What are we missing? Rethinking the evidence on smoking, vaping and health inequalities

Population-level smoking and vaping trends are central to how we understand tobacco use, evaluate policy and identify who is being left behind. But what if the evidence we rely on does not tell the whole story?

This talk brings together two strands of our work that show how our understanding of tobacco use can be distorted in different ways. First, I will consider the debate over whether e-cigarette use acts as a gateway to smoking. Using triangulation of evidence across different study designs and analytical approaches, we will examine whether the observed relationship between vaping and subsequent smoking initiation is consistent with a causal gateway effect. I will explore what these different sources of evidence tell us, where they converge and where important uncertainties remain, emphasising the need to look beyond any single study or apparent association when addressing complex causal questions. I will then turn to a less visible problem: the people who are missing from the data altogether. Our recent research estimates that around 1.9 million adults in the UK live outside the sampling frames of household surveys, including people living in communal establishments and those experiencing homelessness or temporary accommodation. Smoking prevalence in these populations is substantially higher than in the general population, meaning their exclusion may lead us to underestimate both the scale of smoking and the inequalities associated with it. Together, these examples raise a broader question: what do we need to consider when interpreting the evidence we use to guide tobacco control? I will argue that we need to look beyond headline prevalence estimates and apparently straightforward associations, paying greater attention to causality, population coverage, measurement and who is, and is not, represented in our data. If we want tobacco control to reduce inequalities rather than simply improve averages, we need to understand not only what the data tell us, but also what they might be missing.