This study employed a YouTube crawling program, ContextMiner [33], to retrieve the videos related to e-cigarettes available on YouTube as of June 30, 2013. To identify e-cigarette-related videos, we first compiled a list of 70 e-cigarette keyword rules (see Multimedia Appendix 1). Our keyword rules were developed through expert consensus and were expanded via an iterative process wherein an initial set of limited keywords, such as e cig and electronic cigarettes, were used to retrieve content and identify co-occurring words, which were then tested for relevance and added to form new keyword rules [34]. For each keyword rule, daily YouTube crawls were performed for 1 year between July 1, 2012, and June 30, 2013; metadata—title, description, tags, channels, posting date—for matching videos were retrieved and downloaded to a database. Videos were retrieved on the basis of YouTube's relevance algorithm, which ranks videos in descending order of presumed relevance for a given keyword query. Since YouTube limited the number of videos retrieved by each crawl, we also conducted separate crawls where matching videos were retrieved on the basis of posting dates. Through the combination of two YouTube search techniques—relevance and posting date—and repeated daily crawls to overcome YouTube limits, we built a set of videos that YouTube’s algorithm deemed related to our e-cigarette keywords.

While the YouTube Web interface provides an approximation of the number of videos retrieved for a keyword search (eg, typing electronic cigarettes into the YouTube search bar yielded 89,800 results on April 8, 2015), this number is an approximation and may include numerous irrelevant videos. We reviewed our YouTube e-cigarette video database and excluded duplicate videos using the YouTube video ID, a unique identifier assigned by YouTube. Deduplication yielded 42,484 unique videos retrieved by one or more of our keyword rules over the study period. Unique videos were then reviewed to assess relevance of each video to electronic cigarettes. During this process, we noted that our database included many videos—mainly music videos—that did not contain e-cigarette-related keywords in any collected metadata fields and that did not include e-cigarette content in the videos themselves. This finding likely reflects that YouTube’s search algorithms incorporate additional metadata not collected in our crawls, and possibly indicates that some videos were uploaded in order to boost the relevance of related videos (ie, posted in response to e-cigarette videos) [35]. To exclude videos not relevant to e-cigarettes, we used a two-step method. First, two coders viewed the top 50 most viewed videos in our database to assess whether they were relevant to e-cigarettes. The reason for manual review of the top 50 most viewed videos was to ensure that our view calculation was not artificially inflated by nonrelevant videos, which tended to be associated with higher view counts. Five videos were deemed irrelevant by both coders during this process. In the second step, we searched metadata fields—description, tags, and title—using a list of e-cigarette-specific terms (see Multimedia Appendix 2) and classified relevant videos as those that contained such terms in their metadata fields. Our keyword algorithm was highly accurate in discerning between relevant and irrelevant content in a random sample of 500 retrieved videos, with 95% of allocations agreeing with a human coder. Approximately one-third (34%) of videos were classified as irrelevant in this step and were excluded from our database. The final database contained 28,089 e-cigarette-relevant videos.

For all e-cigarette-related videos in our database, two main types of metadata were collected: static video characteristics and, where available, dynamic engagement data. Static video characteristics included video title and description, any tags the uploader provided, YouTube channel with which the video was associated (eg, Tech, HowTo, and Entertainment), posting date, name of uploader account, and video URL. Dynamic daily engagement data included view count, number of comments, number of ratings, and average rating.

The regulatory status of e-cigarettes has major implications for their marketing and promotion. In 2008, a federal district court ruled that the FDA could not regulate e-cigarettes as drugs or devices unless they were marketed for therapeutic purposes (ie, smoking cessation). This ruling was later upheld by the US Court of Appeals. However, the appeals court clarified that e-cigarettes may be subject to regulation as “tobacco products” since they contain nicotine, which is derived from tobacco [21]. Such regulation could make e-cigarettes subject to the marketing guidelines that govern traditional tobacco products, including ingredient listing; good manufacturing practice; mandated health warnings; and prohibitions against television and radio advertising, event sponsorship, and youth-targeted advertising [39]. In April 2014, the FDA issued a proposed rule to deem electronic cigarettes, among other products, as tobacco products. If finalized, the FDA’s proposed Deeming Rule would extend FDA regulatory authority to e-cigarettes and other tobacco products, allowing the FDA to propose rules that restrict the manufacture, distribution, and marketing of e-cigarettes, including advertising and promotion restrictions. Until then, e-cigarette manufacturers and distributors can continue to make relatively unrestricted advertising appeals. Despite prohibitions, promotional claims about the role of e-cigarettes in smoking cessation may be common and may have contributed to increased e-cigarette use [23]. On YouTube, over 2500 e-cigarette videos mention smoking cessation in text fields and thus may be retrieved by consumer searches related to quitting.

Our analyses suggest that YouTube is heavily utilized for promotional and networking purposes, with 70% of videos including Web addresses; however, the platform appears to be very unevenly utilized for brand-specific promotion. For example, we noted that only 15% of videos included mentions of the top 10 best-selling brands, with only Blu and NJOY representing greater than 1% of total videos or views. Further, the vast majority of brand mentions were not made through company accounts. For example, videos from the account BluCigs accounted for only 32 videos and approximately 500,000 views, representing 0.8% of total videos mentioning Blu and 4% of total views. Videos uploaded by NJOYecigs accounted for 0.3% of videos mentioning NJOY and for 4% of total views. Efforts by some vendors, nonetheless, had potential to change the landscape of available content; one website, EcigsFreeTrialOffer.com (no longer active), accounted for 80% of new videos posted in April 2012. This spike coincided with the launch of the first Tips from Former Smokers campaign, a major antismoking media effort sponsored by the US Centers for Disease Control and Prevention. Keyword searches for additional brands revealed variable presence on YouTube: following Blu, eGo and V2 comprised the second and third highest fraction of content. Most e-cigarette liquids contain 6, 12, 18, or 24 mg/mL nicotine levels, but concentrations of 36 mg/mL and 100 mg/mL solutions for making e-liquid also are available [40]. Given this wide variability in nicotine delivery and manufacturing standards for e-cigarettes, heavy promotion of certain brands may lead to use of devices not optimized for nicotine delivery and thus ineffective for smoking cessation. It is unknown how marketing appeals of smaller brands may differ from those of larger companies.

