Centre for Population Health SciencesUniversity of Edinburgh

A systematic review of the effectiveness of policies and interventions to reduce socio-economic inequalities in smoking among youth.

Report March 2013Amanda AmosTamara BrownStephen Platt

1

SILNE - Tackling socio-economic inequalities in smoking: learning from natural experiments by time trend analyses and cross-national comparisons

Project team

Amanda Amos, Professor of Health Promotion

Tamara Brown, Research Fellow

Stephen Platt, Professor of Health Policy Research

Centre for Population Health Sciences

School of Molecular, Genetic and Population Health Sciences

The University of Edinburgh

Medical School

Teviot Place

Edinburgh

Scotland

EH8 9AG

Phone: (+44)-(0)131-650-3237

Fax: (+44)-(0)131-650-6909

Acknowledgements

The project team would like to thank members of the SILNE project and members of the European

Network for Smoking and Tobacco Prevention (ENSP) who helped in the search for grey literature.

2

Table of Contents

EXECUTIVE SUMMARY.............................................................................................................. 4

1 INTRODUCTION...................................................................................................................... 6

1.1 Background........................................................................................................................................ 6

1.2 Aims and objectives............................................................................................................................ 8

2 METHODS............................................................................................................................... 10

2.1 Search strategy................................................................................................................................. 10

2.2 Study selection................................................................................................................................. 112.2.1 Study selection process...................................................................................................................112.2.2 Inclusion criteria..............................................................................................................................112.2.3 Data extraction................................................................................................................................132.2.4 Quality assessment.........................................................................................................................132.2.5 Data synthesis.................................................................................................................................14

3 RESULTS................................................................................................................................. 16

3.1 Introduction..................................................................................................................................... 16

3.2 Impact of population-level policies and interventions on smoking inequalities in youth....................213.2.1 Smoking restrictions in cars, schools, workplaces and other public places..........................................213.2.1 Controls on advertising, promotion and marketing of tobacco.......................................................293.2.2 Mass media campaigns...................................................................................................................323.2.3 Increases in price/tax of tobacco products.....................................................................................333.2.4 Controls on access to tobacco products..........................................................................................373.2.5 School-based prevention.................................................................................................................433.2.6 Multiple policy interventions..........................................................................................................48

3.3 Impact of individual level cessation services and support on smoking inequalities in youth...............51

4 DISCUSSION........................................................................................................................... 55

5 CONCLUSIONS....................................................................................................................... 60

6 REFERENCES......................................................................................................................... 61

7 APPENDICES.......................................................................................................................... 657.1 Appendix A Search strategies: electronic searches, handsearching and searching for grey literature..657.2 Appendix B WHO European countries and other stage 4 countries....................................................777.3 Appendix C Inclusion/exclusion form................................................................................................787.4 Appendix D Included studies-Youth..................................................................................................807.5 Appendix E Excluded studies-Youth..................................................................................................837.6 Appendix F Data extraction - Youth...................................................................................................867.7 Appendix G Quality assessment......................................................................................................1547.8 Appendix H Summary of equity impact of youth polices/interventions............................................1567.9 Appendix I Equity impact model of youth policies/interventions by SES measure............................163

3

EXECUTIVE SUMMARY Smoking is the single most important preventable cause of premature mortality in

Europe and a major cause of inequalities in health.

While there is good evidence on what types of tobacco control policies are effective

in reducing smoking uptake in young people, little is known about what is effective

in reducing inequalities in smoking in young people.

The aim of this report was to undertake a systematic review of the effectiveness of

policies and interventions in reducing socioeconomic inequalities in smoking among

youth.

The systematic review included primary studies involving young people (aged 11-

25), published between January 1995 and January 2013, which assessed the impact of

smoking prevention policies and interventions, and smoking cessation support, by

socioeconomic status.

4

Any type of tobacco control intervention, of any length of follow-up, with any type

of smoking-related outcome was included. A broad range of smoking related

outcomes and socioeconomic variables was included.

The equity impact(s) of each intervention/policy on smoking-related outcomes was

assessed as either being positive (reduced inequality), neutral (no difference by

socioeconomic status) or negative (increased inequality).

Very few studies were found to have assessed the equity impact of the

policy/intervention and all were from tobacco control. Thirty-three studies were

included in the review, of which 31 were population level tobacco control

policies/interventions and two were individual level cessation support interventions.

The types of policies/intervention included were: smoking restrictions in cars,

schools, workplaces and other public places (9); controls on the advertising,

promotion and marketing of tobacco (3); mass media campaigns (1); increases in

price/tax of tobacco products (6); controls on access to tobacco products (5); school-

based prevention programmes (5); multiple policy interventions (3) and individual

cessation support (2). (One study was included in two types of policies/intervention

category).

Assessing the overall equity impact of different types of interventions/policies was

complicated by studies having different outcome measures and length of follow-up.

However, overall there was no consistent equity effect for each type of tobacco

control policy/intervention. Most interventions had, on balance, either a negative (11)

or neutral (15) equity impact. One had a mixed impact.

Only six of the 31 population level prevention studies showed the potential to

produce a positive equity impact. These included three US studies of increasing the

price/tax of tobacco products, two US studies on age-of-sales laws and one UK study

of a smoking prevention programme (ASSIST). The two smoking cessation studies

both used text-messaging interventions. The New Zealand study had a short-term

neutral equity impact and the US study had a short-term positive equity impact.

Very few studies have assessed the equity impact of policies and interventions on

smoking prevention or cessation in youth. There is therefore little available evidence

to inform tobacco control policy and interventions that are aimed at reducing

socioeconomic inequalities in youth smoking. There is a need to strengthen the

evidence base for the equity impact of tobacco control interventions which target

young people.

5

6

1 INTRODUCTION1.1 BackgroundSmoking prevalence rates differ substantially within European countries according to

people’s educational level, occupational class and income level; and smoking is the largest

single contributor to socioeconomic inequalities in mortality in Northern Europe. The

patterning of smoking by socioeconomic status (SES) within a country reflects the stage of

the tobacco epidemic in that country. In general smoking is initially taken up by higher SES

groups, followed by lower SES groups. Higher SES groups are then the first to show

declines in smoking, followed by lower SES groups.1 The tobacco epidemic is also gendered

in that men first take up smoking, followed by women.2 Most countries in the European

Union (EU) are characterised as being in the fourth (last) stage of the epidemic. In these

countries lower SES groups have higher rates of smoking prevalence, higher levels of

cigarette consumption and lower rates of quitting.3;4 Some EU countries are at a slightly

earlier stage. This is reflected in the differential patterning of smoking by SES and gender,

where the clear relationship between low SES and smoking found in men is only starting to

emerge in women.

SES is an important determinant of smoking uptake in young people. Parental smoking

status, which is related to SES, is a predictor of smoking uptake in young people.5;6

However, the relationship between SES and smoking uptake is generally less clear than that

for adult smoking, reflecting the difficulty of assessing SES among adolescents. Commonly

used adult measures of SES such as educational attainment, occupation and income are not

relevant for adolescents. However, some surveys have developed measures of youth SES,

including the Health Behaviour in School-aged Children survey (HBSC). The HBSC, which

is carried out in 39 countries, mostly in Europe, uses a measure of ‘family affluence’ (FAS)

to assess participants’ SES. The 2005/6 survey found that, as with adult smoking, the

relationship between youth smoking and SES varied between countries depending on their

stage of the tobacco epidemic and gender.7 Low family affluence was significantly

associated with weekly smoking among girls in nearly half the countries, but in only a few

countries among boys. This pattern was strongest for girls in countries in stage four of the

tobacco epidemic (North and Western Europe, Canada, USA). In Eastern and Southern

Europe (mostly Stage 3 countries such as Ukraine, Estonia, Russia), family affluence was

7

generally not associated with smoking. Fifteen year old girls from low affluent families in

North Europe were also more likely to have started smoking earlier i.e. at age 13 or younger.

8

Since the 1990s, many European countries have implemented new and stronger tobacco

control policies including smokefree legislation covering smoking in public places, bans on

tobacco advertising and promotion, and tax increases. There is good evidence on what is

effective in reducing adult smoking amongst the general population. A review of the

international evidence by the World Bank in 20038 identified six cost-effective policies

which they concluded should be prioritised in comprehensive tobacco control programmes:

price increases through higher taxes on cigarettes and other tobacco products including measures to combat smuggling

comprehensive smokefree public and work places better consumer information including mass media campaigns comprehensive bans on the advertising and promotion of all tobacco products, logos

and brand names large, direct health warnings on cigarette packs and other tobacco products treatment to help dependent smokers stop, including increased access to medications

These priorities have been endorsed by World Health Organisation (WHO)9 and form the

basis of the Framework Convention on Tobacco Control (FCTC), the first international

public health treaty.10

Reviews on smoking prevention in young people have endorsed the importance of these

measures for preventing smoking uptake, though the evidence on effective youth cessation

support is less strong than that for adults.5 The recent US Surgeon General’s report on

Preventing Tobacco Use Among Youth and Young Adults6 stated that the evidence is

sufficient to conclude that mass media campaigns, comprehensive community programmes,

comprehensive statewide tobacco control programmes and increases in cigarette prices

reduce smoking initiation and prevalence in youth (and taxes also reduce prevalence among

young adults). They also concluded that certain types of school programmes can produce at

least short-term effects in reducing youth smoking prevalence.

9

What is much less certain is how ‘real world’ policies and interventions that reduce overall

smoking prevalence within the general population impact on socioeconomic inequalities in

smoking. Tackling these socioeconomic inequalities in smoking is central to reducing the

health inequalities gap and is the fundamental underpinning aim of the “SILNE” project,11

“Tackling socioeconomic inequalities in smoking: learning from natural experiments by time

trend analyses and cross-national comparisons”. SILNE is a three-year European project, co-

ordinated by the University of Amsterdam, Department of Public Health, Academic Medical

Centre, the Netherlands, with financial support from the European Commission Seventh

Framework Programme; ‘Developing methodologies to reduce inequities in the determinants

of health’ programme (grant agreement no. 278273). The SILNE project involves twelve

European partners who will deliver the seven work packages which make up the project.

This systematic review is part of Work Package 6 of the SILNE project.

There have been two previous reviews on the equity impact of tobacco control

interventions.12;13 In 2008 the Centre for Reviews and Dissemination (CRD) at the University

of York published a systematic review of the equity impact of tobacco control on young

people and adults,12 focussing on population level interventionsa (not individual-level

smoking cessation interventions) published up to January 2006. In 2010 the Department of

Health’s Policy Research Programme, through the Public Health Research Consortium

(PHRC), funded a study of tobacco control and inequalities in health in England.13 This

study included a review of the evidence on the effectiveness of interventions to reduce adult

smoking amongst socioeconomically deprived populations, which built on the CRD review

and included evidence published from January 2006 until September 2010. It included both

population-level interventions and individual-level cessation support interventions. The

PHRC review concluded that there was limited evidence to inform tobacco control policy

and interventions that are aimed at reducing socioeconomic inequalities in smoking

behaviour.

a Population level control interventions have been defined as ‘those applied to populations, groups, areas, jurisdictions or institutions with the aim of changing the social, physical, economic or legislative environments to make them less conducive to smoking.’

10

While considerable progress has been made in tobacco control in many countries in the EU

in recent years, there is considerable variation in the strength and comprehensiveness of

tobacco control policies and their implementation.14 However, while overall smoking

prevalence is reducing; the social gradient is not. Addressing inequalities in smoking is a key

public health priority, starting with improving our understanding of the equity impact of

existing policies and interventions.

1.2 Aims and objectivesThe overarching aims of Work Package 6 are to undertake a systematic review of the

effectiveness of policies and interventions to reduce socioeconomic inequalities in smoking

among youth and adults, and to assess the implications of this evidence for understanding the

effects of such policies and interventions in countries within the EU.

This report focuses on the findings of the systematic review of the effectiveness of policies

and interventions to reduce socio-economic inequalities in smoking among youth. It has two

objectives:

1. To identify and review the strengths and limitations of the published evidence on the

effectiveness of policies (at the population level) to prevent and/or reduce smoking

amongst socioeconomically deprived populations as compared to higher

socioeconomic groups, and implications for European and other countries at stage 4b

of the tobacco epidemic.

2. To identify and review the strengths and limitations of the published evidence on

the effectiveness of tobacco control interventions (at the individual level) to prevent

and/or reduce smoking amongst socioeconomically deprived populations as

compared to higher socioeconomic groups, and implications for European and other

countries at stage 4 of the tobacco epidemic.

b The 4 stages of the tobacco epidemic are described: Stage 1, characterised by low uptake of smoking and low cessation rates; Stage 2, characterised by increases in smoking rates among women and an increase to 50% or more among men; Stage 3, typified by a marked downturn in smoking prevalence among men, and a plateau and then gradual decline in women; and Stage 4, marked by further declines in smoking prevalence among men and women, with numbers of new smokers starting to decrease. Richmond, R. Addiction 2003;98 (5).

11

2 METHODS2.1 Search strategy A comprehensive search strategy was developed to encompass studies published from

January 1995 to May 2012. The search included published papers identified through

searches of relevant electronic databases, and papers pending publication identified through

handsearching of key journals, and contacting key tobacco control experts. A database of

relevant references was produced using Reference Manager 12 software package. Details of

the search strategies, including hand searching and searching for grey literature, can be

found in Appendix A.

The following databases were searched:

BIOSIS

CINAHL Plus

Cochrane Library (Cochrane Database of Systematic Reviews; Database of Abstracts of Reviews of Effects; Cochrane Central Register of Controlled Trials; Health Technology Assessment Database)

EMBASE

ERIC

Conference Proceedings Citation Index

MEDLINE

PsycINFO

Science Citation Index Expanded

Social Science Citation Index.

This search was supplemented by handsearching of four key journals from January 2012 to

the end of July 2012 to identify articles ‘in press’ published on the journals’ websites:

Addiction

Nicotine and Tobacco Research

Social Science and Medicine

Tobacco Control

12

Three key reviews were also searched for relevant primary studies: the York review,12 the

PHRC review,13 and a report by the US Surgeon General on Preventing Tobacco Use Among

Youth and Young Adults6 which was published during the production of this review.

Bibliographies of included studies were also searched for further relevant studies. Members

of SILNE and members of the ENSP were asked to identify any relevant studies not

identified by the extensive searching of the electronic databases and the handsearching.

Update search

The electronic search strategy was rerun in the three databases which yielded the majority of

the included studies from the initial search (EMBASE, MEDLINE and PsycINFO) to

identify studies published between May 2012 and end of January 2013. In February 2013,

the same four key journals were handsearched to identify articles published on the journals’

websites (but not yet listed in electronic databases) for publication in journal issues up to

April 2013. See appendix A for details.

2.2 Study selection

2.2.1 Study selection processArticles retrieved from the searches were screened by title and abstract, to identify potentially

relevant studies. An initial screen of the first 200 references imported into Reference Manager

from MEDLINE were screened by title and abstract by two reviewers (AAc and TBd) to clarify

inclusion and exclusion criteria and establish consistency. The remaining references were screened

by title and abstract by one reviewer (TB) and checked by a second reviewer (AA). A second

screen of full text articles was then carried out by one reviewer (TB) and checked by a

second reviewer (AA). Any disagreements between reviewers were resolved by discussion at

each stage and, if necessary, a third reviewer (SPe) was consulted.

2.2.2 Inclusion criteriaAll primary study designs based in a WHO European country or non-European country at

stage 4 of the tobacco epidemic were eligible for inclusion (see Appendix B for list of

included countries).

c AA=Amanda Amos d TB=Tamara Browne SP=Stephen Platt

13

The inclusion ages for the youth review were 11-25 years and, for the adult review, 18+

years. Smoking uptake continues until around the age of 25 years, which is why this cut-off

was chosen for the youth review; it also enables comparisons to be made across studies set

within different countries where age of leaving secondary education can vary considerably.

However, many adult focused interventions target smokers aged 18 years and older. Thus 18

years and older was used to categorise adult interventions. In the rare cases where studies

straddled both age categories they were included in both the youth and adult reviews.

When the inclusion ages for the youth review were defined this was with a focus on studies

relating to smoking initiation. This inclusion criterion was later modified for studies

evaluating smokefree legislation in light of studies that included all ages of children.

In order to assess the equity impact of tobacco control measures in the general population,

we included both population-level policies and interventions, and individual-level

interventions which aimed to reduce adult smoking or to prevent youth starting to smoke.

Studies of population-level policies and interventions cover secondhand smoke (SHS)

exposure by SES, the strength or reach of policy coverage by SES, and the impact by SES of

the 'voluntary' adoption/spread/strength of smokefree policies, i.e., where countries do not

have comprehensive legislation.

In order to be included in the reviews an article must have assessed the equity impact of a

tobacco control intervention or policy, and have presented results with a differentiation

between high and low socioeconomic groups. In other words, the review only included

studies which reported differential smoking-related outcomes for at least two socioeconomic

groups.

Any type of tobacco control intervention, of any length of follow-up, with any type of

smoking-related outcome was included. A broad range of smoking related outcomes, either

self-reported or observed/validated, was included: initiation and cessation rates, quit

attempts, intentions to smoke/quit, prevalence, exposure to SHS, policy reach, social

norms/attitudes, use of quitting services and sources of smoking (i.e. vending machines).

Socioeconomic variables included income, education, and occupational social class, area-

level socio-economic deprivation (including neighbourhood and school-level SES), housing

tenure, subjective social status and health insurance. Proxy measures for youth SES were

also included, such as free school meals, parental educational, occupation and income.

14

A measure of SES had to be reported in the abstract of the electronic references in order to

be included. Evidence identified through handsearching, searching of key reviews, or

contacting experts, could be included if a measure of SES was reported in the main body of

the text even if the abstract did not report that SES was assessed. If grey literature, such as

reports not published as journal articles, was identified by experts as assessing equity impact

then this evidence could be included even if the abstract did not report that SES was

assessed. In addition, such reports that were written in non-English were included if an

English synopsis was provided (and otherwise met the inclusion criteria). Only studies

published since 1995 in full-text and in English language were included. No settings were

excluded. See Appendix C for inclusion/exclusion form.

The SILNE review excluded interventions targeted exclusively at one socioeconomic group

and also excluded studies which reported socio-demographic data only (without any

socioeconomic data). For example, ethnicity alone was not considered to be an appropriate

indicator of SES for this review as the smoking patterns associated with ethnicity differ from

one country to another. Interventions that focused solely on tobacco products other than

cigarettes (e.g. cigars, smokeless tobacco, waterpipes) or tobacco replacement products were

excluded, unless used as part of a smoking cessation programme. Interventions that focused

solely on outcomes for providers of a smoking cessation intervention were excluded unless

results were also reported for high versus low socioeconomic participant groups. Papers

reporting study protocol and design only without reporting the impact of the intervention or

policy were excluded.

2.2.3 Data extractionData from the included studies were extracted by one reviewer (TB) and independently

checked by another reviewer (AA). Data relating to population characteristics, study design

and outcomes were extracted into data extraction forms. Data from studies presented in

multiple publications were extracted and reported as a single study with all other relevant

publications listed in the report. Data extraction from non-English reports (grey literature) was

limited because it was derived from an English synopsis provided by an expert; therefore the

synopsis is reported directly in the text (not in data extraction tables).

2.2.4 Quality assessment

15

All included studies were assessed for methodological quality by one reviewer (TB) and

independently checked by another reviewer (SP). The exception to this was non-English

language reports (grey literature); where any reference to quality was derived from an

English synopsis and reported directly in the text. Methodological quality was assessed by

adapting the method used in the York review.12 Each study was assessed on a scale of quality

of execution using the six item checklist of quality of execution adapted from the criteria

developed for the Effective Public Health Practice Project in Hamilton, Ontario.15 Certain

items of quality are not applicable to all study designs, for example, randomisation and

comparability are not applicable to cross-sectional study designs. We added a new criterion

of ‘generalisability’ (external validity) and assessed whether the findings of each study were

generalisable at a national, regional, or local level.

2.2.5 Data synthesisGiven the variations in study methodologies, intervention types and outcome measures, the

results are presented in the form of a narrative synthesis and according to intervention type

(population level policies/interventions and individual level cessation support interventions).

In order to provide a simple basis for comparing the methodology of each study a typology

of study designs was devised (Table 1).

Table 1 Typology of study designsCode Study design

1.0 Population-based observational1.1 Cross-sectional1.2 Repeat cross-sectional1.3 Cohort longitudinal1.4 Econometric analyses (cross-sectional data)2.0 Intervention-based observational2.1 Single intervention (before and after, same participants)2.2 Single intervention with internal comparison2.3 Comparison between different types of intervention3.0 Intervention-based experimental3.1 Randomised controlled trial (individual or cluster)3.2 Non-randomised controlled trial3.3 Quasi-experimental trial4.0 Qualitative4.1 Cross-sectional4.2 Repeat cross-sectional4.3 Longitudinal

16

The equity impact of each intervention/policy is summarised by adapting a model used in the

York review16:

The null hypothesis that for any given socio-economic characteristic related to education, occupation or income, there is no social gradient in the effectiveness of the intervention i.e. a neutral equity impact.

The hypothesis of a positive equity impact defined as evidence that groups such as lower occupational groups, those with a lower level of educational attainment, the less affluent, those living in more deprived areas, are more responsive to the intervention.

The hypothesis of a negative equity impact defined as evidence that groups such as higher occupational groups, those with a higher level of educational attainment, the more affluent, or those who live in more affluent areas are more responsive to the intervention.

The main strengths and limitations of each study, particularly internal and external validity,

are considered when discussing the equity impact of each intervention. Particular attention is

given to the issue of generalisability: to what extent are results from interventions and

policies carried out in various countries transferable across Europe despite differences in

tobacco control policies, stage of the tobacco epidemic, socioeconomic conditions, and other

factors? We draw conclusions about the strengths and weaknesses of the current evidence of

the impact of tobacco control and other policy interventions on reducing socioeconomic

inequalities in smoking in youths and adults (equity impact) and identify the most effective

and promising interventions.

17

3 RESULTS3.1 IntroductionThe initial electronic search produced 12,605 references after duplicates were removed. Two

hundred and eighty-seven references were identified as potentially relevant to the reviews

and 286 references were successfully obtained as full-text journal articles. Of these 286 full-

text articles, 171 were excluded. Sixteen of the remaining 115 studies focused on young

people and were included in the youth review. In addition to these 16 studies, a further 10

studies (11 papers) were identified through handsearching, searching of key reviews and

contacting experts. Three of these 10 studies17-19 were identified in one paper by Mercken et

al20 which included secondary analyses of these three primary studies, and these four papers

are classed as three studies. An update of the searches was carried out in January 2013

which included both electronic searching, handsearching and contact with experts, which

identified a further seven relevant studies .

In summary, a total of 33 studies were included in the youth review; of which 31 studies

were population level polices/interventions and two studies were individual level cessation

support interventions. Appendix D contains bibliographic details for all the included youth

studies including details of source. The details of studies that were excluded at the stage of

screening the full-text articles, for the initial electronic search (n=13) and for the updated

electronic search (n=7) are listed in Appendix E with reasons for exclusion.

The findings of these 33 included studies are presented by intervention type. A summary of

studies by design and type of intervention are summarised in Table 2. Population-level

interventions (which aimed to change social norms, smoking behaviour and/or access to

tobacco) included: smoking restrictions in cars, schools, workplaces and other public places;

controls on advertising, promotion and marketing of tobacco; anti-tobacco mass media

campaigns; increases in price/tax of tobacco products; controls on access to tobacco

products, school-based prevention programmes, and multiple policy interventions.

Individual-level cessation support interventions included two interventions using mobile

phone text messaging.

Data extraction tables and quality assessment, grouped by intervention type, can be found in

Appendices F and G, respectively. Textual and visual summaries of the data can be found in

18

Appendices H and I, respectively. It should be noted that whilst the equity impact graph

(Appendix I) is meant to provide a visual representation of the equity impact of the various

population-level policies/interventions; it should be interpreted in conjunction with the

narrative descriptions of the results.

19

Figure 1 Study selection flow chart

*9 papers assessed in more than 1 review; **1 paper = secondary analyses of 3 papers, so 4 papers classed as 3 studies;***2 papers assessed in more than 1 review

20

Electronic search May 2012Titles and abstracts screened

n = 12,605

excluded from title and abstract

n = 12,318

Full papers ordered n = 287

screenedn = 286

INCLUDEDN = 115

EXCLUDED (full text)n = 171

(13 youth + 34 adult policy + 133 adult cessation)*

youth includedn = 16

youth handsearching, reviews,

expertsn = 11

update youthn = 2

update youth handsearch, experts

n = 5

total number youth studiesn = 33**

individual-level cessation studiesn = 2

population-level studiesn = 31

update electronic search January 2013

titles and abstractsn = 1149

update full papers screenedn = 42

update includedn = 16

update excluded n = 26

(7 youth + 13 adult policy + 8)***

Table 2 Summary of studies by design and intervention type*

21

Design code Intervention typeSmoking restrictions in cars, schools, workplaces, and other public places1.2 Akhtar 20101.1 Galan 20121.2 MacKay 20101.2 Millett 20131.2 Moore 20111.2 Moore 20121.1 Nabi-Burza 20121.1 Noach 20122.1 Woodruff 2000Controls on advertising, promotion and marketing of tobacco1.1 Gilpin & Pierce 19973.1 Hammond 20111.1 Pucci 1998Mass media campaigns1.2 Vallone 2009Increases in price/tax of tobacco products1.1 Biener 19981.1 Gilpin & Pierce 19971.3 Glied 20021.4 Gruber 20001.4 Madden 20071.1 Perretti-Watel 2010Controls on access to tobacco products1.3 Kim 20061.1 Lipperman-Kreda 20121.2 Millett 20111.2 Schneider 20111.1 Widome 2012School-based prevention programmes3.1 Bacon 20013.1 Crone 2003**3.1 De Vries 2006**3.1 Campbell 2008**3.3 Menrath 2012Multiple policy interventions***1.2 Helakorpi 20081.3 Pabayo 20121.2 White 2008Individual cessation support

22

3.1 Rodgers 20053.3 Ybarra 2013* Studies can be categorised in more than one intervention type; **Study identified in Mercken 2012; ***Interventions that have several elements and/or papers that try to assess the relative impact of several policy interventions over a period of time

23

3.2 Impact of population-level policies and interventions on smoking inequalities in youth

3.2.1 Smoking restrictions in cars, schools, workplaces and other public placesA total of nine studies assessed the socio-economic impact of smoking restrictions in public

places; one intervention study21 five repeat cross-sectional studies22-26 and three single cross-

sectional studies.27-29 Three studies explored the impact of national comprehensive smokefree

legislation on primary school children’s exposure to secondhand smoke (SHS), one of which

was set in Scotland22 one in Wales25and one study pooled data from Scotland, Wales and

Northern Ireland.26 Two studies examined whether smokefree legislation was associated with

change in hospital admissions for childhood asthma in Scotland23 and England.24 One study

examined smoking behaviour in cars with children present, amongst smoking parents in the

US.27 A further two studies explored the impact of voluntary compliance with smoking

restrictions on smoking behaviour in secondary school children, one of which was set in

Spain29 and one in Israel.28 An intervention study assessed the impact of an organisational

(workplace) smokefree ban in 19 year old female US Navy recruits, using a before and after

experimental study design.30

Only the two school-based studies of comprehensive smokefree legislation22;25 scored the

maximum according to study design. All the study samples except two27;28 were

representative of the study population. All cross-sectional studies used credible data

collection methods, and all repeat cross-sectional studies had a sufficient number of

participants included in analysis in each wave. The US intervention study30 had partially

validated data collection instruments and an acceptable level of attrition for post-intervention

data but not at the 3-month follow-up. It is reasonably likely that the observed effects of

smokefree legislation in Scotland, Wales and Northern Ireland; and the smokefree workplace

ban30 were attributable to the interventions under investigation.

