A Chatbot Intervention for Reducing HPV Vaccine Hesitancy

The goal of this project is to test the efficacy of a chatbot intervention for reducing HPV vaccine hesitancy among African American parents. An online survey-experiment will be conducted. In the survey-experiment, participants will first answer questions related to their attitudes toward childhood vaccines and see a brief CDC message about HPV vaccine. Then they will be directed to interact with a chatbot designed to deliver personalized HPV vaccine messages (tailored to the participant's personality) or a similar chatbot with non-personalized messages. There will also be a control condition where participants will see the brief CDC message about HPV vaccine but will not interact with any chatbot. All participants will answer questions related to their attitudes toward HPV vaccination, intentions to vaccinate their children, as well as other message response measures.

• Personalized Chatbot Intervention (Behavioral): In this intervention, participants will first see a standard CDC HPV vaccine message. After that they will be invited to interact with a chatbot designed to deliver additional HPV vaccine messages, which are tailored to their personalities.
• Non-Personalized Chatbot Intervention (Behavioral): In this intervention, participants will first see a standard CDC HPV vaccine message. After that they will be invited to interact with a chatbot designed to deliver additional HPV vaccine messages, which are not tailored to their personalities.
• No Chatbot Control (Behavioral): In this intervention, participants will see a standard CDC HPV vaccine message only and will not interact with any chatbot.

Primary outcomes

• Attitudes Toward HPV Vaccination [baseline posttest (test took approximately 5 minutes to complete)]: Participants' attitudes toward HPV vaccination were measured by seven questions such as "The HPV vaccine is beneficial for adolescents." The responses were recorded on 1 (strongly disagree) to 5 (strongly agree) scales. Final score was the average of all 7 questions.
• Intentions Toward HPV Vaccination [baseline posttest (test took approximately 5 minutes to complete)]: Participants' intentions toward vaccinating their children against HPV were measured by three questions such as "I am likely to get my child/children the HPV vaccine." Responses were indicated on 1 (strongly disagree) to 5 (strongly agree) scales. The final score was the average of the three questions.

Secondary outcomes

• Ratings of the Chatbot [baseline posttest (test took approximately 5 minutes to complete)]: Ratings of the Chatbot were measured by nine questions such as "How would you describe the HPV chatbot you just interacted with? - Engaging" on 1 (strongly disagree) to 5 (strongly agree) scales. The final score was the average of the nine questions.

Sex: All, ages: 18 Years to any, accepts healthy volunteers

Inclusion Criteria:

* 18 years or older
* Self-identify as Black or African American
* Parents of children under 18 years old
* Access to computer/laptop/mobile device and the Internet

Exclusion Criteria:

* Under 18 years
* Does not self-identify as Black or African American
* Not a parent of at least a child under 18 years old
* No access to computer/laptop/mobile device or the Internet

Participant flow

Outcome results

• Attitudes Toward HPV Vaccination (Primary), Mean, score on a scale: Personalized Chatbot Condition 3.86 +/- .84; Non-Personalized Chatbot Condition 3.96 +/- .86; No Chatbot Control Condition 3.41 +/- .85
• Intentions Toward HPV Vaccination (Primary), Mean, score on a scale: Personalized Chatbot Condition 3.87 +/- 1.17; Non-Personalized Chatbot Condition 3.95 +/- 1.08; No Chatbot Control Condition 3.56 +/- 1.18
• Ratings of the Chatbot (Secondary), Mean, score on a scale: Personalized Chatbot Condition 4.10 +/- .79; Non-Personalized Chatbot Condition 4.16 +/- .70; No Chatbot Control Condition

Adverse events

Limitations and caveats: The online survey-experiment may not fully reflect how individuals typically encounter health messages in the real world. There is concern about ecology validity. Also the long-term effects of message exposure were not determined.

• Study Protocol (PDF, 2022-11-16)
• Statistical Analysis Plan (PDF, 2023-01-14)
• Informed Consent Form (PDF, 2022-11-16)

• Xiaoli Nan, Principal investigator, University of Maryland, College Park