Our search of 5 online news sites and 9 policy blogs revealed 36 media pieces that mentioned the CDC’s “Tips from Former Smokers” campaign in the months following the campaign’s release. All the blog and news outlets we searched included at least one piece mentioning Tips with the exception of the National Review and RedState. Collected pieces had posting dates between March 14 and June 16, 2012, with the majority of stories accompanying campaign launch in March. Of these pieces, 23 were news articles, 11 were blog postings, and 2 were video or image reels (see Multimedia Appendix 1). Nine pieces mentioned but were not primarily focused on the Tips campaign, for example, two stories from the Wall Street Journal that summarized a number of news items (“AM Vitals”) and one from Huffington Post that mentioned Tips in the context of racial disparities in smoking rates (“Smoking Rates Increase With Perceived Racial Discrimination, Study Says”). The nine “non-Tips focused” pieces are included in Multimedia Appendix 2, but we did not analyze the public/audience engagement they generated.

Most articles (both those posted on news as well as blog sites) shared considerable content, suggesting that they were closely based on content from a press release or from a single wire (eg, Associated Press) story. Stories routinely reported on the fact that the Tips campaign is the largest ever federal program of its kind, as well as the cost of the campaign (US $54 million), the need for the campaign due to the stall in reduction of adult smoking rate (at around 20%), and the number of smokers (50,000) who might be expected to quit smoking after seeing the ads. Calling upon a wide variety of terminology, the news and blog pieces universally included a reference to the graphic nature of the ads (gruesome, difficult to watch, powerful, hard hitting, grisly, emotional, brutally honest, harsh). The actual ads were often either embedded in the story or described in detail in the text.

Assessment of the presentation of the campaign goals revealed some differences in coverage. Several stories focused on the ads’ potential to prompt quit attempts, whereas others identified their potential to prevent youth initiation. Stories typically related the ads to a desire to “shock”, “jolt”, or “scare” the audience, and act as a “wake-up call”. In contrast, few stories highlighted the potential for the ads to “educate”. The campaign was often presented in the context of other tobacco control policies whose impact has been limited or now seems to have stalled; tobacco taxes and smoking bans were specifically mentioned in this regard. The framing of campaign cost also differed somewhat between stories; some highlighted the extreme expense of the campaign while others contrasted the US $54 million spent by the federal government with the tobacco companies’ US $10 billion marketing budget.

Context for campaign coverage was provided by inclusion of quotes from government officials (namely, Kathleen Sebelius from US Health and Human Services and Thomas Frieden from the Centers for Disease Control and Prevention), as well as representatives of public health agencies and tobacco control organizations (John Seffrin from the American Cancer Society and Matthew Myers from Campaign for Tobacco Free Kids). A few stories mentioned that Philip Morris declined to officially comment, and one story provided a quote from an RJ Reynolds spokesperson comparing the ad campaign to the issue of graphic warnings on cigarette packs.

Engagement with online coverage of the CDC Tips campaign was tabulated from audience interaction data provided alongside the stories, although not all forms of engagement were reported by all outlets (see Multimedia Appendix 1). Across the 27 stories primarily focused on Tips, the most common form of engagement was commenting on the host site, with 9547 comments collected from the 21 pieces where commenting was enabled. Of these comments, 4040 began new threads, with the remainder responding to existing threads (average reply rate = 1.3). “Liking” stories on Facebook was the second most common form of engagement, with a total of 8891 likes reported for 18 stories where data were available. In total, 22 stories reported data related to Twitter for a total of 1027 tweets of story URLs; 10 stories reported a total of 505 shares of story URLs on Facebook.