24

The comprehensive smokefree legislation studies including two studies of hospital

admissions for asthma are all generalisable on a national level (all UK based). The Spanish

study29 results of voluntary compliance are likely to be generalisable at the regional level.

The study population in the Israeli study28 of voluntary compliance, was heterogeneous; with

a broad range of ethnic, religious and socioeconomic subpopulations and is not generalisable

to other WHO European or stage 4 countries. It was unclear how generalisable the results

were from the study of smoking in cars amongst US parents.27 The workplace study

population was specific to female young US Navy recruits only30.

National smokefree policies

Three studies from the changes in child exposure to environmental tobacco smoke (CHETS)

study were included: CHETS Scotland,22 CHETS Wales25 and a CHETS UK study.26

Individual data from the Scottish22 and Welsh25 studies are described separately and are also

included in the pooled analyses of UK data along with data from Northern Ireland.26 The

Scottish, Welsh and Northern Irish studies applied repeat cross-sectional class-based

surveys, in order to explore the impact of smokefree legislation on 11 year old children’s

exposure to SHS; using biochemical measures (salivary cotinine levels).

The smokefree legislation in Scotland22 was associated with a decline in cotinine levels

across all socio-economic groups. The greatest absolute decline in cotinine levels was among

the lowest self-reported family socioeconomic classification (SEC) and family affluence

scale (FAS) groups, even after adjusting for parental smokers (e.g. 0.10ng/ml in SEC1 vs

0.28ng/ml in SEC4). However, a linear regression model suggests that relative inequality

between socio-economic groups had widened; the decline in SHS exposure among children

from lower SES households was greater in absolute terms but smaller in relative terms,

compared with changes in SHS exposure among children from higher SES households.

Cotinine levels remained the highest in children from the lowest SEC/FAS groups.

25

The likelihood of providing a sample containing an undetectable level of cotinine increased

significantly after smokefree legislation in Wales25 among children from high SES

households [relative risk ratio (RRR) = 1.44, 95% CI = 1.04–2.00, p=0.03] and medium SES

households (RRR = 1.66, 95% CI = 1.20–2.30, p<0.01), while exposure among children

from lower SES households remained unchanged (RRR=0.93, 95% CI=0.62-1.40, p=0.72).

Parental smoking in the home, car-based SHS exposure, and perceived smoking prevalence

were highest among children from low SES households. Parental smoking in the home and

children’s estimates of adult smoking prevalence declined only among children from higher

SES households. Children’s estimates of people smoking in the streets outside buildings

declined greatest and approached statistical significance amongst children from high-SES

households only.25

In summary, in Wales25 post-legislation reductions in SHS exposure were limited to children

from higher SES households whose exposure was already significantly lower prior to

smokefree legislation. Children from lower SES households continued to have high levels of

exposure (though these had not increased), particularly in homes and cars, and to perceive

that smoking is the norm among adults. Therefore the smokefree legislation was potentially

associated with increased socioeconomic disparity in terms of SHS exposure amongst

children. Average cotinine concentrations among children in the Scottish study were

substantially higher than in the Welsh study, and children’s SHS exposure outside of the

home was perhaps greater in Scotland, with impacts of the smokefree legislation therefore

greater overall in Scotland than in Wales, and distributed among all socio-economic groups.

One UK study26 pooled data from the Scottish, Welsh and Northern Irish CHETS studies.

The pooled data were used to examine socioeconomic patterning (using the FAS) in

children’s SHS exposure, and parental restrictions on smoking in private spaces (cars,

home). Participants were non-smokers (self-reported non-smokers providing saliva samples

containing <15ng/ml cotinine) in their final year at 304 primary schools in Scotland (n =

111), Wales (n = 71) and Northern Ireland (n = 122). Multinomial regressions were used to

assess change in SHS exposure as measured by cotinine levels; and change in home-smoking

restrictions. Binary logistic regression models examined car-based smoking. The pooled data

was adjusted for country and age, and clustering was accounted for. The data set comprised

10, 867 children (5347 baseline/5520 follow-up), average age was 11.2 years. SES varied

significantly between survey years, with affluence being higher at follow-up survey.

26

Percentages of children with undetectable concentrations of cotinine increased from 31.0%

(n = 1715) to 41.0% (n = 2251) following legislation overall, and from 20.1% to 34.2%,

44.9% to 51.0% and 38.6% to 42.9% in Scotland, Wales and Northern Ireland, respectively. 26 Regression analysis indicated that the relative risk of children’s samples containing no

detectable cotinine increased significantly following legislation. However this was accounted

for by decreases in samples containing low levels of cotinine rather than decreases in

samples containing higher levels of cotinine; and this was the case in all three countries and

after adjusting for parental smoking and smoking restriction levels in homes and cars. 26

Children of high SES were significantly more likely to have no detectable cotinine and

significantly less likely to have high levels of cotinine following the smokefree legislation

compared to lower SES children, and this remained significant following adjustment for

country, parental smoking and private smoking restrictions. The study26 author’s report that

the gap between low and high SES children appears to have widened following the

legislation, in terms of children with no detectable cotinine levels. A trend towards widening

inequality was also seen within each individual country for no detectable cotinine levels.

Gradients for higher cotinine levels remain unchanged.

Two studies evaluated the impact of national smokefree legislation on emergency hospital

admissions for asthma in children aged less than fifteen years: one set in Scotland23 and one

set in England.24 Both study samples were representative of the general population and

generalisable on a national scale. Both studies used SES quintiles based on the Index for

Multiple Deprivation and both studies applied binomial regression models to assess hospital

admissions. The English study24 also produced admission rate ratios, which is the ratio of the

actual admission rate in relation to the rate projected by the underlying trend.

27

A Scottish study23 assessed the impact of national smokefree legislation on hospital

admissions for childhood asthma by linking data from the Scottish Morbidity Record and

death-certificate data to identify all hospital admissions and deaths before arrival at the

hospital that occurred from January 2000 through October 2009. Before the legislation was

implemented, admissions for asthma were increasing at a mean rate of 5.2% per year (95%

confidence interval [CI], 3.9 to 6.6). After implementation of the legislation, there was a

reduction of 18.2% (95% CI, 14.7 to 21.8; P<0.001) in the annual rate of asthma admissions,

resulting in a net reduction in asthma admissions of 13.0% per year (95% CI, 10.4 to 15.6).

The study accounted for asthma deaths and showed that the decrease in admissions was not

due to an increase in the incidence of deaths before arrival at the hospital. There were no

significant interactions between hospital admissions for asthma and quintile of SES. All SES

subgroups were associated with significant reduction in admissions.

An English study24 assessed the impact of national smokefree legislation on hospital

admissions for childhood asthma, using Hospital Episode Statistics over 8.5 years (April

2002 to November 2010). Before the implementation of the legislation, there was a mean

increase in the admission rate for asthma of 2.2% per year (adjusted rate ratio 1.02; 95% CI:

1.02–1.03). After implementation of the legislation, there was a significant immediate

reduction in the admission rate of 8.9% (adjusted rate ratio 0.91; 95% CI: 0.89–0.93) and a

reduction in time trend of 3.4% per year (adjusted rate ratio 0.97; 95% CI: 0.96–0.98).

Overall, the legislation was associated with a net 12.3% reduction of hospital admissions for

childhood asthma in the first year. This change was equivalent to 6802 fewer hospital

admissions in the first 3 years after implementation. The results were very similar when

based on admissions data alone, as there were few recorded deaths prior to admission.

Reductions in asthma admissions did not differ by SES.

Both studies23;24 were sufficiently similar to enable comparison and show that both the

English and Scottish smokefree legislation were associated with significant reductions in

admissions for asthma across all SES subgroups i.e. a neutral equity impact. The relative

rate of admissions before the legislation was higher in Scotland compared to England, and

relative reductions in hospital admissions after the legislation were higher in Scotland

compared with England, however the net overall reduction in hospital admissions was

similar in both studies (12-13%).

28

Neither study determined the extent to which the observed reduction in asthma was due to

reduced exposure to SHS by setting (public places, home, car) or reduction in smoking

among children. The impact on results of changes in the treatment of asthma and diagnostic

coding of asthma cannot be ruled out. However both studies assessed asthma which required

hospitalisation (i.e. severe asthma).

Smokefree car policies

Pooled data from the CHETS26 study showed that in the UK as a whole and also within

England, Northern Ireland and Wales, as SES increased, the likelihood of partial or no home

smoking restrictions (rather than full smoking restrictions) decreased significantly, whilst the

odds of smoking being allowed inside the family car also decreased significantly. These

trends remained after adjustment for parental smoking and there was no change in inequality

following legislation i.e. a neutral equity impact.

A US study27 determined the prevalence of parents smoking in their cars with children

present and how often paediatric health care providers advised parents to have smoke-free

cars. The study used baseline data from 10 control sites (in 8 US states) from a cluster RCT

‘Clinical Efforts Against Secondhand Smoke Exposure’ which was an intervention to

address parental tobacco use within the paediatric clinic setting. The study sample were

parents or legal guardians who accompanied a child to the visit; were at least 18 years old;

spoke English; had smoked at least a puff of a cigarette in the past 7 days and completed a

baseline enrolment survey for which they received $5 cash.

Parents who smoked were asked about smoking behaviours in their car and receipt of smoke-

free car advice at the visit. Parents were considered to have a “strictly enforced smoke-free

car policy” if they reported having a smoke-free car policy and nobody had smoked in their

car within the past 3 months. The measure of SES used was level of education (high school

or less versus some college or college graduates). Analyses were limited to parents who

smoked and who reported having a car that they owned or travelled in frequently, it was

unclear how representative this study sample was of the SES of the general population.

29

Twenty-nine percent of 795 parents reported a smokefree car policy and 48% reported that

smoking occurred with children present in the car. Fourteen percent of smoking parents

reported being asked if they had a smoke-free car, and 12% reported being advised to have a

smoke-free car policy by a paediatric health care provider. Of those who smoked with

children present in the car, only 5% were counselled about having a smoke-free car.

No significant association was found between parents education level and having a strictly

enforced smokefree car policy. However, parents of children aged less than one year were

more likely to have strict smoke-free car policies if they were college educated (OR:2.42;

95% CI: 1.21 to 4.83, p = 0.013). Strict smoke-free car policies were more common when

parents were both light smokers (smoked 10 cigarettes or less per day) and college educated

(OR: 2.88; 95% CI: 1.24 to 6.66, p = 0.013).

Voluntary compliance with smoking restrictions in schools

Two cross-sectional studies explored the impact of voluntary compliance with smoking

restrictions on smoking behaviour in secondary school children, one of which was set in

Spain29 and one in Israel.28 The smoking outcomes were not biochemically validated and

were based on self-report.

In Madrid smoking has been banned in schools since August 2002 however at the time of

this survey29 among smokers aged 15 to 16 years, 50.6% had smoked on school premises

during the last thirty days with significant variability (0% to 100%) between schools. A

lower probability of smoking on school premises was found among adolescents whose

fathers had a university education (OR 0.43; 95% CI: 0.19 to 0.96) or among those who did

not know the level of studies of their father (OR 0.39; 95% CI: 0.16 to 0.94) compared with

those with fathers who had a very low level of educational attainment. A lower probability of

smoking on school premises was found for state subsidized private schools (OR 0.20; 95%

CI: 0.11 to 0.35) and non-subsidized private schools (OR 0.30; 95% CI: 0.14 to 0.62) when

compared with that for public schools. Employment status of either parent, educational level

of the mother, SES of the school census tract, written reference to a smoking control policy

and educational activities about smoking prevention were not significantly associated with

smoking on school premises among student smokers.

30

In Israel28 there was no comprehensive smokefree ban at the time of the survey and most

Israeli adolescents (average age 15 years) were exposed to SHS (total: 85.6%; home: 40%;

school: 31.4%; entertainment: 73.3%; other: 16.3%). Parental education was not a significant

determinant of smoking in school but correlates of exposure at school differed from those at

home. Adolescents whose fathers had less than 12 years of education were more exposed to

SHS at home, than were teenagers whose fathers had a degree from a university or college

(OR = 1.48; CI: 1.09 to 1.99, p = 0.0111). Adolescents with less-educated mothers were

more exposed to SHS at home than teenagers with mothers with degrees from a university or

college (OR = 1.39; CI: 1.02 to 1.90, p = 0.0366). The high levels of SHS exposure among

Israeli adolescents were characterized by different patterns of exposure among different

population subgroups. Israel is a heterogeneous country; with a broad range of ethnic,

religious and socioeconomic populations and the results are not generalisable to other WHO

European or stage 4 countries.

Workplace smokefree policies

One intervention study assessed the impact of an organisational (workplace) smokefree ban

(24-hours, 8-weeks) in 19 year old female US Navy recruits, using a before and after

experimental study design.30 Among the 4393 recruits who provided entry (before) and

graduation (after) survey data, 41.4% (n = 1819) reported any smoking in the 30 days before

entering compared with 25% that reported being a smoker at graduation (after), which was a

significant reduction. Slightly over two-thirds (n = 724) of “smokers” who responded to the

follow-up survey had resumed smoking three months after graduation, and 32% (n = 340)

reported not smoking. Among past month smokers at entry (before), the relapse rate at the

three month follow-up after graduation was 81%. Daily smokers at entry (before) had the

highest relapse rate (89%) at the three month follow-up after graduation. The study did not

aim to assess differential impact by SES but reported that education did not significantly

predict smoking relapse. It was not reported whether there was a difference by SES in

change over time.

31

A response bias is present in this study; there was a low response rate (39%) at the 3-month

follow-up, and non-respondents had a slightly higher past 30 day smoking rate at baseline

than did respondents. In addition, the definition of ‘smoker’ differed at graduation (post 8

weeks) from baseline and 3-month follow-up. The group of smokers assessed for relapse was

broadly defined and included daily smokers, occasional smokers, experimenters, or former

smokers. As well as these quality-related issues, the study only included female recruits and

results may not be generalisable to a civilian population or setting.

Summary

The evidence relating to smokefree restrictions is limited to eight cross-sectional studies and

an intervention study of a workplace 24-hour 8-week smoking ban.

National comprehensive smokefree restrictions are associated with declines in SHS exposure

in primary school children but the equity effect may vary according to how exposure is

measured (absolute levels or relative levels), on the pre-ban level of exposure and the

balance between sources of exposure i.e. public places versus home. Prior to the CHETS

studies, scant attention has been paid to whether adoption of private smoking restrictions

following smokefree legislation has been patterned by SES.

Pooled data from Scotland, Wales and Northern Ireland following national smokefree

legislation showed that declines in exposure occurred predominantly among children with

low exposure before legislation, and from more affluent families, leading to increased

socioeconomic disparity (negative equity impact). Substantial socioeconomic gradients in

proportions of children with higher SHS exposure levels remained unchanged. Children from

lower SES households continued to perceive that smoking is the norm among adults whereas

smoking as a perceived norm declined amongst high-SES children.

32

Pooled data from Scotland, Wales and Northern Ireland following national smokefree

legislation showed that there was no change in inequality following legislation. As SES

increased, the likelihood of partial or no home smoking restrictions (rather than full smoking

restrictions) decreased significantly, whilst the odds of smoking being allowed inside the

family car also decreased significantly. Only one US study was included, of parental

smoking behaviour in cars it was found that parent’s education level interacted with a child’s

age and the number of cigarettes smoked per day, both of which were significant predictors

of car smoking policy. Parents with higher SES that were light smokers were more likely to

have a strict no smoking car policy and higher SES parents with children less than one year

were also more likely to have a smokefree car policy.

English and Scottish national smokefree legislation was associated with a significant

reduction in childhood asthma admissions which did not differ by SES (neutral equity

impact).

When reviewing whether students comply with smoking restrictions in secondary schools

where there is no enforced and comprehensive smokefree ban, it is apparent that parental

education may influence smoking behaviour of adolescents and smoking behaviour amongst

adolescents is also influenced by the setting (home/school). Two school-based studies in two

very different countries showed conflicting results. A study in Israel where there was no

comprehensive smokefree ban showed high levels of SHS exposure among Israeli

adolescents which were characterized by different patterns of exposure among different

religious groups; however parental education was not a significant determinant of smoking

in schools. Second-hand smoke exposure from outside the home and school settings was

sizeable and overall SHS exposure and SHS exposure at home was greater among lower SES

adolescents. In a study in Spain where there were school smoking bans but variable

enforcement; adolescents whose fathers had a lower level of educational attainment were

more likely to smoke on school premises.

A 24-hour 8-week workplace ban in the US Navy did reduce the proportion of women

smoking immediately post-ban but most had relapsed by 3-month follow-up. Education did

not significantly predict smoking relapse however the response rate to the follow-up was low

and non-respondents were more likely to be smoking.

3.2.1 Controls on advertising, promotion and marketing of tobacco

33

Three very different US studies assessed the equity impact of controls on the advertising,

promotion and marketing of tobacco products including; a retrospective survey31 of the

impact on smoking initiation of cigarette prices and tobacco industry marketing budgets

conducted in the US in 1993 of nearly 141,00 respondents aged 17 to 38 years that would

have been aged between 14 and 21 years old between 1979 and 1989, an RCT32 of a short

online survey of brand appeal of cigarette packaging, and an observational field study of

advertising density with school ‘buffer zones’.33 The RCT consisted of a convenience

internet sample and it was not clear if it was representative of the study population. In

addition there were some significant differences among the women at baseline between

treatment groups which may have affected the results: education varied by condition, with

the highest level of education in the standard pack condition, and number of cigarettes

smoked per day was significantly higher in the plain pack condition compared with the

standard pack condition among current smokers. All three studies used credible data

collection methods. It is reasonably likely that the observed effects of cigarette packaging

were attributable to the intervention under investigation and that these results are likely to be

generalisable at a national level. The observational field study of advertising density with

school buffer zones may only be generalisable at the local level as the study population were

limited to neighbourhoods in Boston, Massachusetts, US and no details of the 6 Boston

neighbourhoods were provided.

One retrospective survey31 conducted in the US in 1993 of nearly 141,00 respondents aged

17 to 38 years that would have been aged between 14 and 21 years old between 1979 and

1989 examined trends in smoking initiation by cigarette prices and tobacco industry

marketing budget. Adolescent initiation rates decreased from 1979 to 1984 but increased

thereafter. Initiation rates were highest among high school dropouts and lowest amongst

those who eventually attended college. In 1988 the initiation rate was 9.9% for those who

did not graduate from high school, 6.9% for high-school graduates reporting no college and

3.7% for those reporting at least some college education. The equity results from the study

can only be tentative because the study does not directly assess the effect of changes in the

tobacco marketing budget or cigarette prices on smoking initiation rates by education level.

The study simply highlights that cigarette prices and tobacco marketing budget increased

during this decade as did smoking initiation rates amongst adolescents, and that marketing

expenditure may be associated with an increase in smoking initiation especially in young

people with lower levels of education.

34

A recent RCT of a short online survey intervention32 examined brand appeal of cigarette

packaging amongst women aged 18 to 19 years in the US. The convenience sample was

randomised to four experimental conditions which viewed eight cigarette packages one at a

time displayed in random order and according to the four experimental conditions: (1)

female-oriented packages (standard condition); (2) female-oriented packages with brand

imagery, including colours and graphics, but with descriptors (e.g. slims) removed; (3)

female-oriented packages without brand imagery and descriptors (i.e., plain packages); and

(4) popular U.S. brands of “ regular ” or non – female- oriented packages.

Women in the high income and high education categories endorsed a greater number of

positive smoker traits (female/male, glamorous/not glamorous, cool/not cool, popular/not

popular, attractive/unattractive, slim/overweight, and sophisticated/not sophisticated) than

those in the low income and low education categories. High income respondents were more

likely to endorse smoking and weight control beliefs compared with respondents reporting

low (OR = 1.70, 95% CI = 1.12 – 2.60) and medium income (OR = 1.73, 95% CI = 1.09 –

2.73) and those who did not state their income (OR = 2.17, 95% CI = 1.29 – 3.65). The

reactions to and perceptions of the different types of packs was the same by SES for nearly

all the measures. No significant differences in pack selection were observed for smoking

status, age, income, education, ethnicity, or weight concerns.

An observation field study33 assessed youth exposure to stationary outdoor tobacco

advertising density within FDA 1,000 foot buffer zones around schools in 6 Boston

neighbourhoods in the US. The overall advertising density for schools in all neighbourhoods

combined was higher for middle (10.1) and high schools (9.9) than for elementary schools

(6.3). The majority of outdoor tobacco advertising was in the neighbourhoods with the

lowest median household incomes. The study probably underestimated advertising density

because it does not include point-of-purchase advertising, advertising inside stores that is

seen from the street, or advertising on taxis and buses.

Summary

Three very different US-based studies assessed the equity impact of controls on the

advertising, promotion and marketing of tobacco products.

35

One study showed that initiation rates of smoking amongst adolescents varied by level of

education; initiation rates were highest amongst high-school dropouts and lowest amongst

those who eventually attended college. Marketing expenditure may be associated with an

increase in smoking initiation especially in young people with lower levels of education.

Very tentatively, controlling the promotion of cigarettes through plain packaging might have

a positive effect on all young women and have a neutral equity effect for young women

because reactions to/perceptions of different types of packs were the same regardless of SES

for nearly all the measures.

Despite the FDA buffer zone policy, one study showed that tobacco advertising is targeted at

adolescents of low SES inside school buffer zones, particularly middle and high school

adolescents, and this has the potential to increase inequality in smoking behaviour amongst

youth. Banning all outdoor tobacco advertising would reduce exposure particularly in

children of lower SES.

3.2.2 Mass media campaignsOne telephone survey34 evaluated the impact of the US truth® campaign on awareness and

receptivity among youth aged 12 to 17 years. The truth® campaign is a branded counter

tobacco marketing campaign designed to prevent smoking among at-risk youth, primarily

through edgy television advertisements with an anti-tobacco industry theme. Seven waves of

Legacy Media Tracking Survey data were collected from September 2000 through to

January 2004. It was unclear how representative the study sample was of the study

population because response rates declined over the seven waves of data collection, from

60% to 30%.

36

Youth who lived in zip codes in which the median household income was less than or equal

to US$ 35,000 had a lower level of confirmed awareness of the campaign than respondents

in each of the other income categories (p< 0.05). There were no statistically significant

differences in confirmed awareness by median level of education, though there was a pattern

in which the proportion of confirmed awareness increased with education. There were no

differences in receptivity by median household income or median household education,

though there was a pattern of increasing receptivity with greater income and education.

During the campaign there was a gradual shift towards cable TV ownership and education is

positively associated with cable TV ownership. However the authors report that SES

differences were concentrated in the early years of the campaign when it was aired mainly

through network TV. The study controlled for year of survey administration and the effect of

the intervention over the seven waves of survey data. It is not reported whether the effect of

the intervention differed by SES over time.

Summary

This one study of a relatively large, lengthy and well-funded anti-tobacco mass media

campaign, using repeat cross-sectional data over four years, showed that youth who lived in

zip codes in which the median household income was less than or equal to US$ 35,000 had a

lower level of confirmed awareness than respondents in each of the other income categories.

Zip code level median household education was not associated with confirmed awareness

and there were no differences in receptivity by zip code level income or education. The

equity impact of the mass media campaign is unclear as the effect on campaign awareness

varied according to the SES variable that was measured (income/education) and the equity

impact in terms of receptivity appeared neutral.

3.2.3 Increases in price/tax of tobacco products

37

Six studies evaluated the equity impact of increases in the price or tax of cigarettes, the

majority of which were US-based studies using retrospective survey data. Two studies35;36

were econometric studies (report price elasticities), one of which used both longitudinal and

cross-sectional data.35 One study used retrospective cohort data37 and the remaining three

studies were single cross-sectional studies.31;38;39 Four of the study samples were

representative of the study populations and for two studies it was unclear if the samples were

representative.35;37 For three studies it was unclear if credible methods of data collection had

been used, due to lack of reported information in one case38 and unpublished data in the other

two studies.37;39 Two studies38;39 were likely to be generalisable at the regional level and two

studies31;35;36 at a national level.

A retrospective survey38 examined smokers aged 12 to 17 years perceptions of the impact of

statewide tobacco taxes in Massachusetts, USA. Teenage smokers from low income

households were much more likely than more affluent teenagers to report cutting the costs of

their smoking (by cutting down the amount smoked or, less often, by switching to cheaper

brands) in response to the price increase, rather than do nothing (OR 7.57; 95%CI: 1.55 to

36.98) or cutting costs rather than consider quitting (OR 14.72; 95%CI: 2.55 to 84.95).

Household income was unrelated to the choice between considering quitting and doing

nothing (OR 0.51; 95% CI: 0.13 to 2.77). Young low income smokers were not more likely

than wealthier teenagers to consider quitting. There appeared to be a positive equity impact

on smoking less and a neutral equity impact on quitting behaviour of statewide tobacco tax

increases. It should be noted that 53% of the teenagers who continued to smoke denied

having had any of the 3 potential reactions to price increase and so it is possible that the

study failed to measure an important variable.

One US retrospective survey31 examined trends in smoking initiation by cigarette prices and

tobacco industry marketing budget; results are reported in section 3.2.1. Initiation rates were

highest among high school dropouts and lowest amongst those who eventually attended

college. The study highlights that cigarette prices and tobacco marketing budget increased

during this decade as did smoking initiation rates amongst adolescents, and that price

increases did not reduce smoking initiation.

38

One US econometric study35 tested the assumption that policies targeting youth to reduce

smoking initiation will reduce lifetime smoking propensities. Estimates of the effect of

current taxes (taxes in the year of interview) on current adult smoking measured in 1984

(aged 19 to 28), 1992 (aged 27 to 35), and 1994 (aged 29 to 37) revealed that the age

coefficients were positive (measured in 1979) showing that there was a positive secular trend

in youth smoking. Youth from higher income families were less likely to smoke, whereas the

results were inconsistent for level of education between different types of analyses (probit

marginal effects and linear regression fixed effects). Participation elasticities for the three tax

current tax variables (1984, 1992, and 1994) using probit marginal effects or linear

regression fixed effects were −0.1 and−0.09, respectively.

The study estimated the effect of cigarette taxes at age 14 years (in 1979) on future overall

smoking behaviour, quitting and initiation using prospective longitudinal cohort data with

cross-sectional analyses. Cigarette tax at age 14 had the most effect on low income people at

ages 19-28 for current smoking but not late initiation or quitting according to longitudinal

data. The effect of cigarette tax at age 14 on subsequent smoking (at follow-up in 1992 and

1994) was not significant. Elasticities declined over time for low income people indicating

that by age 39 the effect of taxes at age 14 had largely disappeared. Low income (< $12,000

median in 1979) elasticity was -0.65, p<0.10 (at age 14), -0.33 (at age 24), -0.01 (at age 34),

and 0.15 (at age 39). Cigarette tax increases at age 14 reduced smoking and had a positive

equity effect on young people in their 20’s.

