The patterns of predictors identified for EC, ES and EC+S in men and women suggest that reactions differ by tobacco type. Our findings imply that it is not just reactions to nicotine that influence future use as reactions differ by route of administration. Snus appeared to be more tolerable and induced Volasertib clinical more favorable reactions than did cigarettes, and this was further supported by analyses conducted among dual users, which suggested future use of either tobacco product was associated with a favorable initial reaction to that product. We acknowledge that unmeasured behavioral, social, and environmental factors play a role in adoption and may also be associated with initial reactions. Strengths of this study include its population-based design, assessment of both cigarette and snus use patterns, and excellent statistical power.

Applying BRP to these data was a novel and appropriate method to determine combinations of reactions that predict future use for several reasons. First, unlike logistic regression, only individuals missing all predictor values are excluded. This reduced the chance for selection bias to impact results. Second, the predictive power of BRP is superior to logistic regression, particularly in large sample sizes. Finally, BRP reveals combinations of reactions predicting future use, allowing for a more intuitive interpretation of the data. Admittedly, BRP is a hypothesis-generating approach and we regard our findings as preliminary. Limitations are acknowledged. Participants missing all reaction data were excluded from this analysis, the majority of who were exclusive cigarette smokers.

It is unclear how missing initial reaction data among exclusive smokers would impact our results; however, our findings are consistent with prior reports. All tobacco use data were obtained through retrospective self-report and use of either tobacco product at interview was not confirmed through biochemical testing. It is possible that recall bias and subject self-selection could have affected our findings. One might expect differential recall bias, with regular users recalling more pleasant initial reactions and non-users recalling more unpleasant initial reactions, which would result in exaggerated effect sizes. Positive associations between retrospective and prospective reports of initial reactions to cigarettes suggest that recall bias may be minimal in studies of small sample sizes (Perkins et al.

, 2008; Pomerleau et al., 2005). Because ability to detect subtle differences improves as sample size increases, and our study was substantially larger than prior reports, we acknowledge our results may be more sensitive to recall bias. Future studies are needed to validate our results. Our findings highlight the need for more research on the progression AV-951 to regular snus use and dual use.