Data Driven Public Health Awareness Campaigns for COVID-19 and Beyond

We use proven advertising technology and behavior-change communications expertise to measure, refine, and optimize public health messaging campaigns. 

Brennan Lake, Senior Director, Cuebiq

The COVID-19 pandemic laid bare the immense challenges of implementing evidence-based public health interventions - namely social distancing, testing, and vaccination to mitigate the impact of COVID-19.  

Despite ample evidence and public sector attempts at intervention through mobility restrictions and an unprecedented vaccine rollout, misinformation and active disinformation led to varying degrees of non-compliance (social distancing and testing) and hesitancy (vaccination). Moreover, there are divergent root causes of vaccine hesitancy and motivations for noncompliance, including political resistance, ideological opposition, and a response to systemic inequities within health systems. 

 Targeted public health Behavior Change Communication (BCC) campaigns delivered in the form of multimedia Public Service Announcements (PSAs) are already reaching large audiences with messaging designed to promote awareness of proven interventions, such as social distancing and vaccination. The effectiveness of such campaigns are often measured through coarse aggregate outcomes or conventional survey approaches using self-reporting, which are subject to various biases and often under-represent respondents from marginalized communities. As a result, there is a lack of empirical and real-time measurement at scale to indicate which messaging, creative content, media channels, and demographic targeting strategies are most effective at driving behavior change.

The primary target audience for our public reports, papers and datasets are public health decision-makers who will seek to refine communication strategies for improving vaccine uptake, while also benefiting from improved epidemic modeling. 

Moreover, while our active pilot in the United States is being distributed through multiple media channels to the entire US population, the PSA’s creative assets are targeted towards key audience segments that represent responsive groups of individuals in terms of their potential for overcoming hesitancy, i.e. groups who are “skeptical” and “open but uncertain”, as opposed to segments that self-identify as adamantly resistant. Within these segments, the PSAs are targeting the following demographic subsegments with corresponding demographic and psychographic skews: 

• African Americans subsegments who: 

  • Are employed as essential workers

  • Reside in high-density Black neighborhoods

  • Have limited Access to high quality healthcare

  • Have low trust in Government

• Latinx subsegments who:

  • Are employed as essential workers

  • Live in multi-generational households

  • Have limited Access to high quality healthcare

  • Have low trust in Government

• Caucasian Americans who:

  • Are Age 34-49

  • Live in Rural or suburban areas

  • Have Low educational attainment

  • Have Limited to no experience with COVID-19

Our program will result in the following public goods: 

• Academic papers on the use of attribution technology for behavior change communications and the impact of heterogeneous hesitancy rates and vaccine accessibility on vaccine inequality and disease burden.

• Publicly accessible dashboard providing and indices on county’s mobility and physical proximity contact two weeks pre and post vaccination. 

• Publicly accessible and downloadable database of vaccination POIs in the USA

• Creation of publicly accessible reports disseminated to relevant public health decision-makers in order to inform response strategies while addressing health disparities. 

• Publicly accessible dashboard providing improved COVID-19 forecasts and alternative scenario analyses that depend on vaccine availability, vaccines hesitancy rates, and effectiveness of non-pharmaceutical interventions

• Publicly accessible dashboard providing information on mobility, commuting, and contact patterns across the United States

Our program directly impacts vaccine-hesitant populations, including African Americans and Latinx communities in the USA through our pilot. Our campaign focuses on essential workers, women, and those with limited access to healthcare - target demographics that have bore the brunt of the pandemic’s negative health and economic impacts, while also being underrepresented amongst Americans who have received the vaccine to date. 

We focus on these communities through two key analyses: vaccine hesitancy and accessibility. First, we analyze the effectiveness of PSAs in driving vaccination-center visitation among hesitant groups in order to identify the strategies (creative content, channel, etc) that lead to the highest incremental visit rates. We will use this data to create audiences for retargeting demographically similar segments that may have a higher propensity to overcome hesitancy. 

Second, we will analyze vaccine accessibility by measuring key mobility metrics related to vaccine center visitors, including distance traveled from home, density of vaccine centers near target neighborhoods, and transportation methods used to access vaccination centers. We intend for these findings, published in openly accessible academic papers and via disseminations to local governments, to improve accessibility of vaccines for both the current pandemic and ongoing vaccination programs. 

Year 1

• Our pilot program continues to measure the current COVID-19 vaccine campaigns, scaling from an already measured 150 million impressions to 7 billion impressions by the end of 2021. 

• Launch an additional campaign in the United Kingdom, together with Facebook, integrating learnings from our first pilot and defining retargeting strategies.

• Develop a manuscript with initial results from our analysis. 

• Develop the epidemic scenario online dashboard showing GLEAM projections under different vaccine hesitancy scenarios

Year 2: 

• Submit the first paper with analytical results

• Run additional campaigns for COVID-19 boosters and influenza vaccines in the US and UK, while also planning a pilot launch in an emerging-market country.

