A Behaviourally Informed Chatbot Increases Vaccination Rates in Argentina More than a One-way Reminder

"Finding low-cost, scalable and effective interventions to increase vaccination is a key challenge that governments around the world are facing beyond COVID-19. Emerging technologies such as two-way chatbots are a promising avenue to address this challenge."

Maintaining COVID-19 vaccine demand was key to ending the global health emergency. However, in Chaco, a disadvantaged province in northeast Argentina, only 33% of the eligible population had received a booster dose 18 months into the vaccine rollout. By way of comparison, in the city of Buenos Aires, 91% of residents had received a booster dose at this time. In response, this group of researchers designed a WhatsApp chatbot to help ease the process of getting vaccinated. They then ran a large-scale preregistered randomised controlled trial (RCT) with 249,705 participants to understand whether two-way interactive messaging incorporating behaviourally informed functionalities could perform better than one-way message reminders.

The chatbot featured five core behavioural functionalities. First, users received individual-level personalised information on their eligibility based on the government's vaccinations database. Second, they could use a locator tool to find their nearest vaccination centres on the basis of their postcode or by using WhatsApp's Share Location functionality. Third, they were prompted to make a plan for when and where they would get vaccinated. Fourth, they received a reminder message the day before their chosen date, and fifth, they were provided a link with Google Maps directions on how to get to their chosen centre. In each case, users could reply to the chatbot's messages by choosing an answer from a set menu of options.

These features were designed to address a set of barriers to vaccine uptake that the researchers identified through exploratory qualitative research among the local population in Chaco, including low visibility of when and where vaccines were available, a lack of direct government communications on eligibility, and out-of-date online information on vaccination centre locations and hours. The design of the features was based on existing evidence that suggests that making a behaviour easier by providing personalised information, prompting users to make a plan, and sending reminders can encourage behaviour change.

Participants for the trial were selected from one of three phone databases owned by the Ministry of Health. The researchers restricted the sample to include individuals over age 18 who had received at least one dose of the COVID-19 vaccine. Shortly before the launch of the trial, 83% of the Chaco population had received at least one dose of the vaccine, indicating that anti-vaccination sentiment was not widespread. However, uptake of the second dose was 70%, and only 33% had received a booster dose. Individuals who had not received any doses by this point in the pandemic were probably unwilling to get vaccinated, and so the researchers believed they would not benefit from this chatbot service. Rather, the intervention intended to make the process of getting a vaccination easier for individuals who were not strongly vaccine-resistant. 

The data used to construct the eligible sample and to conduct randomisation were extracted on August 22 2022. Among the final sample of 249,705 individuals, the behaviourally informed chatbot more than tripled COVID-19 vaccine uptake compared with the control group (a 1.6 percentage point (pp) increase (95% confidence interval (CI), 1.36 pp, 1.77 pp) and nearly doubled uptake compared with the one-way message reminder (a 1 percentage point increase (95% CI, 0.83 pp, 1.17 pp).

In absolute terms, 1,882 individuals in the chatbot group received their next vaccine dose in the 4-week observation period, which is 1,300 more than in the control group. Had the chatbot been distributed across the full sample, it would have led to nearly 6,000 vaccinations. The chatbot was impactful across a wide range of demographics, with vaccination increasing for men and women of all age groups.

There was a high level of engagement with the chatbot, as 25% of users clicked through the initial message, and only 3.6% expressed a lack of interest. Per the researchers, additional functionalities give people more intense support in navigating the vaccination user journey. But if the flow becomes too long, it risks disengagement. In future research, they plan to separately test the effects of different functionalities to help optimise chatbot design.

The study suggests that chatbots can be effective even with low vaccination demand. The researchers launched the chatbot in September 2022, when COVID-19 infection rates were low. However, because the chatbot aims to make the vaccination process easier for those who are already motivated to do it, it is possible that the effects will be even larger in contexts when demand is high, such as the initial rollout of a new vaccine.

In conclusion: "Communications tools designed with behaviourally informed functionalities that simplify the vaccine user journey can increase vaccination more than traditional message reminders and may have applications to other health behaviours." The researchers believe that "well-designed chatbots that incorporate behavioural functionalities can work and may be a promising [and low-cost] tool to increase the uptake of routine vaccinations for governments around the world."