Using the Health Belief Model to Analyze Instagram Posts about Zika for Public Health Communications

We analyzed Instagram posts about Zika by using the Health Belief Model. We found a high presence of threat messages, yet little engagement with these posts. Public health professionals should focus on posting messages to increase self-efficacy and benefits of protective behavior, especially when a vaccine becomes available.


Research Questions
To better understand how to communicate with the public about the Zika virus and about Zika preventative measures on Instagram using HBM constructs, it is necessary to identify how persons respond to existing Zika messages on the platform. Given the current growth of Instagram as a platform for public discussions and sharing of information, exploring users' engagement with HBM constructs as part of Zika-related content can yield useful insights for the development of timely and relevant health messages. Therefore, our 2 research questions are 1) What HBM constructs are present in Zika-focused posts on Instagram, and 2) How do users engage with posts containing HBM constructs?

Methods
We analyzed Zika-related posts on the social media platform Instagram by using quantitative content analysis. During August 1-31, 2016, we collected Instagram posts by using #Zika and #ZikaVirus hashtags by using a web-based social media mining tool Applied to Zika and Zika-preventative behaviors (because no treatment except for supportive care is currently known for Zika), the HBM constructs are operationalized as follows: perceived benefits (Appendix Table 1) of the Zika preventative measures, such as protection against disease; perceived barriers (Appendix Table 2) to Zika preventative measures, such as complex instructions and the presence of conspiracy theo ries; perceived susceptibility to the Zika virus; perceived severity of Zika (Appendix Table 3) including expressed fear of Zika; selfefficacy related to Zika preventative measures; and cues to action related to Zika preventative measures, such as a physician's recommendation or an advertisement encouraging the use of mosquito repellent.
Two coders were trained and coded 10% of 100 posts for intercoder reliability. After pretesting and subsequent changes to the coding protocol, the intercoder reliability test with the ReCal statistical program (http://dfreelon.org/utils/recalfront/) showed an average Scott (5) π value of 0.82. The individual coefficients were all considered to be reliable, and the lowest coefficient was 0.71 (a complete list is available from the authors). Once reliability was reached, the first coder coded the remaining 900 posts for all study variables.

Statistical Analyses
We used Mann-Whitney U tests to check for differences in Instagram engagement between posts with versus without a range of dichotomous variables. We used Kruskal-Wallis

Perceived Susceptibility
The following variables were used to measure perceived susceptibility (Appendix Table   4) to the Zika virus: high chance of contracting Zika when in an area with a lot of mosquitoes, living in an area with ongoing local Zika transmission, traveling to an area with ongoing local Zika transmission, living in an area with a lot of mosquitoes, mention of travel restrictions, mention of sexual transmission (Appendix Figure 3), and mention that getting infected with Zika is currently a possibility. We show the presence of each of these variables in the Instagram posts of this sample (Appendix Figure 4). A total of 59.9% (599) of Instagram posts in this sample mentioned the perceived susceptibility of Zika virus.

Perceived Benefits
The following variables were used to measure perceived benefits of Zika preventative measures: benefits of use of mosquito repellent (Appendix Figure 2), benefits of postponing travel to Zika-infected areas, benefits of avoiding travel to Zika-infected areas, benefits of wearing long sleeves and long pants when outdoors, and benefits of using condoms. We determined the presence of each of these variables in the Instagram posts of this sample (Appendix Table 1). In total, 21.7% (217) of Instagram posts in this sample mentioned the perceived benefits of Zika preventative measures. However, of the number of posts that mentioned Zika preventative measures in general, 91.2% (217) mentioned perceived benefits of these measures.

Perceived Barriers
The following variables were used to gauge perceived barriers to Zika prevention: lack of funds for Zika prevention, complex instructions for Zika prevention, questioning the safety of Zika preventative measures, Zika preventions being unrealistic, Zika being hard to prevent, and conspiracy theories related to Zika. We determined the presence of each of these variables in the Instagram posts of this sample (Appendix Table 2). Of particular interest is that 10% of the sample discussed conspiracy theories related to the Zika virus (Appendix Figure 4).

Cues to Action and Self-Efficacy
Mentions of cues to action (Appendix Figure 1 The second research question asked how users engaged with Zika-related Instagram posts that contain HBM constructs. We performed Mann-Whitney U tests to determine whether there were differences in likes and comments between posts with the presence versus absence of HBM constructs (i.e., perceived benefits, perceived barriers, perceived susceptibility, perceived severity, perceived self-efficacy, and cues to action). Like (p = 0.023) and comment (p = 0.010) frequencies were significantly lower in posts that mentioned perceived benefits of Zika prevention and that mentioned the perceived severity of Zika versus posts that did not (p<0.001 for both likes and comments). In addition, like (p<0.001) and comment (p<0.001) frequencies were significantly higher in posts that mentioned conspiracy theories related to the Zika virus than in posts that did not. There were no major differences between Instagram engagement metrics for perceived susceptibility, self-efficacy, and cues to action (Appendix Table 3).