Mode and level of engagement varied considerably by story (see Multimedia Appendix 1) and by website (see Figures 1 and 2); CNN, Yahoo, and the Huffington Post were associated with considerably more engagement than other outlets. On CNN, Facebook likes were the most common form of engagement (n=5974), followed by commenting (n=3221) and tweets (n=237). On Yahoo, where Facebook likes were not enabled, there were a total of 5493 comments and 418 tweets. The Daily Beast and the Huffington Post both enabled Facebook shares, but only the Huffington Post was associated with a sizeable number of shares (n=484). The six Huffington Post stories were also associated with 2290 Facebook likes, 406 comments, and 133 tweets. Although the New York Times stories allowed sharing links through Facebook and Twitter, the number of shares was not reported. Politico, RedState, and the National Review did not include Tips-focused stories, and the LA Times coverage garnered a negligible level of engagement; these outlets are not represented in Figure 2.

Figures 3 and 4 depict comment valence and frequency of campaign mentions in the sample of primary comments from four outlets, selected on the basis that each had a high number of public comments posted across multiple stories. The valence of comments varied by outlet/site, with the greatest proportion of supportive comments found on Yahoo (43.2%, 380/880) and the lowest on CNN (19.6%, 64/327) and MSNBC (24%, 9/38). The proportion of “on topic” comments mentioning the Tips campaign was highest on Huffington Post (71%, 29/41) and MSNBC (53%, 20/38) and lowest on CNN (39.8%, 130/327). Only one New York Times story allowed comments to be posted, and this piece was unusual in that it was a learning blog where school children responded to a prompt about the potential efficacy of the Tips campaign. The school children’s comments were largely formulaic and supportive, making this distinct from other comments sections.

We generated a thematic coding frame for comments by a process of open coding the first 20 primary comments for each story (n=270) (see Table 2). Below, we outline four predominant criticisms of the campaign within the comments that were coded as oppositional to the campaign as well as primary themes we identified within supportive comments. Among oppositional comments, we extracted themes of questioning: (1) the appropriateness of federal government’s involvement in influencing individuals’ health behaviors, (2) the allocation of funds and efforts to tobacco issues specifically, (3) the evidence upon which anti-smoking arguments are based, or (4) the effectiveness of campaigns. Comments supporting the campaign less frequently made an explicit case for the campaign, likely reflecting that most news/blog stories had already provided the campaign’s rationale. These comments instead tended to (1) disclose personal stories related to smoking, (2) express support for quitting or disapproval of smoking, and (3) convey reactions to the campaign.

Among oppositional comments, arguments regarding message appropriateness typically depicted governmental health promotion campaigns as a threat to individual liberties, often tying programs to the expanding role of government (eg, “...It used to be that, when a person was an adult, then he/she was in charge of his/her own actions. Now the Nanny State wants to control our actions...”). Arguments regarding allocation questioned why anti-smoking efforts were deemed more worthy of governmental attention than other health or policy issues such as other drugs or obesity (eg, “And just WHEN will we be seeing the same type anti-alcohol use billboards???”). Others questioned the evidence linking smoking to serious health outcomes (eg, “...everyone I have known to get cancer never smoked a cigarette in their lives...”) or questioned the efficacy of the ads (eg, “Everyone in the USA has known that smoking is hazardous to your health since the 1960s at least”). In contrast to oppositional comments, which typically developed explicit arguments against the campaign, supportive comments tended to endorse the campaign indirectly, often through personal anecdotes and reactions. Many commenters shared detailed narratives about their own or others’ experiences with smoking, smoking-related diseases, and with quitting. Others expressed the importance of not smoking, but varied greatly in their tone and civility; some offered support and encouragement to those attempting to quit while others expressed disdain for smoking. Finally, many responded to the content of the campaign, noting its personal impact or potential effectiveness.

Table 3 depicts the overall levels of supportive and oppositional comments across the 27 media pieces that focused on the campaign. We excluded 297 comments from analysis on the basis that their valence was coded as unclear or irrelevant, leaving 1073 comments. Overall, 555 comments (51.72%) were coded as oppositional while 518 (48.28%) were coded as supportive of tobacco control. The fraction of oppositional comments was higher among those that mentioned Tips specifically (57.6%, 316/548), for stories posted to news sections of websites (73.6%, 290/394), and for stories that did not include videos from the Tips campaign (65.6%, 242/369). Reply rates were higher for comments with oppositional valence (1.6 vs 1.3 replies per comment). Reply rates were also higher for comments that did not mention the campaign specifically (1.7 vs 1.3 replies per comment).

