4) | 864 |
Somewhat unlikely | 59 (23) | 109 (42) | 28 (11) | 32 (13) | 3 (1) | 28 (11) | 259 |
Unsure | 38 (14) | 20 (7) | 158 (56) | 16 (6) | 11 (4) | 37 (13) | 280 |
Somewhat likely | 16 (5) | 46 (15) | 8 (3) | 114 (37) | 38 (12) | 86 (28) | 308 |
Very likely | 19 (7) | 1 (<1) | 6 (2) | 9 (4) | 50 (19) | 171 (67) | 256 |
Likelihood categories combined | Very unlikely, somewhat unlikely, or unsure | Somewhat likely or very likely | Vaccinated | Total | |||
Very unlikely, somewhat unlikely, or unsure | 1199 (85) | 101 (7) | 103 (7) | 1403 | |||
Somewhat likely or very likely | 96 (17) | 211 (37) | 257 (46) | 564 |
Table 2. Change in Intent in June/July 2021 Among Unvaccinated Respondents Who Did Not Intend to Get a Vaccine in April/May 2021 (Weighted N = 1403)a.
Characteristic | Participants who intend to get a vaccine in June/July 2021b | ||
---|---|---|---|
No. | Adjusted rate, % (95% CI)c | Adjusted risk ratio (95% CI) | |
Sex | |||
Men | 778 | 14.3 (9.3-21.9) | 1 [Reference] |
Women | 626 | 13.6 (8.4-22.1) | 0.95 (0.65-1.41) |
Age, y | |||
18-49 | 906 | 11.2 (7.3-17.2) | 1 [Reference] |
50-64 | 326 | 17.8 (10.4-30.7) | 1.59 (1.03-2.46) |
≥65 | 170 | 13.6 (7.8-23.8) | 1.21 (0.71-2.08) |
Education | |||
High school or less | 712 | 15.1 (9.7-23.5) | 1 [Reference] |
Some college | 398 | 9.5 (5.8-15.6) | 0.63 (0.39-1.03) |
Bachelor’s degree or more | 293 | 18.9 (10.8-33.2) | 1.25 (0.80-1.96) |
Geographic location | |||
Rural | 342 | 9.2 (4.8-17.7) | 1 [Reference] |
Urban/suburban | 1062 | 21.2 (15.2-29.5) | 2.31 (1.23-4.33) |
Race and ethnicityd | |||
Hispanic | 198 | 10.5 (5.4-20.6) | 0.92 (0.47-1.81) |
Non-Hispanic Asian | 28 | 36.5 (20.1-66.2) | 3.20 (1.76-5.84) |
Non-Hispanic Black | 185 | 13.6 (7.7-24.2) | 1.19 (0.67-2.14) |
Non-Hispanic White | 942 | 11.4 (7.9-16.4) | 1 [Reference] |
Other | 51 | 8.9 (2.6-30.5) | 0.78 (0.22-2.70) |
Political affiliatione | |||
Republican party | 630 | 10.6 (6.5-17.1) | 1 [Reference] |
Democratic party | 216 | 19.4 (11.3-33.5) | 1.84 (1.11-3.06) |
Other | 353 | 13.3 (8.3-21.3) | 1.26 (0.79-2.01) |
Did not intend in April/May 2021: individuals who responded “unsure,” “somewhat unlikely,” or “very unlikely” to get a vaccine in April/May.
Intend to get a vaccine in June/July 2021: individuals who responded “vaccinated” or “very likely” or “somewhat likely” to get a vaccine in June/July.
Mutually adjusted for all the factors in the table.
Race and ethnicity were selected as a category by the Understanding America Study survey developers because of known disparities in health metrics resulting from structural racism. Survey respondents self-reported responses to these survey questions. Race and ethnicity were selected in this analysis because racial and ethnic minoritized groups are known to be disproportionately infected by SARS-CoV-2, and because prior studies have shown lower COVID-19 vaccination rates among these groups. The “other” category refers to individuals who self-reported a different race (American Indian or Alaska Native [n = 6], Native Hawaiian or Other Pacific Islander [n = 1], or multiple races [n = 44]); sample sizes were too small to analyze separately.
Individual characteristics were missing for less than 0.1% of respondents, with the exception of political affiliation, which was missing for 14.5% of respondents.
Discussion
For most individuals, reported likelihood of receiving the COVID-19 vaccine remained stable between April 2021 and July 2021. However, individuals who were unsure or somewhat/very unlikely in April/May 2021 and who were middle-aged, in an urban/suburban area, Asian, and Democrat were most likely to report being vaccinated or switching to somewhat/very likely by July 2021, suggesting that some groups are “moveable” toward vaccination. Structural barriers may remain because many individuals who were somewhat/very likely in April/May remained unvaccinated in June/July.
Study limitations include providing only English-language and Spanish-language surveys, self-reported metrics, and small sample sizes for subgroups.
Although outreach, education, and reducing barriers may nudge “moveable” demographic groups toward vaccination, more intensive strategies (eg, mandates) may be needed for resistant groups.
Section Editors: Jody W. Zylke, MD, Deputy Editor; Kristin Walter, MD, Associate Editor.
References
- 1.Baack BN, Abad N, Yankey D, et al. COVID-19 vaccination coverage and intent among adults aged 18-39 years: United States, March-May 2021. MMWR Morb Mortal Wkly Rep. 2021;70(25):928-933. doi: 10.15585/mmwr.mm7025e2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Buttenheim AM. SARS-CoV-2 vaccine acceptance: we may need to choose our battles. Ann Intern Med. 2020;173(12):1018-1019. doi: 10.7326/M20-6206 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Al-Jayyousi GF, Sherbash MAM, Ali LAM, et al. Factors influencing public attitudes towards COVID-19 vaccination: a scoping review informed by the socio-ecological model. Vaccines (Basel). 2021;9(6):1-27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Szilagyi PG, Thomas K, Shah MD, et al. National trends in the US public’s likelihood of getting a COVID-19 vaccine: April 1 to December 8, 2020. JAMA. 2020;325(4):396-398. doi: 10.1001/jama.2020.26419 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Kapteyn A, Angrisani M, Bennett D, et al. . Tracking the effect of the COVID-19 pandemic on American households. Survey Research Methods. 2020;14(2):179-186. doi: 10.18148/srm/2020.v14i2.7737 [DOI] [Google Scholar]