Flavored e-cigarette use among U.S. adults: results from two national surveys
Flavored e-cigarettes have generated much controversy and interest. To examine flavored e-cigarette use among U.S. adults, we performed analyses of two national surveys. Population Assessment of Tobacco and Health (PATH) provides a representative, weighted sample of 32,320 adults (1,575 current established e-cigarette users), surveyed in-person from September 2013 to December 2014; the National Tobacco Behavior Monitor (NTBM) provided a weighted sample of 46,637 adults (4,845 past-30-day e-cigarette users), surveyed online from January 2014 to June 2015. The two surveys define current e-cigarette use differently, yet provided similar findings. Across surveys, two-thirds of e-cigarette users reported using flavored varieties (PATH: 67 % / NTBM: 67 %). Flavored e-cigarette use was lowest among non-Hispanic Caucasians (63 % / 62 %). In NTBM, African Americans reported the highest rate of flavor use (86 %), due to a high percentage using menthol e-cigarettes; 78 % of African Americans in PATH reported using flavored varieties, but the flavor was not identified. Flavor use declined steadily with age in both surveys, from 83 % / 85 % for 18-24-year-olds to 50 % / 41 % after age 65. The relationship between flavored e-cigarette use and frequency was modest (varying from 66 % to 73 % in PATH and 60 % to 76 % in NTBM, across a range of use frequencies), and was less consistent between surveys. E-cigarette users who were former smokers were more likely to use flavors (70 % / 74 %), compared to current smokers (66 % / 66 %); among current smokers, those who smoked less frequently were more likely to use flavors (80 % / 82 % of less-than-weekly smokers, 63 % / 58 % of daily smokers). These patterns suggest adoption of flavored e-cigarettes increases with decreasing smoking (including cessation), consistent with prior reports that shifts toward flavored e-cigarettes are often part of a transition away from smoking. The relatively close agreement between the two surveys - despite differences in definitions of use, sampling, and data collection methods - suggests these findings are robust, and that similar estimates can be obtained via different methods.