We examined online earned media generated by the 2012 CDC Tips From Former Smokers campaign by compiling a diverse sample of leading online news outlets and blogs. We sought to summarize coverage of the campaign as well as indicators of public engagement. The volume of earned media coverage generated about Tips was fairly limited; we identified 36 pieces that mentioned the campaign, of which 27 were focused on Tips. Across our sample of 14 leading online news sources and blogs, most coverage clustered closely within a day of campaign launch. Our content analysis indicated that the news and blog pieces did not deviate greatly from the campaign press release. Media coverage of the campaign (including blog postings) generally included a set of core content with many pieces using visuals (photos or video) seemingly to provide exposure to or understanding of the ads’ graphic nature. The nature of the coverage did not suggest deep engagement with the issue on the part of journalists or bloggers. Earned media is sometimes seen as a way to develop understanding of an issue beyond what can be achieved in a 30-second or 60-second ad, but the news and blog pieces themselves provided little evidence of such development.

While the extent of earned media coverage about Tips may have been limited, even this small set of articles and postings generated notable audience engagement in the form of social media shares and comments posted on news/blog sites. The stories yielded over 9000 comments and almost 9000 Facebook likes, in addition to diffusion and dissemination of content via a variety of other social media channels. Such levels of audience engagement may indicate that the campaign and the issues it raises are salient to a sizable number of people [34]. The social media and comment data available adjacent to a story may also serve to indicate story importance (agenda-setting) to other audience members, with a potential cyclical effect. No directly comparable studies are available with which to compare the engagement levels found in this research. However, although drawn from different sources and related to different topics, two prior studies did suggest a much lower level of commenting on average than was found in this study [25,35]. Our research may provide a baseline that future studies can build upon to analyze audience engagement.

Levels of engagement varied considerably between sources, ranging from fewer than 100 shares/likes/comments to several thousand for a single piece (on the CNN site). Notably, not all sites facilitated user engagement in the same way; FoxNews and USA Today, for example, enabled comments only on a subset of the stories they posted. If audience engagement is to become part of how we conceptualize earned media, it is important to better understand why some sites generate far more active engagement than others, and the rationale behind decisions by some sites not to facilitate certain sharing or commenting mechanisms. Such insight could lead to targeted efforts to generate interest on the part of key journalists/writers who serve as gatekeepers to extensive social networks for information diffusion.

The review of user-submitted public comments revealed that they did not universally reference the campaign directly and many had unclear valence. Only about half of comments explicitly referenced the Tips campaign, although most others did include more general discussion of the role of government, the value of government-funded health campaigns, tobacco control policy, and health effects of tobacco use (see Multimedia Appendix 1). In some cases, no clear link to tobacco control (very broadly defined) could be discerned. Furthermore, we could not discern the valence of 297 of 1370 coded comments (21.68%) either because they were ambiguous or off-topic. Some of these off-topic comments might be seen as “trolling” or “spam” but, in any case, it was difficult to assess any relevance to the campaign other than the presence of negative noise [14].

