An Online Randomized Pilot Evaluation of AI-Driven Public Health Campaign Messages among People with a History of Drug Use

Background: Integrated public health messaging has the potential to make public health campaign more efficient and efficacious. However, there is limited evidence on whether these integrated messages are perceived as actionable and persuasive. Additionally, it is unclear whether integrating messaging about multiple health issues (i.e., HIV, HCV, and overdose) has a similar influence on behavioral intentions as single-condition messaging. Objective: This pilot study aims to (1) assess the perceived quality of vetted HIV/HCV/OD campaign messages and (2) compare individuals' behavioral intentions after being exposed to either a vetted HIV/HCV/OD campaign or a vetted HIV-only campaign.Methods: We recruited 136 adults with a history of drug use through Prolific. Participants were randomly assigned in a 1:1 ratio to either the vetted integrated HIV/HCV/OD campaign or the vetted HIV-only campaign. Each participant viewed 10 randomly selected messages from a pool of 11 and rated the characteristics of each message using 5-point Likert scales. For the analysis, we averaged the ratings across the 10 messages per participant. At the end of the study, participants completed seven intention items on a 5-point scale (1 = definitely no to 5 = definitely yes).Results: Across both conditions, the mean ratings for all message characteristics were above the midpoint of 3, ranging from 3.55 to 4.04. Compared to the vetted HIV-only messages, the vetted HIV/HCV/OD messages were associated with a higher intention to use naloxone nasal spray (d = 0.80, p < .001), use syringe services programs (d = 0.39, p = .0273), and engage with HCV prevention behaviors (d = 0.34, p = .0508). Other intentions showed favorable trends of the vetted HIV/HCV/OD messages, but did not reach statistical significance in this pilot study.