COVID-19 messaging on social media for American Indian and Alaska Native communities: Thematic analysis of audience reach and web behavior

Description: 

Background: During the COVID-19 pandemic, tribal and health organizations used social media to rapidly disseminate public health guidance highlighting protective behaviors such as masking and vaccination to mitigate the pandemic's disproportionate burden on American Indian and Alaska Native (AI/AN) communities. Objective: Seeking to provide guidance for future communication campaigns prioritizing AI/AN audiences, this study aimed to identify Twitter post characteristics associated with higher performance, measured by audience reach (impressions) and web behavior (engagement rate). Methods: We analyzed Twitter posts published by a campaign by the Johns Hopkins Center for Indigenous Health from July 2020 to June 2021. Qualitative analysis was informed by in-depth interviews with members of a Tribal Advisory Board and thematically organized according to the Health Belief Model. A general linearized model was used to analyze associations between Twitter post themes, impressions, and engagement rates. Results: The campaign published 162 Twitter messages, which organically generated 425,834 impressions and 6016 engagements. Iterative analysis of these Twitter posts identified 10 unique themes under theory- and culture-related categories of framing knowledge, cultural messaging, normalizing mitigation strategies, and interactive opportunities, which were corroborated by interviews with Tribal Advisory Board members. Statistical analysis of Twitter impressions and engagement rate by theme demonstrated that posts featuring culturally resonant community role models (P=.02), promoting web-based events (P=.002), and with messaging as part of Twitter Chats (P<.001) were likely to generate higher impressions. In the adjusted analysis controlling for the date of posting, only the promotion of web-based events (P=.003) and Twitter Chat messaging (P=.01) remained significant. Visual, explanatory posts promoting self-efficacy (P=.01; P=.01) and humorous posts (P=.02; P=.01) were the most likely to generate high-engagement rates in both the adjusted and unadjusted analysis. Conclusions: Results from the 1-year Twitter campaign provide lessons to inform organizations designing social media messages to reach and engage AI/AN social media audiences. The use of interactive events, instructional graphics, and Indigenous humor are promising practices to engage community members, potentially opening audiences to receiving important and time-sensitive guidance.