It should be noted that in some models (i.e. effect of cigarette tax at age 14 on current

smoking), results presented for the low income subgroup include a control for ‘current’ tax

(taxes in the year of interview), whereas other models (i.e. effect of cigarette tax at age 14 on

late initiation, quitting) did not control for current tax in low-income subgroup. It is difficult

to see how an effect of tax at age 14 could be determined if there is no adjustment for tax at

other subsequent time points.

39

A US econometric analysis36 using repeated cross-sectional data, evaluated the impact of

prices, clean air regulations and youth access restrictions on youth (13 to 18 years) smoking

in the 1990’s. Price was the only significant determinant of smoking. Price was the most

important determinant of smoking by 16-18 year olds but not for younger teenagers.

Sensitivity to price suggested cross-elasticity between price and income: for 16 to 18 year

olds: sensitivity to prices increased for teenagers with less educated parents, however

sensitivity to smoking intensity increased for those with more educated parents. For 16 to 18

year olds, the elasticity of participation was -4.39 (p<0.05) for those whose parents were

high school dropouts or graduates and -0.24 for parents with some college education. For

smoking intensity this trend was reversed with elasticities of -0.40 for high school and -2.39

(p<0.05) for college education. There was no pattern for younger teenagers (<16 years),

although participation elasticity was positive and statistically significant for high school

educated parents (2.72, p<0.05).

A survey39 conducted between 2005 and 2006 on a random sample of 2455 university

students in South-Eastern France, investigated young smokers’ (mean age 19.5 years)

retrospective reactions to an increase in cigarette prices. Daily smokers with low educated

parents were less likely to report reacting to the price increase, daily smokers who had at

least one parent that completed high school were more prone to report reacting to higher

cigarette price (OR 2.5; 95% CI: 1.6 to 4.0 for cheaper smoking versus no reaction; and OR

2.1; 95% CI: 1.4 to 3.3 for smoking less versus no reaction; in multivariate analysis, p <

0.001 and p< 0.01, respectively). Students who reported difficulties in financing their studies

were significantly more likely to purchase cheaper cigarettes (OR 1.9; 95% CI: 1.0 to 3.7; p<

0.1). It should be noted that overall, 32% said that they did not react to price increase, the

survey was regional rather than national and the reactions to price increase are only relevant

to daily smokers who did not quit, all of which may which may limit study generalisability.

We can’t tell whether these reactions to a price increase impacted on quitting but there

appeared to be a negative equity impact on smoking less.

40

An Irish study37 used retrospective cohort data to investigate the role of tobacco taxes from

1960 to 1998, in starting and quitting smoking and how this differed by level of education.

The data was derived from a single cross-sectional survey on women’s knowledge,

understanding and awareness of lifetime health needs, but mainly focussed on hormone

replacement therapy as part of an unpublished MA thesis at the University College Dublin.

The sample consisted of just over 700 women, mean age was 35 years and mean age started

smoking was 19 years. The SES measure used was education level (‘primary cert’/’junior

cert’/’leaving cert’/’third level’).

Higher cigarette tax levels were associated with later initiation of smoking which differed by

education level. Taxes had the greatest positive effect in terms of delaying smoking initiation

for women with intermediate level education and weakest effect among women with the

lowest education. The results were tentative because of the potential for recall bias (going

back 40 years in some cases) and the results are specific to a sample of Irish women aged 48

years or younger. The measure of education level used in this study may not be

generalisable across time and to other countries. The SES subgroups were relatively small,

and during the study period cigarette tax was relatively low and there was increasing

awareness of the harms of smoking. Therefore study findings cannot be directly attributed to

the effects of increasing cigarette tax.

It should be noted that whilst data has been extracted for this review on smoking initiation

(because this is the outcome of relevance for youth), the study also reported smoking

cessation and showed inconsistent equity impact results for how tax effect differed by

education level, depending on the outcome measure (initiation and cessation). Cigarette

taxes had the greatest positive effect in terms of delaying smoking initiation for women with

intermediate level education and the weakest effect among women with the lowest

education. However cigarette taxes had the strongest effect on cessation among women with

the lowest education, and an equal impact on those with other levels of education.

Summary

The majority of evidence is from the US, and suggests there is variation in the evidence of

the equity impact of increases in cigarette tax or price on youth smoking behaviour and

variation in smoking behaviour amongst youth of different ages and different SES groups.

41

Two retrospective surveys showed contrasting results; one survey showed that low income

teenagers were more likely than more affluent teens to cut costs by cutting down smoking or

(less often) by switching to cheaper brands but were not more likely than more affluent

teenagers to consider quitting. However, only 53% of the teenagers who continued to smoke

denied having had any of the 3 potential reactions to the price increase. A regional survey of

French university student smokers showed that students with a lower SES were less likely to

have reacted to the cigarette price increase which included smoking less, however 32% of

students reported that they did not react to the price increase.

An Irish study showed that cigarette taxes were associated with later smoking initiation in

women with intermediate education but not for women with only a primary education.37

Two econometric studies showed contrasting results; one study showed that cigarette tax at

age 14 had a statistically significant negative effect on current smoking for low income

people but by age 39 years, the effect of taxes at age 14 had largely disappeared. In the other

study, the equity impact varied according to the age of the teenagers and there was no pattern

for younger teenagers. For older teenagers: sensitivity to prices increased for teenagers with

less educated parents, and sensitivity to smoking intensity increased for those teenagers with

more educated parents.

It does not appear that low income youth are consistently more responsive to tax/price

increases than high income youth groups: youth of lower SES are not more likely to stop

smoking when cigarette prices/taxes increase.

3.2.4 Controls on access to tobacco productsA total of five studies assessed the socio-economic impact of controls on access to tobacco

products. Three studies assessed the impact of legislation on age of sale of cigarettes. Two

single cross-sectional studies40;41 examined the impact of age-of-sale laws in the US on

retailer compliance and whether the impact differed by SES. One repeat cross-sectional

study examined the impact of UK legislation which increased the minimum age for the legal

purchase of cigarettes, and was set in secondary schools in England.42 A German study used

observational field data of new electronic locking devices on cigarette vending machines to

prevent underage purchasing of cigarettes in Cologne43. A prospective cohort study based in

the US, examined whether young, especially low SES females, are influenced by tobacco

control policies in terms of smoking initiation and transition.44

42

An English study examined whether there was any differential impact of UK legislation

which increased the minimum age for the legal purchase of cigarettes from 16 years to 18

years and which came into force in October 2007.42 The SES variable employed was

eligibility for free school meals (FSM) which is assessed on the basis of parental

employment status and income levels. Annual survey data was collected before and after the

legislation; from 2003 to 2008. There were baseline differences in age, gender and ethnicity

but these differences were controlled for in analyses.

Increasing the minimum age for purchase was associated with a significant reduction in

regular smoking among youth aged between 11 and 15 years (adjusted OR 0.67; 95% CI

0.55 to 0.81, p=0.0005). This effect was not significantly different in pupils eligible for FSM

compared with those who were not eligible (adjusted OR 1.29; 95% CI 0.95 to 1.76, p=0.10

for interaction term). Regular smoking was not significantly different in pupils eligible for

FSM compared with those that were not (adjusted OR 1.29; 95% CI 0.95 to 1.76, p=0.10).

The percentage of regular smokers who usually bought cigarettes from a vending machine

decreased significantly in the non-FSM but not in the FSM group. The percentage of regular

smokers who usually bought cigarettes from friends and relatives or from other people

increased significantly in the non-FSM but not the FSM group after the introduction of age

restriction. Regular smokers eligible for FSM were significantly more likely to be given

cigarettes by their parents in 2006 (p<0.001) but this was no longer the case in 2008

(p=0.42). The percentage of pupils who stated that they found it difficult to buy cigarettes

from a shop did not increase in those eligible for FSM (25.2% to 33.3%; p=0.21) but did

increase significantly in others (21.2% to 36.9%; p<0.01) between 2006 and 2008. The

percentage of regular smokers who were successful in buying cigarettes from a shop during

their latest attempt decreased significantly in the non-FSM but not the FSM group between

2006 and 2008. No differences in ease of purchase were found between pupils eligible for

FSM and those not before or after the legislation (2006: p=0.34, 2008: p=0.55).

It should be noted that although the response rate for schools was only 58% in 2008, the

sampling frame ensured that schools participating in the survey closely reflect the

composition of schools in England generally. However, the national smokefree legislation

and alcohol restrictions were also introduced during this time which may confound these

results.

43

The German Sources of Tobacco for Pupils (STOP) study43 compared the number of vending

machines and other commercial sources before and after new legislation which involved

electronic locking devices on vending machines to prevent underage (<16 years) purchasing

of cigarettes in Germany. Three geocoders made an inventory of commercial cigarette

sources in 2005, 2007 and 2009 and mapped using Geographic Information System to

produce a density of sources before and after the legislation. Cologne was selected as the

area of study because it had existing sociogeographical data, however the authors report data

to show that Cologne data appears comparable with Germany as a whole.

The number of commercial sources declined by 12% from 2005 to 2009, resulting mainly

from the removal of 44% of outdoor cigarette vending machines (indoor machines decreased

by 5%). The lower the income level in a district, the higher the availability of cigarettes

(Pearson’s r = .595; p = .009). Convenience cigarette sources reduced by only 0.9%, and

supermarket and drug stores increased by only 2.6%. The study did not report whether the

decline in commercial source by retail category (outdoor and indoor vending machines,

convenience stores, supermarkets and drug stores) varied by the income level of districts.

The same occurred for the alternative indicators such as youth unemployment (Pearson’s r

= .548; p = .019), the percentage of people receiving social welfare (Pearson’s r = .485; p

= .041), and the percentage of pupils attending low-qualifying schools (Pearson’s r = .473; p

= .048).

In 2005 as well as in 2009, there were significantly fewer commercial cigarette sources in

districts with above average SES than in districts with below average SES. This can be seen

in terms of absolute as well as relative numbers. The density of commercial cigarette sources

in 2005 in districts with above average SES was 3.20 per 1,000 inhabitants and 4.84 per

1,000 inhabitants in the districts with below average SES. In 2009, the numbers were 2.63

per 1,000 inhabitants and 4.44 per 1,000 inhabitants, respectively. The differences between

socially advantaged and disadvantaged districts appeared to be significant in both years

(2005: t(15) = 9.017, p < .001 and 2009: t(17) = 6.915, p < .001). This study showed that

electronic locking devices on vending machines to prevent underage (<16 years) purchasing

of cigarettes in Germany was not associated with a decrease in inequalities of access to

cigarettes, for youth.

44

A US study41 evaluated the relationship of point-of-sale tobacco advertising and

neighbourhood characteristics (including 150% below the poverty level) to underage sales of

tobacco. Study authors used three data sources: observations of the advertising environment

in stores; records of age-of-sale tobacco checks where an undercover minor working with

law enforcement attempted to purchase tobacco; and demographic data from the Year 2000

U.S. census. Analyses were conducted on 467 of 655 licensed tobacco vendors in Minnesota,

USA. Compliance failure was defined as the sale of tobacco to a youth, regardless of

whether the store clerk examined the minor’s ID.

The study did not find a significant association between store advertising characteristics or

neighbourhood poverty level and stores’ compliance check failure. Of a total of 467 stores,

48 failed the compliance check. Tobacco shops were most likely to fail compliance checks

(44%) and supermarkets were least likely to fail (3%). The poverty level of stores ‘block

group’ was not associated with compliance failure. Stores in ‘block groups’ with a greater

percentage of people living in poverty were not more likely to fail the compliance check.

The study sample was representative and the results are generalisable at a regional level.

Only vendors with a current license can sell tobacco in state of Minnesota but this is not the

case across all US states. Also stores who repeatedly violate youth access laws have their

license rescinded. The study authors report that compliance checks may not be a very valid

measure of commercial tobacco accessibility for minors.

Another US study40 examined contextual, community and retail characteristics associated

with youth access to tobacco through commercial sources. Data sources were access surveys

carried out by four buyers who were over 18 years of age (mean age 19 years) but who were

judged to appear younger by an independent panel. Purchase attempts were made at 997

tobacco outlets in 50 mid-sized California cities by a team of two buyers. At each outlet a

single buyer attempted to purchase a pack of Marlboro or Newport cigarettes (the most

popular cigarette brands among high school-aged students). If asked about their age they

stated that they were over 18 years old, and if asked for an age ID they indicated they had

none. If a sale was refused, the buyers left without attempting to pressure the clerk. The main

outcome measure was retailer compliance with underage tobacco sales laws.

45

Overall, the rate of retailer non-compliance with underage tobacco sales laws in the 997

selected outlets was 14.3%. Buyer’s actual age, being a male clerk and asking young buyers

about their age were each positively associated with successful cigarette purchases. Buyer’s

actual age and minimum age signs increased the likelihood that clerks requested

identification (ID). A greater percentage of residents (within each city) with at least a college

degree was associated with increased likelihood of non-compliance with underage tobacco

sales laws. A lower percentage of residents with at least a college degree was associated with

retailers asking for an ID. Higher cigarette prices of Marlboro but not Newport were

associated with higher median household income.

Although the study authors state that there were no significant differences between the

sampled and the un-sampled cities in relation to population size, ethnic diversity, household

size and median household incomes, there was no data reported to clarify the

representativeness of the study sample and therefore the generalisability of the study results.

A US national longitudinal study of adolescent health (Add Health) was a school based

survey of the health related behaviours of adolescents using follow-up in-home surveys.44

‘Add Health’ used state level tobacco policy on age of sale scores developed by the US

National Cancer Institute, evaluating 9 items for each state each year (statewide

enforcement, random inspections, graduated penalties, photo identification, free distribution,

minimum age, packaging, vending machines, and clerk intervention).

The analyses were restricted to female adolescents, and showed that stronger state level

tobacco policies were associated with lower likelihood of smoking initiation and adverse

transition among low SES women, although the effect sizes were small. The positive policy

effects for initiation were strongest for low SES females, whose odds ratio was 0.95 (0.98

for middle SES, 1.00 for high SES). For initiation, school level smoking rates did not vary

substantially across low, middle, and high SES groups (OR=1.01, 0.99 and 1.00,

respectively. For statewide enforcement, the odds ratios of initiation were significantly lower

for the low (0.89) and middle (0.91) SES female groups; on the other hand, the policy had no

effect on the high SES female group (OR=1.00). For random inspections the odds ratios of

initiation were significantly lower for low (0.88) and middle (0.90) SES female groups.

Photo identification had a significant positive effect on the low SES female group

(OR=0.85), but not on the middle SES female group (OR=0.95, NS) and on high SES

females (OR=1.10, NS). Other policies had a pattern similar to the significant ones.

46

It should be noted that this US cohort uses longitudinal data with a seven year gap in the data

used to assess transition from adolescence to young adulthood, and this gap may have

missed other important mediators.

Summary

Five studies of controls on access to tobacco products showed mixed results for equity

impact. Although four of the five studies focussed on age of sale legislation, the German

study of vending machines was unique, and in addition, the range of outcomes reported with

the studies varied.

Increasing the minimum age for the purchase of tobacco in England was associated with a

significant reduction in overall youth smoking and regular smoking was not significantly

different in pupils eligible for FSM compared with those that were not and so the legislation

was neutral with regard to equity. However smokefree legislation also came into force

during the time of this study and could have contributed to the reduction in overall youth

smoking. In addition there were significant differences in the percentages of adolescents

eligible for FSM compared to those not eligible for FSM in terms of higher rates of

accessing cigarettes from a variety of sources, which showed negative equity impact

differences.

New legislation which involved electronic locking devices on vending machines to prevent

underage purchasing of cigarettes in Germany has not been associated with a decrease in

inequalities of access to cigarettes, for youth. The supply density of cigarette vending

machines in Germany was greater in socially disadvantaged areas, both before and after new

legislation to prevent underage access; there were also greater decreases in the number of

vending machine sources in socially advantaged areas.

Two US studies reporting retailer compliance with age-of-sale laws showed inconsistent

results for SES. A US study41 evaluating the relationship of point-of-sale tobacco advertising

and neighbourhood characteristics to underage sales of tobacco did not find a significant

association between store advertising characteristics or poverty and stores’ compliance check

failure. A study of compliance with underage tobacco sales laws in California40 showed that

higher education was a significant predictor of underage tobacco sales and youth in

communities with higher educational levels may have easier access to cigarettes from

commercial sources.

47

A US national longitudinal study of adolescent health showed that stronger state level

tobacco policies on age of sale were associated with lower likelihood of smoking initiation

and adverse transition among low SES adolescent girls, although the effect sizes were small.

It is difficult to ascertain how access to tobacco translates into smoking prevalence and how

stricter enforcement of access laws would help to reduce the gap between low and high SES

in terms of smoking prevalence. Increasing age of sale and restricting youth access do not

appear to be widening the gap between high and low SES but the evidence is limited and

only two studies report smoking outcomes rather than supply outcomes.

3.2.5 School-based preventionFive RCTs assessed the socio-economic impact of school-based smoking prevention

programmes. Two interventions were drug prevention programmes which included elements

of smoking prevention.45;46 One RCT examined the effects of a school-based drug prevention

programme which included smoking prevention in school children aged 11 years in Florida,

US.45 One quasi-randomised trial46 in 53 public secondary schools in northern Germany

evaluated the effects of two validated life skills programmes: ‘Fit and Strong for Life’ and

‘Lions Quest’.

Three intervention studies focused on smoking prevention. One RCT investigated whether a

peer group pressure and social influence intervention reduced the percentage of adolescents

who start to smoke, in the Netherlands.17 The European Smoking Prevention Framework

(ESFA) study assessed the impact of a social influence school-based intervention with

parental and community involvement on smoking uptake amongst adolescents in six

European countries.19 The ‘A Stop Smoking in Schools Trial’ (ASSIST) assessed the

effectiveness of a peer-led intervention that aimed to prevent smoking uptake in secondary

schools in England and Wales.18 Two of the studies did not report socioeconomic impact on

initial analyses; however a paper by Mercken et al.20 was identified which performed

secondary analyses of the socioeconomic impact of these three intervention studies17-19 using

the SES variables reported within the original primary studies.

48

The secondary analysis included a review to identify ‘high-quality European intervention

studies with clear overall effects that could be selected for secondary analyses'. Included

intervention studies had to be published in the international scientific literature in English

language, since 1995 and conducted in Europe since 1990. This procedure resulted in the

inclusion of three school-based intervention studies. The three studies were reanalysed using

the definitions of variables as defined in the original studies. Multilevel modelling

techniques were used; models were estimated using the restricted iterative generalized least

squares (RIGLS) estimation procedure combined with first-order penalized quasi-likelihood

within MLWin 2.10 beta. The multilevel model was tested separately for adolescents in each

of the categories of the included SES indicators.

It is unclear how representative all five study samples were of the respective study

populations. The groups in three of the studies17-19 had comparable characteristics at baseline.

Attrition rates were acceptable for three studies18;45;46 but relatively high for the other two

studies17;19. The ASSIST study was the only study to biochemically validate measures of self-

reported smoking, and scored highest for quality.18 It is likely that the observed effects of

each of the five interventions were attributable to the interventions.

Two studies evaluated school-based programmes which included elements of smoking

prevention: one based in the US and one in Germany.46 One RCT examined the effects of a

school-based drug prevention programme which included smoking prevention in school

children.45 The study was published as a paper presented at the Annual Conference of the

American Educational Research Association and assesses the impact of a school-based drug

prevention programme ‘Too Good for Drugs II’ (TGFD II) on student’s behaviours and risk

and protective factors. Students in six middle schools in Florida, US were randomised to 9

lesson units (40 minutes each) taught by a trained classroom teacher or TGFD II instructor;

including social and emotional competencies, reducing risk factors and building protective

factors; emphasising cooperative learning activities, role-play and skills building methods.

Students were followed-up 20 weeks after the 9 week intervention. The school-based

curriculum also involved community partners and parents; and the theoretical basis included

Social Learning Theory, Problem Behaviour Theory and Social Development Theory.

49

At the end of the intervention, 8% (48/588) of students in the intervention group indicated

greater likelihood of actual tobacco use compared with 12% (45/375) of students in the

control group, and this difference was statistically significant. There was no statistically

significant difference between the groups at 20 weeks follow-up. The overall findings of the

comparison of change scores for treatment students indicated the programme was similarly

effective in impacting on students risk and protective factors regardless of economic status

(perception of peer resistance skills; positive attitudes toward non-drug use, perceptions of

peer normative substance use, perceptions of peer disapproval of substance use, association

with prosocial peers, perceptions of locus of control self-efficacy).

A significant interaction effect for treatment students was seen between level of risk and

protective factor scores and SES (measured by free/reduced lunch status) at the end of

intervention and 20-week follow-up. Significant trends appeared between low and high SES

in the areas of ‘perceived peer norms’ and ‘perceived peer approval of substance use’ at the

end of the intervention and in addition with ‘association with prosocial peers’ at 20-week

follow-up. The direction of the effect by SES is not reported.

One RCT investigated whether a peer group pressure and social influence intervention

reduced the percentage of adolescents who start to smoke, in 26 junior secondary education

schools in the Netherlands.17 The intervention consisted of three lessons on knowledge,

attitudes, and social influence, followed by a class agreement not to start or to stop smoking

for five months and a class based competition.

At five months 9.6% of the non-smokers at baseline had started to smoke in the intervention

group, whereas 14.2% started to smoke in the control group (N = 1388, OR = 0.61, 95% CI

= 0.41–0.90). After 1-year follow-up, the effect was no longer significant. At 5 months,

smoking behaviour was significantly lower in adolescents who indicated that their parents

had mid to high completed education (OR = 0.35, 95% CI = 0.13–0.95). The intervention did

not result in smoking fewer cigarettes among adolescents who indicated that their parents

had lower education (OR = 0.80, 95% CI = 0.37–1.72). The additional analyses stratified by

gender and SES showed that the intervention was only effective at 5 months follow-up

among boys with higher parental educational levels (OR = 0.24, 95% CI = 0.07–0.79). All

significant intervention effects disappeared at 12 months follow-up.17

50

The ESFA study assessed the impact of a social influence school-based intervention with

parental and community involvement on smoking uptake amongst adolescents in six

European countries.19 In Finland, Denmark, UK and Portugal schools or regions were

randomly assigned whereas in Spain and The Netherlands the study design was quasi-

randomised. Since the strongest and significant long-term effects after 24 and 30 months

were found in the Portuguese sample, only data of the ESFA study in Portugal were

reanalysed on the impact by SES and so only results for Portugal are discussed within this

review.

The Portuguese intervention consisted of lessons on the effects of tobacco, reasons for (not)

smoking, social influence processes, refusal skills and decision making and a smoke-free

competition. Due to the fact that peer-led programmes were uncommon in the ESFA

countries, programmes were teacher-led. Teachers received 48 hours of training, a manual

and smoking cessation material. Schools received the ESFA no-smoking policy manual and

non-smoking posters. For the parents, information was offered on how to discuss non-

smoking with their adolescents. Pharmacists furthermore offered cessation courses for 150

parents. At the community level, the Portuguese Health Minister and mayor of the

community introduced the ESFA study on the national no smoking day.19

51

At 30 months, 41.8% of the never smokers at baseline had started to smoke in the

intervention group, compared to 53.8% of the never smokers at baseline in the control group

(N = 1304, OR = 0.62, 95% CI = 0.48–0.80). The results were mixed depending on the SES

indicator used (mother/father and full-time/not full-time jobs were not included as a measure

of SES in our review). The intervention was significant in reducing smoking uptake among

adolescents who indicated having no to only a low amount of spending money (OR = 0.62,

95% CI = 0.46–0.84). This statistically significant effect was not seen among adolescents

reporting receiving mid to high amounts of spending money (OR = 0.57, 95% CI = 0.32–

1.03). Although the actual odds ratio is smaller for the ‘mid to high’ spending money

subgroup compared with the ‘none to low’ spending money subgroup, the lack of

significance here is due to the wider confidence intervals, which are explained by the

relatively small numbers in the subgroup with ‘mid to high’ spending money (n=182).

Additional analyses stratified by gender and SES showed that the intervention was mostly

effective among girls.19 Pocket money was used as a proxy measure and there may not be a

strong association between adolescents’ pocket money and household income. As Mercken

et al.20 state; those adolescents with less pocket money may well have parents with higher

levels of education or income.

The ASSIST study assessed the effectiveness of a peer-led intervention that aimed to prevent

smoking uptake in secondary schools in S.W. England and Wales.18 Influential students were

trained by external professionals to act as peer supporters during informal interactions

outside the classroom to encourage peers not to smoke. During the 10-week intervention

period, peer supporters undertook informal conversations about smoking with their peers

when travelling to and from school, in breaks, at lunchtime and after school in their free

time.

At 1-year follow-up, the OR of being a smoker in intervention compared with control group

was 0·77 (95% CI 0·59–0·99). At 2-year follow-up, the corresponding OR of 0·85 (0·72–

1·01) was not significant (p=0·067). For the high-risk group (occasional, experimental, or

ex-smokers at baseline), the OR at 1-year follow-up was 0·75 (0·56–0·99) and at 2-year

follow-up was 0·85 (0·70–1·02). In a three-tier multi-level model using data from all three

follow-ups the odds of being a smoker in the intervention group compared with the control

group was 0.78 (95% CI = 0.64–0.96).18 The original primary study paper found no evidence

of a differential effect by FSM entitlement (0.99 (95% CI =0.65-1.51)).

52

The secondary data analysis20 combined data from the three follow-up periods and conducted

a multi-level analysis using three measures of SES: FAS, FSM and school located in the

Valleys (which are areas of deprivation). No significant main effect of the intervention was

found for FAS or FSM entitlement, though a trend was visible for FSM. The intervention

was significant among schools located in the valleys but not in schools in other locations.

Additional analyses showed that in Valley schools the intervention was also effective among

those with low FAS score, and a gender analysis showed that the intervention was mostly

effective among lower SES girls.

Summary

The overall findings from the five school-based studies are mixed in terms of the impact by

SES, the results also varied by the type of SES used to measure effect, and over time

(shorter-term benefit appeared to attenuate over time).

The findings from a substance abuse prevention programme set in schools were equally

effective for students regardless of SES, however the study did not ask about current

smoking behaviour and the outcome was intention to smoke in the next 12 months rather

than actual smoking behaviour.45 The results of this prevention programme relate to scores

for substance use which includes (but is not limited to) tobacco use and so this limits study

findings. A German study46 of two life skills programmes had a positive effect on smoking

prevention regardless of SES; with socially disadvantaged children benefitting equally

(neutral equity impact).

The Netherlands study had a significant effect among higher SES adolescents only and in the

short-term only, and appeared to widen inequalities (negative equity impact).17 The ESFA

study showed mixed results depending on the specific SES indicator used; when using

spending money as a SES indicator, the intervention did appear to decrease inequalities in

smoking.19 However, the amount of spending money which an adolescent has may not be

strongly associated with household income. For example, in Scotland low SES adolescents

have higher levels of disposable income than higher SES adolescents.47 ESFA interventions

differed between countries and Portugal received the most intensive teacher training; so

results may only be generalisable to that type of intervention in that country. Process

evaluation of ESFA included self-report of exposure to each element of the intervention and

showed it was reasonably likely that the observed effects were attributable to the school-

based elements of the intervention rather than outside school elements.

