As we enter the third year of the COVID-19 pandemic, vaccination rates hover around 65% nationwide. Overcoming patients’ barriers to COVID-19 vaccination remains a top public health and behavior change priority, and one that we are actively working on at Lirio with Precision Nudging™. Our commitment to using technology to scale behavioral science in order to move people toward better health means we were thrilled to read the recent mega-study of nudges to encourage vaccination in pharmacies, conducted by Dr. Katherine Milkman and a team of talented scientists from leading academic and health institutions in order to inform the US approach to COVID-19 vaccination.

Lirio’s Chief Behavioral Officer, Amy Bucher, Ph.D., explains, “This large-scale study offers valuable insights that support the development of behavioral science-informed interventions like ours. Like the Milkman et al. team, we looked to flu vaccination as an analogue and leveraged existing research in that area to inform our intervention development. Now that we have collected our own data on COVID-19 vaccination, we’re encouraged that there are in fact similarities between the behavioral science interventions that work to drive flu and COVID-19 vaccination.”

Intervention Design

At Lirio, we rely on research like this mega-study (as well as the same team’s 2021 study about flu shots in providers’ offices) to inform our program development, and see Precision Nudging™ as an important complementary source of insights for prompting health behavior change. It is important to recognize what each project can contribute to our knowledge about driving vaccination. The differences in intervention design and intervention recipients preclude direct comparisons but also allow for a richer picture of how vaccination behaviors happen in the real world.

  Milkman et al. (2022) Lirio’s Precision Nudging™ for COVID-19 Vaccination
Message recipients 689,693 Walmart pharmacy consumers in 50 US states 550,000 patients of a health system in a Southern US state
Messaging channels used Text Text (56%) and email (44%) based on patient communication preference
Target behavior (outcome) Flu vaccination COVID-19 vaccination
# Behavior Change Techniques (BCTs) 22 (including variations in message numbers and timing) 35 (primary series) and 16 (booster shots)
Message Types per Recipient One (randomized into one of 22 conditions) Up to 35/16 (reinforcement learning selects new message until action completed)

Key characteristics of Milkman et al. Intervention and Lirio’s Precision Nudging™ for COVID-19 Vaccination

When we see similar sorts of techniques work in a controlled intervention designed to draw statistically meaningful, generalizable conclusions and in a real-world intervention deployed to maximize patient reach with research insights as a secondary goal, we are encouraged that there is a “there there.” Let’s talk about some of the behavior change techniques (BCTs) that worked well in both interventions, and where they differed.

Winning Tactics

One of the stated objectives of the mega-study team was to generate insights around vaccination behavior that could be applied to support COVID-19 vaccination efforts. In fact, one of the challenges Lirio’s behavioral designers encountered in creating the Precision Nudging™ solution for COVID-19 vaccination was the lack of an evidence base specific to the novel virus and related prevention behaviors. Among the BCTs in common between Milkman et al.’s study and Lirio’s Precision Nudging™ library, we find that some BCTs performed well in both interventions, suggesting that they may be applicable for any vaccination intervention. At the same time, there are some BCTs that performed well in Lirio’s solution only. Milkman et al. also tested a cadence of multiple messages on different days that Lirio does not offer.

High-Performing Behavior Change Techniques for Driving Vaccination
Both Milkman et al. (2022) & Lirio’s Precision Nudging™ for COVID-19 Vaccination

·       Endowment effect

·       Commitment and consistency

·       Prosocial orientation

·       Gain frame

Milkman et al. (2022) ·       Multiple texts sent on separate days
Lirio’s Precision Nudging™ for COVID-19 Vaccination

·       Anticipated regret

·       Future orientation

BCTs in Milkman et al.’s study coded using Lirio’s taxonomy

One reason why we might see some differences has to do with the frequency and framing of each vaccination type. COVID-19 vaccination is not a one-time behavior, with the primary series for mRNA vaccines including two doses and boosters recommended every five months thereafter. This means that a successful intervention must account for patients’ barriers and motivations for vaccination over time. In an early evaluation of Lirio’s solution (i.e., on a sample of 5,000 patients), we found differences in the BCTs that were associated with either 1 or 2 shots. Prompts associated with capability support, such as action planning, were associated with receiving the first vaccine dose, whereas prompts associated with motivation support, such as identification of self as a role model, were associated with receiving the second dose. We hypothesize that different types of support are needed to begin the vaccination series compared with completing the behavior. Being able to adapt the specific message content based on where a patient is in their vaccination journey and their prior behavioral responses should contribute to better behavioral outcomes.

The Importance of Behavioral Science at Scale to Drive Vaccination

The commonalities between effective BCTs in the Milkman et al. study and Lirio’s Precision Nudging™ suggest that there are some relevant, reusable nudges that can be applied across flu and COVID-19 vaccination interventions. This is very good news.

Both vaccination interventions also demonstrate the importance of scalable technology to reach large populations efficiently. By using text messages to deliver their intervention, Milkman et al. positively influenced hundreds of thousands of people to engage in an important health behavior. Similarly, at Lirio we leveraged our reinforcement learning platform to message hundreds of thousands of Louisianans through scalable communication channels of email and text, personalize, and adapt in real-time.

We are particularly encouraged by our own data that we seem able to drive behavioral outcomes at roughly equal levels across demographic groups, including race/ethnicity, income level, educational level, and political affiliation. For example, while national data shows Black Americans underrepresented among the vaccinated, we see statistically indistinguishable differences between Black and White recipients of Precision Nudging™, with about 14% vaccination rates for both. Bucher notes, “We believe that one reason we are able to reach different population segments is our scalable personalization technology. By designing intervention content that addresses a range of barriers and then optimizing the message/recipient match, we can truly tailor to individuals and effectively move more people to action.”

We find ourselves encouraged that Milkman et al.’s conclusions are consistent in so many ways with what we see in our Precision Nudging™ data for COVID-19 vaccination. It is clear that the combined talents of behavioral scientists, public health professionals, medicine, and more are required to combat the COVID-19 pandemic. With evidence-driven insights and the application of scalable personalization technology like Precision Nudging™, we at Lirio are ready to do our part.

 

 

Find Out More About Lirio

 

 


Follow Lirio on Facebook: Facebook.com/lirio.llc, LinkedIn: LinkedIn.com/company/lirio, and Twitter: @Lirio_LLC.

Other readers viewed:

Lirio Launches Hyper-Personalized Vaccination Journey to Encourage Vaccine Adoption

Communication Strategies to Overcome Vaccine Hesitancy

Let’s Get Hyper-Personal: How Behavior Change AI Supports Population Health

Using Behavioral Interventions to Overcome Vaccine Hesitancy

Reach Out

Want to learn more about how Lirio’s behavioral engagement solution utilizes behavioral science and machine learning to help organizations motivate the people they serve to achieve better outcomes?