Using Location Intelligence to Evaluate the COVID-19 Vaccination Campaign in the United States: Spatiotemporal Big Data Analysis

• Published in: JMIR public health and surveillance 2023-02, Vol.9, p.e39166-e39166
• Main Authors: Li, Qingfeng, Peng, James Cheng, Mohan, Diwakar, Lake, Brennan, Euler, Alex Ruiz, Weir, Brian, Kan, Lena, Yang, Cui, Labrique, Alain
• Format: Article
• Language: English
• Subjects: Bayes Theorem; Collaboration; Councils; COVID-19; COVID-19 - epidemiology; COVID-19 - prevention & control; COVID-19 Vaccines; Data Analysis; Data collection; Exposure; Herd immunity; Humans; Immunization; Immunization Programs; Intelligence; Mass media; Minority & ethnic groups; Original Paper; Pandemics; Political advertising; Political identity; Privacy; Public health; Public service advertising; United States - epidemiology; Vaccines
• ISSN: 2369-2960; 2369-2960
• DOI: 10.2196/39166

Highly effective COVID-19 vaccines are available and free of charge in the United States. With adequate coverage, their use may help return life back to normal and reduce COVID-19-related hospitalization and death. Many barriers to widespread inoculation have prevented herd immunity, including vaccine hesitancy, lack of vaccine knowledge, and misinformation. The Ad Council and COVID Collaborative have been conducting one of the largest nationwide targeted campaigns ("It's Up to You") to communicate vaccine information and encourage timely vaccination across the United States. More than 300 major brands, digital and print media companies, and community-based organizations support the campaigns to reach distinct audiences. The goal of this study was to use aggregated mobility data to assess the effectiveness of the campaign on COVID-19 vaccine uptake. Campaign exposure data were collected from the Cuebiq advertising impact measurement platform consisting of about 17 million opted-in and deidentified mobile devices across the country. A Bayesian spatiotemporal hierarchical model was developed to assess campaign effectiveness through estimating the association between county-level campaign exposure and vaccination rates reported by the Centers for Disease Control and Prevention. To minimize potential bias in exposure to the campaign, the model included several control variables (eg, age, race or ethnicity, income, and political affiliation). We also incorporated conditional autoregressive residual models to account for apparent spatiotemporal autocorrelation. The data set covers a panel of 3104 counties from 48 states and the District of Columbia during a period of 22 weeks (March 29 to August 29, 2021). Officially launched in February 2021, the campaign reached about 3% of the anonymous devices on the Cuebiq platform by the end of March, which was the start of the study period. That exposure rate gradually declined to slightly above 1% in August 2021, effectively ending the study period. Results from the Bayesian hierarchical model indicate a statistically significant positive association between campaign exposure and vaccine uptake at the county level. A campaign that reaches everyone would boost the vaccination rate by 2.2% (95% uncertainty interval: 2.0%-2.4%) on a weekly basis, compared to the baseline case of no campaign. The "It's Up to You" campaign is effective in promoting COVID-19 vaccine uptake, suggesting that a nationwide targeted mass media camp