What types of interventions reduce or worsen inequalities in adolescent smoking? A review of the literature and reanalysis of intervention studies.

In many developed countries including the UK, health outcomes improve significantly as socioeconomic position (SEP) increases. Poorer health behaviours among lower SEP individuals contribute to these inequalities. In particular, while smoking rates have fallen in recent times, smoking remains more common among lower SEP individuals. Most smoking uptake occurs in adolescence (1 in 9 English adolescents are daily smokers by age 15), with socioeconomic inequalities emerging in adolescence before tracking into adulthood. Hence, action to reduce inequalities in adolescent smoking may play a vital role in reducing inequality. There is increasing debate regarding what types of actions might increase or decrease health inequalities. Many argue from perspectives such as socio-ecological theory or the theory of fundamental causes, that interventions based upon education or individual choice may widen inequality, while targeting more structural factors, such as pricing and availability may reduce inequality. However, there is currently limited empirical evidence for these propositions, as few studies have analysed effects by SEP, or attempted to identify what types of actions reduce or increase existing inequalities, in what contexts. The study will involve a review of current published evidence for impacts of such intervention on inequalities and reanalysis of published studies which collected measures of SEP.

Phase 1 will be largely descriptive and will aim to examine the extent to which socioeconomic inequalities have been considered in the design of evaluations and interventions to reduce adolescent smoking. Systematic searches of a range of databases will be used to identify intervention studies in developed countries since the turn of the century. How commonly, and in what ways, differential effectiveness was examined (e.g whether via subgroup analysis, use of interaction terms, or targeting of interventions toward lower SEP groups) within such studies will be examined. How commonly, and in what ways, impacts on inequalities in adolescent smoking are considered in intervention development will also be examined. The extent and nature of consideration of impacts on inequality in intervention and evaluation design will be compared across intervention types (i.e individual level, community level, or society level intervention). Harvest plot methodology will be used to combine studies which examine impacts of different levels of intervention on socioeconomic inequality. A database of studies which collected measures of SEP will be compiled, with studies labelled according to intervention type, study quality and country. Studies to be reanalysised in Phase 2 will be sampled from this database.

Phase 2 will aim to advance our understandings of the impacts of various types of public health interventions upon socioeconomic inequalities through reanalysis of datasets from intervention studies which collected measures of SEP. Up to 40 studies will be selected from the database developed during Phase 1, representing a range of i) individual, ii) community, iii) society and iv) multi-level interventions from a range of developed countries. Corresponding authors will be contacted by email and asked whether any unpublished analyses of differential effectiveness were conducted, and if so, if they are willing to share these. Where they were not, authors will be asked to supply original data, which will be reanalysed for differential effectiveness by SEP, or to conduct reanalysis and supply syntax and log files. The harvest plot will then be extended to incorporate these additional analyses, with effects combined using meta-analysis where feasible. The study will provide policymakers with information on how their actions might impact inequalities, advance social science theory for reducing inequalities, and highlight the importance of designing evaluations in a manner which captures impacts on inequality.

Phase 1 (the review) will identify and synthesise evidence for impacts of interventions on inequalities in adolescent smoking. CINAHL, Cochrane Central, Embase, Medline, Medline In-Process, PsycINFO and Web of Science will be searched for evaluations of interventions to reduce or prevent smoking amongst adolescents, categorised as acting at the level of the individual, community, society, or at multiple levels. Percentages reporting impact on inequality will be presented by intervention type. A database will be compiled of studies which collected SEP measures, but did not report impacts on inequality, providing the sampling frame for the programme of reanalysis in Phase 2. Finally, the review will draw tentative conclusions regarding the impact of interventions on inequalities in adolescent smoking, through synthesising studies reporting impacts on inequality, using Ogilvie and colleagues' (2008) harvest plot methodology.

Phase 2 will explore which types of interventions impact on inequalities through reanalysis of evaluations which collected SEP measures without examining differential effectiveness. Up to 40 studies will be selected for reanalysis, including a range of intervention types (individual, community, society level or multi-level interventions) and contexts. Authors will be asked whether unpublished analyses of impacts on inequality were conducted, and if they are willing to share these. Where they were not, authors will be asked to provide an anonymised dataset, or reanalyse data by SEP and provide syntax and output. Analyses will divide samples into equal high and low SEP subgroups, presenting effect sizes in each subgroup before entering an interaction term for intervention*SEP. Attention will be paid to how well powered studies are to detect interactions. The harvest plot developed in Phase 1 will be extended to incorporate Phase 2 studies. Meta-analyses will be conducted where feasible.

Who will benefit from this research?

The immediate intended beneficiaries of this research are academic researchers with a specific interest in developing theory and empirical evidence for actions to reduce inequalities in health behaviours, and academics with more general interests in promoting population health, without worsening existing inequalities. However, in the medium term, the study will provide policymakers and practitioners concerned with the formulation and implementation of public health interventions to reduce inequalities or improve population health without worsening inequalities, with information which will guide the allocation of resources to effective and equitable public health action. Ultimately, the long-term beneficiaries of the research are lower SEP adolescents, who will benefit from improved tobacco control policy and practice. Given that smoking is the leading cause of preventable ill health, particularly among low income groups, this will also contribute to a significant lowering in NHS healthcare costs.

How will they benefit?

Disseminating the findings of the review stage of the research will highlight the extent to which inequalities are currently considered in evaluation and intervention design, and will draw academic stakeholders attention to the difficulties inherent to understanding impacts of public interventions upon health inequalities within traditional approaches to outcomes evaluation which focus largely on aggregate impacts. This is intended to influence the design and analysis of future evaluations, with evaluation design and analysis plans more routinely including emphasis on health inequalities.

The second phase of the research will play a meaningful role in informing the development of theory for actions to reduce inequalities, through increasing the volume of published studies for which differential effects are examined. Hence, this phase is likely to be of greater interest to stakeholders with responsibility for developing and implementing public health interventions, given that it will provide greater detail on how their actions might reduce inequalities in health, or improve population health without worsening inequalities. Hence, in the medium term, the research will have the potential to increase the extent to which public health decision making is able to take into consideration likely impacts on inequalities, leading to adoption of policies and practices which are more likely to reduce inequalities than to widen them. Ultimately therefore, adolescents, particularly those from more disadvantaged backgrounds, will benefit from improved and more equitable tobacco control practices, and reduced risk of associated diseases. Public Health organisations will be able to direct resources more efficiently towards effective and equitable policies and practices, while reductions in smoking will lead to lower healthcare costs to the NHS.