53

The most promising findings in terms of equity impact were for ASSIST which used a social

network approach in which adolescents delivered the intervention. While this intervention

also showed mixed results depending on the specific SES indicator used, it was effective at

one year and most effective for adolescents in deprived areas, particularly among low SES

girls (positive equity impact). However, the beneficial effects of the intervention seemed to

attenuate over time.18

3.2.6 Multiple policy interventionsThree studies assessed the socio-economic impact of multiple policy interventions: two

repeat cross-sectional studies48;49 and a prospective cohort study.50 One repeat cross-sectional

study49 was set in Australia and examined the impact of tobacco control policy on smoking

prevalence. The other repeat cross-sectional study48 assessed the impact of the 1976

Tobacco Control Act (TCA) on smoking initiation across socioeconomic groups of Finnish

youth. A prospective cohort study described the association between smoking intolerance in

schools, restaurants and corner shops near secondary schools, and the initiation of smoking

in a convenience sample of adolescents (mean age 13 years) in Montreal, Canada.50

The cohort study50 used a convenience sample and it was unclear whether the study sample

was representative of the study population or whether the study results are generalisable. The

Australian study49 reported changing retention rates which meant that the characteristics of

the student sample in school years 11 and 12 were likely to differ systematically across the

survey years, which could have affected the prevalence rates (instead of, or as well as,

tobacco control policy). Both the Australian and Finnish studies were large population

surveys with results that are likely to be generalisable at the national level.

54

An Australian national survey49 examined whether SES was associated with changes in

smoking prevalence among adolescents during three phases of tobacco control activity: low

tobacco-control funding (1992-1996) and high tobacco-control activity (1984-1991 and

1997-2005) which included smoking restrictions and increased tax. Random samples of

students aged 12 to 17 years from each Australian state and territory and three main

education sectors, completed anonymous surveys of cigarette use as part of a larger survey

assessing the use of alcohol and illicit drugs between 1987 and 2005. There was a significant

and substantial reduction in the likelihood of smoking among all SES groups for older (16-

17 years) and younger students (12-15 years) between 1987 and 2005. Overall, for younger

students (12-15) the reductions differed by SES (interactions p <0.01), with reductions in all

smoking behaviours, greater for students from higher SES groups. Among older students

(16-17), only the reductions in committed smoking (cigarette use on at least three of the

previous seven days) differed across SES groups (interaction p < 0.01), and again reductions

were greater among students from higher SES groups.

Between 1990 and 1996 the proportion of younger and older students involved with smoking

increased significantly. Among younger students, the increase in monthly and weekly

smoking was greater among lower SES students. Between 1996 and 2005 the prevalence of

monthly and weekly smoking decreased significantly among both younger and older

students, and these decreases were consistent across SES groups. The magnitude of the

decreases in smoking prevalence between 1996 and 2005 did not differ significantly between

SES groups for most indicators of smoking behaviour. For committed smoking, the

interaction between year and SES was of borderline significance for students from both age

groups, suggesting that the decrease may not be consistent across SES groups. It should be

noted that co-operation rates of the schools declined over time from 85% in 1987 to 63% in

2005 and the changing prevalence estimates might be the result of different survey samples.

55

A Finnish study48 assessed the impact of the 1976 TCA on smoking initiation across

socioeconomic groups. The TCA prohibited smoking in most public places, including public

transport; and the sale of tobacco products to those below 16 years of age; and required

obligatory health warnings on packages. The study used annual cross-sectional postal survey

data from 1978 to 2002 to assess the impact of the TCA on smoking prevalence (defined as

ever smoked daily for at least a year). The study authors defined the critical age range for

smoking initiation as 13 to 20 years. Most of the analyses were focussed on the three largest

socioeconomic groups (upper white collar workers, lower white collar workers, blue collar

workers manual workers).

Amongst men the secular cohort trend showed a decline in smoking only in upper white

collar workers before the TCA (stable for lower white collar and blue collar) and this trend

remained unchanged after the TCA, with no difference for the interaction between SES and

trend. Among women the secular cohort trend was increasing in each SES group before the

TCA and was reversed after the TCA, evenly across SES groups. For women, the general

cohort trend after the TCA differed from the secular cohort trend before the TCA, and this

differed by SES. In cohorts reaching the smoking initiation age after the TCA, the

prevalence of ever smoking remained relatively stable among white collar female workers

but tended to decline among blue collar female workers (OR 0.88; 95% CI: 0.72 to 1.02), in

contrast to the sharply increasing trend in older cohorts.

In terms of study validity, the average response rate during 1978 to 2002 was 70% among

men and 79% among women and the response rate declined during this period, in both

genders and all age groups. The decline was faster among men than women, and in younger

than older age groups, which may have biased the study results. Other tobacco control

policies came into force during the study period which could have influenced the study

results and explain some of the variability in smoking initiation by SES: the 1976 TCA was

supplemented by a total tobacco advertising ban in 1978, and the environmental tobacco

smoke amendment (of the TCA) in 1995. In addition, tobacco prices rose substantially (real

price increase 27%) in 1975–1976.

56

A Canadian study described the association between smoking intolerance (the extent to

which smoking is socially unacceptable) in schools, restaurants and corner stores near

schools and the initiation of smoking in adolescents. ‘The Natural History of Nicotine

Dependence in Teens Study’50 involved completion of questionnaires administered in the

classroom, every 3 months from 1999 to 2005 by students average age 13 years, in seven

English and three French language secondary public schools in Canada. The study used a

convenience sample which produced a 55% student response rate.

Students in smoking-intolerant schools (access and restrictions) were less likely to initiate

smoking than students in smoking-tolerant schools (Hazard ratio [HR] = 0.83; 95% CI: 0.68,

1.01). Students attending schools located in neighbourhoods with smoking-intolerant

restaurants were less likely to initiate smoking (HR 0.85; 95% CI: 0.68 to 1.07). There was

no association between corner store smoking intolerance and smoking initiation. The HR’s

for cigarette use initiation for low SES schools were not significant. However, there was a

25% loss to follow-up of students over the five years and these students were more likely to

attend a low SES school, which may have impacted on the results.

Summary

The Australian survey49 showed that the magnitude of the decreases in smoking prevalence

between 1996 and 2005 did not differ significantly between SES groups for most indicators

of smoking behaviour, but there may be differences between younger and older youth.

However, there appeared to be an association between level of tobacco control funding and

smoking prevalence. There was also some evidence that low tobacco control funding had a

negative equity effect on smoking prevalence among 12-15 year olds but not older students.

The Finnish TCA48 was associated with a reduction in smoking initiation across all SES

groups. Among men, the 1976 TCA appears to have had the greatest impact on male white

collar employees. Among women, the apparent effect was very pronounced in all

socioeconomic groups and among blue collar female workers the cohort trend tended to

decline.

A convenience sample of pupils in Canada50 showed that cigarette use initiation was

associated with levels of smoking intolerance in schools and communities but that this did

not differ by SES. But there was evidence of response bias by SES which may have

impacted on the results.

57

3.3 Impact of individual level cessation services and support on smoking inequalities in youth

There were only two individual cessation support interventions identified for youth which

assessed smoking outcomes by SES, one set in New Zealand and one in USA: both of which

used text-messaging as the mode of intervention.

The New Zealand study51 aimed to determine the effectiveness of a mobile phone text

messaging smoking cessation programme which provided advice, support and distraction for

smokers who owned a mobile phone and who wanted to quit smoking. Participants were

aged 16 years and over, with a mean age of 25 years. The intervention included five free

personalised text messages per day for one week prior to a negotiated quit date and for four

weeks after the quit date. The control participants received one free month of text messaging

if they participated until 26 weeks. A total of 1,705 smokers were recruited from adverts on

websites, media, email and text messaging mailing lists; and posters at tertiary education

institutions.

The RR of not smoking in the past week was 2.20 (95% CI 1.79 to 2.70) at 6 weeks, 1.55

(95% CI 1.30 to 1.84) at 12 weeks and 1.07 (95% CI 0.91 to 1.26) at 26 weeks (when all

participants with missing status were assumed to be smoking). The RR of not smoking (in

the past week) at six weeks by income level was presented as a forest plot and showed no

difference in effect by income level; all income levels showed significant positive effects of

the intervention.

Biochemically verified abstinence was only performed on a random selection of participants

and showed over-reporting of quit rates but this over reporting was not different between the

intervention and control group. The quit rate at 6 weeks was 28.1% in the intervention group

compared with 12.8% in the control group. Assuming the rate of true quitters was the same

as in the sample assessed for cotinine levels, then the quit rate at 6 weeks was 13.9% in the

intervention group compared with 6.2% in the control group and the absolute difference in

quit rates at 6 weeks is reduced to 7.7% from 15.3%.

58

Only 74% (n=1265) of participants were followed-up at 26 weeks and the attrition rate

differed significantly between the groups at 12 weeks and at 26 weeks (69% in intervention

group vs 79% in control group at 26 weeks). This meant there was some uncertainty about

between group differences at 26 weeks. In addition reported quit rates increased amongst the

control group from 13% at 6 weeks to 24% at 26 weeks, however this would have led to an

underestimation of treatment effects and all methods of data analyses showed a significant

difference in quit rates in favour of the intervention.

The US study52 targeted a diverse sample of motivated daily smokers aged 18 to 25 years,

owning their own mobile phone and ‘seriously thinking about quitting in the next 30 days’.

Two hundred and eleven young adults were randomised from 585 eligible participants and

the final sample included 164 participants: 101 in the intervention group and 63 in the

control group; mean age 22 years, 56% male, with 43% reporting an annual household

income of less than $15,000.

The 6-week text-messaging intervention was tailored to each young adult smoker based on

their quitting stage. Intervention participants also had access to a ‘Text Buddy’ similar to

that used in the New Zealand study51 and ‘Text Crave’ (immediate, on-demand messages

aimed at helping the participant through a craving); and a project website

(StopMySmoking.com). The control group received a similar number of text messages, but

message content was aimed at improving sleep and exercise habits within the context of how

it would help the participant quit smoking. Control group messages were not tailored nor

were Text Buddy and Text Crave components available.

Intervention participants were significantly more likely to have quit at 4 weeks post quit

(39%) than those in the control group (21%; adjusted odds ratio [aOR] = 3.33, 95% CI:

1.48, 7.45); and this was also the case for 7-day point prevalence (44% vs. 27%; aOR = 2.55,

95% CI: 1.22, 5.30). However the impact was not sustained, and 40% of the intervention

participants had a quit status verified by a ‘significant other’ compared with 30% in the

control arm at 3 months post-quit, which was not statistically significant (OR = 1.62, 95%

CI:0.82, 3.21). Cessation rates among intervention participants were stable between 4 weeks

and 12 weeks, but increased among control participants

59

The intervention appeared to be more effective in young adults not currently enrolled in

higher education settings (45% intervention vs. 26% control had quit at 3 months, p = .07;

aOR of verified quit at 3 months = 2.7, 95% CI: 1.0 to 7.4). The US study52 was a feasibility

study with a relatively small sample size so it was not sufficiently powered particularly to

detect differences in subgroup results. Eight participants were manually assigned to

treatment groups (rather than randomly) due to an imbalance within study subgroups.

Summary

Two studies of text messaging smoking cessation interventions were included, one set in the

USA and one in New Zealand. Participants in both studies were mobile phone owners in

their late teens to early twenties, who were motivated to quit smoking. The New Zealand

study control participants received a passive control (one month free text messaging) and US

control participants received a text-messaging service that was not tailored (intervention

participants received a tailored text-messaging service).

The New Zealand study showed personalised mobile phone text messaging support could

double quit rates at 6 weeks amongst young adult smokers who wanted to quit, irrespective

of income level. The effect was still significant at 12 weeks but not at 26 weeks, in addition

there was an increase in quit rates amongst the control group and significantly more

intervention participants were lost to follow-up at 26 weeks.

The US study of a tailored text-messaging intervention compared to a non-tailored text-

messaging intervention, showed a significant increase in quit rates in intervention group

participants compared with control group participants at 6 weeks that were not sustained at

12 weeks. Quit rates increased in control group participants. However youth not enrolled in

higher education (i.e. lower SES) appeared to benefit from the tailored text messaging

intervention with significantly positive quit rates at 12 weeks compared to youth enrolled in

higher education.

The New Zealand study showed a short-term neutral equity effect and the US study showed

a short-term positive equity effect. Quit rates increased in the control groups in both studies.

It is unclear how representative either study samples were of each study population, however

both studies cut across all settings and all locations.

60

4 DISCUSSIONOnly one review, the CRD review, has previously assessed the equity impact of tobacco

control policies on youth smoking. No intervention, including restrictions on smoking in

schools and restrictions on sales to minors, provided any evidence about possible differential

effects by parental income, occupation or educational level for the youth population. The

review presented in this report has systematically assessed the available evidence on the

impact of population- and individual-level tobacco control interventions on socioeconomic

inequalities in youth smoking. It identified 31 studies which have evaluated the impact of

population level prevention policies/interventions and two individual level cessation support

interventions, on smoking in young people by SES, measured by income, occupation or

education. Before presenting the main review findings it is important to consider the

strengths and limitations of both the review and the available evidence.

4.1 Strengths and limitations of the review

Considerable attempts were made to include published and ‘in press’ studies. However, it is

possible that some relevant studies might have been missed which had not been published in

the peer reviewed literature and/or which were not published in English. It is also possible

that papers which undertook analyses by SES were not included because these analyses were

not mentioned in their abstract. However, this review goes beyond the previous CRD review

in including all types of youth interventions (prevention and cessation, population and

individual levels) and also searching for non-tobacco control interventions and polices (eg

education, social policy) which assessed any smoking-related equity impacts. It also

included ‘in press’ articles from four key journals and asked European tobacco control

experts to provide any other relevant peer reviewed articles (non-English language) or grey

literature. We also developed a modified quality assessment tool which was designed to

enable us to assess the quality of the diverse range of types of interventions and study

designs encompassed in the included studies.

4.2 Strengths and limitations of the available evidence

There are major limitations in the available evidence, most importantly the very small

number of studies which have considered the equity impact of tobacco control interventions

aimed at young people. In addition, there was a lack of consistency on the reported outcome

measures and length of follow-up. There was also considerable variation in the quality of the

studies (Section 7.6 Appendix G). Several of the studies were pilot or feasibility studies 61

and/or involved small numbers of participants. Thus, their findings may not be replicable.

For several important areas of youth tobacco control eg social marketing, multifaceted

community programmes, mass media approaches using social media, combating

smuggling/reducing the black market, smokefree homes interventions and most forms of

cessation support, we found no evidence on equity impact. Nearly half the studies were from

North America (US and Canada) and a third from the UK, which raises concerns about their

generalisability and potential transferability to, or relevance for, countries in Europe which

have different social and cultural contexts and/or different levels of tobacco control. Finally,

a range of indicators of SES was used in papers (e.g. education level, income, area

deprivation, and other indicators) which made comparisons between studies difficult. Most

studies used education income level as a measure of SES but levels of educational attainment

and income vary between countries and generations.

4.3 Main findings and conclusions

Relatively few intervention studies have assessed their impact on socioeconomic inequalities

in youth smoking or other smoking-related outcomes (eg exposure to second-hand smoke).

Out of the original 12, 605 identified papers (which also included papers focusing on adults)

only 33 studies met the inclusion criteria and were included in the review and none were

from outwith tobacco control (Figure 1 and Table 2). The literature was international, with

nearly half of the studies being carried out in North America. Studies also included the UK,

the Netherlands, France, Spain, Finland, Israel, New Zealand and Australia.

Of the 33 studies included in the review 31 were population level tobacco control

policies/interventions and 2 were individual level cessation support interventions. The types

of policies/intervention included were: smoking restrictions in cars, schools, workplaces and

other public places (9); controls on the advertising, promotion and marketing of tobacco (3);

mass media campaigns (1); increases in price/tax of tobacco products (6); controls on access

to tobacco products (5); school-based prevention programmes (5); multiple policy

interventions (3) and individual cessation support (2). (One study was included in two types

of policies/intervention category).

62

4.3.1 Positive equity impactOnly six of the 31 population-level studies showed the potential to produce a positive equity

impact i.e. to reduce inequalities in youth smoking. These ‘positive’ studies included three

US studies of increasing the price/tax of tobacco products,35;36;38 two US studies on age-of-

sales laws,40;44 and UK one school-based smoking prevention programme.18 Three US studies

of cigarette price/tax increases35;36;38 demonstrated a positive effect on low SES youth of

increasing price/tax to reduce smoking. A US prospective cohort study44 showed that

stronger state level tobacco policies on age of sale were associated with a lower likelihood of

smoking initiation and adverse transition among low SES adolescent girls, although the

effect sizes were small. A study of compliance with underage tobacco sales laws in

California40 showed that higher education was a significant predictor of underage tobacco

sales and youth in communities with higher educational levels may have easier access to

cigarettes from commercial sources. One school-based smoking prevention study (ASSIST),

using a peer-delivered intervention through social networks, appeared to reduce smoking

inequalities in school-children in England and Wales. However, results were mixed

depending on the specific SES indicator used.

4.3.2 Equity impact by type of tobacco control policy/intervention

Assessing the overall equity impact of different types of interventions/policies was

complicated by studies having different outcome measures and length of follow-up. In some

studies different outcomes varied in equity impact or the same SES measure and outcome

varied by gender or by setting. For example, one school-based prevention programme

showed a positive effect only in high SES girls and had the potential to widen inequalities.

Which specific measures of SES were used appeared to influence the results across all types

of policy interventions. The equity impact could also vary depending on the timing of the

outcome measurement. For example, two of the school-based prevention programmes found

that the effect varied across time points; with beneficial intervention effects attenuating over

time. Similarly both cessation interventions using text-messaging showed a significant

beneficial effect that was not sustained in the longer-term. Thus, the summary of the equity

impact of policies/interventions was derived ‘on balance’ (Appendices H and I).

63

Overall there was no consistent equity effect for each type of tobacco control

policy/intervention. Most interventions had, on balance, either a negative (11) or neutral (15)

equity impact. One had a mixed impact. However, it should be borne in mind that studies of

policies associated with a neutral equity effect indicate that these policies have benefits for

youth across SES groups. For example, both the English and Scottish national smokefree

legislation were associated with significant reductions in admissions for asthma across all

SES subgroups.

Smoking restrictions in cars, schools, workplaces and other public places- None of

the nine studies showed a positive equity impact. Four had a negative equity impact,

four had a neutral impact, and one had both negative and neutral impacts. The studies

indicate that the equity impact of comprehensive smoking legislation in public places

may differ depending on the pre-ban level of exposure and the balance between

sources of exposure i.e. public places versus the home. While comprehensive

smoking restrictions can reduce overall SHS exposure across all SES groups of

children. Changes in smoking restrictions in homes and cars following UK smokefree

legislation did not appear to be patterned by SES in pooled analyses, however

smoking in homes and cars remains more prevalent amongst children from low SES

families. Evidence shows that there is significant variation by SES in levels of

exposure prior to smokefree legislation with higher levels of exposure in lower SES.

Whether exposure is measured in relative or absolute terms appears to influence the

equity impact results. However, there is some evidence that smokefree legislation can

also have a neutral equity impact in terms of increasing voluntary smoking

restrictions in cars. The evidence also suggests that where there are no

comprehensive smoking restrictions in schools or where there is variable compliance

with voluntary bans; inequity in smoking will continue.

Controls on the advertising, promotion and marketing of tobacco- Two of the studies

had a negative equity impact and one had a neutral impact. The three studies were

very different with one indicating that tobacco companies marketing expenditure may

be associated with an increase in smoking initiation especially in young people with

lower levels of education. Another study found that despite an FDA buffer zone

policy to protect children from tobacco advertising, tobacco advertising was

specifically targeted at adolescents of low SES inside school buffer zones and that

this has the potential to increase inequality in smoking amongst youth. This would

64

indicate that banning all tobacco advertising would be particularly beneficial for low

SES children. There was also some tentative support from one study that introducing

plain packaging would have a similar impact across all SES groups.

Mass media campaigns- only one study assessed the equity impact of a mass media

campaign, the Truth campaign. The overall equity impact was difficult to assess but

there was a neutral equity impact in terms of receptivity.

Increases in price/tax of tobacco products - the majority of evidence on price/tax is

from the US, and suggests that there is variation in the equity impact of increases in

cigarette tax or price on youth smoking behaviour and variation in smoking

behaviour amongst youth of different ages and different SES groups. Low income

youth were not consistently more responsive to tax/price increases compared with

high income youth: youth of lower SES were not more likely to stop smoking when

cigarette prices/taxes increased. The inconsistency within the evidence could reflect a

true effect or measurement errors such as failure to capture youth behavioural

reactions in retrospective recall studies.

Controls on access to tobacco products- Reducing access to cigarettes through

increasing the minimum age of sale, including vending machines sales, may impact

on youth sales but the inconsistent evidence from the UK and US studies make it

difficult to draw conclusions about whether they also reduce youth smoking

inequalities.

School-based prevention programmes-only one study (ASSIST) had promising

findings in terms of a positive equity impact. The other studies findings were

inconsistent, varied by type of SES measure used and attenuated over time.

Multiple policy interventions- these were three very different studies (two national

and one at community level) in three different countries looking at different types of

policies which makes it difficult to draw any conclusions about the equity impact of

multiple policy interventions.

Individual cessation support- only two studies were included which evaluated

individual level smoking cessation support for youth. Both of these interventions

used text messaging. The US study showed a short-term (12 weeks) neutral equity

impact and the New Zealand study showed a short-term (12 weeks) positive equity

impact but this was not significant at 26 weeks. Their equity impacts should be

viewed with caution given that the representativeness of both study samples were

unclear: both sample participants were motivated young adults who owned a mobile

65

phone. However text messaging interventions have the potential to reach large

numbers of young smokers.

5 CONCLUSIONSThirty-three studies were included which evaluated the effect of policies and interventions to

prevent or stop youth smoking by SES. Only six of the 31 population level studies showed

the potential to reduce inequalities in youth smoking; including three on increasing the

price/tax of cigarettes, enforcing strong policies on age-of-sale, and one school-based

prevention study (ASSIST). There were only two individual level cessation support

interventions identified for youth which assessed smoking outcomes by SES. Both cessation

studies used text messaging. One showed a neutral equity impact and the other showed a

positive equity impact. There was variation in the equity impact of each type of tobacco

control policy/intervention.

The limited nature and extent of the evidence base considerably constrains what conclusions

can be drawn about which types of tobacco control polices/interventions are likely to reduce

inequalities in youth smoking. Very few studies have assessed the equity impact of policies

and interventions on smoking prevention or cessation in youth. There is therefore little

available evidence to inform tobacco control policy and interventions that are aimed at

reducing socioeconomic inequalities in youth smoking. There is a need to strengthen the

evidence base for the equity impact of tobacco control interventions aimed at young people.

The review provides very little evidence to suggest that any specific policies would be able

to reduce inequalities in smoking initiation.

66

6 REFERENCES(1) Lopez AD, Collishaw NE, Piha T. A descriptive model of the cigarette epidemic in

developed countries. Tobacco Control 1994; 3: 242-247.

(2) Amos A, Greaves L, Nichter M, Bloch M. Women and tobacco: A call for including gender in tobacco control research, policy and practice. Tobacco Control 2012; 21:236-243.

(3) Hiscock R, Bauld L, Amos A, Fidler JA, Munafo M. Socioeconomic status and smoking: a review. Annals of the New York Academy of Sciences 2012; 1248:107-123.

(4) Hiscock R, Bauld L, Amos A, Platt S. Smoking and socioeconomic status in England: the rise of the never smoker and the disadvantaged smoker. Journal of Public Health 2012; 34:390-396.

(5) Amos A, Angus K, Fidler J, Hastings G. A Review of Young People and Smoking in England. 2009. York, Public Health Research Consortium.

(6) U.S.Department of Health and Human Services. Preventing Tobacco Use Among Youth and Young Adults: A Report of the Surgeon General. 2012. Atlanta,GA, U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health.

(7) Currie C, Nic GS, Godeau E, Roberts C, Smith R, Currie D et al. Inequalities in Young People's Health- HBSC International Report from the 2005/2006 Survey. Health Policy for Children and Adolescents, No. 5. 2008. Copenhagan, Denmark, WHO Regional Office for Europe.

(8) Joosens L, Raw M. The Tobacco Control Scale: a new scale to measure country activity. Tobacco Control 2006; 15:247-253.

(9) World Health Organisation. MPOWER: Six Policies to Reverse the Tobacco Epidemic.WHO report on the global tobacco epidemic, 2008. 2008. Geneva, World Health Organisation.

(10) Conference of the Parties to the WHO FCTC. WHO Framework Convention on Tobacco Control. 2003. Geneva, World Health Organisation.

(11) SILNE consortium. Project: SILNE – Tackling socio-economic inequalities in smoking: learning from natural experiments by time trend analyses and cross-national comparisons. 2012.

(12) Fayter D, Main C, Misso K, Ogilvie D, Petticrew M, Sowden A et al. Population tobacco control interventions and their effects on social inequalities in smoking. York: Centre for Reviews and Dissemination 2008;322.

(13) Public Health Research Consortium. A9-10R: Tobacco control, inequalities in health and action at the local level in England. 2011.

67

(14) Joosens L, Raw M. The Tobacco Control Scale 2010 in Europe. 2011. Chaussée de Louvain 479, B-1030 Brussels, Belgium, Association of the European Cancer Leagues.

(15) Thomas H. Quality assessment tool for quantitative studies. 2003. Hamilton, Ontario, Effective Public Health Practice Project.

(16) Thomas S, Fayter D, Misso K, Ogilvie D, Petticrew M, Sowden A et al. Population tobacco control interventions and their effects on social inequalities in smoking: systematic review. Tobacco Control 2008; 17(4):230-237.

(17) Crone MR, Reijneveld SA, Willemsen MC, van Leerdam FJ, Spruijt RD, Sing RA. Prevention of smoking in adolescents with lower education: a school based intervention study. J Epidemiol Community Health 2003; 57(9):675-680.

(18) Campbell R, Starkey F, Holliday J, Audrey S, Bloor M, Parry-Langdon N et al. An informal school-based peer-led intervention for smoking prevention in adolescence (ASSIST): a cluster randomised trial. Lancet 2008; 371:1595-1602.

(19) de Vries H, Dijk F, Wetzels J, Mudde A, Kremers S, Ariza C et al. The European Smoking prevention Framework Approach (ESFA): effects after 24 and 30 months. Health Education Research 2006; 21(1):116-132.

(20) Mercken L, Moore L, Crone MR, de VH, De Bourdeaudhuij I, Lien N et al. The effectiveness of school-based smoking prevention interventions among low- and high-SES European teenagers. Health Education Research 2012; 27(3):459-469.

(21) Woodruff SI. The epidemiology of smoking among united states navy women recruits: Prevalence, correlates, and short-term effects of involuntary cessation. Dissertation Abstracts International: Section B: The Sciences and Engineering 1998; .59(5-B).

(22) Akhtar PC, Haw SJ, Levin KA, Currie DB, Zachary R, Currie CE. Socioeconomic differences in second-hand smoke exposure among children in Scotland after introduction of the smoke-free legislation. Journal of Epidemiology & Community Health 2010; 64(4):341-346.

