E-cigarette use is rapidly increasing among adolescents in the United States, with some suggesting that e-cigarettes are the cause of declining youth cigarette smoking. We hypothesized that the decline in youth smoking changed after e-cigarettes arrived on the US market in 2007.
Data were collected by using cross-sectional, nationally representative school-based samples of sixth- through 12th-graders from 2004–2014 National Youth Tobacco Surveys (samples ranged from 16 614 in 2013 to 25 324 in 2004). Analyses were conducted by using interrupted time series of ever (≥1 puff) and current (last 30 days) cigarette smoking. Logistic regression was used to identify psychosocial risk factors associated with cigarette smoking in the 2004–2009 samples; this model was then applied to estimate the probability of cigarette smoking among cigarette smokers and e-cigarette users in the 2011–2014 samples.
Youth cigarette smoking decreased linearly between 2004 and 2014 (P = .009 for ever smoking and P = .05 for current smoking), with no significant change in this trend after 2009 (P = .57 and .23). Based on the psychosocial model of smoking, including demographic characteristics, willingness to wear clothing with a tobacco logo, living with a smoker, likelihood of smoking in the next year, likelihood of smoking cigarettes from a friend, and use of tobacco products other than cigarettes or e-cigarettes, the model categorized <25% of current e-cigarette–only users (between 11.0% in 2012 and 23.1% in 2013) as current smokers.
The introduction of e-cigarettes was not associated with a change in the linear decline in cigarette smoking among youth. E-cigarette–only users would be unlikely to have initiated tobacco product use with cigarettes.
Comments
We agree on the importance of contextual and temporal accuracy when studying novel tobacco products
We agree with Manderski et al that contextual and temporal accuracy is important when studying not just novel but all tobacco products, and carefully considered both issues that they raise in their comment when conducting our research.
As they acknowledged in their comment, 2009 is the closest year that NYTS data are available to the advent of e-cigarettes in the US, which occurred around 2007. As they note, as of 2009, e-cigarette use was very low and grew over time. We explicitly model this fact by quantifying this change with a slope (rather than an intercept) change in the regression model. (This is the approach shown in Figure 2b of reference 11 of their comment). To ensure that the specific year that we selected did not substantially affect the results, we conducted a sensitivity analysis using different years for the hinge point: 2011 (first year e-cigarettes measured in NYTS), 2012 (e-cigarette use takes off among youth2), and 2013 (evidence of high adolescent e-cigarette advertising exposure3). None were associated with significant changes in the downward secular trend in ever cigarette smoking (2011: p=0.32; 2013: p=0.36; 2013: p=0.26) or current cigarette smoking (2011: p=0.29; 2012, p=0.31; 2013, p=0.21). As a result, we are confident in our results that there was not a significant change in cigarette smoking prevalence when e-cigarettes arrived on the market in the US or when US adolescents started using them. This information had been deleted from the manuscript due to length restrictions.
We recognized the issues raised by the change in wording of the e-cigarette question between the 2013 and 2014 NYTS questionnaires likely impacted the estimated prevalence of e-cigarette use between 2013 and 2014. As noted in Arrazola et al,4 cited in our paper, “In 2014, current use of e-cigarettes was assessed by the question, ‘During the past 30 days, on how many days did you use e-cigarettes such as Blu, 21st Century Smoke, or NJOY?’, and in 2011 to 2013, such use was assessed by the question, ‘In the past 30 days, which [tobacco products] have you used on at least 1 day?’” As the comment notes, not only was the format of the question different, but the list of brands of e-cigarettes was more limited between 2011 and 2013 (“Electronic Cigarettes or E-cigarettes, such as Ruyan or NJOY”) than in 2014. Although this limitation is important when interpreting changes in e-cigarette use between 2013 and 2014, the wording of this question has no impact on the change in smoking prevalence between 2011 and 2013 and, only limited impact on the overall trend. This is another point that we were not able to discuss in detail in the manuscript because of word count limitations. More important, the change in the e-cigarette question has no effect on the trend in cigarette smoking, which is the key point in the paper.
In addition, if anything, the changes in definition of e-cigarettes in 2014 (when the highest use was reported) make the analysis of the psychosocial determinants of e-cigarette consumption in our paper more reliable. This analysis concluded, consistent with earlier smaller longitudinal studies5, 6 that e-cigarette-only users had risk profiles that made them unlikely to initiate tobacco product use with cigarettes.
References
1. Lopez Bernal J, Cummins S, Gasparrini A. Interrupted time series regression for the evaluation of public health interventions: a tutorial. Int J Epidemiol. 2016.
2. Arrazola RA, Dube SR, King BA. Tobacco product use among middle and high school students -- United States, 2011 and 2012. MMWR: Morbidity and Mortality Weekly Report. 2013;62(45):893-897.
3. Duke JC, Lee YO, Kim AE, Watson KA, Arnold KY, Nonnemaker JM, et al. Exposure to electronic cigarette television advertisements among youth and young adults. Pediatrics. 2014;134(1):e29–36.
