We demonstrated the feasibility of novel digital surveillance for tobacco control, yielding the first population estimates of WNTD’s effectiveness in the first program evaluation to use digital surveillance. WNTD coincided with large increases in cessation news coverage and cessation Internet search queries in seven Latin American nations. WNTD increases eclipsed New Year’s Day increases and approximated the effectiveness of the combination of New Year’s and a 2.8% cigarette tax hike. There was little correlation between news trends and query trends when comparing across the entire time series, but higher correlation when comparing trends around WNTD. This may indicate that news plays a minor role in increasing cessation queries generally but is magnified when the population is primed by WNTD [33
This report builds on a programmatic agenda complementing tobacco control surveillance with digital data streams [34
]. Epidemiologists have demonstrated the potential of digital surveillance for monitoring chickenpox [36
], dengue [25
], gastritis [36
], influenza [18
] including H1N1 [20
], kidney stones [40
], listeriosis [23
], Lyme disease [42
], methicillin-resistant Staphylococcus aureus
], and salmonella outbreaks [22
]. However, applications to health behaviors are very rare [20
]. With smoking accounting for almost 6 million deaths each year [45
], the application of digital surveillance to tobacco control may be of great significance. Herein we present our approach as a viable method to evaluate WNTD where traditional methods were ill equipped to do so.
While the infoveillance approach has many strengths, it is not without limitations. We only considered increase in news coverage around WNTD and did not consider the placement of news coverage (eg, lead versus minor story), but this may be captured by news volume, where increased volume is associated with premium placement. Cessation interest was only assessed for smokers with access to the Internet, but we assume that changes in cessation interest among those with access mirrors those without access [46
]. Recent studies encourage this assumption, suggesting queries are valid proxies for a host of health outcomes, even in disparate regions [25
]. For example, depression queries correlate with suicide rates [47
] and cancer queries correlate with cancer-specific incidence rates [48
]. We compared WNTD with cigarette excise taxes in Mexico (the only nation with several cigarette excise tax increases), and all tax increases occurred on New Year’s Day; hence, WNTD’s approximation to a 2.8% tax increase might be larger in other nations and when taxes do not increase during New Year’s.
Global health awareness campaigns are difficult to evaluate given sparse transnational data streams, and contemporary surveillance does not collect data on the same time dimensions as the campaigns that will likely effect change (eg, annual surveillance versus daily impacts) [12
]. Tobacco control can learn much from adapting novel techniques that are gaining traction in biosurveillance [38
]. Here we have built on validated approaches used in disease surveillance [25
] and highlighted the potential advantages of these techniques over other tobacco control surveillance methods. Survey-based tobacco control surveillance has high financial costs, lacks timeliness, and may not reach some populations (eg, cell phone-only households) [49
]. Also, responses gathered may be biased by strong social desirability [50
]. Alternatives using passive data generation are only reliable in certain geographic areas and still require costly data processing (eg, nicotine replacement therapy sales) [51
]. Digital surveillance, as such, is a critical forefront for tobacco control with real-time, public, and low-cost data streams useful for program and policy evaluations. Analyses of media and search query trends also afford greater transparency, as scientists may quickly replicate each other’s work, downloading data from regularly updated online archives. Extending our approach can serve as a continual evaluation system for global WNTD effects.
In part, these results may outline how well WNTD’s yearly themes resonate with Latin American populations. In 2009 and 2010 when WNTD focused on tobacco health warnings and smoking among women, respectively, cessation news coverage was the highest. News coverage increases were smallest when WNTD focused on general campaigns, such as promoting the World Health Organization Framework Convention on Tobacco Control (2011) and promoting awareness of the health effects of nonsmoking-related tobacco harms (2006). The variability in cessation search queries was much smaller year to year, but increases were generally larger during WNTD’s clean indoor air (2007) and restricting youth access to tobacco (2008) campaigns.
WNTD planners, and those responsible for tobacco cessation initiatives more broadly, may also use our results and techniques to scale up and refine the online components of their campaigns [52
]. Since many search engines rank links according to the number of times searchers visit the page, or how many other pages link to a page, it is unrealistic to assume that creating cessation webpages, blogs, or YouTube videos will necessarily reach large numbers of tobacco users, even if they are seeking these resources. Search engine advertisements triggered by user-specific queries may be purchased for as little as US $0.01 per click. Advertised links appear on the first page of search results, ensuring that cessation queries meet webpages with current, health-conscious, and objective discussion of cessation when program planners advertise online. Search queries are a rare opportunity to reach smokers when they are thinking about cessation, while at the same time taking into account their stage in the cessation process with links individually tailored to optimize effectiveness [53
]. By appropriately purchasing ad links, planners can ensure that evidence-based approaches maintain a strong first-page presence in cessation search results, especially around WNTD when many smokers appear interested in cessation. This is especially crucial given that most smokers attempt to quit without professional counseling or pharmacologic therapy [55
The effectiveness of WNTD has been widely speculated [8
], but to our knowledge these are the first empirical estimates of its impact. In evaluating this intervention, we highlight the importance of digital surveillance for performing difficult evaluations using real-time, low-cost, and transparent methods. Herein we find that WNTD is producing large increases in cessation news coverage and cessation Internet search queries, rivaling the effects of Mexico’s recent tobacco tax increases and New Year’s resolutions. Because WNTD is constantly evolving, our work is also a framework for real-time surveillance and potential improvement in WNTD.