(23) Mackay D, Haw S, Ayres JG, Fischbacher C, Pell JP. Smoke-free legislation and hospitalizations for childhood asthma. New England Journal of Medicine 2010; 363(12):1139-1145.

(24) Millett C, Lee JT, Laverty AA, Glantz SA, Majeed A. Hospital admissions for childhood asthma after smoke-free legislation in England. Pediatrics 2013; 131(2):e495-e501.

(25) Moore GF, Holliday JC, Moore LA. Socioeconomic patterning in changes in child exposure to secondhand smoke after implementation of smoke-free legislation in Wales. Nicotine & Tobacco Research 2011; 13(10):903-910.

68

(26) Moore GF, Currie D, Gilmore G, Holliday JC, Moore L. Socioeconomic inequalities in childhood exposure to secondhand smoke before and after smoke-free legislation in three UK countries. Journal of Public Health 2012; 34(4):599-608.

(27) Nabi-Burza E, Regan S, Drehmer J, Ossip D, Rigotti N, Hipple B et al. Parents smoking in their cars with children present. Pediatrics 2012; 130(6):December.

(28) Noach MB, Steinberg DM, Rier DA, Goldsmith R, Shimony T, Rosen LJ. Ethnic Differences in Patterns of Secondhand Smoke Exposure Among Adolescents in Israel . Nicotine and Tobacco Research 2012; 14(6):648-656.

(29) Galan I, Diez-Ganan L, Mata N, Gandarillas A, Cantero JL, Durban M. Individual and contextual factors associated to smoking on school premises. Nicotine and Tobacco Research 2012; 14(4):2012.

(30) Woodruff SI, Conway TL,Edawrds CC. Effect of an eight week smoking ban on women at US Navy recruit training command. Tobacco Control 2009; 9:40-46.

(31) Gilpin EA, Pierce JP. Trends in adolescent smoking initiation in the United States: is tobacco marketing an influence? Tobacco Control 1997; 6(2):122-127.

(32) Hammond D, Doxey J, Daniel S, Bansal-Travers M. Impact of Female-Oriented Cigarette Packaging in the United States. Nicotine Tobacco Res 2011; 13(7):579-588.

(33) Pucci LG, Joseph HM, Jr., Siegel M. Outdoor tobacco advertising in six Boston neighborhoods. Evaluating youth exposure. American Journal of Preventive Medicine 1998; 15(2):155-159.

(34) Vallone DM, Allen JA, Xiao H. Is socioeconomic status associated with awareness of and receptivity to the truth campaign? Drug & Alcohol Dependence 2009; 104:Suppl-20.

(35) Glied S. Youth tobacco control: reconciling theory and empirical evidence. Journal of Health Economics 2000; 21:117-135.

(36) Gruber J. Youth smoking in the US:Prices and policies, working paper 7506. 2000. Cambridge, MA. National Bureau of Economic Research (NBER) Working Paper Series.

(37) Madden D. Tobacco taxes and starting and quitting smoking: does the effect differ by education? Applied Economics 2007; 39:613-627.

(38) Biener L, Aseltine RH, Jr., Cohen B, Anderka M. Reactions of adult and teenaged smokers to the Massachusetts tobacco tax. American Journal of Public Health 1998; 88(9):1389-1391.

(39) Peretti-Watel P, Guagliardo V, Combes J-B, Obadia Y, Verger PE-MA, Peretti-Watel Ppif. Young smokers' adaptation to higher cigarette prices: How did those daily smokers who did not quit react? The case of students of South-Eastern France. Drugs: Education, Prevention & Policy 2010; .17(5).

69

(40) Lipperman-Kreda S, Grube JW, Friend KB. Contextual and community factors associated with youth access to cigarettes through commercial sources. Tobacco Control Online First . 2012. R

(41) Widome R, Brock B, Noble P, Forster JL. The Relationship of Point-of-Sale Tobacco Advertising and Neighborhood Characteristics to Underage Sales of Tobacco. Evaluation and the Health Professions 2012; 35(3):September.

(42) Millett C, Lee JT, Gibbons DC, Glantz SA. Increasing the age for the legal purchase of tobacco in England: impacts on socio-economic disparities in youth smoking. Thorax 2011; 66(10):862-865.

(43) Schneider S, Gruber J, Yamamoto S, Weidmann C. What happens after the implementation of electronic locking devices for adolescents at cigarette vending machines? A natural longitudinal experiment from 2005 to 2009 in Germany. Nicotine & Tobacco Research 2011; 13(8):732-740.

(44) Kim H, Clark PI. Cigarette smoking transition in females of low socioeconomic status: impact of state, school, and individual factors. Journal of Epidemiology & Community Health 2006; 60:Suppl-9.

(45) Bacon TP, Hilderbrand JA. Impact of a School-Based Drug Prevention Program on Students' Behaviors and Risk and Protective Factors. 1-4-2001.

(46) Menrath I, Mueller-Godeffroy E, Pruessmann C, Ravens-Sieberer U, Ottova V, Pruessmann M et al. Evaluation of school-based life skills programmes in a high-risk sample: A controlled longitudinal multi-centre study. Journal of Public Health (Germany) 2012; 20(2):April.

(47) West P, Sweeting H, Young R. Smoking in Scottish youths: personal income, parental social class and the cost of smoking. Tobacco Control 16, 329-335. 2007.

(48) Helakorpi S, Martelin T, Torppa J, Vartiainen E, Uutela A, Patja K. Impact of the 1976 Tobacco Control Act in Finland on the proportion of ever daily smokers by socioeconomic status. Preventive Medicine 2008; 46(4):340-345.

(49) White VM, Hayman J, Hill DJ. Can population-based tobacco-control policies change smoking behaviors of adolescents from all socio-economic groups? Findings from Australia: 1987-2005. Cancer Causes & Control 2008; 19(6):631-640.

(50) Pabayo R, O'Loughlin J, Barnett TA, Cohen JE, Gauvin L. Does Intolerance of Smoking at School, or in Restaurants or Corner Stores Decrease Cigarette Use Initiation in Adolescents? Nicotine Tobacco Res 2012; first published online February 21, 2012(7).

(51) Rodgers A, Corbett T, Bramley D, Riddell T, Wills M, Lin RB et al. Do u smoke after txt? Results of a randomised trial of smoking cessation using mobile phone text messaging. Tobacco Control 2005; 14(4):255-261.

(52) Ybarra ML, Holtrop JS, Prescott TL, Rahbar MH, Strong D. Pilot RCT Results of Stop My Smoking USA: A Text Messaging–Based Smoking Cessation Program for Young Adults. Nicotine & Tobacco Research Advance Access. 2013.

70

7 APPENDICES7.1 Appendix A Search strategies: electronic searches, handsearching

and searching for grey literatureElectronic searches

Ovid MEDLINE(R) In-Process & Other Non-Indexed Citations and Ovid MEDLINE(R) 1946 to May 04 2012, search date 09/05/2012; also Ovid MEDLINE(R) 1946 to January week 3, 2013, search date 23/01/2013

1. smoking/2. smoking cessation/3. tobacco/4. "Tobacco Use Disorder"/5. nicotine/6. tobacco, smokeless/7. tobacco use, cessation/8. (smokers or smoker).ti,ab.9. cigar$.mp.10. smoking.ti,ab.11. or/1-1012. smoking cessation/13. tobacco use, cessation/14. tobacco use, cessation products/15. smoking/pc16. smoking/dt17. smoking/th18. ((smok$ or anti smok$ or tobacco or cigarette$) adj3 (ban or bans or prohibit$ or restrict$ or discourage$)).ti,ab.19. ((smok$ or anti smok$ or tobacco or cigarette$) adj3 (workplace or work place or work site or worksite)).ti,ab.20. ((smok$ or anti smok$ or tobacco or cigarette$) adj3 (public place$ or public space$ or public area$ or office$ or school$ or institution$)).ti,ab.21. ((smok$ or anti smok$ or tobacco or cigarette$) adj3 (legislat$ or government$ or authorit$ or law or laws or bylaw$ or byelaw$ or bye law$ or regulation$)).ti,ab.22. ((tobacco free or smoke free) adj3 (hospital or inpatient or outpatient or institution$)).ti,ab.23. ((tobacco-free or smoke-free) adj3 (facilit$ or zone$ or area$ or site$ or place$ or environment$ or air)).ti,ab.24. ((tobacco or smok$ or cigarette$) adj3 (campaign$ or advertis$ or advertiz$)).ti,ab.25. ((billboard$ or advertis$ or advertiz$ or sale or sales or sponsor$) adj3 (restrict$ or limit$ or ban or bans or prohibit$)).ti,ab.26. (tobacco control adj3 (program$ or initiative$ or policy or policies or intervention$ or activity or activities or framework)).ti,ab.27. ((smok$ or tobacco) adj (policy or policies or program$)).ti,ab.28. ((retailer$ or vendor$) adj3 (educat$ or surveillance$ or prosecut$ or legislat$)).ti,ab.29. test purchas$.ti,ab.30. voluntary agreement$.ti,ab.31. health warning$.ti,ab.32. ((tobacco or cigarette$) adj3 (tax or taxes or taxation or excise or duty free or duty paid or customs)).ti,ab.33. ((cigarette$ or tobacco) adj3 (packaging or packet$)).ti,ab.

71

34. ((cigarette$ or tobacco) adj3 (marketing or marketed)).ti,ab.35. ((cigarette$ or tobacco) adj3 (price$ or pricing)).ti,ab.36. point of sale.ti,ab.37. vending machine$.ti,ab.38. (trade adj (restrict$ or agreement$)).ti,ab.39. (contraband$ or smuggl$ or bootleg$ or cross border shopping).ti,ab.40. (tobacco control act or clean air or clean indoor air).ti,ab.41. ((reduce$ or prevent$) adj3 (environmental tobacco smoke or passive smok$ or secondhand smok$ or second hand smok$ or SHS)).ti,ab.42. ((population level or population based or population orientated or population oriented) adj3 (intervention$ or prevention or policy or policies or program$ or project$)).ti,ab.43. (community adj3 (intervention$ or prevention or policy or policies or program$ or project$)).ti,ab.44. ((sale or sales or retail$ or purchas$) adj3 (minors or teenage$ or underage$ or under-age$ or child$)).ti,ab.45. (youth access adj3 restrict$).ti,ab.46. (smoking cessation or cessation support).ti,ab.47. (smokefree or smoke-free or smoke free).ti,ab.48. ((stop$ or quit$ or reduc$ or give up or giving up) adj3 (cigarette$ or tobacco or smoking)).ti,ab.49. quit attempt$.ti,ab.50. tobacco quit.ti,ab.51. quit rate$.ti,ab.52. (quitline$ or quit line$ or quit-line$).ti,ab.53. ((smok$ or tobacco or nicotine or cigarette$) adj2 (abstinence or cessation)).ti,ab.54. or/12-5355. (socioeconomic or socio economic or socio-economic).ti,ab.56. inequalit$.ti,ab.57. depriv$.ti,ab.58. disadvantage$.ti,ab.59. educat$.ti,ab.60. (social adj (class$ or group$ or grade$ or context$ or status)).ti,ab.61. (employ$ or unemploy$).ti,ab.62. income.ti,ab.63. poverty.ti,ab.64. SES.ti,ab.65. demographic$.ti,ab.66. (uninsur$ or insur$).ti,ab.67. minorit$.ti,ab.68. poor.ti,ab.69. affluen$.ti,ab.70. equity.ti,ab.71. (underserved or under served or under-served).ti,ab.72. occupation$.ti,ab.

72

73. (work site or worksite or work-site).ti,ab.74. (work place or workplace or work-place).ti,ab.75. (work force or workforce or work-force).ti,ab.76. (high risk or high-risk or at risk).ti,ab.77. (marginalised or marginalized).ti,ab.78. (social$ adj (disadvant$ or exclusion or excluded or depriv$)).ti,ab.79. exp socioeconomic factors/80. exp public assistance/81. exp social welfare/82. vulnerable populations/83. or/55-8284. 11 and 5485. 83 and 8486. limit 85 to (abstracts and english language and yr="1990 -Current")

73

Embase; Excerpta Medica Database Guide, 1980 to 2012 Week 18, search date 09/05/2012; also 1980 to 2013 week 3, search date 23/01/2013

1. smoking/2. smoking cessation/3. tobacco/4. nicotine/5. tobacco, smokeless/6. "smoking and smoking related phenomena"/7. cigarette smoking/8. cigarette smoke/9. tobacco smoke/10. (smokers or smoker).ti,ab.11. cigar$.mp.12. smoking.ti,ab.13. or/1-1214. smoking cessation program/15. ((smok$ or anti smok$ or tobacco or cigarette$) adj3 (ban or bans or prohibit$ or restrict$ or discourage$)).ti,ab.16. ((smok$ or anti smok$ or tobacco or cigarette$) adj3 (workplace or work place or work site or worksite)).ti,ab.17. ((smok$ or anti smok$ or tobacco or cigarette$) adj3 (public place$ or public space$ or public area$ or office$ or school$ or institution$)).ti,ab.18. ((smok$ or anti smok$ or tobacco or cigarette$) adj3 (legislat$ or government$ or authorit$ or law or laws or bylaw$ or byelaw$ or bye law$ or regulation$)).ti,ab.19. ((tobacco free or smoke free) adj3 (hospital or inpatient or outpatient or institution$)).ti,ab.20. ((tobacco-free or smoke-free) adj3 (facilit$ or zone$ or area$ or site$ or place$ or environment$ or air)).ti,ab.21. ((tobacco or smok$ or cigarette$) adj3 (campaign$ or advertis$ or advertiz$)).ti,ab.22. ((billboard$ or advertis$ or advertiz$ or sale or sales or sponsor$) adj3 (restrict$ or limit$ or ban or bans or prohibit$)).ti,ab.23. (tobacco control adj3 (program$ or initiative$ or policy or policies or intervention$ or activity or activities or framework)).ti,ab.24. ((smok$ or tobacco) adj (policy or policies or program$)).ti,ab.25. ((retailer$ or vendor$) adj3 (educat$ or surveillance$ or prosecut$ or legislat$)).ti,ab.26. test purchas$.ti,ab.27. voluntary agreement$.ti,ab.28. health warning$.ti,ab.29. ((tobacco or cigarette$) adj3 (tax or taxes or taxation or excise or duty free or duty paid or customs)).ti,ab.30. ((cigarette$ or tobacco) adj3 (packaging or packet$)).ti,ab.31. ((cigarette$ or tobacco) adj3 (marketing or marketed)).ti,ab.32. ((cigarette$ or tobacco) adj3 (price$ or pricing)).ti,ab.33. point of sale.ti,ab.34. vending machine$.ti,ab.35. (trade adj (restrict$ or agreement$)).ti,ab.36. (contraband$ or smuggl$ or bootleg$ or cross border shopping).ti,ab.37. (tobacco control act or clean air or clean indoor air).ti,ab.

74

38. ((reduce$ or prevent$) adj3 (environmental tobacco smoke or passive smok$ or secondhand smok$ or second hand smok$ or SHS)).ti,ab.39. ((population level or population based or population orientated or population oriented) adj3 (intervention$ or prevention or policy or policies or program$ or project$)).ti,ab.40. (community adj3 (intervention$ or prevention or policy or policies or program$ or project$)).ti,ab.41. ((sale or sales or retail$ or purchas$) adj3 (minors or teenage$ or underage$ or under-age$ or child$)).ti,ab.42. (youth access adj3 restrict$).ti,ab.43. (smoking cessation or cessation support).ti,ab.44. (smokefree or smoke-free or smoke free).ti,ab.45. ((stop$ or quit$ or reduc$ or give up or giving up) adj2 (cigarette$ or tobacco or smoking)).ti,ab.46. tobacco quit.ti,ab.47. quit attempt$.ti,ab.48. quit rate$.ti,ab.49. (quit line$ or quitline$ or quit-line$).ti,ab.50. ((smok$ or tobacco or nicotine or cigarette$) adj2 (abstinence or cessation)).ti,ab.51. or/14-5052. (socioeconomic or socio economic or socio-economic).ti,ab.53. inequalit$.ti,ab.54. depriv$.ti,ab.55. disadvantage$.ti,ab.56. educat$.ti,ab.57. (social adj (class$ or group$ or grade$ or context$ or status)).ti,ab.58. (employ$ or unemploy$).ti,ab.59. income.ti,ab.60. poverty.ti,ab.61. SES.ti,ab.62. demographic$.ti,ab.63. (uninsur$ or insur$).ti,ab.64. minorit$.ti,ab.65. poor.ti,ab.66. affluen$.ti,ab.67. equity.ti,ab.68. (underserved or under served or under-served).ti,ab.69. occupation$.ti,ab.70. (work site or worksite or work-site).ti,ab.71. (work place or workplace or work-place).ti,ab.72. (work force or workforce or work-force).ti,ab.73. (high risk or high-risk or at risk).ti,ab.74. (marginalised or marginalized).ti,ab.75. (social$ adj (disadvant$ or exclusion or excluded or depriv$)).ti,ab.76. exp socioeconomics/77. public assistance/78. welfare, social/79. exp social status/80. social security/81. vulnerable population/82. or/52-8183. 13 and 5184. 82 and 8385. limit 84 to (abstracts and english language and yr="1990 -Current")

75

PsycInfo (OVID) 1987 to May Week 1 2012, search date 10/05/2012; also 1987 to January week 3 2013, search date 23/01/2013

76

1. exp tobacco smoking/2. exp smoking cessation/3. nicotine/4. tobacco, smokeless/5. (smokers or smoker).ti,ab.6. tobacco.ti,ab.7. nicotine.ti,ab.8. cigar$.mp.9. smoking.ti,ab.10. or/1-911. exp smoking cessation/12. ((smok$ or anti smok$ or tobacco or cigarette$) adj3 (ban or bans or prohibit$ or restrict$ or discourage$)).ti,ab.13. ((smok$ or anti smok$ or tobacco or cigarette$) adj3 (workplace or work place or work site or worksite)).ti,ab.14. ((smok$ or anti smok$ or tobacco or cigarette$) adj3 (public place$ or public space$ or public area$ or office$ or school$ or institution$)).ti,ab.15. ((smok$ or anti smok$ or tobacco or cigarette$) adj3 (legislat$ or government$ or authorit$ or law or laws or bylaw$ or byelaw$ or bye law$ or regulation$)).ti,ab.16. ((tobacco free or smoke free) adj3 (hospital or inpatient or outpatient or institution$)).ti,ab.17. ((tobacco-free or smoke-free) adj3 (facilit$ or zone$ or area$ or site$ or place$ or environment$ or air)).ti,ab.18. ((tobacco or smok$ or cigarette$) adj3 (campaign$ or advertis$ or advertiz$)).ti,ab.19. ((billboard$ or advertis$ or advertiz$ or sale or sales or sponsor$) adj3 (restrict$ or limit$ or ban or bans or prohibit$)).ti,ab.20. (tobacco control adj3 (program$ or initiative$ or policy or policies or intervention$ or activity or activities or framework)).ti,ab.21. ((smok$ or tobacco) adj (policy or policies or program$)).ti,ab.22. ((retailer$ or vendor$) adj3 (educat$ or surveillance$ or prosecut$ or legislat$)).ti,ab.23. test purchas$.ti,ab.24. voluntary agreement$.ti,ab.25. health warning$.ti,ab.26. ((tobacco or cigarette$) adj3 (tax or taxes or taxation or excise or duty free or duty paid or customs)).ti,ab.27. ((cigarette$ or tobacco) adj3 (packaging or packet$)).ti,ab.28. ((cigarette$ or tobacco) adj3 (marketing or marketed)).ti,ab.29. ((cigarette$ or tobacco) adj3 (price$ or pricing)).ti,ab.30. point of sale.ti,ab.31. vending machine$.ti,ab.32. (trade adj (restrict$ or agreement$)).ti,ab.33. (contraband$ or smuggl$ or bootleg$ or cross border shopping).ti,ab.34. (tobacco control act or clean air or clean indoor air).ti,ab.35. ((reduce$ or prevent$) adj3 (environmental tobacco smoke or passive smok$ or secondhand smok$ or second hand smok$ or SHS)).ti,ab.36. ((population level or population based or population orientated or population oriented) adj3 (intervention$ or prevention or policy or policies or program$ or project$)).ti,ab.37. (community adj3 (intervention$ or prevention or policy or policies or program$ or project$)).ti,ab.

77

38. ((sale or sales or retail$ or purchas$) adj3 (minors or teenage$ or underage$ or under-age$ or child$)).ti,ab.39. (youth access adj3 restrict$).ti,ab.40. (smoking cessation or cessation support).ti,ab.41. (smokefree or smoke-free or smoke free).ti,ab.42. ((stop$ or quit$ or reduc$ or give up or giving up) adj3 (cigarette$ or tobacco or smoking)).ti,ab.43. quit attempt$.ti,ab.44. tobacco quit.ti,ab.45. quit rate$.ti,ab.46. (quitline$ or quit line$ or quit-line$).ti,ab.47. ((smok$ or tobacco or nicotine or cigarette$) adj2 (abstinence or cessation)).ti,ab.48. or/11-4749. (socioeconomic or socio economic or socio-economic).ti,ab.50. inequalit$.ti,ab.51. depriv$.ti,ab.52. disadvantage$.ti,ab.53. educat$.ti,ab.54. (social adj (class$ or group$ or grade$ or context$ or status)).ti,ab.55. (employ$ or unemploy$).ti,ab.56. income.ti,ab.57. poverty.ti,ab.58. SES.ti,ab.59. demographic$.ti,ab.60. (uninsur$ or insur$).ti,ab.61. minorit$.ti,ab.62. poor.ti,ab.63. affluen$.ti,ab.64. equity.ti,ab.65. (underserved or under served or under-served).ti,ab.66. occupation$.ti,ab.67. (work site or worksite or work-site).ti,ab.68. (work place or workplace or work-place).ti,ab.69. (work force or workforce or work-force).ti,ab.70. (high risk or high-risk or at risk).ti,ab.71. (marginalised or marginalized).ti,ab.72. (social$ adj (disadvant$ or exclusion or excluded or depriv$)).ti,ab.73. exp socioeconomic status/74. poverty/75. disadvantaged/76. or/49-7577. 10 and 4878. 76 and 7779. limit 78 to (english language and abstracts and yr="1990 - 2012")

78

Cochrane Library 2012 (Cochrane Database of Systematic Reviews; Database of Abstracts of Reviews of Effects; Cochrane Central Register of Controlled Trials; Health Technology Assessment Database), search date 10/05/12

79

#1 MeSH descriptor Smoking, this term only#2 MeSH descriptor Tobacco Use Cessation explode all trees#3 MeSH descriptor Tobacco explode all trees#4 MeSH descriptor Tobacco Use Disorder, this term only#5 MeSH descriptor Nicotine, this term only#6 (smoking or smokers or smoker or tobacco or cigar* or nicotine)#7 (#1 OR #2 OR #3 OR #4 OR #5 OR #6)#8 (smok* or anti-smok* or tobacco or cigarette*) near3 (ban or bans or prohibit* or restrict* or discourage*)#9 (smok* or anti-smok* or tobacco or cigarette*) near3 (workplace or work place or worksite)#10 (smok* or anti-smok* or tobacco or cigarette*) near3 (public next place*)#11 (smok* or anti-smok* or tobacco or cigarette*) near3 (public next space)#12 (smok* or anti-smok* or tobacco or cigarette*) near3 (public next area*)#13 (smok* or anti-smok* or tobacco or cigarette*) near3 (office* or school* or institution*)#14 (smok* or anti-smok* or tobacco or cigarette*) near3 (legislat* or government* or authorit* or law or laws or bylaw* or byelaw* or bye-law* or regulation*)#15 (tobacco-free or smoke-free) near3 (hospital* or inpatient* or outpatient* or institution*)#16 (tobacco-free or smoke-free) near3 (facility* or zone* or area* or site* or place* or environment* or air)#17 (tobacco or smok* or cigarette*) near3 (campaign* or advertis* or advertiz*)#18 (billboard* or advertis* or advertiz* or sale or sales or sponsor*) near3 (restrict* or limit* or ban or bans or prohibit*)#19 (tobacco next control) near3 (program* or initiative* or policy or policies or intervention* or activity or activities or framework)#20 (smok* or tobacco) next (policy or policies or program*)#21 (retailer* or vendor*) near3 (educat* or surveillance or prosecut* or legslat*)#22 test next purchas* in All Fields or (voluntary next agreement*)#23 (sale or sales or retail* or purchas*) near3 (minors or teenage* or underage* or under-age* or child*)#24 (youth near3 access) near3 restrict*#25 health next warning*#26 (tobacco or cigarette*) near3 (tax or taxes or taxation or excise or duty-free or duty-paid or customs)#27 (cigarette* or tobacco) near3 (packaging or packet*)#28 (cigarette* or tobacco) near3 (marketing or marketed)#29 (cigarette* or tobacco) near3 (price* or pricing)#30 "point of sale"#31 vending next machine*#32 trade near3 (restrict* or agreement*)#33 contraband* or smuggl* or bootleg* or (cross-border next shopping)#34 "tobacco control act" or "clean air" or "clean indoor air"#35 reduce* near3 "environmental tobacco smoke" or (passive next smok*) or (secondhand next smok*) or (second next hand next smok*) or SHS#36 prevent* near3 "environmental tobacco smoke" or (passive next smok*) or (secondhand next smok*) or (second next hand next smok*) or SHS#37 (population next level) near3 (intervention* or prevention or policy or policies or program* or project*)

80

#38 (population next based) near3 (intervention* or prevention or policy or policies or program* or project*)#39 (population next orientated) near3 (intervention* or prevention or policy or policies or program* or project*)#40 (community next level) near3 (intervention* or prevention or policy or policies or program* or project*)#41 (community next based) near3 (intervention* or prevention or policy or policies or program* or project*)#42 (community next orientated) near3 (intervention* or prevention or policy or policies or program* or project*)#43 (community next oriented) near3 (intervention* or prevention or policy or policies or program* or project*)#44 smoking next cessation or cessation next support#45 smokefree or smoke-free or smoke next free#46 (stop* or quit* or reduc* or give next up or giving next up) near3 (cigarette* or tobacco or smoking)#47 quit next attempt*#48 tobacco next quit#49 quit next rate*#50 quitline* or quit-line* or quit next line*#51 (smok* or tobacco or nicotine or cigarette*) near2 (abstinence or cessation)#52 (#8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19 OR #20 OR #21 OR #22 OR #23 OR #24 OR #25 OR #26 OR #27 OR #28 OR #29 OR #30 OR #31 OR #32 OR #33 OR #34 OR #35 OR #36 OR #37 OR #38 OR #39 OR #40 OR #41 OR #42 OR #43 OR #44 OR #45 OR #46 OR #47 OR #48 OR #49 OR #50 OR #51)#53 socioeconomic or socio next economic or socio-economic#54 inequalit*#55 depriv*#56 disadvantage*#57 educat*#58 social next (class* or group* or grade* or context* or status)#59 employ* or unemploy*#60 income#61 poverty#62 SES#63 demographic*#64 insur* or uninsur*#65 minorit*#66 poor#67 affluen*#68 equity#69 underserved or under next served or under-served#70 occupation*#71 work next site or worksite or work-site#72 work next place or workplace or work-place#73 work next force or workforce or work-force#74 high next risk or high-risk or at next risk#75 marginalised or marginalized#76 social* next (disadvant* or exclusion or excluded or depriv*)

81

#77 MeSH descriptor Socioeconomic Factors explode all trees#78 MeSH descriptor Public Assistance, this term only#79 MeSH descriptor Social Welfare, this term only#80 MeSH descriptor Vulnerable Populations, this term only#81 (#53 OR #54 OR #55 OR #56 OR #57 OR #58 OR #59 OR #60 OR #61 OR #62 OR #63 OR #64 OR #65 OR #66 OR #67 OR #68 OR #69 OR #70 OR #71 OR #72 OR #73 OR #74 OR #75 OR #76 OR #77 OR #78 OR #79 OR #80)#82 (#7 AND #52)#83 (#81 and #82), from 1990 to 2012

82

Science Citation Index Expanded, Social Sciences Citation Index, Conference Proceedings Citation Index (Science, and Social Science & Humanities), in Web of Science hosted on ISI Web of Knowledge, search date 10/05/12

(TS=(smoking or smokers or smoker or tobacco or cigar* or nicotine) AND TS=(abstinence or cessation or quit*) AND TS=(socioeconomic or socio economic or socio-economic)) AND Language=(English), Timespan=1990-2012

BIOSIS Previews hosted on ISI Web of Knowledge, search date 10/05/12(TS=(smoking or smokers or smoker or tobacco or cigar* or nicotine) AND TS=(abstinence or cessation or quit*) AND TS=(socioeconomic or socio economic or socio-economic)) AND Language=(English), Timespan=1990-2012

CINAHL Plus (EBSCO host) search date 10/05/12

S8 S5 AND S9, Limiters - Published Date from: 19900101-20121231S9 S6 OR S7 OR S8 S8 TX social* W1 (disadvantage* or exclusion or excluded or depriv*)S7 TX social W1 (class* or group* or grade* or context* or status)S6 (MH "Socioeconomic Factors") OR "SOCIOECONOMIC" OR (MH "Poverty") OR "POVERTY" OR "EQUITY"S5 S1 OR S2 OR S3 OR S4S4 TX (stop* or quit* or reduc* or give up or giving up) W3 (cigarette* or tobacco or smoking)S3 TX Smoking W1 cessationS2 (MH "Tobacco, Smokeless") OR (MH "Tobacco Abuse Control (Saba CCC)") OR (MH "Risk Control: Tobacco Use (Iowa NOC)") OR (MH "Passive Smoking")S1 (MH "Smoking Cessation Programs") OR (MH "Smoking Cessation") OR (MH "Smoking Cessation Assistance (Iowa NIC)")

ERIC (EBSCO Host) search date 11/05/12S10 S8 and S9S9 S4 or S5 or S6 or S7S8 S1 or S2 or S3S7 AB Socioeconomic OR AB Poverty OR AB equityS6 ((DE "Socioeconomic Background" OR DE "Socioeconomic Influences" OR DE "Socioeconomic Status") OR (DE "Poverty")) AND (DE "Disadvantaged Environment" OR DE "Economically Disadvantaged" OR DE "Socioeconomic Influences")S5 TX social* W1 (disadvantage* or exclusion or excluded or depriv*)S4 TX social W1 (class* or group* or grade* or context* or status)S3 TX (stop* or quit* or reduc* or give up or giving up) W3 (cigarette* or tobacco or smoking)S2 TX Smoking W1 cessationS1 DE SMOKING

83

Handsearching:

1. Addiction 2012 volume 107 issues 1 to 8 (August 2012) and Early View, search date 31/7/12; also ‘Accepted Articles’, ‘Early View’, search date 14/2/13 and 2012 volume 107 issues 12 and S2, volume 108 issues 1 to 2 search date 18/2/13.

