Publishing in the journal Tobacco Control a group of scientists have examined how use of e-cigarettes is related to subsequent smoking behaviour, and report that those who use e-cigarettes were more likely to start smoking cigarettes. This analysis accompanied Roundup comments.
Title, Date of Publication & Journal
Longitudinal study of e-cig use and onset of cigarette smoking among high school students in Hawaii, British Medical Journal (BMJ)
Study’s main claims – and are they supported by the data
The press release title “Teens who use e-cigs more likely to try the real thing a year later” is potentially misleading as it implies that using an e-cig leads to using real cigarettes. This paper does not show that.
The study does provide evidence that teenagers who tried e-cigs were more likely to try cigarettes a year later, but it cannot say whether e-cigs cause smoking. Does one lead to the other? Or does using e-cigs define a certain population of teens – i.e. the ones most likely to start smoking anyway? That’s a crucial question which cannot be answered by this study.
The press release states: “This was irrespective of other factors that influence smoking uptake”. But the authors didn’t measure other factors that influence smoking uptake. In other words there are important possible confounders left unaddressed – e.g. parents’ attitudes to smoking.
Therefore it would be completely wrong to conclude from this study that e-cigs lead to smoking.
The definition of “use” e-cigs means anyone that has ever tried an e-cig, even just once.
The press release importantly notes that “subsequent regular smoking is linked only to higher levels of e-cig use at the outset”. That is correct and important to highlight.
The authors highlight that the rate of cigarette smoking was relatively low among the children in this study – in other words, most of them still didn’t smoke a year later, including the ones who had tried ecigs.
One cannot use this paper to gain any insight into whether e-cig use was related to a reduction in smoking in adults or adolescents. The authors themselves say “the present study did not provide a strong test of the question because the sample contained relatively few persons who only smoked cigarettes”.
In total 2338 children answered the questionnaire in 2013 but only 1302 (56%) of them answered the questionnaire again in 2014. That’s a lot of dropouts and could skew the data.
“3 times more likely” relates to 5% of non-smokers becoming smokers in the year and 14% of e-cig users become smokers.
Adolescents with higher parental support and from families with more education were less likely to transition from never-ever to dual-user.
68% of the children in 2013 considered e-cigs healthier than cigarettes.
They took into account that children were nested within schools i.e. those from the same school are likely to be more related to one another than children from a different school – be clearer
The study uses over 1000 school children
Crucially, the study didn’t ask about other potential confounders – e.g. parents’ smoking habits or parents’ attitudes to smoking.
The authors noted that the measure of e-cig use was relatively simple – i.e. their own definition of “use” goes from ever tried an e-cig to using one every day.
Beware of multiple testing! A lot of tests were carried out and p-values produced – which inflates the probability of finding a significant result when there isn’t one.
The authors noted there was some error in the data. Some children who were “users” of e-cigs from time 1 (2013) defined themselves as “never used” in time 2 (2014) although exact numbers of these errors aren’t reported in the paper.
This study was based in Hawaii and may not be representative of children in other countries (31% of children having used an e-cig is a bit higher than other studies have shown).
Longitudinal study – one where data is gathered for the same subjects over a period of time
Before The Headlines is a service provided to the SMC by volunteer statisticians: members of the Royal Statistical Society (RSS), Statisticians in the Pharmaceutical Industry (PSI) and experienced statisticians in academia and research. A list of contributors, including affiliations, is available here.