Estimated Ages of JUUL Twitter Followers

JUUL is the most popular electronic nicotine delivery system (ENDS) in the United States. JUUL’s discreet design, availability in flavors such as mango, and use of nicotine salt solutions that deliver a high dose of nicotine with minimal harshness may explain its appeal. In a study of high school students, 35.6% had ever used e-cigarettes. Among these ever e-cigarette users, 83.6% had ever tried at least 1 of the 4 e-cigarette devices presented to them (including JUUL); of these 83.6%, 64.2% had ever used JUUL and 47.1% currently used JUUL.¹ Social media may have fueled JUUL’s popularity. The number of Tweets about JUUL rose dramatically in 2015 to 2017, corresponding with the sharp increase in JUUL retail sales.² Online conversations about JUUL were rarely about quitting smoking and more commonly about pods, purchasing, and use (eg, “juuling” in school).³ JUUL’s online advertising strategy was youth focused; they used young adult models, marketed sweet and fruity flavors, and used social media influencers to promote the brand.⁴ Youth exposure to ENDS advertising increases susceptibility to and trial of ENDS among potential new users.⁵ To date, and to our knowledge, no study has examined whether youth and young adults are exposed to JUUL marketing on social media. This study used a previously developed computational age-prediction algorithm⁶ to determine the extent to which underage youth are following the JUUL brand account on Twitter.

In April 2018, we used Twitter’s application programing interface to collect data on all public followers of JUUL’s Twitter account (@JUULvapor) with at least 1 public Tweet. We obtained these followers’ metadata and up to their last 200 Tweets to create variables about their account characteristics (eg, number of followers), language use (eg, school, college), and Tweeting behavior (eg, use of emojis). The institutional review board of RTI International does not require ethical approval of studies analyzing public data, and informed consent was not required. All data were analyzed in aggregate, and potentially identifiable information such as Twitter handle names were not included in our analysis. We ran classification models developed by Morgan-Lopez et al⁶ to assess whether the account belonged to an individual and the age of the account owner. Our methodological approach is described elsewhere.⁶ The models had high accuracy for predicting youth aged 13 to 17 years (precision, 71%; recall, 75%), young adults aged 18 to 24 years (precision, 80%; recall, 73%), or youth younger than 21 years (precision, 83%; recall, 83%).

Of the 11 861 public active accounts following @JUULvapor, 9077 (76.5%) were estimated to be individuals. Using the 3-age category model on the 9077 individuals, we predicted that 4078 (44.9%) were youths (aged 13-17 years), 3957 (43.6%) were young adults (aged 18-24 years), and 1042 (11.5%) were adults (21 years or older) (Table). Using the 2-age category model, we predicted that 7313 (80.6%) of the followers were aged 13 to 20 years and that 1764 (19.4%) were 21 years or older.

Most JUUL followers on Twitter were predicted to be younger than the legal age to purchase ENDS products and may be exposed to JUUL’s marketing on social media. This finding suggests that the US Food and Drug Administration and state regulatory agencies should mandate that JUUL and other tobacco companies report data on their social media followers to ensure that youth are being restricted from accessing age-restricted content online, and physicians and other health care professionals should routinely ask about teenagers’ ENDS and social media use during medical visits to discourage use. Medical and public health organizations may need to work closely with government regulatory agencies to require social media companies and brands to develop more stringent methods for protecting youth from age-restricted content online. This study has several limitations, including the possibility of misclassification and lack of generalizability to social media platforms other than Twitter. In conclusion, our findings show that youth were likely exposed to JUUL’s social media marketing practices and underscore the need for stringent age-gating procedures online.

Accepted for Publication: March 1. 2019.

Published Online: May 20, 2019. doi:10.1001/jamapediatrics.2019.0922

Correction: This article was corrected on July 1, 2019, to clarify some misstated data in the first paragraph.

Corresponding Author: Annice E. Kim, PhD, Center for Health Policy Science and Tobacco Research, RTI International, 3040 E Cornwallis Rd, Research Triangle Park, NC 27709 (akim@rti.org).

Author Contributions: Messrs Chew and Wenger had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Kim, Chew, Wenger, Bukowski, Hair.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Kim, Chew, Wenger, Cress, Bukowski.

Critical revision of the manuscript for important intellectual content: Kim, Chew, Farrelly, Hair.

Statistical analysis: Chew, Wenger, Cress.

Obtained funding: Kim, Farrelly, Hair.

Administrative, technical, or material support: Kim, Wenger, Cress, Bukowski, Farrelly.

Supervision: Kim, Farrelly, Hair.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported by a contract from the Truth Initiative Schroeder Institute for Tobacco Research and Policy Studies (Dr Kim, Messrs Chew, Wenger, and Bukowski, and Ms Cress).

Role of the Funder/Sponsor: The sponsor was a coauthor on the study and collaborated on interpretation of data; review and approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: The authors thank Ashley Feld, MPH, and James Nonnemaker, PhD, RTI International, Research Triangle Park, North Carolina, Alison Cuccia, MPH, and Minal Patel, PhD, Truth Initiative, Washington, DC, and Jeffrey Willett, PhD, American Heart Association, Kansas City, Missouri, provided feedback on an earlier version of the manuscript. None of these contributors were compensated for this work,