2. Nicotine and Tobacco Research 2012, volume 14, issues 1 to 6, search date 30/7/12; also 2013 volume 15 issues 1 to 3 and ‘Advance Access’ search date 18/2/13.

3. Social Science and Medicine 2012, volume 74 issues 1 to 12, volume 75 issues 1 to 7, articles ‘in press’ search date 31/7/12; also 2013 volumes 74 to 82 ‘in progress’, and ‘articles in press’, search date 18/2/13.

4. Tobacco Control 2012, volume 21, issues 1 to 4, ‘online first’ search date 31/7/12; also volume 21 issue 6, volume 22 issues 1 to 2 and ‘online first’, search date 18/2/13.

84

Searching for grey literature

23/11/12Dear All,

As you know, ENSP is an Associated Partner in the SILNE project (http://www.ensp.org/node/738).

In order to support the implementation of Work Package 6: Review & Synthesis by Amanda Amos and Tamara Brown, our colleagues from the University of Edinburgh, and help them to identify any grey literature, we would be grateful if you could inform them of any such literature that they may be able to include in their review, particularly government reports that they may not have identified through their searching.

They are now at the stage where they have a complete list of included studies both for the review of youth policies and the review of adult policies. Please see the attached inclusion/exclusion criteria. Attached are also the reference lists of these studies.

Amanda and Tamara are specifically interested in any reports of the socio-economic impact of policies which are written in non-English and which an English synopsis could be provided.

Please do not hesitate to contact them should you need any further clarification:

Tamara BrownResearch FellowCentre for Population Health SciencesUniversity of EdinburghTeviot PlaceEdinburghEH8 9AGScotland, UKTel: 0131 650 3237Fax: 0131 650 6909Email: tbrown23@staffmail.ed.ac.uk

It would be great if you could not remain simply silent. So, even if you have no available information, a simple negative reply would be appreciated. The deadline is 31/12/12.

Thanking you in advance,

Best regardsFrancisFrancis GrognaSecretary GeneralENSP - European Network for Smoking and Tobacco Prevention

85

10/12/12

To all members of SILNE,

I am pleased to tell you that the youth report for Work Package 6: Review & Synthesis is nearly complete and the adult policy review is well under way.

Amanda and I look forward to presenting the initial results of these reviews when we all meet in Brussels in January.

Do you know of any grey literature that we may be able to include in our review, particularly government reports that we may not have identified through our searching? We are specifically interested in any reports of the socio-economic impact of policies which are written in non-English and which an English synopsis could be provided.

I attach reference lists of included studies both for the review of youth policies and the review of adult policies. I also attach our inclusion/exclusion criteria.

Our deadline for receiving literature is 31/12/12.

Please let me know if you require any further information and I look forward to some hopeful replies and meeting you again in January.

Very best wishes Tamara

Tamara Brown Research Fellow Centre for Population Health Sciences University of Edinburgh Teviot Place Edinburgh EH8 9AG Scotland, UK Tel: 0131 650 3237 Fax: 0131 650 6909 Email: tbrown23@staffmail.ed.ac.uk

86

7.2 Appendix B WHO European countries and other stage 4 countriesAlbania AndorraArmeniaAustriaAzerbaijanBelarusBelgiumBosnia and HerzegovinaBulgariaCroatiaCyprusCzech RepublicDenmarkEstoniaFinlandFranceGeorgiaGermanyGreeceHungaryIceland

87

IrelandIsraelItalyKazakhstanKyrgyzstanLatviaLithuaniaLuxembourgMaltaMonacoMontenegroNetherlandsNorwayPolandPortugalRepublic of MoldovaRomaniaRussian FederationSan MarinoSerbiaSlovakiaSloveniaSpainSwedenSwitzerlandTajikistanThe Former Yugoslav Republic of MacedoniaTurkeyTurkmenistanUkraineUnited Kingdom of Great Britain and Northern IrelandUzbekistanOther stage 4 countries: Australia, United States, New Zealand, Canada

7.3 Appendix C Inclusion/exclusion formRef ID

FIRST AUTHOR YEAR

CODE

ANSWER

TYPE QUESTION

1 population Is the study population 11 years of age or older?2 Is it based in a WHO European country or non-

European country at stage 4 of the tobacco epidemic?

3 intervention/policy Is it an intervention or policy to reduce adult smoking or to prevent youth starting to smoke?

4 socio-economic inequalities

Does it report outcomes for high vs. low socio-economic group?*

What type of study design is it? (highlight) Review RCT

88

Non-randomised controlled study Observational cohort Qualitative Other

What type of intervention is it? (highlight) taxation/pricing tobacco advertising and marketing bans smoking cessation support smoke free policies (public places, workplaces, home) school-based interventions mass media campaigns community programmes educational policies social and welfare policies employment policies multifaceted lifestyle interventions/policies (not just smoking cessation) other

What type of SES indicator does it report? (highlight) Income Education Occupational social class Area-level socio-economic deprivation Housing tenure Subjective social class Health insurance Proxy measures for youth, i.e. Free School Meals, Family Affluences Scale (FAS)

What type of outcomes does it report? (highlight) quit rates initiation rates changes in initiation/cessation or abstinence rates uptake and reach use of quitting aids/services smoking status (self-reported/validated) number of quit attempts exposure prevalence changing attitudes passive smoking policy reach/awareness/comprehensiveness attitude/social norms intentions to smoke sources (i.e. vending machines) second hand smoke exposure other

What is the length of follow up? (highlight)<3 months3 months6 months12 monthsOtherIs the interventionYouth or adult or both? (highlight)Individual support or population/policy or both? (highlight)

89

What is the type of analyses?Population-level or individual level or both? (highlight)*INCLUDE? YES/NO/UNCLEAR (highlight)

*To be included a paper must be rated as YES to 1 + 2 + 3 + 4

REVIEWER COMMENTS

90

7.4 Appendix D Included studies-YouthReference Source

Akhtar PC, Haw SJ, Levin KA, Currie DB, Zachary R, Currie CE. Socioeconomic differences in second-hand smoke exposure among children in Scotland after introduction of the smoke-free legislation. Journal of Epidemiology & Community Health 2010; 64(4):341-346.

MEDLINE

Bacon TP, Hilderbrand JA. Impact of a School-Based Drug Prevention Program on Students' Behaviors and Risk and Protective Factors. Paper presented at the Annual Conference of the American Educational Research Association (Seattle, WA, April 10-14, 2001).

ERIC

Biener L, Aseltine RH, Jr., Cohen B, Anderka M. Reactions of adult and teenaged smokers to the Massachusetts tobacco tax. American Journal of Public Health 1998; 88(9):1389-1391.

MEDLINE

*Campbell R, Starkey F, Holliday J, Audrey S, Bloor M, Parry-Langdon N et al. An informal school-based peer-led intervention for smoking prevention in adolescence (ASSIST): a cluster randomised trial. Lancet 2008; 371:1595-1602.

Mercken 2012

*Crone MR, Reijneveld SA, Willemsen MC, van Leerdam FJ, Spruijt RD, Sing RA. Prevention of smoking in adolescents with lower education: a school based intervention study. Journal of Epidemiology & Community Health 2003; 57(9):675-680.

Mercken 2012

*de Vries H, Dijk F, Wetzels J, Mudde A, Kremers S, Ariza C et al. The European Smoking prevention Framework Approach (ESFA): effects after 24 and 30 months. Health Education Research 2006; 21(1):116-132.

Mercken 2012

Galan I, Diez-Ganan L, Mata N, Gandarillas A, Cantero JL, Durban M. Individual and contextual factors associated to smoking on school premises. Nicotine and Tobacco Research 2012; 14(4):2012.

HAND SEARCH

Gilpin EA, Pierce JP. Trends in adolescent smoking initiation in the United States: is tobacco marketing an influence? Tobacco Control 1997; 6(2):122-127.

MEDLINE

Glied S. Youth tobacco control: reconciling theory and empirical evidence. Journal of Health Economics 2000; 21:117-135.

YORK REVIEW

Gruber J. Youth smoking in the US:Prices and policies, working paper 7506. 2000. Cambridge, MA. National Bureau of Economic Research (NBER) Working Paper Series.

YORK REVIEW

Hammond DDJDSB-TM. Impact of Female-Oriented Cigarette Packaging in the United States. Nicotine & Tobacco Research 2011; 13(7):579-588.

EXPERT

Helakorpi S, Martelin T, Torppa J, Vartiainen E, Uutela A, Patja K. Impact of the 1976 Tobacco Control Act in Finland on the proportion of ever daily smokers by socioeconomic status. Preventive Medicine 2008; 46(4):340-345.

MEDLINE

91

Kim H, Clark PI. Cigarette smoking transition in females of low socioeconomic status: impact of state, school, and individual factors. Journal of Epidemiology & Community Health 2006; 60:(Suppl II):ii13-ii19.

MEDLINE

Lipperman-Kreda S, Grube JW, Friend KB. Contextual and community factors associated with youth access to cigarettes throughcommercial sources. Tobacco Control Online First. 2012.

HANDSEARCH

Menrath I, Mueller-Godeffroy E, Pruessmann C, Ravens-Sieberer U, Ottova V, Pruessmann M et al. Evaluation of school-based life skills programmes in a high-risk sample: A controlled longitudinal multi-centre study. Journal of Public Health (Germany) 2012; 20(2):159-170.

EXPERT

Madden D. Tobacco taxes and starting and quitting smoking: does the effect differ by education? Applied Economics 2007; 39:613-627.

PHRC REVIEW

Mackay D, Haw S, Ayres JG, Fischbacher C, Pell JP. Smoke-free legislation and hospitalizations for childhood asthma. New England Journal of Medicine 2010; 363(12):1139-1145.

MEDLINE

*Mercken L, Moore L, Crone MR, de VH, De Bourdeaudhuij I, Lien N et al. The effectiveness of school-based smoking prevention interventions among low- and high-SES European teenagers. Health Education Research 2012; 27(3):459-469.

EXPERT

Millett C, Lee JT, Gibbons DC, Glantz SA. Increasing the age for the legal purchase of tobacco in England: impacts on socio-economic disparities in youth smoking. Thorax 2011; 66(10):862-865.

MEDLINE

Millett C, Lee JT, Laverty AA, Glantz SA, Majeed A. Hospital admissions for childhood asthma after smoke-free legislation in England. Pediatrics 2013; 131(2):e495-e501.

EXPERT

Moore GF, Currie D, Gilmore G, Holliday JC, Moore L. Socioeconomic inequalities in childhood exposure to secondhand smoke before and after smoke-free legislation in three UK countries. Journal of Public Health 2012; 34(4):599-608.

EXPERT

Moore GF, Holliday JC, Moore LA. Socioeconomic patterning in changes in child exposure to secondhand smoke after implementation of smoke-free legislation in Wales. Nicotine & Tobacco Research 2011; 13(10):903-910.

MEDLINE

Nabi-Burza E, Regan S, Drehmer J, Ossip D, Rigotti N, Hipple B et al. Parents smoking in their cars with children present. Pediatrics 2012; 130(6):e1471.

EMBASE

Noach MB, Steinberg DM, Rier DA, Goldsmith R, Shimony T, Rosen LJ. Ethnic Differences in Patterns of Secondhand Smoke Exposure Among Adolescents in Israel. Nicotine & Tobacco Research 2012; 14(6):648-656.

HAND SEARCH

Pabayo R, O'Loughlin J, Barnett TA, Cohen JE, Gauvin L. Does intolerance of smoking at school, or in restaurants or corner stores

HAND SEARCH

92

decrease cigarette use initiation in adolescents? Nicotine & Tobacco Research 2012; first published online February 21, 2012(7).

Peretti-Watel P, Guagliardo V, Combes J-B, Obadia Y, Verger P. Young smokers' adaptation to higher cigarette prices: How did those daily smokers who did not quit react? The case of students of South-Eastern France. Drugs: Education, Prevention & Policy 2010; 17(5): 632-640.

PsycINFO

Pucci LG, Joseph HM, Jr., Siegel M. Outdoor tobacco advertising in six Boston neighborhoods. Evaluating youth exposure. American Journal of Preventive Medicine 1998; 15(2):155-159.

MEDLINE

Rodgers A, Corbett T, Bramley D, Riddell T, Wills M, Lin RB et al. Do u smoke after txt? Results of a randomised trial of smoking cessation using mobile phone text messaging. Tobacco Control 2005; 14(4):255-261.

MEDLINE

Schneider S, Gruber J, Yamamoto S, Weidmann C. What happens after the implementation of electronic locking devices for adolescents at cigarette vending machines? A natural longitudinal experiment from 2005 to 2009 in Germany. Nicotine & Tobacco Research 2011; 13(8):732-740.

MEDLINE

Vallone DM, Allen JA, Xiao H. Is socioeconomic status associated with awareness of and receptivity to the truth campaign? Drug & Alcohol Dependence 2009; 104:Suppl-20:S15-S20.

MEDLINE

White VM, Hayman J, Hill DJ. Can population-based tobacco-control policies change smoking behaviors of adolescents from all socio-economic groups? Findings from Australia: 1987-2005. Cancer Causes & Control 2008; 19(6):631-640.

MEDLINE

Widome R, Brock B, Noble P, Forster JL. The relationship of point-of-sale tobacco advertising and neighborhood characteristics to underage sales of tobacco. Evaluation and the Health Professions 2012; 35(3):331-345.

EMBASE

Woodruff SI. Effect of an eight week smoking ban on women at US Navy recruit training command. Tobacco Control 2009; 9:40-46.

PsycINFO

Ybarra ML, Holtrop JS, Prescott TL, Rahbar MH, Strong D. Pilot RCT results of Stop My Smoking USA: a text messaging–based smoking cessation program for young adults. Nicotine & Tobacco Research Advance Access. 2013.

HANDSEARCH

*Mercken 2012 is a secondary analysis paper of Campbell 2008, Crone 2003 and de Vries 2006.

93

7.5 Appendix E Excluded studies-YouthReference Reason for exclusion

Alwan N, Siddiqi K, Thomson H, Cameron I. Children's exposure to second-hand smoke in the home: a household survey in the North of England. Health & Social Care in the Community 2010; 18(3):257-263.

Not an intervention or policy to reduce adult smoking or to prevent youth starting to smoke

Centers for Disease Control and Prevention (CDC). Response to increases in cigarette prices by race/ethnicity, income, and age groups--United States, 1976-1993. MMWR - Morbidity & Mortality Weekly Report 1998; 47(29):605-609.

Does not report outcomes for high versus low socio-economic group (for youth)

El Ansari W, Stock C. Factors associated with smoking, quit attempts and attitudes towards total smoking bans at university: a survey of seven universities in England, Wales and Northern Ireland. Asian Pacific Journal of Cancer Prevention 2012; 13(2):2012.

Not an intervention or policy to reduce adult smoking or to prevent youth starting to smoke – attitudes towards possible total smoking ban

Fardy PS, White RE, Clark LT, Amodio G, Hurster MH, McDermott KJ et al. Health promotion in minority adolescents: a Healthy People 2000 pilot study. Journal of Cardiopulmonary Rehabilitation 1995; 15(1):65-72.

Does not report outcomes for high versus low socio-economic group

Flynn BS, Worden JK, Secker-Walker RH, Pirie PL, Badger GJ, Carpenter JH. Long-term responses of higher and lower risk youths to smoking prevention interventions. Preventive Medicine 1997; 26(3):389-394.

Does not report outcomes for high versus low socio-economic group

Hamilton G, Cross D, Resnicow K, Hall M. A school-based harm minimization smoking intervention trial: outcome results. Addiction 2005; 100(5):689-700.

Does not report outcomes for high versus low socio-economic group

Hawkins SS, Chandra A, Berkman L. The impact of tobacco control policies on disparities in children's secondhand smoke exposure: a comparison of methods. Maternal and Child Health Journal 2012; 16:S70-77.

Included in adult policy review – examines tobacco use among households with school-age children and adolescents

Herbert RJ, Gagnon AJ, O'Loughlin JL, Rennick JE. Testing an empowerment intervention to help parents make homes smoke-free: a randomized controlled trial. Journal of Community Health 2011; 36(4):650-657.

Does not report outcomes for high versus low socio-economic group

Hublet A, Schmid H, Clays E, Godeau E, Gabhainn SN, Joossens L et al. Association between tobacco control policies and smoking behaviour among adolescents in 29 European countries. Addiction 2009; 104(11):1918-1926.

Does not report outcomes for high versus low socio-economic group

Jensen R, Lleras-Muney A. Does staying in school (and not working) prevent teen smoking and drinking? Journal of Health

Not based in a WHO European country or

94

Economics 2012; 31(4):644-657. non-European country at stage 4 of the tobacco epidemic – Dominican Republic.

Linetzky B, Mejia R, Ferrante D, De Maio FG, Diez Roux AV. Socioeconomic status and tobacco consumption among adolescents: A multilevel analysis of Argentina's global youth Tobacco survey. Nicotine and Tobacco Research 2012; 14(9):1092-1099.

Not based in a WHO European country or non-European country at stage 4 of the tobacco epidemic - Argentina.

Mata HJ. Development and evaluation of a personalized normative feedback intervention for Hispanic youth at high risk of smoking. Dissertation Abstracts International Section A: Humanities and Social Sciences 73[4-A], 1295. 2012.

Does not report outcomes for high versus low socio-economic group

Poulin CC. School smoking bans: do they help/do they harm? Drug & Alcohol Review 2007; 26(6):615-624.

Does not report outcomes for high versus low socio-economic group

Schmitt CL. The effect of decision heuristics and ethnicity on cigarette sales to minor girls. Dissertation Abstracts International: Section B: The Sciences and Engineering 2002; .62(9-B).

Not an intervention or policy to reduce adult smoking or to prevent youth starting to smoke

Sims M, Bauld L, Gilmore A. England's legislation on smoking in indoor public places and work-places: Impact on the most exposed children. Addiction 2012;107(11): 2009-2016.

Does not report outcomes for high versus low socio-economic group - regression analyses adjust for SES

Straub DM, Hills NK, Thompson PJ, Moscicki AB. Effects of pro- and anti-tobacco advertising on non-smoking adolescents' intentions to smoke. Journal of Adolescent Health 2003; 32(1):36-43.

Does not report outcomes for high versus low socio-economic group

Tangari AH, Tangari AH, Burton S, Andrews JC, Netemeyer RG. How do anti-tobacco campaign advertising and smoking status affect beliefs and intentions? Some similarities and differences between adults and adolescents. Journal of Public Policy & Marketing 2007; .26(1):60-74.

Does not report outcomes for high versus low socio-economic group

Veldwijk J, Hoving C, van Gelder BM, Feenstra TL. Potential reach of effective smoking prevention programmes in vocational schools: Determinants of school directors' intention to adopt these programmes. Public Health 2012; 126(4):338-342.

Not an intervention or policy to reduce adult smoking or to prevent youth starting to smoke - about theoretical intention to adopt a schools programme

Weinman ML, Weinman ML. A comparison of three groups of young fathers and program outcomes. School Social Work Journal 2007; .32(1):1-13.

Does not report outcomes for high versus low socio-economic group

95

Wildey MB, Clapp EJ, Woodruff SI, Kenney EM. Retailer education to reduce the availability of single cigarettes. Journal of Health Education 1995; 26(5):297-302.

Does not report outcomes for high versus low socio-economic group

96

7.6 Appendix F Data extraction - YouthDetails Method Results CommentsSmoking restriction in cars, workplaces, schools and other public placesAuthor, yearAkhtar, 2010

Age (years)11

Study designRepeat cross-sectional surveys in same schools before and after legislation

ObjectiveExplore socioeconomic differences in child exposure to environmental tobacco smoke (CHETS) after Scottish smoke-free legislation

SettingPrimary schools, Scotland

InterventionSmoke-free public and workplaces

SES variables usedSelf-reported family socioeconomic classification (SEC) and family affluence scale (FAS)

Data sourcesTwo surveys of 11 year olds, January 2006 and January 2007

Participant selection11 year olds, final year of primary school. 2006 n=2559 (86%) 2007 n=2424 (85%), 116/170 (68%) schools participated at baseline and 111/170 (65%) schools at follow up

Participant characteristics% low/medium/high FAS divided evenly. SEC mostly SEC1 and 2.FAS score for 96.6% of pupils in 2006 and 94.5% in 2007, and meaningful family SEC (family SEC 1-4) 79.1% in 2006 and 76.7% in 2007

Intervention detailsSmoking prohibited in almost all public and work places in Scotland from March 2006.

Outcomes measuredParental smoking statusPupil’s smoking statusSalivary cotinine levels

General population impactAfter legislation cotinine levels fell across all groups

Impact by SES variableThe greatest absolute decline in cotinine levels was among the lowest SEC and FAS groups even after adjusting for parental smokers (e.g. 0.10ng/ml in SEC1 v 0.28ng/ml in SEC4).However a linear regression model suggests that relative inequality between groups has widened. SHS exposure declined among children from lower SES households, higher in absolute terms but lower in relative terms than among children from higher SES households.

Author’s conclusion of SES impactSmoke-free legislation has reduced exposure to SHS among all children. Although the greatest absolute reduction in cotinine is observed in the lowest SEC/FAS group, cotinine levels remain highest for this group and there is a suggestion of possible increases in inequalities, which may warrant longer-term monitoring

Internal validityFAS and SEC determined using child’s answers, including parental occupation and material affluence although questions on family affluence seem fairly simple, so should lead to few being incorrectly categorised.Biochemical measure of smoking.

External validityNo bias detected in non-participation rates. Students absent from school on day of data collection were not included, though these represent a small proportion. Ignores those excluded from school, most likely to be low SES.Narrow age group limits generalisability.Same linear analyses as CHETS Wales (Moore 2011).Average cotinine concentrations among children in the Scottish CHETS were substantially higher than in

97

Details Method Results CommentsSmoking restriction in cars, workplaces, schools and other public placesStudy analysisAnalysis of variance and linear regression analyses, accounts for cluster and stratification. All results based on confirmed non-smoking pupils

Wales and children’s SHS exposure outside of the home was perhaps greater in Scotland, with impacts of legislation therefore greater overall than in Wales and distributed among all groups

Validity of author’s conclusionValid. Reviewer ratio calculations using reported mean concentrations also suggest a widening of relative inequality by FAS, but not by SEC.The impact of comprehensive smoking bans may differ depending on the pre-ban level of exposure and the balance between sources of exposure i.e. public places v home.

98

Details Method Results CommentsSmoking restriction in cars, workplaces, schools and other public placesAuthor, yearGalan 2012

Age (years)15-16

Setting79 secondary schools, Madrid, Spain

Study designCross-sectional study

ObjectiveTo evaluate the relationship of contextual factors with smoking on school premises

SES variables usedSchool-level:Socioeconomic status of the census tracts in which the school is located, estimated from an index based on aggregate data (unemployment, temporary workers, manual workers, and low educational level among the overall adult [age >15 years] and young adult [age 16–29 years] populations).

Individual variables:Educational attainment of parents

Data sourcesSurveillance System of Risk Factors associated with non-communicable diseases targeting the adolescent population in the fourth year of compulsory secondary education in the Madrid region (15–16 years), 2004-2005.

Participant selectionTwo-stage cluster sampling with stratification in the selection of schools according to area and type (public or private). Overall response rate (schools and students) was 83.1%.

