UMD Study Finds Simple Reminders Boost Vaccination Rates More Than Free Ride

In a recent study featured in Nature, Rayyan S. Mobarak, a PhD student in the Department of Agricultural and Resource Economics (AREC) at the University of Maryland, College Park, contributed to groundbreaking research that tested various incentives and nudges to increase vaccination rates. Conducted in collaboration with CVS pharmacy, the megastudy involving millions of participants, explored the effectiveness of eight different interventions, ranging from free Lyft rides to simple reminder messages, aimed at encouraging COVID-19 booster uptake.

All tested interventions, whether free Lyft rides or simple reminder messages, significantly increased vaccination rates compared to a control group that received no messages. However, contrary to the predictions of both experts and laypeople, the much cheaper intervention of sending reminder messages proved to be just as effective, if not more so, than offering free rides. Prior to the study, behavioral scientists, economists, psychologists, and laypeople predicted that the Free Lyft ride incentive would outperform other interventions. Surprisingly, the study revealed that free rides did not outperform the simpler, less costly reminder messages.

While the primary focus was on increasing COVID-19 booster uptake, the interventions also had a significant, albeit smaller, effect on flu vaccination rates. This finding highlights the broader applicability of these interventions in promoting public health. As the world continues to grapple with COVID-19, encouraging routine vaccinations remains a critical policy challenge. Ensuring high vaccination rates is essential to prevent unnecessary hospitalizations and deaths. The study's findings underscore the importance of rigorously testing and implementing cost-effective strategies that can be widely deployed. The research demonstrates that simple, cost-effective interventions like reminder messages can significantly boost vaccination rates. This insight is invaluable for policymakers and public health officials seeking to design efficient vaccination campaigns. Additionally, the study highlights the need for ongoing research to identify and implement the most effective strategies for promoting public health.

Rayyan S. Mobarak is a dedicated PhD student at UMD's AREC department, where he focuses on agricultural and health economics and data analysis. His work on this study exemplifies the department's commitment to addressing real-world challenges through rigorous research and innovative solutions. We had the opportunity to interview Mobarak to delve deeper into the research and its impact. In this interview, Rayyan shares insights into his involvement in the study, the surprising outcomes, and the potential future applications of these findings.

AREC: How did you get involved in this research, and why is it important to you?
Rayyan S. Mobarak (RSM): Prior to joining UMD, I worked as a predoctoral candidate at the University of Pennsylvania under the Behavior Change for Good Initiative (BCFG). The initiative conducts megastudies, which are designed to test various interventions on a large scale to create sustained behavior change. Vaccination behavior is a focal point for this initiative, which is how I became involved in this research. As an aspiring economist, I find the decision to get vaccinated an interesting setting to study economic and behavioral theory. It brings forth questions of risk attitudes, present bias and irreversible decisions. Furthermore, it comes down to a question of saving lives and understanding the barriers that deter people when making this decision will allow us to design incentives that maximize their welfare.     

AREC: What surprised you the most about the findings of the study?
RSM: I find pleasantly surprising the positive spillover effects on flu shots, interventions that targeted COVID-19 bivalent booster uptake. This seems to suggest that people maximized their benefit of taking the trip to the Pharmacy by bundling their vaccinations. From an intervention design perspective, we were greatly constrained by space when constructing the treatment arms which were all delivered via text messages. It would have been difficult to create messages that targeted multiple vaccinations and so the increased flu shot uptake is a welcome surprise. 

AREC: Can you explain the significance of using simple reminder messages over more costly incentives like free rides?
RSM: Every policy has its opportunity cost. Designing policies that maximize welfare under budget constraints is an optimization problem with answers that are not at all obvious. If you look at the results of our prediction studies, seasoned researchers who also think about policy design expected Lyft rides to have much larger effect sizes and also outperform all other interventions. Under such priors, enacting such a policy would be inefficient for our study population. Absorbing the costs of the rides would make reaching a large population infeasible. Furthermore, coordination and enforcement costs between the ride-sharing company and pharmacies would require the benefits of such a policy to be even larger to make this policy worth it. Simple reminder messages are significantly cheaper, can target a larger population and have produced a larger impact on booster rates in our study compared to Lyft rides. Without our megastudy, such a conclusion would not have been obvious and highlights the importance of using large scale empirical testing as a guide to policy implementation.     

AREC: How do you think these findings will impact public health policy and vaccination campaigns in the future?
RSM: I hope these findings can inform policymakers of the impact that light touch nudges can have under certain contexts. BCFG has shown multiple times that such nudges excel when it comes to vaccination behavior (see Milkman et al. 2021 and 2022 published in PNAS). The tradeoff with the easy to implement and cost-effective nudges are the fact that effect sizes and benefits are also likely to be small compared to higher-cost incentives. However, if you look at booster rates in our control group (close to 5%), you will see that we are dealing with a population that is vaccine-hesitant. As we have shown in our study, high cost policies such as free Lyft rides for such a population would have been a misallocation of resources. I am optimistic that future policy will take into account the years of research BCFG has conducted in vaccination behavior and will begin incorporating light touch nudges in future vaccination campaigns with the assurance that the impact of such nudges have been replicated across multiple empirical studies. 

That being said, I fear our research would be misinterpreted as an absolute statement on the dominance of interventions targeting attentional hurdles over those targeting transportational hurdles. The context matters, as well as the study population. This point is illustrated by a paper by Meriggi and Mobarak et al. that was also published in Nature this year. Bringing mobile vaccination teams to rural areas of Sierra Leone create large increases in immunization rates. The population studied here is shown to be much more elastic towards interventions that directly reduce the marginal cost of getting vaccinated than the population that we have studied in our paper. I urge policy-makers and readers to be aware of such context when extrapolating the results of any such studies.     

AREC: What challenges did you face during the research, and how did you overcome them?
RSM: A challenge that we faced early on was defining our population of interest. Due to budgetary constraints, we had to pick locations in the US with a high density of pharmacies and a strong presence of the Lyft market while maintaining a large enough population to support the external validity of results. This defined population would then be randomized into all our interventions. We started with the core cities of some of the largest metropolitan statistical areas in the US, and moved outward in concentric circles by adding neighboring zip codes into our program. This was a difficult task which needed a combination of both spatial econometrics and some manual work to get done. Although challenging, this was a valuable experience and a good insight into the production process behind the polished research papers that readers get to enjoy.  

For more information about the study, please visit the full article here.