Is e-cigarette use in non-smoking young adults associated with later smoking? A systematic review and meta-analysis

Objective The aim of this review was to investigate whether e-cigarette use compared with non-use in young non-smokers is associated with subsequent cigarette smoking. Data sources PubMed, Embase, Web of Science, Wiley Cochrane Library databases, and the 2018 Society for Research on Nicotine and Tobacco and Society for Behavioural Medicine conference abstracts. Study selection All studies of young people (up to age 30 years) with a measure of e-cigarette use prior to smoking and an outcome measure of smoking where an OR could be calculated were included (excluding reviews and animal studies). Data extraction Independent extraction was completed by multiple authors using a preprepared extraction form. Data synthesis Of 9199 results, 17 studies were included in the meta-analysis. There was strong evidence for an association between e-cigarette use among non-smokers and later smoking (OR: 4.59, 95% CI: 3.60 to 5.85) when the results were meta-analysed in a random-effects model. However, there was high heterogeneity (I2=88%). Conclusions Although the association between e-cigarette use among non-smokers and subsequent smoking appears strong, the available evidence is limited by the reliance on self-report measures of smoking history without biochemical verification. None of the studies included negative controls which would provide stronger evidence for whether the association may be causal. Much of the evidence also failed to consider the nicotine content of e-liquids used by non-smokers meaning it is difficult to make conclusions about whether nicotine is the mechanism driving this association.


Data Collection Process
For each paper, we extracted administrative details, study details and participant characteristics. Specifically, these included: author names; year of publication; country of the study; study design; study name (if applicable); sex of included participants, percentage of males included in the total sample and in the case and control groups; number of cases, controls and the size of the cohort; year(s) of data collection; age of the total sample, cases and controls; follow up length (if applicable); comparison group; exposure; outcome; covariates; definition of e-cigarette use and smoking; and type of assessment of e-cigarette use and smoking. We also extracted exposure and control details, outcome details, and results and conclusions. Specifically, these included: stratification information; direction of effect; effect estimate reported; number of individuals included in specific analyses; number of individuals exposed and unexposed in the analysis and number of subsequent smokers for each group; effect size, confidence intervals, standard errors and p-values for both unadjusted and adjusted analyses; and the conclusion regarding support for the gateway hypothesis.

Risk of Bias Assessment
Thresholds were applied to convert the Newcastle Ottowa Scale (NOS) for study quality to Agency for Health Research and Quality standards (whereby a good quality rating indicates low risk of bias and a poor rating indicates high risk of bias). Good quality ratings were determined by 3 or 4 stars in the selection domain and 1 or 2 stars in the comparability domain and 2 or 3 stars in the outcome/exposure domain. Fair quality was determined by 2 stars in the selection domain and 1 or 2 stars in the comparability domain and 2 or 3 stars in the outcome/exposure domain. Poor quality was determined by 0 or 1 star in the selection domain or 0 stars in the comparability domain or 0 or 1 stars in outcome/exposure domain.

Causality Criteria
Four Bradford-Hill criteria were selected to assess the evidence provided by the studies for a causal association: strength of association, specificity, temporality and dose responsivity. The specific thresholds and assessment techniques used are detailed below.
Strength of association. Strong associations were defined as having an adjusted odds ratio of two or more.
BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s)

Specificity.
Studies were considered specific if they adjusted for more than basic demographics (i.e. sex, age, socioeconomic position).
Temporality. The temporality criterion was met if studies were assessed longitudinally (i.e., ecigarette use was measured at time point one with a measure of smoking prior to measuring later smoking at time point 2)retrospective measures did not meet this criterion. Dose Responsivity. Studies which measured and took into account frequency of e-cigarette use, length of time the product was used for, or how much nicotine was in the e-liquid used, were considered to meet the dose responsivity criterion.  1 or 2 stars in comparability domain AND 2 or 3 stars in outcome/exposure domain. Fair quality = 2 stars in selection domain AND 1 or 2 stars in comparability domain AND 2 or 3 stars in outcome/exposure domain. Poor quality = 0 or 1 star in selection domain OR 0 stars in comparability domain OR 0 or 1 stars in outcome/exposure domain. *Odds ratios described as strong if more the 2 and weak if less than or equal to 2. BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s)