Participant characteristics15-16 year olds in fourth year of compulsory education, N=1179

Outcomes measuredProbability of smoking in school

Intervention detailsSurvey of smoking behaviour and individual and school-level contextual variables

General population impactAmong smokers, 50.6% had smoked on school premises during the last thirty days with significant variability (0% to 100%) between schools

Impact by SES variableModel with school-level and individual-level variables: a lower probability of smoking on school premises was found among adolescents whose fathers had a university education (OR 0.43, 95% CI: 0.19 to 0.96) or among those who did not know the level of studies of their father (OR 0.39, 95% CI: 0.16 to 0.94) compared with those with fathers who had a very low level of educational attainment.Adolescents with low academic achievement showed an OR of 1.51 (95% CI: 1.00–2.29).Employment status of either parent or educational level of mother was not significant. SES of the census tracts of the school was not significant, nor was written reference to smoking control policy or educational activities about smoking prevention. A lower probability of smoking on school premises was found for state subsidized private schools (odds ratio [OR]: 0.20; 95% CI: 0.11–0.35) and nonsubsidized private schools (OR: 0.30; 95% CI: 0.14–0.62) when

Internal validityThe self-completed questionnaire was filled out in the classroom in the presence of previously trained staff. Smoking variable previously validated but potential for response bias. Some subgroups are not sufficiently powered. SES measured by census tracts of the school rather than the location of the home of the student so may not be accurate.

External validityShould be representative and generalisable to other secondary schools in Spain and other similar countries. However, presence or absence of smoking policy did not include an evaluation of whether policy was implemented which limits applicability of this study.

Validity of author’s conclusionAgreed, census tract of school may not be sensitive to school-level SES

99

Details Method Results CommentsSmoking restriction in cars, workplaces, schools and other public placesEmployment status of parents

Study analysisMultilevel logistic regression models of smoking population (n=1,179=32.6% analyses on 1,116=94.7% of sample of smokers)

compared with that for public schools

Author’s conclusion of SES impactA higher probability of smoking on school premises was found among adolescents whose fathers had a lower level of educational attainment. However, at the contextual level, no relationship was found with socioeconomic status

100

Details Method Results CommentsSmoking restriction in cars, workplaces, schools and other public placesAuthor, yearMacKay 2010

Age (years)Less than 15 years: 0 to 4 and 5 to 14 years

SettingHospitals, Scotland

Study designRepeat cross-sectional (before and after)

Objectiveto determine whether the risk of a hospital admission for childhood asthma has changed since the introduction of comprehensive smoke-free legislation in Scotland.

SES variables usedarea deprivation score: quintiles 1 to 5 Index of Multiple Deprivation 2006

Study analysisnegative binomial regression

Data sourcesScottish Morbidity Record (SMR01) collects information on all admissions to acute care hospitals in Scotland, General Register Office for Scotland collects death-certificate data on all deaths that occur in Scotland. The admission and death databases are linked at an individual level so that records relating to the same person can be identified. Combined SMR01 and death-certificate data to identify all hospital admissions and deaths before arrival at the hospital that occurred from January 2000 through October 2009

Participant selectionEmergency hospital admissions (plus deaths occurring before arrival at hospital) for childhood (0-14 years) asthma, from January 2000 through October 2009

Participant characteristics21,415 hospital admissions: 11,796 (55.1%) occurred among preschool children and 9619 (44.9%) among school-age children.

Outcomes measuredadmission rate,adjusted admission rate

Intervention detailsNational smokefree legislation in Scotland March 2006

General population impactBefore the legislation was implemented, admissions for asthma were increasing at a mean rate of 5.2% per year (95% confidence interval [CI], 3.9 to 6.6). After implementation of the legislation, there was a reduction in the annual rate of 18.2% relative to the rate on March 26, 2006 (95% CI, 14.7 to 21.8; P<0.001), resulting in a net reduction in asthma admissions after implementation of the legislation of 13.0% per year (95% CI, 10.4 to 15.6).After adjustment for the potential confounding effects of sex, age group, urban or rural residence, and quintile of socioeconomic status, admissions for asthma before implementation of the legislation increased by a mean of 4.4% per year (95% CI, 3.3 to 5.5) relative to the rate in January 2000. After implementation of the legislation, there was a reduction of 19.5% (95% CI, 16.5 to 22.4; P<0.001) relative to the rate on March 26, 2006, resulting in a net reduction in admissions for asthma of 15.1% per year (95% CI, 12.9 to 17.2).The trends before the legislation varied according to age group, with a mean annual increase of 9.1%

Internal validityAlso accounts for deaths - the decrease in admissions was not due to an increase in the incidence of deaths before arrival at the hospital.

External validityComparable with English study on childhood asthma admissions.

Validity of author’s conclusionCannot determine the extent to which the observed reduction in asthma was due to reduced exposure to environmental tobacco smoke in the home, reduced exposure to environmental tobacco smoke in public places, or a reduction in active smoking among school-age children. Cannot rule out impact of change in asthma treatments.

101

Details Method Results CommentsSmoking restriction in cars, workplaces, schools and other public places

among preschool children, as compared with no significant change over time among school-age children. However, the change after legislation was similar in the two groups, with a reduction of 18.4% among preschool children and 20.8% among school-age children relative to the rate on March 26, 2006Very similar results based on admissions alone as few deaths.Impact by SES variableThere were no significant interactions between hospital admissions for asthma and quintile of SES. All SES subgroups associated with significant reduction in admissions.

Author’s conclusion of SES impactThe additional change after implementation of the legislation was significant in all subgroups

102

Details Method Results CommentsSmoking restriction in cars, workplaces, schools and other public placesAuthor, yearMillett 2013

Age (years)preschool (0–4 years) and school age (5–14 years)

SettingHospitals, England

Study designInterrupted time series (before and after legislation)

ObjectiveTo assess whether the implementation of English smokefree legislation in July 2007 was associated with a reduction in hospital admissions for childhood asthma.

SES variables usedarea deprivation score: quintiles 1 to 5 Index of Multiple Deprivation

Study analysisnegative binomial regression, multivariate

Data sourcesHospital Episode Statistics (HES): national administrative database for hospital activity in England

Participant selectionNonplanned (emergency) hospital admissionsfor childhood (0-14 years) asthma, from April 1, 2002 and November 30, 2010 (8.5 years)

Participant characteristics217 381 hospital admissions for childhood asthma, evenly distributed between preschool (50.1%) and school age children (49.9%). The number of admissions was higher in boys (63.4%) than girls (36.6%). Most admissions occurred in children living in urban locations (86.5%), and there were a higher number of admissions in children living in the most deprived areas.

Outcomes measuredAdmission rate,Adjusted admission rate ratio, (ratio of the actual admission rate in relation to the rate projected by the underlying trend)

Intervention detailsNational smokefree legislation in England July 2007

General population impactBefore the implementation of the legislation, the admission rate for childhood asthma was increasing by 2.2% per year (adjusted rate ratio 1.02; 95% confidence interval [CI]: 1.02–1.03). After implementation of the legislation, there was a significant immediate change in the admission rate of -8.9% (adjusted rate ratio 0.91; 95% CI: 0.89–0.93) and change in time trend of -3.4% per year (adjusted rate ratio 0.97; 95% CI: 0.96–0.98). Overall, the legislation was associated with a net 12.3% reduction of hospital admissions for childhood asthma in the first year. This change was equivalent to 6802 fewer hospital admissions in the first 3 years after implementation.

Impact by SES variableDuring the study period there were a higher number of admissions in children living in the most deprived areas.There were similar reductions in asthma admission rates among children from different SES groups.

Author’s conclusion of SES impactThe findings suggest immediate as well as cumulative benefits over time

Internal validityITS - estimates both the immediate change and change in time trend after policy implementation.Changes in diagnostic coding over the study period, may have underestimated the effect of smoke-free legislation if coding of childhood asthma admissions improved over the study period.

External validityComparable with Scottish study on childhood asthma admissions.

Validity of author’s conclusionOne of few studies to report longer-term outcomes for children but does not measure SHS exposure. Cannot rule out impact of change in asthma treatments.

103

Details Method Results CommentsSmoking restriction in cars, workplaces, schools and other public places

applying across SES.

104

Details Method Results CommentsSmoking restriction in cars, workplaces, schools and other public placesAuthor , yearMoore 2011

Age (years)11

SettingPrimary schools, Wales

Study designRepeat cross-sectional surveys of 10-11 year old children in same schools before and after legislation

ObjectiveTo assess socioeconomic patterning in changes in salivary cotinine concentrations, reports of parental smoking in the home and car and estimates of population-level smoking prevalence following introduction of smoke-free legislation

InterventionSmoke-free legislation in Wales

SES variablesFamily Affluence Scale (bedroom occupancy, car ownership, holidays, computer ownership)

Data sourcesCHETS Wales study

Participant selectionIn 2007, 1,611 pupils of an eligible 1,761 pupils within 75 schools completed the smoking questionnaire (91.5%), compared with 1,605 of an eligible 1,775 children within the same 75 schools in 2008 (90.4%). In total, 1,447 children pre-legislation (82.2% of those eligible) and 1,461 children post-legislation (82.3% of those eligible) from 71 schools provided useable saliva samples

Participant characteristicsMean age 11 years. Pre-legislation, 422 (27.1%), 606 (39.0%), and 527 (33.9%) of children were assigned to low-, medium-, and high-SES tertiles, respectively. Post-legislation, a slightly smaller proportion of children were assigned to the low-SES group (n = 360, 23.6%), with 621 (40.6%) and 547 (35.8%) assigned to medium- and high-SES groups, respectively.

OutcomesSalivary cotinine levelsParental smoking in the home

General population impactThere was no significant increase in inequality in the relative likelihood of a child’s sample containing a high level of cotinine (RRR = 1.03; 95% CI = 0.91–1.17).

Impact by SES variableThe likelihood of providing a sample containing an undetectable level of cotinine increased significantly after legislation among children from high [relative risk ratio (RRR) = 1.44, 95% CI = 1.04–2.00,p=0.03] and medium SES households (RRR = 1.66, 95% CI = 1.20–2.30, p<0.01), while exposure among children from lower SES households remained unchanged (RRR=0.93, 95% CI=0.62-1.40, p=0.72).

Parental smoking in the home, car-based SHS exposure, and perceived smoking prevalence were highest among children from low SES households. Parental smoking in the home and children’s estimates of adult smoking prevalence declined only among children from higher SES households.

Author’s conclusion of SES impactPost-legislation reductions in SHS exposure were limited to children from higher SES households. Children from lower SES households continue to have

Internal validityBiochemical measure of smoking. No significant differences between characteristics of pre- and post-legislation samples, nor were there significant differences between those providing useable saliva samples and those providing only questionnaire responses.Required imputation of random values for 47% of cases which limits reliability.

External validityGeneralisability limited by narrow age group and analyses restricted to children attending school and living with parents, a parent and step-parent or a single parent. Same linear analyses as CHETS Scotland (Akhtar 2010).Average cotinine concentrations among children in the Scottish CHETSwere substantially higher than in Wales (Holliday et al., 2009) and children’s SHS exposure outside of the home was perhaps greater in

105

Details Method Results CommentsSmoking restriction in cars, workplaces, schools and other public placesStudy analysisMultinomial logistic regression analysis accounting for clustering and adjusted for age, year and time of data collection.Analyses are limited to children living with parents, a parent and step-parent or a single parent, and who completed the FAS (smoking questionnaire n = 1,555/1,528; salivary cotinine n = 1,397/1,390 pre/post-legislation). Cotinine analyses are limited to children classified as non-smokers [i.e., who both reported being a non-smoker and provided saliva with a cotinine concentration <15 ng/ml (n = 1,362/1,364)].

Car-based SHS exposure

Intervention detailsQuestionnaire plus cotinine assay

high levels of exposure, particularly in homes and cars, and to perceive that smoking is the norm among adults.Children’s SHS exposure did not worsen for any SES subgroup after introduction of legislation in Wales. However, the unanticipated reductions in children’s SHS exposure following legislation appear limited to children from more affluent households in Wales, whose exposure was already significantly lower prior to legislation, leading to increased socioeconomic disparity.

Scotland, with impacts of legislation therefore greater overall than in Wales and distributed among all groups.

Validity of author’s conclusionThe impact of comprehensive smoking bans may differ depending on the pre-ban level of exposure and the balance between sources of exposure i.e. public places v home.

106

Details Method Results CommentsSmoking restriction in cars, workplaces, schools and other public placesAuthor , yearMoore 2012

Age (years)11.2

SettingPrimary schools, Scotland, Northern Ireland, Wales

Study designRepeat cross-sectional surveys of children in same primary schools before and after legislation

ObjectiveTo pool data from 3 countries in order to assess socioeconomic patterning in SHS exposure and parental restrictions on smoking in homes and cars before and after smokefree legislation

InterventionSmoke-free legislation in Scotland, Northern Ireland, Wales

SES variablesFamily Affluence Scale (bedroom occupancy, car ownership, holidays, computer ownership)

Data sourcesCHETS Scotland, Northern Ireland and Wales studies, questionnaire plus cotinine assay

Participant selectionOf 586 schools approached, 320/304 (54/51%) participated at baseline/follow-up.

Participant characteristics10 867 non-smokers (self-reported nonsmokers providing saliva samples containing <15 ng/ml cotinine) in their final year at 304 primary schools in Scotland (n = 111), Wales (n = 71) and NI (n = 122).

OutcomesSalivary cotinine levelsSmoking restrictions in the homeSmoking restrictions in the car

Intervention detailsNational smokefree legislation prohibiting smoking in enclosed public places andworkplaces (Scotland March 2006, Wales March 2007, Northern Ireland (NI) April 2007

General population impactRelative risk of children’s samples containing no detectable cotinine increased significantly following legislation. Percentages of children with undetectable concentrations increased from 31.0 (n = 1715) to 41.0% (n = 2251) following legislation overall, and from 20.1 to 34.2, 44.9 to 51.0 and from 38.6 to 42.9% in Scotland, Wales and NI, respectively. Relative risk of providing a sample containing a ‘high’ cotinine concentration also increased significantly.

Impact by SES variableRelative risk of children’s samples containing no detectable cotinine increased significantly as SES increased, whilst the relative risk of samples containing a ‘high’ cotinine concentration fell. These associations were almost identical in all countries, remaining significant after entry of terms for parental smoking and private smoking restrictions.This inequality appears to have widened following legislation (in the combined data set and trend in individual countries), with percentages of samples above the limit of detection ranging from 96.9 to 38.2% for the least and most affluent children, respectively, after legislation. Gradients for higher exposure levels remained relatively unchanged.

Internal validityBiochemical measure of smoking. Children reported on smoking restrictions in homes and cars.SES varied significantly between survey years (affluence higher at follow-up).

External validityGeneralisability limited by narrow age group and analyses restricted to children attending school and living with parents, a parent and step-parent or a single parent. However pools data from 3 CHETS studies.

Validity of author’s conclusionValid. Impact may differ between individual countries because baseline cotinine concentrations differed between countries. Difficult to compare results by SES pertaining to individual countries with other CHETS papers because analyses are different.

107

Details Method Results CommentsSmoking restriction in cars, workplaces, schools and other public places

Study analysisMultinomial logistic regression analysis accounting for clustering and adjusted for age and country. Binomial logistic regression for car-based smoking.

In all countries, and the combined data set, as SES increased, the likelihood of partial or no home smoking restrictions (rather than full smoking restrictions), decreased significantly, whilst the odds of smoking being allowed inside the family car also decreased significantly. These trends remained after adjustment for parental smoking No change in inequality following legislation for home and car-based smoking restrictions (socioeconomic patterning remained stable).

Author’s conclusion of SES impactSocioeconomic inequality in the likelihood of a child’s sample containing detectable traces of cotinine increased. Hence, declines in exposure occurred predominantly among children with low exposure before legislation, and from more affluent families. Substantial socioeconomic gradients in proportions of children with higher SHS exposure levels remained unchanged. Post-legislation changes in smoking restrictions in cars or homes were not patterned by socioeconomic status.

108

Details Method Results CommentsSmoking restriction in cars, workplaces, schools and other public placesAuthor , yearNabi-Burza 2012

Age (years)26% (n=214) aged less than 1 year;35% (n=288) aged 1 to 4 years;19% (n=158) aged 5 to 9 years);18% (n=147) aged 10 years or over

SettingPaediatric clinics in 8 US states

Study designCross-sectional study

ObjectiveTo determine prevalence and factors associated with strictly enforced smoke-free car policies among smoking parents.

SES variableeducation (high schoolor less versus some college or college

Data sourcesBaseline data collected at paediatric practices enrolled in the control arm of a cluster, randomized controlled trial, Clinical Effort Against Secondhand Smoke Exposure.

Participant selectionParticipants were eligible to enrol in the study if they had accompanied a child to the office visit, had smoked at least a puff of a cigarette in the past 7 days, were the parent or legal guardian of the child seen that day, were at least 18 years old, and spoke English. Enrolled parents received $5 in cash for completing the baseline enrolment survey.Screening continued until 100 eligible parents were enrolled at each practice.

Participant characteristics817/981 parents reported having a car. The majority (70%) of the parents were in the age group 25 to 44 years, 77% were females, mostly mothers (98% vs 2% legal guardians), and 68% were non-Hispanic whites. Many parents (42%) had only a high school degree, and 16% had completed college. Most of the children (60%) were covered by Medicaid

OutcomesSmokefree car policy

Intervention details

General population impactOf 795 parents, 73% reported that someone had smoked in their car in the past 3 months. Less than 1 in 3 parents who had a smoke-free car policy reported that it was violated in the past 3 months. Of the 562 parents who did not report having a smoke-free car policy, 48% reported that smoking occurred with children present in the car. Approximately one-fifth of all enrolled parents reported being asked by a paediatric health care provider about their smoking status. Only 14% of smoking parents reported being asked if they had a smoke-free car, and 12% reported being advised to have a smoke-free car policy by a paediatric health care provider.

Impact by SES variableNo association between parent’s age, race and ethnicity, education, and intention to quit smoking with having a strictly enforced smokefree car policy.Exploratory analyses assessed possible interactions between the 4 parent demographic variables (age, gender, race, and education) and the 3 significant predictors of car policy (child’s age, number of

Internal validityUnable to ascertain how representative the study sample was. Self-reported outcome data.

External validitySample excludes non-car owners. Sample is derived from 8 US states.

Validity of author’s conclusionEducated was not significantly associated with smokefree car policy on its own, only significant in interaction with child age and amount smoked.

109

Details Method Results CommentsSmoking restriction in cars, workplaces, schools and other public placesgraduates)

Study analysisLogistic regression

Questionnaire of smoking behaviour in cars and home

cigarettes smoked per day by the parent, and having another smoker at home). Parent gender and education interactedwith child’s age: parents of children aged <1 year were more likely to have strict smoke-free car policies if they were female (OR: 3.00 [95% CI: 1.22–7.38], P = .016) or college educated (OR:2.42 [95% CI: 1.21–4.83], P = .013). Strict smoke-free car policies were more common when parents were both light smokers (smoked 10 or less cigarettes per day) and college educated (OR: 2.88 [95% CI: 1.24–6.66], P = .013).

Author’s conclusion of SES impactCollege educated parents of children aged <1 year were more likely to have strict smoke-free car policies.

110

Details Method Results CommentsSmoking restriction in cars, workplaces, schools and other public placesAuthor , yearNoach 2012

Age (years)15

SettingSchools, Israel

Study designCross-sectional study

ObjectiveTo examine determinant of SHS exposure in Israeli adolescents

SES variableMaternal and paternal education (<12 years, 12 years, College or University, other degree)

Study analysisLogistic regression models

Data sourcesIsrael National Health and Nutrition Youth survey among 7-12th grade, 2003-2004

Participant selectionIsrael Ministry of Education provided list of approximately 1,000 schools from the state-sponsored educational system. Sample stratified by population group (Jewish, Arab + Bedouin, Druze), stream (state, state religious), school level (7–9, 10–12), and SES (defined by the Ministry of Education as related to the school, as low, high). For each school chosen as a primary sampling unit, the grade level and then the class within each grade level were randomly selected. Response rates were high (school: 91.8%, child: 87.9%), with 6,274 participants.

Participant characteristicsAverage age was 15 years (11–19 years). N=6,274 students: 55.7% girls, 44.3% boys; 70% Jews, 30% non-Jews (18% Moslem Arab, 4% Christian Arab, 6% Druze, and 1% Christian).

OutcomesCorrelates of exposure to SHS at home, school, entertainment, ‘other’ places

Intervention detailsSurvey

General population impactMost Israeli adolescents were exposed to SHS (total: 85.6%; home: 40%; school: 31.4%; entertainment: 73.3%; other: 16.3%).

Impact by SES variableParental education is not a significant determinant of smoking in schoolHome:Teenagers whose fathers had less than 12 years of education (OR = 1.48; CI: 1.09, 1.99; p = .0111) were more exposed than were teenagers whose fathers had a degree from a university or college. Teenagers with less-educated mothers (OR = 1.39; CI: 1.02, 1.90; p = .0366) were more exposed than teenagers with mothers with degrees from a university or college.

Author’s conclusion of SES impactThe high levels of SHS exposure among Israeli adolescents were characterized by different patterns of exposure among different population groups

Internal validityThe smoking question in the survey has not been validated by biochemical measures in Israel. Main aim of the survey was to assess nutritional status and so smoking question is basic yes/no.

External validitySchools from the ultraorthodox Jewish, independent and private sectors were excluded, as were boarding schools.Israel is heterogeneous with broad range of ethnic, religious and socioeconomic populations and is not generalisable to other WHO European or stage 4 countries.

Validity of author’s conclusionNo comprehensive smokefree bans at time of survey so survey is not assessing impact of specific policy implementation but lending support to National Tobacco Control plan recently approved by Israeli government

111

Details Method Results CommentsSmoking restriction in cars, workplaces, schools and other public placesAuthor, yearWoodruff 2000

Age (years)19

SettingRecruit Training Command, US

Study designBefore and after experimental study

ObjectiveTo examine the effect of a US Navy smoking ban in female recruits

InterventionOrganisational smoking ban for 8 weeks, unique as 24-hour ban with ‘live-in’ recruits

SES variableEducation (less than a high schooleducation, high school, and greater than a high school education);

Study analysisPredictors of changes in perceptions of being a smoker: stepwise logistic regression to determine the

Data sourcesIntervention study

Participant selection5505/5197=93% response amongst recruits who volunteered to take part;86% response rate among 5129 eligibles, n=4411/5129, ever-smokers at entry=2820/4411, 39% response rate for 3 month follow-up, n=1077/2748 (72 left Navy before follow-up); volunteers entering Recruit Training Command Illinois March 1996-March 1997

Participant characteristicsAll female. The mean (SD) age was 19 (2.75) years. The majority (94.5%) had at least a high school education. Recruits were ethnically diverse, with 42% belonging to ethnic groups other than white non-Hispanic.

OutcomesPerceived smoking statusSmoking relapse

Intervention details8-week 24-hour smoking ban

General population impactAmong the 4393 recruits who provided entry and graduation survey data, 41.4% (n = 1819) reported being smokers at entry (that is, reported any smoking in the 30 days before entering). Twenty five per cent (n = 1110) of all women recruits reported being a smoker at graduation, a significant reduction from the 41% smoking rate at entry into RTC (McNemar ÷2 = 665.7, p < 0.001).Slightly over two thirds (n = 724) of “smokers” who responded to the follow up survey had resumed smoking three months after graduation, and 32% (n = 340) reported not smoking. Among past month smokers at entry to RTC, the relapse rate at the three month follow up was 81%.Daily smokers at entry had the highest relapse rate (89%)=11% follow-up cessation rate

Impact by SES variableEducation did not significantly predict relapse

Author’s conclusion of SES impactNone- Study did not aim to assess differential impact by SES

Internal validityResponse bias is present; low response rate at 3 month follow-up, nonrespondents had a slightly higher past 30 day smoking rate at baseline than did respondents.Definition of ‘smoker’ differed at graduation (post 8 weeks) from baseline and 3 month follow-up. Group of smokers assessed for relapse was broadly defined and included daily smokers, occasional smokers, experimenters, or former smokers.

External validityNot generalizable to civilian population or setting. All female.

Validity of author’s conclusionStudy did not aim to assess differential impact by SES.

112

independent correlates of graduation smoking status.Predictors of relapse at 3 month follow-up: multivariate logistic analysis

113

Details Method Results CommentsControls on advertising, promotion and marketing of tobaccoAuthor, yearGilpin & Pierce 1997

Age (years)14-21 in 1979 to 1989

SettingUS

Study designCross-sectional study

ObjectiveTo investigate the possible association between increased tobacco marketing and increased smoking initiation by adolescents

InterventionExamines trends in smoking initiation by inflation-adjusted cigarette prices and tobacco industry budget for marketing

SES variables usedEducation (less than 12 years, 12 years with no college education, more than 12 years with some college education as adults)

Study analysis

Data sourcesCombined data from 3 Current Population Surveys (September 1993, January 1993, May 1993) that contained special supplement on tobacco use. One quarter=in person interviews and ¾ = telephone interviews. Tobacco Institute for weighted average pack prices, US Federal Trade Commission for marketing expenditure; adjusted to 1989 dollars using Consumer Price Index

Participant selectionCivilian non-institutionalised population aged 15 years+, surveyed about 56,000 households per month

Participant characteristicsAnalysis restricted to respondents 17-38(n=140,975) that would have been 14-21 in 1979 to 1989

Outcomes measuredInitiation rates by education (rate calculated as number in an age group who reported starting smoking regularly in a year, divided by number of never-smokers at start of the year)

General population impact1979 to 1984 adolescent initiation rates decreased but increased thereafter

Impact by SES variableInitiation rates highest among high school dropouts and lowest amongst those who eventually attended college. Only quadratic model significant for dropouts (p=0.035). Neither model was significant for high-school graduates and both models were significant for ‘some college’ (p=0.081 linear, p=0.014 quadratic).In 1988 initiation rate was 9.9% for those who did not graduate from high school, 6.9% for high-school graduates reporting no college and 3.7% for those reporting at least some college.

Author’s conclusion of SES impactMarketing expenditure may be associated with an increase in smoking initiation especially in young people with lower levels of education.

Internal validityRespondents are asked about how old they were when they started smoking (retrospective so potential for recall bias and underreporting)

External validityInitiation rates are for decade 1979 to 1989 so relatively older study which limits its generalisability to current youth

Validity of author’s conclusionTentative because study links overall initiation rates by marketing budget but doesn’t assess marketing budget impact on initiation rates by education level

114

Details Method Results CommentsControls on advertising, promotion and marketing of tobaccoLinear or quadratic models fitted to initiation rates

Intervention detailsExamines trends in smoking initiation by inflation-adjusted cigarette prices and tobacco industry budget for marketing

115

Details Method Results CommentsControls on advertising, promotion and marketing of tobaccoAuthor, yearHammond 2011

Age (years)18-19

SettingUS

Study designRandomised controlled trial

ObjectiveTo examine the impact of cigarette pack design among young women

InterventionShort online survey intervention looking at brand appeal of tobacco packaging

SES variablesEducation level categorized as “low ” (grade school or some high school), “ medium ” (high school, technical school , or community college), and “ high ” (university).Income

Study analysis

Data sourcesOnline survey of 18-19 year old women in US in February 2010

Participant selectionConvenience sample of 826 female smokers and non-smokers aged 18-19 years, recruited via email from a consumer panel through Global Market Insite Inc. (panel reach 2.8 million) participants received approximately $2USD for completing the survey. Randomised to four experimental conditions after ascertaining smoking status

Participant characteristics18-19 year old women, education varied by condition, with the highest level of education in the standard condition ( χ 2 = 18.0, p = .04),

Outcomes measuredPacks rated by participants on measures of appeal and health risk, also behavioural pack selection task

Intervention detailsOnline survey intervention. Participants viewed eight cigarette packages, one at a time, displayed in a random order. Packages were

General population impactFully branded female packs were rated significantly more appealing than the same packs without descriptors, “plain” packs, and non – female- branded packs. Female- branded packs were associated with a greater number of positive attributes including glamour, slimness, and attractiveness and were more likely to be perceived as less harmful. Approximately 40% of smokers and non-smokers requested a pack at the end of the study; female- branded packs were 3 times more likely to be selected than plain packs.