• Develop pilot in the US (and potentially UK) for applying framework to Noncommunicable Disease campaign, potentially Type II Diabetes

• Develop GLEAM extensions to include COVID-19 boosters and influenza vaccines in the US

• Development of a high resolution county/metropolitan level GLEAM-COVID model for the UK

Year 3: 

• Continuously run campaigns for COVID-19 boosters and influenza vaccines in the US and UK 

• Run campaigns for NCDs 

• Scale pilot campaign in emerging-market countries across LATAM and South&Southeast Asia. 

Our program’s core value proposition lies in our ability to measure the impact of behavior change communications strategies in driving positive outcomes, such as increased visitation to vaccination centers among vaccine-hesitant communities. Beyond offering a simple monitoring and evaluation tool, however, our program also provides actionable insights and recommendations for refining and optimizing communication strategies in real-time.

By deriving metrics such as uplift and incrementality, we are able to glean actionable insights on what campaign strategies (messaging, channel, geography, demographic) are most effective in achieving desired outcomes, i.e. visitation to vaccination centers. We then use this data to refine and optimize campaign strategies. 

Therefore, we are able to evaluate the effectiveness of our recommendations by observing changes in key metrics, such as uplift, incrementality, and advertising cost-per visit in order to measure return on advertising spend over time through subsequent iterations of campaign strategies. In order to further validate our model, we will also be engaging existing partners to disseminate surveys to PSA-exposed individuals.

We operate nation-wide across the United States. Within year one, we will replicate our pilot in the UK. Beyond the USA and the UK, we are keenly interested in launching a pilot in the Global South. To successfully deploy in any region, it is paramount to build data supply in order to obtain a statistically significant panel of opted-in individuals. For the UK, we already engaged existing mobile-app partners to expand our current user panel to ~500,000 daily-active-users. While a collaboration with national or local public health authorities in the UK would be ideal, we are prepared to activate our existing advertising agency partners who operate in the UK to run pro-bono PSAs related to vaccine hesitancy.  

In order to successfully replicate the program in a developing region, it is also important to consider inherent biases that arise in smartphone-based data collection methodologies. While mobile phones are ubiquitous across income-segments, smartphone penetration can be lower in LDCs (Least Developed Countries) as opposed to other emerging market countries across the Global South. We are interested in piloting our program in emerging-markets across LATAM and South&Southeast Asia, and plan to engage Trinity Challenge Members with active networks in these regions.

• Johns Hopkins University Bloomberg School of Public Health
• Modeling of Biological and Socio-technical Systems (MOBS Lab), Northeastern University
• AdCouncil
• US Centers for Disease Control
• UNICEF
• World Bank

The pandemic led to an unprecedented demand for mobility data across all sectors. Leveraging pro-bono data-sharing frameworks developed through Cuebiq's Data for Good program, we created a COVID-19 Mobility Data Collaborative to harness the expertise of dozens of institutions. Among these organizations, MOBS Lab and JHSPH stood out not only for their superlative respective areas of expertise, but more importantly for their strong collaborative spirit. 

It is no coincidence that all three organizations became Trinity Challenge members, with a firm commitment to leveraging data and evidence towards better collaborative decision making. We joined the Trinity Challenge to make our data and expertise openly available to the community. 

Likewise, in addition to our funding proposal, we seek support from the Trinity Challenge secretariat and community to:

1. Ideate additional areas of analysis within our program that can unlock added impact for target communities and public decision-makers, alike. 

2. Connect with organizations who may benefit from our program, both within the context of COVID-19, and related to NCDs in a post-pandemic world, as well as public sector entities in the UK and emerging economies primed for replication. 

3. Share the results of our work and public goods with the Trinity Challenge network.

Our team is composed of three Trinity Challenge members: Johns Hopkins University, Cuebiq, and Northeastern University. 

We have also had highly productive conversations with Facebook Health, regarding the potential to measure the effectiveness of their campaigns related to COVID-19 vaccination. Given Facebook’s unrivaled reach in terms of targeting, there is the potential to perform true randomized control trials with our methodology, which are currently not feasible due to the inability to control random assignment in programmatic advertising environments.

In order to co-develop, measure and optimize public health campaigns in emerging-market countries across LATAM and South & Southeast Asia, we are interested in continuing preliminary discussions with Internews, as well as initiating exploratory conversations with the Gates Foundation and Clinton Health Access Initiative.

We are also interested in connecting with the McGovern Foundation to share and receive input on our responsible data sharing practices and ethics frameworks. 

Finally, Northeastern’s MOBS Lab GLEAM model relies on a cloud-based computational pipeline built on top of the Google Cloud Platform and Dr. Chinazzi is a member of the inaugural cohort of Google Cloud Research Innovators program. Therefore, we would like to keep partnering with Google to scale up our epidemic modeling simulation efforts.