“World Vape Day” 2014: Lessons from the Twittersphere

Colditz J.B., Pang K., James A.E., Stanley C., Brook J., & Primack B.A. (April 2015). “World Vape Day” 2014: Lessons from the Twittersphere. Oral presentation in Toronto, ON: 38th Annual Meeting of the Society of General Internal Medicine.

Background: The use of “vaping” with electronic nicotine delivery systems (ENDS) has been steadily increasing since these products entered the US consumer market about a decade ago. ENDS devices — more commonly referred to as e-cigarettes, e-hookahs, hookah pens, or vapor pens — present dilemmas for public health advocates and regulatory agencies. For example, while ENDS may be valuable for cigarette smoking cessation among established adult smokers, thousands of non-cigarette smoking individuals are now experimenting with ENDS, which may lead to subsequent sensitization to nicotine and transition to cigarette smoking. ENDS are available in flavorings and packaging which may be attractive to youth. Furthermore, there are limited data on the toxicant load associated with use of these devices. For example, while many studies suggest that they are associated with a low toxicant load, other research has associated their use with irreversible pulmonary damage. Assessing public sentiment surrounding these products may help us guide public health interventions by discovering information such as apparent knowledge gaps or misconceptions. Open-ended qualitative assessments may be particularly valuable because of the relative lack of prior similar work in this new area. Additionally, assessment of publicly available discussion such as that on social media may help focus researchers on information to which millions of people may be exposed. Therefore, the purpose of this study was to qualitatively analyze publicly available information on Twitter related to ENDS.

Methods: We selected Twitter for our data source not only because of its extreme popularity (about 400 million “Tweets” consisting of no more than 140 characters each are exchanged each day) but also because of ease of data collection via a well-supported programming interface. We focused data collection on one particular day – Thursday, September 18, 2014 – which was promoted as “World Vape Day” by multiple businesses, organizations, and individuals. By focusing on this particular day, we hoped to access a valuable perspective on public discourse and perceptions surrounding ENDS use. We retrieved all Tweets and meta-data from the Twitter live-stream using customized software that filtered by relevant keywords, including “e-cig”, “e-hookah,” “vape,” “vapes,” and “vaping.” Two independent coders coded a random sample of 1000 Tweets related to ENDS and developed a comprehensive codebook through an iterative process based on grounded theory. This approach allowed us to (1) better understand public discourse surrounding ENDS use and the nature of World Vape Day specifically, (2) develop a structured content analysis framework (i.e., codebook) that future research projects may utilize or adapt, and (3) provide an open source software tool for other researchers to more-easily utilize live Twitter data for similar projects.

Results: The filtered Twitter stream resulted in 5,205 tweets over a strategically selected 12 hour span. Five random subsamples of 200 tweets were double-coded and discussed in order to develop and refine the coding framework. This resulted in broad coding categories including positive and negative sentiment, health- and policy-related, cigarette-related, and marketing. The coders then independently coded a randomized subsample of 1000 different ENDS-related tweets. Inter-rater agreement for each coded variable was good-to-perfect (Cohen’s kappa ranged from 0.56 to 1.00 depending on the variable). Marketing-related messages were highly prevalent (41%), while only 8% of Tweets were related to health and only 3% were related to policy or law. Of non-marketing tweets, coders detected more positive than negative sentiment (24% vs. 13%). Open-ended content analysis yielded additional nuance to each of the broad coding categories and revealed valuable insights into contextual associations with ENDS in public discourse. For example, marketing messages focused on diversity of ENDS devices and associated parts/accessories for modifying and refilling them (e.g., high-power batteries, unique shapes and colors, novel flavors of “vape juice” or “e-liquid”). Users frequently referred to flavorings, social settings and social approval, and modifications (i.e., “mods” such as building, modifying, or personalizing ENDS devices). Young adults and teenage users seemed to focus on general social approval (e.g., “coolness”) and use at school.

Conclusions: Examination of Tweets exchanged on “World Vape Day” provided a valuable window into discourse from ENDS marketers and end-users. After an appropriate iterative codebook development process, double-coding was internally consistent. Available data suggest that ENDS represents a highly active area ripe with technical innovation by marketers and consequent excitement and invigoration among users. While the medical and public health community are actively interested in potential health implications and legal loopholes potentially created by this phenomenon, these themes were not commonly manifested in our sample of Tweets. These are trends of which researchers and practitioners should be aware when engaging in dialogue related to ENDS usage. Continued research and effective translational approaches are needed to inform the medical community of these emerging trends in ENDS use, which is a growing medical and public health concern.