There was an almost even split between the valence of comments across all Tips-focused pieces between those that were oppositional to a public health position (51.72%, 555/1073) and those that presented a supportive stance (48.28%, 518/1073). Thus, these data suggest that raw counts of audience engagement would be of limited utility to assess the likely effectiveness of earned media in supporting a campaign; by no means did all comments indicate support for the campaign or the issues it touched upon (including acknowledgement of the harmful effects of tobacco). The fact that comments explicitly referencing the Tips campaign were more likely to espouse an oppositional perspective suggests that the campaign itself may be more challenging than tobacco control more broadly conceptualized, or perhaps that oppositional comments reflected a more considered effort at argumentation. It is possible that the Tips campaign may be effectively pushing against oppositional commenters’ opinions and knowledge in such a way as to move public opinion over time. Alternatively, recent research suggests that some common public health messaging approaches not only fail to influence target populations as intended but may in fact strengthen opponents’ resolve [36]. The variety of online public forums may provide an efficient avenue for such parties to quickly register their opposition and to advance counter-arguments. Accordingly, when crafting press releases, it is important to consider the arguments that will inevitably surface in relation to tobacco and tobacco control (smoking as an issue of individual liberty, for example) and address these in relation to any specific story. This point also speaks to the importance of agenda-setting and framing of messages about the campaign; for example, a story may or may not compare levels of campaign advertising spending to the amount spent on tobacco advertising and may or may not address the issue of individual liberties, and so forth.

Our analysis of the relationship between website and story features and levels of supportive and oppositional engagement revealed both surprising and unsurprising findings. Consistent with our expectations, we found that stories that included video from the Tips campaign were associated with more supportive commentary, suggesting that exposure to the ads themselves may have achieved positive influence. In contrast, we were surprised to find that oppositional comments were more prevalent in response to news stories as compared with blog posts. Further comparison of the nature of commentary in response to blogs (which are explicitly opinionated) versus news pieces is warranted to fully understand the value of each to shaping public discourse. There were also differences in comment valence by website, with comments on Yahoo more likely to be supportive than those on MSNBC or CNN. Such differences are difficult to explain from this analysis, and the findings do not align easily with the political perspective espoused by these various news sources. Perhaps these findings point to more spam/troll activity on some sites than on others.

Developing effective methods for tracking news media and consumer responses to public health campaigns is critical for public health efforts. An analysis of the interactive online environment provides an opportunity for the public health community to monitor sentiment in real time, identify the most productive platforms and pathways for information sharing, and anticipate and address points of resistance. This paper is focused on earned media as it manifests in online outlets.

We faced several limitations in conducting this research. First, because we limited our analysis to a subset of news and blog websites, we are unable to provide a comprehensive view of online earned media related to the Tips campaign. In measuring public engagement, we were constrained by the fact that the sample websites did not have uniform ways of reporting engagement metrics, and some metrics were not reported at all by some sites. Although the New York Times stories we collected were likely associated with sharing via social media such as Twitter, the number of shares was not reported alongside these stories at the time of data collection. We also did not characterize login procedures or standards of anonymity among the websites we examined, although these features may affect decisions of whether and how readers respond [26,37,38]. Further, although we characterized the major themes appearing in campaign coverage, we did not examine whether different ways of framing the campaign affected the level and type of audience engagement. Given the websites’ varying standards of reporting commenters’ names and the possibility that a commenter might post under multiple names or accounts, we also could not characterize the distribution of total comments across unique commenters. Our content analysis of commentary was limited by our difficulty in interpreting some comments, especially given their variety and informality of language and potential for sarcasm, perhaps contributing to our marginal kappa score for the unclear/irrelevant valence category [39]. Replies to existing comment threads were particularly ambiguous, and thus we excluded them entirely. Furthermore, there is some difficulty in interpreting the levels of support and opposition we identified since it is not known to what extent commenters represent the general population or even each outlet's readership. We also do not know whether commentary reflected any organized efforts to promote an agenda with regard to tobacco control or the Tips campaign, although we did not detect evidence for this such as repeated blocks of text across comments. Finally, the content and valence of comments are likely to influence how stories are interpreted, but it is further unknown what fraction of readers consults comments, and this is likely to vary by outlet.

The large scale and graphic nature of the Tips campaign gave it potential to obtain important levels of earned media on both traditional and online platforms. While the campaign may have received limited coverage on either online news or blog sources, earned media that did cover Tips generated significant engagement among online audiences. To extend their reach and impact, future tobacco control campaigns should include advocacy efforts to capture the attention of a wider set of journalists and opinion leaders in order to build earned media and facilitate productive public engagement.