4. Arrazola RA, Singh T, Corey CG, Husten CG, Neff LJ, Apelberg BJ, et al. Tobacco use among middle and high school students - United States, 2011-2014. MMWR: Morbidity and Mortality Weekly Report. 2015;64(14):381–385.
5. Barrington-Trimis JL, Berhane K, Unger JB, Cruz TB, Huh J, Leventhal AM, et al. Psychosocial factors associated with adolescent electronic cigarette and cigarette use. Pediatrics. 2015;136(2):308–317.
6. Wills TA, Sargent JD, Gibbons FX, Pagano I, Schweitzer R. E-cigarette use is differentially related to smoking onset among lower risk adolescents. Tobacco Control. 2016;[Epub ahead of print].
The importance of contextual and temporal accuracy when studying novel tobacco products
The study by Dutra and Glantz examined cigarette and electronic cigarette (e-cigarette) trends among U.S. middle and high school students via an interrupted time-series analysis of the National Youth Tobacco Survey (NYTS). The authors concluded that the rate of decline in smoking prevalence did not change after the introduction of e-cigarettes into the US market. We vehemently disagree with their choice to test change in slope at 2009 and note additional problems with their methodology and subsequent interpretations.
An interrupted time-series study design, “requires a clear differentiation of the pre-intervention period and the post-intervention period…This does not necessarily require the intervention to be introduced overnight but the period of implementation should be well defined so that it can be considered separately.” [1] Although e-cigarettes first appeared in the U.S. as early as 2007, we question the authors’ choice of 2009 as a reference point given temporal trends in sales and availability of e-cigarettes in the U.S. Indeed, the electronic cigarette market was in an embryonic state in 2009, making up less than 0.1% of the non-cigarette tobacco market in the US. [2] E-cigarettes’ rapid expansion in the market occurred in 2012 and coincided with a shift away from internet sales to traditional retailers. Even then, less than one-third of traditional tobacco retailers carried e-cigarettes in 2012, limiting their accessibility, particularly to youth. [3] As such, there should be no expectation that cigarette smoking prevalence trends would have changed because of e-cigarette availability as early as 2009, as the authors suggest.
Importantly, the authors note that combined cigarette and e-cigarette use increased in 2014 as a result of increased e-cigarette use; however they neglected to mention that the questions assessing e-cigarette use on the NYTS changed dramatically from 2013 to 2014. From 2011 through 2013, e-cigarette use was assessed by a check-all-that-apply question, as the authors described. However, in 2014, e-cigarettes were instead assessed by specific forced-choice questions (e.g., “Have you ever tried an electronic cigarette or e-cigarette such as blu, 21st Century Smoke or NJOY?”). Survey methodology research indicates that forced-choice questions yield higher estimates than check-all-that-apply questions among adults and youth alike. [4] Our own analysis of New Jersey Youth Tobacco Survey data, which assessed e-cigarette use using both check-all-that-apply and forced-choice questions, found that e-cigarette prevalence estimates calculated from forced-choice measures were significantly higher than those calculated from check-all-that-apply measures. [5] Therefore, the increase in prevalence reported from 2013 to 2014 should be interpreted with substantial caution given that some of the increase in e-cigarette use found in the NYTS data is attributed to methodological artifact.
Drawing conclusions about the impact of a marketplace event on tobacco use behaviors necessitates familiarity with marketplace trends, and more specifically, temporal precision. Moreover, analyses that describe prevalence changes using serial cross-sectional data should consider the impact of changes in question wording on estimates between years. We urge tobacco control researchers to consider these important limitations when interpreting the results of this study and most importantly, when designing future studies.
Michelle T. Bover Manderski, MPH
Daniel P. Giovenco, PhD MPH
Cristine D. Delnevo, PhD MPH
References
1. Lopez Bernal J, Cummins S, Gasparrini A. Interrupted time series regression for the evaluation of public health interventions: a tutorial. International Journal of Epidemiology. 2016 Jun 9. pii: dyw098. [Epub ahead of print]
2. Euromonitor Passport - Vapour Devices in the US. August 2015.
3. Rose SW, Barker DC, D'Angelo H, Khan T, Huang J, Chaloupka FJ, Ribisl KM. The availability of electronic cigarettes in U.S. retail outlets, 2012: results of two national studies. Tobacco Control. 2014 Jul;23 Suppl 3:iii10-6. doi: 10.1136/tobaccocontrol-2013-051461.
4. Callegaro M, Murakami MH, Tepan Z, Henderson V. Yes-no answers versus check-all in self-administered modes: A systematic review and analyses. International Journal of Market Research. 2015;57(2), 203-223.
5. Delnevo CD, Gundersen DA, Manderski MTB, Giovenco DP, Giovino GA. Importance of survey design for epidemiology of emerging tobacco product use among youth. American Journal of Epidemiology. (In Press). doi: 10.1093/aje/kwx031