Predicting the long-term effect of e-cigarette use on population health: A systematic review of modelling studies

Background: In absence of epidemiological data on long-term effects of e-cigarette use on population health, mathematical modelling of possible projected outcomes has been the popular alternative. This study aims to systematically review these studies to summarize common findings and evaluate study quality.

Methods: PubMed, Scopus, Web of Science and PsycINFO were systematically searched on 23 August 2021 for modelling studies of e-cigarette use published globally. Data on study characteristics, prevalence and population health outcomes estimates were extracted from each article. Two reviewers assessed the quality of the studies.

Results: Two of twenty-two studies included were affiliated with tobacco industry. Studies were from either the USA (90%), the UK, or Singapore. Most of the studies were cohort-based, comparing projected smoking prevalence and mortality between scenarios of no-vaping (‘null’) and vaping use alongside cigarette smoking (‘alternative’) on adult population. Only one study reported morbidity outcome (quality-adjusted life-years), while only four considered demographic influences and three accessed the effects of tobacco polies like tax and raised minimum legal age in their projections. Studies differed significantly in assumptions regarding risk of e-cigarette use and transition probabilities between smoking states. In most case, introduction of e-cigarette was associated with decreases in smoking prevalence and improvement of health outcomes, except when initiation among otherwise non-smokers was assumed to be very high or when e-cigarettes were modelled as discouraging smoking cessation by a large margin. Mixed results were found on ‘gateway effect’ of e-cigarette.

Conclusion: Mathematical simulations generally suggested that e-cigarettes may contribute to smoking prevalence reduction and population health improvements in the long run, on the condition that their use can be restricted to smoking cessation. This, however, should be viewed in light of the various unverifiable assumptions used and the lack of real-world factors included. Further modelling studies attempting to overcome these limitations are recommended.