Self-reported COVID-19 vaccine hesitancy and uptake among participants from different racial and ethnic groups in the United States and United Kingdom

The study investigates whether racial and ethnic disparities exist in COVID-19 vaccine hesitancy and uptake during the initial rollout in the U.S. and U.K. The context includes rapid vaccine authorization amid logistical challenges, with lower early vaccination rates in the U.S. than the U.K. Racial and ethnic minorities have faced disproportionate COVID-19 morbidity and mortality, yet early vaccine eligibility focused on healthcare workers, older adults, and those with comorbidities, not race/ethnicity. Historical and ongoing discrimination contribute to mistrust and hesitancy among minority communities. The U.K.’s centralized National Health Service contrasts with the U.S.’s fragmented state/local rollout with inconsistent data collection on race/ethnicity and hesitancy. Reports suggested disparities in uptake, but comprehensive community-based data—especially in the U.S.—were limited. Using a common smartphone-based platform, the authors aimed to compare country-specific racial/ethnic differences in vaccine willingness and actual uptake, hypothesizing higher hesitancy among minority groups and potential U.S.-specific access barriers affecting uptake.

Prior work documents disproportionate COVID-19 burden among racial/ethnic minorities and suggests vaccine hesitancy driven by mistrust rooted in historical injustices. Early reports indicated disparities in vaccine uptake in both countries; however, U.S. data were limited by fragmented systems and inconsistent race/ethnicity collection. Studies with smaller samples noted greater hesitancy among minority groups, and experimental evidence showed misinformation reduced vaccination intent with sociodemographic heterogeneity in susceptibility. Differences in healthcare infrastructure between the U.S. and U.K. suggested potential divergence in early uptake equity.

Design and setting: Prospective cohort study using the COVID Symptom Study (CSS) smartphone application (developed by Zoe Ltd. with academic partners) in the U.S. and U.K. Enrollment spanned 24 March 2020 to 1 February 2021. The app, originally for symptom tracking (ClinicalTrials.gov NCT04331509), was adapted to collect vaccine hesitancy and uptake data. Participants: Adults ≥18 years who consented to research use of volunteered data. A total of 4,797,306 individuals enrolled (370,282 U.S.; 4,427,024 U.K.), with 1,605,019 active users in December 2020. After excluding those without race/ethnicity data and restricting to respondents to at least one vaccine questionnaire, 1,341,682 participants comprised the analytic cohort. Vaccine willingness was assessed among 1,228,638 respondents. Data collection timeline: Vaccine uptake questions began on 10 December 2020 in the U.K. (two days after first public vaccination) and on 7 January 2021 in the U.S. Vaccine willingness was queried starting January 2021. Exposures: Self-identified race/ethnicity per NIH (U.S.) and ONS (U.K.) standardized categories; U.S. Hispanic defined as any race with Hispanic/Latino ancestry. Individuals selecting “Prefer not to say” or missing were excluded. Covariates: Age, sex at birth, BMI categories, personal history of diabetes, heart, lung, kidney disease, active malignancy, smoking status (current/prior vs never), prior COVID-19 infection, frontline healthcare worker (HCW) status, geographic region (U.S.) or country (U.K.), and community-level educational attainment and financial deprivation. Outcomes: (1) Vaccine hesitancy (unsure or unwilling vs willing to accept a COVID-19 vaccine if offered). (2) Vaccine uptake (self-reported receipt of a COVID-19 vaccine dose) through 1 February 2021, overall and among the vaccine-willing. Statistical analysis: Logistic regression estimated odds ratios (ORs) and 95% CIs. Models were conditioned on age, sex, and date of study entry, with additional adjustment for comorbidities, smoking, BMI, prior COVID-19, HCW status, region/country, and community-level education/income. Inverse probability weighting (IPW) using country-specific census distributions for age, sex, race/ethnicity addressed sampling bias. Stratified analyses: frontline HCWs vs general community; strata of community-level educational attainment and financial deprivation. Interaction between race/ethnicity and country tested via Wald tests. Localized post-vaccination injection-site symptoms were summarized by race/ethnicity. Two-sided p<0.05 considered significant. Analyses used R 4.0.3 with Bioconductor 3.12. Ethics approvals from Mass General Brigham IRB (2020P000909) and King’s College London Ethics Committee (REMAS ID 18210).