Our results show a high level of user engagement with e-cigarette content, with over 100 million total views for e-cigarette-related videos as of June 2013. To put this finding into context, the e-cigarette TV ads reached 29 million youth and young adults in 2013 [41]. Furthermore, 43% of videos included the keyword review and 10% included keywords indicating product demonstrations, both of which suggest videos originating from consumers or affiliated marketers. Mentions of mods, atomizers, e-liquids, and marijuana suggest that customization plays a large role in e-cigarette discourse on YouTube. Consumers’ ability to choose and manipulate aspects of their e-cigarette experience, including flavor, nicotine content, and battery, may have contributed to their rising popularity, but also raises questions about uneven efficacy in tobacco replacement and the potential gateway to other substance abuse. The availability of flavored juices has been criticized by some, since they may appeal to nonsmokers, including younger consumers [42].

In a sample of 280 e-cigarette videos, we found none to be age restricted by YouTube, indicating that youth can easily view and access e-cigarette videos on YouTube. Since the vast majority of these videos provide links to vendors or branded websites, these videos may enhance opportunities for underage e-cigarette purchase.

While e-cigarette marketing efforts largely leverage new media channels, traditional media plays an increasing and interconnected role. Cigarette advertising has been prohibited from US television and radio since 1971, but in recent years, e-cigarette brands have introduced television-advertising campaigns [43]. Television advertisements may use many strategies employed by cigarette advertisers in the past, including jingles, celebrity endorsers, and mascots [44]. Television advertising may also drive activity on social media channels, including YouTube. For example, we noted that a UK advertisement for E-lites garnered an additional 2 million views when posted on YouTube. Several studies have noted additional television content, including footage of actress Katherine Heigl smoking an e-cigarette on The Late Show with David Letterman and a clip from the program The Doctors [37]. Our search for a list of several celebrities featured in viral and traditional marketing campaigns (ie, Katherine Heigl, Stephen Dorff, Bruno Mars, and Courtney Love) retrieved only 153 unique videos, yet these videos were associated with view counts almost four times higher than average, comprising over 2 million total views.

Our study has several limitations and raises questions for further research. First, the study relies on keywords to capture and categorize content relevant to e-cigarettes, and any set of keywords is necessarily incomplete since new brands and terminology are continually emerging. In particular, we may have overlooked some non-English e-cigarette keywords as well as variations of the slang term vape. As a result, our estimates of e-cigarette videos and their view counts underestimate their true overall presence and impact on YouTube. However, given that the vast majority of videos in our dataset include multiple e-cigarette keywords, we believe we likely captured the majority of relevant content. We cannot exclude the possibility that there are YouTube videos that discuss e-cigarettes but do not reference them in any metadata field, but such videos would be unlikely to represent influential content, since attracting viewers relies on effective retrieval of videos by e-cigarette-related search queries.

We also used keyword rules to characterize themes within e-cigarette videos; again, this method is likely to underestimate the true presence of these themes. Further, we did not analyze variations of messages within each theme, for example, to characterize differences between various health claims or to investigate mentions of smoking-related disease. The keyword query rules applied were simple and thus not sensitive to context, in either the video content or the metadata. We did not undertake content analysis of the videos themselves, a task that fell outside the scope of this study but may have yielded rich results. Certain content elements are of considerable interest, but were not discernible by our methodology; for example, we could not reliably distinguish between promotional efforts and noncommercial consumer perspectives. Even with review of video content, such categorization would be challenging given that those affiliated with e-cigarette brands or companies may represent themselves as consumers as a marketing strategy [5,19].

More work is required to discern whether health, safety, and commercial claims derive from commercial or individual accounts. Characterizing commercial claims may clarify social media marketing guidelines. Identifying consumer experiences may help clarify whether and how e-cigarettes are used for smoking cessation, and thus contribute to public health efforts to optimize these products’ harm reduction potential. Characteristics of influential users may be explored by examining posts and comments data for individual accounts [45]. Finally, we did not examine targeted advertising that accompanies searching for and viewing content on YouTube, which may be another important way in which consumers are exposed to e-cigarette-related content [9].

In summary, our study provides an approximation of the total amount of content and consumer engagement on YouTube related to e-cigarette use. Our analyses suggest uneven use of YouTube for promotional purposes by e-cigarette brands, and a high level of engagement with a small subset of content. Further research is needed to establish the information contained in e-cigarette-relevant YouTube videos and how these videos impact consumers’ attitudes, beliefs, and risk perceptions about e-cigarettes and subsequent decisions regarding use of e-cigarettes, conventional cigarettes, and evidence-based smoking cessation aids. A better understanding about the extent, content, and impact of e-cigarette YouTube videos can aid the public health community and policymakers to ensure appropriate e-cigarette marketing regulations on social media platforms.