A Combination of Factors Related to Smoking Behavior, Attractive Product Characteristics, and Socio-Cognitive Factors are Important to Distinguish a Dual User from an Exclusive E-Cigarette User

Kim A. G. J. Romijnders*, Jeroen L. A. Pennings, Liesbeth van Osch, Hein de Vries, Reinskje Talhout

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review


Although total cessation of nicotine and tobacco products would be most beneficial to improve public health, exclusive e-cigarette use has potential health benefits for smokers compared to cigarette smoking. This study investigated differences between dual users and exclusive e-cigarette users provide information to optimize health communication about smoking and vaping. A cross-sectional survey (n = 116) among 80 current, adult dual users and 36 current, adult-exclusive e-cigarette users was conducted in the Netherlands. The questionnaire assessed four clusters of factors: (1) Past and current smoking and vaping behavior, (2) product characteristics used, (3) attractiveness and reasons related to cigarettes and e-cigarettes, and (4) socio-cognitive factors regarding smoking, vaping, and not smoking or vaping. We used random forest-a machine learning algorithm-to identify distinguishing features between dual users and e-cigarette users. We are able to discern a dual user from an exclusive e-cigarette user with 86.2% accuracy based on seven factors: Social ties with other smokers, quantity of tobacco cigarettes smoked in the past (e-cigarette users) or currently (dual users), self-efficacy to not vape and smoke, unattractiveness of cigarettes, attitude towards e-cigarettes, barriers: accessibility of e-cigarettes, and intention to quit vaping (A). This combination of features provides information on how to improve health communication about smoking and vaping.

Original languageEnglish
Article number4191
Number of pages12
JournalInternational Journal of Environmental Research and Public Health
Issue number21
Publication statusPublished - 1 Nov 2019


  • e-cigarettes
  • dual use
  • public health
  • machine learning
  • random forest
  • smoking behavior
  • attractiveness
  • socio-cognitive factors

Cite this