- Cohort: 1,341,682 participants (U.S. and U.K.) included after exclusions; 1,228,638 answered vaccine willingness. - Vaccine willingness: 91% U.S. and 95% U.K. willing to accept a COVID-19 vaccine if offered. - Hesitancy patterns: In both countries, racial/ethnic minorities were more likely to be unsure or unwilling. U.S. age-adjusted ORs for hesitancy vs white: Black 3.84 (95% CI 3.51–4.21), Hispanic 1.69 (1.53–1.86), Asian 1.22 (1.03–1.38), more than one/other 2.14 (1.82–2.52); multivariable ORs similar (e.g., Black 3.68 [3.35–4.05]). U.K. age-adjusted ORs vs white: Black 3.00 (2.86–3.16), South Asian 1.59 (1.51–1.67), Middle East/East Asian 1.83 (1.70–1.97), more than one/other 1.43 (1.36–1.52); multivariable ORs remained elevated (e.g., Black 2.96 [2.82–3.12]). IPW sensitivity analyses yielded comparable results. - Reasons for hesitancy: Most frequent concerns across groups were long-term side effects (50–57%) and adverse reactions (45–54%). Lack of vaccine knowledge was cited more often by Black and Hispanic participants (45–51%) than white participants (37–42%). - Regional differences: Greater hesitancy in the U.S. South; in the U.K., higher hesitancy in Northern Ireland (age-adjusted OR 1.38 [1.25–1.51]) and Wales (1.10 [1.06–1.15]) vs England. - Vaccine uptake (overall): U.S.—Black participants had lower uptake vs white (multivariable OR 0.71 [0.64–0.79]); Hispanic 0.93 (0.84–1.02); Asian 1.00 (0.93–1.09); more than one/other 0.94 (0.81–1.08). U.K.—after adjustment, disparities largely attenuated: Black 0.98 (0.92–1.04), South Asian 1.18 (1.13–1.23), Middle East/East Asian 1.01 (0.94–1.08), more than one/other 0.99 (0.93–1.04). - Vaccine uptake among the vaccine-willing: U.S.—Black participants still had lower uptake (multivariable OR 0.82 [0.73–0.92]); Hispanic 0.95 (0.86–1.04); Asian 1.01 (0.93–1.09); more than one/other 1.00 (0.86–1.16). U.K.—no consistent significant disparities after full adjustment (e.g., Black 1.07 [1.00–1.14]; South Asian 1.21 [1.16–1.26]). - Heterogeneity by country: Disparity in uptake for Black vs white significantly differed between countries (P_heterogeneity < 0.001), with lower uptake among Black participants more pronounced in the U.S. - Subgroups: Among frontline HCWs, U.S. Black participants had lower uptake vs white (e.g., age-adjusted OR 0.59 [0.38–0.92]); in the U.K., Black HCWs also showed lower uptake vs white in some models. Lower uptake among U.S. Black participants persisted in lower educational attainment communities. - Safety: No consistent differences in localized injection-site symptoms by race/ethnicity were observed.

The study demonstrates that racial and ethnic minority groups in both the U.S. and U.K. had higher COVID-19 vaccine hesitancy during the early rollout, particularly among Black participants. Crucially, in the U.S., lower vaccine uptake among Black participants persisted even among those expressing willingness to be vaccinated, suggesting that structural barriers and inequitable access contributed to early disparities beyond hesitancy alone. In contrast, the U.K.’s centralized approach to vaccine delivery was associated with fewer adjusted disparities in uptake, indicating that system-level organization may mitigate inequities. Elevated hesitancy among frontline HCWs was notable and may relate to prior infection rates and safety concerns. The findings align with prior smaller studies and experimental evidence on the impact of misinformation and differential susceptibility by sociodemographics. Together, results underscore the need to address both trust and access to achieve equitable vaccination coverage.

This multinational cohort study using a common digital platform found significantly higher vaccine hesitancy among racial and ethnic minorities in both the U.S. and U.K., with Black participants most affected. In the U.S., early vaccine uptake was significantly lower among Black participants, including among those willing to be vaccinated, implying access-related inequities during the initial rollout. In the U.K., adjusted analyses showed no consistent disparities in uptake, suggesting centralized delivery may support more equitable distribution. Public health efforts should prioritize targeted, culturally competent education delivered by trusted messengers and implement structural interventions that improve access—such as centralized coordination, equitable allocation, and community-based delivery. Future research should evaluate long-term trends in uptake and hesitancy, assess interventions to reduce access barriers, and incorporate more granular socioeconomic and racial/ethnic identity measures to better understand heterogeneity within groups.

- Self-reported, volunteered data may be subject to measurement and reporting bias; however, validation suggests high accuracy of self-report, including vaccination. - Smartphone app-based cohort may underrepresent socioeconomically disadvantaged or less technologically literate groups, potentially underestimating disparities; adoption is high overall but not universal. - Non-random sample of volunteers likely enriched for individuals more accepting of vaccination, limiting generalizability. - U.S. cohort smaller than U.K. cohort, though still sufficient for country-specific estimates. - Limited granularity for individual-level education and income; reliance on community-level proxies; simplified racial/ethnic categories may mask within-group heterogeneity. - Potential undersampling addressed via inverse probability weighting, which did not materially change findings, but residual bias cannot be excluded.