Impact by SES variableParticipants in the high- income (B = 0 .11, p = .004) and high education (B = 0 .08, p = .05) categories endorsed a greater number of positive smoker traits (female/male, glamorous/not glamorous, cool/not cool, popular/not popular,attractive/unattractive, slim/overweight, and sophisticated/not sophisticated) than those in the low- income and low education categories.High- income respondents were more likely to endorse smoking and weight control beliefs compared with respondents reporting low ( OR = 1.70, 95% CI = 1.12 – 2.60) and medium income ( OR = 1.73, 95% CI = 1.09 – 2.73) and those who did not state their income ( OR = 2.17, 95% CI

Internal validityReports significant sociodemographic predictors only, convenience sampleEducation varied by condition,with the highest level of education in the standard condition ( χ 2 = 18.0, p = .04), and number of smoked cigarettes per day was significantly higher in the plain condition ( M =10.6) compared with the standard condition ( M = 7.7, B = −0.14, p = .046) among current smokers

External validityYoung women only limit generalisability.

Validity of author’s conclusionThe reactions to/perceptions of the different types of packs was the same by SES for nearly all the measures. Thus, very tentatively, plain packaging might have a neutral equity effect for young women. Equity impact was not a main aim of the study.

116

Details Method Results CommentsControls on advertising, promotion and marketing of tobaccoRegression models were used to examine the effect of experimental condition for three primary outcomes: pack ratings, smoker image ratings, and beliefs about smoking. For each outcome, regression models were conducted in two steps. In Step 1, the model included only the “condition” variable. In Step 2 of the model, the following variables were entered as covariates: age, education, income, ethnicity, smoking status, and weight concerns. In Step 3, all two-way interactions with the “condition” variable were tested by entering each interaction term into the model one at a time.

displayed according to eachof the four experimental conditions:( 1) female-oriented packages (standard condition); ( 2) female-oriented packages with brand imagery, including colours and graphics, but with descriptors (i.e., slims) removed; ( 3) female-oriented packages without brand imageryand descriptors (i.e., plain packages); and (4) popular U.S. brands of “ regular ” or non – female- oriented packages

= 1.29 – 3.65).No significant differences in pack selection were observed for smoking status, age, income, education, ethnicity, or weight concerns

Author’s conclusion of SES impactNot stated

117

Details Method Results CommentsControls on advertising, promotion and marketing of tobaccoAuthor , yearPucci 1998

Age (years)5-14, 15-19

SettingBoston, US

Study designCross-sectional study

ObjectiveTo determine the prevalence, type and proximity to public schools of all stationary outdoor tobacco advertising in 6 Boston neighbourhoods

InterventionYouth exposure to tobacco advertising density within FDA 1,000 foot buffer zones around schools

SES variableNeighbourhoods defined by median income per household from Boston Neighborhood Health StatusReport

Data sourcesfield survey using single observations in 1996. Six Boston neighbourhoods—two with the highest, two with middle, and two with the lowest median household incomes—were selected. July to August 1996, four observer teams (one adult and two to three youth), recruited from Boston summer youth-employment programs, participated in 2-day training. Observations made by the teams were validated by randomly selecting 8 sites. An independent observer who had attended the team training conducted follow- up observations within a week of the original observations.

Participant selectionThe neighbourhoods, as defined in the Boston Neighborhood Health Status Report, are (from highest to lowest median income) Beacon Hill ($38,816), West Roxbury, Mattapan, North End, East Boston, Roxbury ($19,351).

Participant characteristics580 advertising units at the 94 sites

Outcomes

General population impactThe greatest number of sites for any neighbourhood was 22 in Roxbury, with Mattapan (21) second and EastBoston (16) third. These three neighbourhoods also shared the top three positions for number of units: Mattapan (169), Roxbury (124), and East Boston (113).The overall advertising density for schools in all neighbourhoods combined was higher for middle (10.1) and high schools (9.9) than for elementary schools (6.3)

Impact by SES variableEast Boston and Roxbury, the two neighbourhoods with the lowest median incomes, showed the highest number of advertising sites inside the buffer zones, 16 and 18, respectively

Author’s conclusion of SES impactThe majority of outdoor tobacco advertisingwas in the neighbourhoods with the lowest median household incomes

Internal validityUses actual observations of tobacco density and links to buffer zones. However study does not include point-of-purchase advertising, advertising inside stores that is seen from the street, or advertising on taxis and buses.

External validityUnable to assess generalisability of these 6 Boston neighbourhoods as no details provided.

Validity of author’s conclusionValid but probably underestimates density.

118

Details Method Results CommentsControls on advertising, promotion and marketing of tobaccoStudy analysisAdvertising sites plotted using MapInfo, density calculated by dividing number of advertising units by area of buffer zone

Advertising density by school level and neighbourhood

Intervention detailsObservational survey identifying advertising sites

119

Details Method Results CommentsMass media campaignsAuthor , yearVallone 2009

Age (years)12-17

SettingUS

Study designCross-sectional study

ObjectiveTo determine whether SES is associated with awareness of and receptivity to the truth® campaign among youth aged 12–17.

InterventionThe truth® campaign is a branded countermarketing campaign

SES variableMedian household income and median household education at the zip code level.

Study analysisThis receptivity analysis using bivariable and multivariable analyses is limited to participants who

Data sourcesSeven waves of Legacy Media Tracking Survey data (telephone survey), collected from September 2000 through January 2004. The LMTS was developed to track awareness of, and receptivity to, American Legacy Foundation’s truth® campaign.

Participant selectionResponse rates for the LMTS ranged from 60% to 30% per wave, with a general pattern of decline over time.The samples for each survey wave were generated by a combination of random digit dial (RDD) and supplementary lists. Listed sample was used to achieve target numbers within geographic and racial/ethnic populations. African American, Hispanic and Asian youth and young adults were oversampled in each survey wave in an effort to generate sufficient sample sizes among racial/ethnic groups. To evaluatethe campaign, oversamples were also drawn in some survey waves from within three sentinel sites, from within states which had strong tobacco countermarketing campaigns, and those with variation across truth® gross ratings points

General population impactN/A

Impact by SES variableYouth who lived in zip codes in which the median household income was less than or equal to US$ 35,000 had a lower level of confirmed awareness than respondents in each of the other income categories (p < 0.05). There were no statistically significant differences in confirmed awareness by median level of education, though there was a pattern in which the proportion of confirmed awareness increased with education. Similarly, there were no differences in receptivity by median household income or median household education, though there was a pattern of increasing receptivity with greater income and education

Author’s conclusion of SES impactFrom 2000 to 2004, both female and male youth living in lower education zip codes had lower odds of having confirmed awareness of truth® as compared with youth living in more highly educated zip codes. Zip code level median household income was not associated with confirmed awareness. However, there were no differences in receptivity to the campaign by zip code level income or education.

Internal validityAppends SES proxy measures (zip codes) to data as LMTS did not measure SES.Intervention methods differed over time: proportion of campaign broadcast on cable increased over time.Survey developed specifically for this campaign but repeated over 7 waves and 4 years.

External validityGeneralisability may be limited due to response rates; which ranged from 60% to30% per wave, with a general pattern of decline over time. An examination of the sample demographics across the seven survey waves indicates that there are some statistically significant differences across waves over time; however, further analyses revealed no systematic bias with regard to demographic characteristics by response rate.Could such a huge, lengthy and expensive campaign be applied outside the US? Only national organisations are likely to run similar mass-media

120

Details Method Results CommentsMass media campaignsdemonstrated confirmed awareness of the campaign. Zip codes were appended to the datafiles by one of two means: (1) for the listed sample, zip codes were linked to street addresses; (2) for the RDD sample, the most probable zip code associated with that telephone exchange was selected.

(GRPs).

Participant characteristics30,512, including 15,335 female and 15,177 male respondents.By age group, 51.5% age 12–14 and 48.5% age 15–17. 57.19% identified as white, 20.0% as Hispanic, 15.0% as African American and 7.9% as Asian American. Most of the sample had never smoked (76.3%); however, 16.1% were former smokers and7.6% were current smokers. Youth watched a mean of 3.3 h of TVper day, and 80.9% had cable access. The median household income distribution by zip code was as follows: 25.0% of respondents lived in zip codes in which the median household income was less than or equal to US$ 35K per year; 26.0% lived in US$ 35–45K zip codes, 24.2% lived in US$ 45–60K zip codes and 24.9% lived in wealthier zip codes. The median household education distribution by zip code was as follows: 17.8% ofrespondents lived in zip codes in which the median household education was less than or equal to 12 years; 39.7% lived in zip codes inwhich the median household education was 13 years, 30.5% lived

campaigns due to prohibitive cost

Validity of author’s conclusionValid but awareness and receptivity do not inform us of changes in smoking behaviour. Difference in results between awareness according to income or education, and between awareness and receptivity outcomes, may indicate measurement issues of how or what study is measuring?

121

Details Method Results CommentsMass media campaigns

in zip codes in which the median household education was 14 years and 12.0% lived in more educated zip codes.

OutcomesConfirmed awareness and receptivity

Intervention detailsThe truth® campaign is a branded countermarketing campaigndesigned to prevent smoking among at-risk youth, primarily through edgy television advertisements with an anti-tobacco industry theme

122

Details Method Results CommentsIncreases in price/tax of tobacco productsAuthor, yearBiener 1998

Age (years)12-17

SettingMassachussetts, US

Study designCross-sectional study

ObjectiveExamines smokers perceptions of the impact of new tobacco taxes

InterventionStatewide (Massachusetts) tobacco control programme

SES variables usedhousehold income is dichotomised at the median, for teenagers=$50,000/y, obtained from the report of an adult household resident

Study analysisMultinomial logistic regression using bivariate and multivariate models

Data sourcesTelephone interviews from sample from random-digit-dialling, 1993-1994

Participant selectionScreening interviews for 78% of sampled households (random-digit dialling), 75% response rate for youths=1606 youth interviews

Participant characteristicsAnalysis restricted to 216 current teenage smokers who reported having ever bought cigarettes

Outcomes measuredSmoking behaviour

Intervention detailsSurvey retrospectively assessing reactions to 1993 tax increase

General population impact26% (10.4,42.0) cut costs, 21% (9.3,31.9) considered quitting, 53% (36.8, 69.6) no response

Impact by SES variableTeenaged smokers from low income households were much more likely to cut costs of their smoking in response to the price increase, rather than do nothing (OR 7.57, 95%CI 1.55,36.98) or cut costs rather than consider quitting (OR 14.72, 95%CI 2.55,84.95), household income was unrelated to the choice between considering quitting and doing nothing (OR 0.51, 95% CI 0.13,2.77), these significant bivariate effects are still significant in multivariate model

Author’s conclusion of SES impactLow-income teenagers more likely than more affluent teens to cut costs by cutting down on smoking or (less often) by switching to cheaper brands. Young low-income smokers were not more likely than wealthier teenagers to consider quitting

Internal validity53% of the teenagers who continued to smoke denied having had any of the 3 potential reactions to price increase. Analysis restricted to small sample.

External validityNo further details of teenager demographics although reports that age and sex not significantly related to reported response to price increase

Validity of author’s conclusionPossible that study failed to measure an important variable.

123

124

Details Method Results CommentsIncreases in price/tax of tobacco productsAuthor, yearGilpin & Pierce 1997

Age (years)14-21 in 1979 to 1989

SettingUS

Study designCross-sectional study

ObjectiveTo investigate the possible association between increased tobacco marketing and increased smoking initiation by adolescents

InterventionExamines trends in smoking initiation by inflation-adjusted cigarette prices and tobacco industry budget for marketing

SES variables usedEducation (less than 12 years, 12 years with no college education, more than 12 years with some college education as adults)

Study analysis

Data sourcesCombined data from 3 Current Population Surveys (September 1993, January 1993, May 1993) that contained special supplement on tobacco use. One quarter=in person interviews and ¾ telephone interviews. Tobacco Institute for weighted average pack prices, US Federal Trade Commission for marketing expenditure; adjusted to 1989 dollars using Consumer Price Index

Participant selectionCivilian non-institutionalised population aged 15 years+, surveyed about 56,000 households per month

Participant characteristicsAnalysis restricted to respondents 17-38(n=140,975) that would have been 14-21 in 1979 to 1989

Outcomes measuredInitiation rates by education (rate calculated as number in an age group who reported starting smoking regularly in a year, divided by number of never-smokers at start of the year)

General population impact1979 to 1984 adolescent initiation rates decreased but increased thereafter

Impact by SES variableInitiation rates highest among high school dropouts and lowest amongst those who eventually attended college. Only quadratic model significant for dropouts (p=0.035). Neither model was significant for high-school graduates nor both models were significant for ‘some college’ (p=0.081 linear, p=0.014 quadratic).In 1988 initiation rate was 9.9% for those who did not graduate from high school, 6.9% for high-school graduates reporting no college and 3.7% for those reporting at least some college.

Author’s conclusion of SES impactIncrease in cigarette taxes did not reduce smoking initiation rates.

Internal validityRespondents are asked about how old they were when they started smoking (retrospective so potential for recall bias and underreporting)

External validityInitiation rates are for decade 1979 to 1989 so relatively older study which limits its generalisability to current youth

Validity of author’s conclusionTentative because study links overall initiation rates by marketing budget but doesn’t assess marketing budget impact on initiation rates by education level

125

Linear or quadratic models fitted to initiation rates

Intervention detailsExamines trends in smoking initiation by inflation-adjusted cigarette prices and tobacco industry budget for marketing

126

Details Method Results CommentsIncreases in price/tax of tobacco productsAuthor, yearGlied 2002

Age (years)14-23 in 1979

SettingUS

Study designProspective longitudinal cohort study with cross-sectional analysis (econometric)

ObjectiveTo test the assumption thatpolicies targeting youth toreduce smoking initiation willreduce lifetime smokingpropensities

InterventionEstimates the effect of cigarette taxes at age 14 on future overall smoking behaviour, quitting and initiation

SES variables usedFamily income in 1979 below the sample median (about $12,000 in

Data sourcesSmoking data from the National Longitudinal Survey of Youth (1979, 84, 92 and 94). Cigarette tax rates and tax policies from the Tobacco Institute 1996Participant selectionNo details

Participant characteristicsN=7,605; Sixty percenthad tried cigarettes by age 16, mean (SD): age 17.5 (2.2), age began smoking 13.6 (3.4); 53% female; 30% black; 18% Hispanic; mean (SD) family income in 1979 $18,270 ($11,747)

Outcomes measuredSmoking participation, quitting, initiation

Intervention detailsAssesses relationship (price elasticity) between tax and smoking behaviour over time and across time

General population impactLongitudinal data: Taxes at age 14 had a significant negative impact on later smoking behaviour (elasticity -0.66, p<0.05) although this effect reduced over time. This result was confirmed by the fixed effect analysis.Cross-sectional data: Taxes at age 14 had a significant negative impact on current smoking at ages 19-28 (elasticity -0.96, p<0.01) and late initiation (p<0.10), but no effect on quitting.

Impact by SES variableLongitudinal dataLow income (< $12,000 median in 1979)-0.65, p<0.10 (at age 14)-0.33 (at age 24)-0.01 (at age 34)0.15 (at age 39)Tax at age 14 had a statistically significant negative effect on current smoking overall, for low income people.Elasticities declined over time for low income people indicating that by age 39 the effect of taxes at age 14 has largely disappeared.Cross-sectional dataCurrent smoking at age 19 to 28-1.00, p<0.05 (low income)Taxes at age 14 had most effect on low income people at ages 19-28 although this

Internal validityLongitudinal data and cross-sectional data appear to be similar and similar for general population and for low-income population

External validityNot clear if National Longitudinal Survey of Youth was representative as minorities were oversampled, and the analysis was restricted to only those surveyed in 1979, 84, 92 and 94.

Validity of author’s conclusionValid.

127

Details Method Results CommentsIncreases in price/tax of tobacco products1979)

Study analysisModel: using longitudinal data: (1) probit model including the effects of time and how taxes change over time, with adjustment for clustering within an individual; (2) ordinary least squares regression using individual fixed effects with an interaction term between tax at and time since age 14.Using cross-sectional data (analysing 1984, 92 and 94 separately) to estimate the effect of taxes at age 14 on overall smoking behaviour, quitting and initiation.

reduced and was no longer significant in later years.QuittingTaxes at age 14 had a positive but not significant effect on quitting by the age of 27 to 37 for low income people.Late initiation (starting after age 16)Taxes at age 14 did not have a significant effect on late initiation for low income people.

Author’s conclusion of SES impactThese results suggest that reducing smoking among teens through tax policy may not be sufficient to substantially reduce smoking in adulthood

128

Details Method Results CommentsIncreases in price/tax of tobacco productsAuthor, yearGruber 2000

Age (years)13-18

SettingUS

Study designCross-sectional study (econometric)

ObjectiveTo provide a comprehensiveanalysis of the impact of pricesand other public policies onyouth smoking in the 1990s

InterventionState-level measures of prices,clean air regulations and youthaccess restrictions

SES variableParental education

Study analysisEconometric analysisModel: linear regression models with standard errors corrected for within

Data sourcesMonitoring the Future (MTF: University of Michigan) providing smoking behaviour, race, age, sex and state data for 8th, 10th, 12th graders (1991-97); Youth Behaviour Risk Survey (YBRS) data (CDC) for 1991, 3, 5, and 7 for 9th-12th graders; Vital Statistics Natality Detail Files from 1989 onwards providing smoking behaviour of women during pregnancy.

Participant selection

Participant characteristicsNumber=641,759 (MTF); 106,556 (YBRS); 3,970 (Natality, aged 13-18)

OutcomesSmoking participationSmoking intensity

Intervention detailsEconometric analysis using repeated cross-sectional data

General population impactN/A – all results stratified by ageThere is no public policy (clean air or access) variable other than price which is significant for either age group in all three data sets, or even in both the data sets representing the full teen population (MTF and YRBS).

Impact by SES variableParental education (YRBS data only)For seniors the elasticity of participation was -4.39* for those whose parents were high school dropouts or graduates and -.24 for parents with some college education. For smoking intensity this trend was reversed with elasticities of -0.40 for high school and -2.39* for college education. There was no pattern for younger teenagers, although participation elasticity was positive and statistically significant for high school educated parents (2.72*). [* p<0.05]

Author’s conclusion of SES impactThese results suggest that the single greatest policy determinant of youth smoking is the price of cigarettes.Older teenagers are more sensitive to prices with a central elasticity estimate of -0.67. This price sensitivity rises for moresocioeconomically disadvantaged

Internal validityParental education is used as a proxy for income.

External validityNo information on high-school dropouts who may be differentially price sensitive. State by year dropout rates were controlled for in regression analysis which suggests no selection bias due to dropout related to tax.

Validity of author’s conclusionIn cross-sectional data it is impossible to disentangle price and policy impacts from other underlying long-run determinants of smoking attitudes. Sensitivity to price suggests cross-elasticity between price and income

129

Details Method Results CommentsIncreases in price/tax of tobacco productsstate-year correlation (to account forvariation across states and years). Separate models built for each dataset. (MTF, YBRS, Natality)Outcome variables: smoking participation (any smokingover past months); conditional intensity (quantity smoked)Explanatory variables: price per pack (including taxes); clean air regulations (private workplace, public workplace, restaurants, schools, other e.g. public transport); youthaccess index (score across 9 categories including minimum purchase age, vending machine availability, which is added to create a total index with high scores indicating more restrictions); state and year (as fixed effects to account for between state and between year price differences).

groups such as blacks or those with less educated parents.

130

Details Method Result CommentsIncreases in price/tax of tobacco productsAuthor, yearMadden, 2007

Age19 (when started smoking)

CountryIrelandDesignSingle cross-sectional survey containing retrospective cohort dataObjectiveTo investigate the role of tobacco taxes in starting and quitting smoking and explores how tax effect differs by educationSES variablesEducation (primary, junior (age 16), secondary (age 18), University)AnalysesDuration analyses – various parametric duration models

Data sourcesRetrospective data from a survey on women’s knowledge, understanding and awareness of lifetime health needs (Saffron Survey, 1998).Participant selectionAll survey respondents who were born after 1950 (so that sample’s exposure matches price data).Participant characteristicsN=703. Average age 34, ex-smokers slightly older. 10% primary education, 27% junior education, 40% secondary, 21% university. Ever-smokers and current smokers more likely to have lower levels of education. 55% employment rate, 47.5% among current smokers.InterventionTobacco taxation from 1960 onwards.Length of study1960 to 1998OutcomesEver smoked, age of initiation, and cessation.

General populationHigher tax levels are associated with later initiation and earlier cessation.SESTaxation has a stronger effect to prevent or delay initiation among those with intermediate education, and weakest among those with the lowest education.Taxation has the strongest effect on cessation among those with the lowest education, an equal impact on those with other levels of education.Author’s conclusion of SES impactResults are extremely tentative, but it appears that the greater impact is among those with intermediate education. Greatest effect on quitting for the lowest levels of education.

Internal validityPotential for recall bias, going back up to 40 years in some cases.Doesn’t capture failed attempts to quit.External validityRevenue Commissioners does not break down the tax component into excise and VAT for the period up to 1973. Thus, authors have taken the total tax component of the retail price and deflated it by the personal consumption deflator to arrive at a real tax on tobacco.Tax was relatively low through the study period, unclear whether the relationship would continue with further increases from current levels of taxation.Potential quitters had less cessation support available.Only covers Irish females.Covers a period of increasing awareness of the impact of smoking, unclear whether cessation was linked to taxation or increased awareness.Validity of author’s conclusionResults are extremely tentative

131

Details Method Results CommentsIncreases in price/tax of tobacco productsAuthor , yearPerretti-Watel 2010

Age (years)19.5

SettingSouth-Eastern France

Study designCross-sectional study

ObjectiveTo investigate young smokers retrospective reactions to an increase in cigarette prices

InterventionPrice increase on tobacco

SES variableStudents report of parental education

Study analysisRestricted to 427 daily smokers,

Data sourcesSurvey on Provencal Students’ Health conducted by The South-eastern Health Regional Observatory between November 2005 and June 2006 (unpublished). random sample of 2455 students stratified by university andacademic department

Participant selectionResponse rate 71%, n=1753, excluded another 30 due to incomplete survey, n=1723

Participant characteristicsFirst year University students in 6 Universities, 58% were girls and 42% were boys (mean age: 19.5 years old,). 32% current smokers (daily smokers: 25%, occasional smokers: 7%), and 6% were former smokers.

OutcomesSmoking behaviour: no reaction, cheaper smoking, smoking less

Intervention detailsSurvey

General population impact32% did not react to price increase, 33% reduced costs of smoking(purchasing in foreign countries/smuggling, cheaper brand, hand-rolled), 35% reduced consumption or tried to quit

Impact by SES variableDaily smokers with low-educated parents were less likely to react to the price increase, daily smokers who had at leastone parent that completed high school were more prone to react to highercigarette price (OR=2.5, 95% CI=1.6,4.0 for cheaper smoking vs no reaction; and OR=2.1, 95% CI 1.4,3.3 for smoking less vs no reaction; in multivariate analysis, p < 0.001 and p< 0.01, respectively)Students who reported difficulties in financing their studies were significantlymore likely to purchase cheaper cigarettes (OR=1.9,95% CI=1.0,3.7; p< 0.1).

Author’s conclusion of SES impactYoung smokers with a lower socio-economic status were less likely to react to the price increase

Internal validityStudent’s report of parent’s educational level may be prone to bias. Study was powered for a response rate of 70% but small sample size for this analysis with statistical significance threshold of 10%. Retrospective questions after price increase.

External validityRegional not national survey so may not be generalisable to whole of France; also reactions to price increase are only relevant to daily smokers who did not quit.

Validity of author’s conclusionValid but small specific sample

132

Details Method Results CommentsControls on access to tobacco productsAuthor, yearKim 2006

Age (years)15

SettingUS

Study designProspective cohort study

ObjectiveTo examine whether young especially low SES females are influenced by tobacco control policies in terms of smoking initiation and transition

InterventionState level tobacco control policies and state cigarette excise tax

SES variablesParents level of education and income (parents questionnaire)

Study analysismultilevel logistic regressionscomparing initiators to never smokers

Data sourcesState level tobacco policy scores developed by US National Cancer Institute, evaluating 9 items for each state each year= statewide enforcement, random inspections, graduated penalties, photo identification, free distribution, minimum age, packaging, vending machines, and clerk intervention. Dataset, the national longitudinal study of adolescent health (Add Health), is a school based survey of the health related behaviours of adolescents.

Participant selectionAdd Health surveys individual adolescents from 132 schools, grades 7 to 12, using a sampling frame stratified by region, level of urbanisation, school type, school size, and by school racial compositions. In 1994–5(wave 1), data from 18 924 adolescents were collected; in1996 (wave 2) and in 2001–2 (wave 3), follow up in-home surveys were conducted to interview again 15 197 of the respondents from wave 1 about their health behaviours and life experience as young adults.

General population impactN/A

Impact by SES variableStronger state level tobacco policies were associated with lower likelihood of smoking initiation and adverse transition among low SES women, although the effect sizes were small. The positive policy effects for initiation were strongest for low SES females, whose odds ratio was 0.95 (0.98 for middle SES, 1.00 for high SES). For initiation, school level smoking rates did not vary substantially across low, middle, and high SES groups (OR=1.01, 0.99 and 1.00, respectively. For statewide enforcement, the odds ratios of initiation were significantly lower for the low (0.89) and middle (0.91) SES female groups; on the other hand, the policy had no effect on the high SES female group (OR=1.00). For random inspections the odds ratios of initiation were significantly lower for low (0.88) and middle (0.90) SES female groups. Photo identification had a significant positive effect on the low SES female group (OR=0.85), but not on the middle SES female group (OR=0.95, NS) and on high SES females (OR=1.10, NS). other policies had a pattern similar to the significant ones

Author’s conclusion of SES impact

Internal validityStudy is longitudinal but 7 year gap in data used to assess transition from adolescence to young adulthood may have missed other important mediators. Missing values for family income were imputed.We don’t know how demographic characteristics at each wave compare.

External validity2697 females only-no further information on how representative this sample is.Study adopted a measure of comprehensive state tobacco control efforts based on a score developed by the National Cancer Institute evaluating nine items for each state each year; enables future studies to use similar rating scores for policy.

Validity of author’s conclusionValid but effect size is small.

133

Details Method Results CommentsControls on access to tobacco products

Participant characteristicsRestricted to female adolescents younger than 18 at baseline=2697 females from 33 states,126 schoolsSample sizes were 1245 for low SES, 812 for middle SES, and640 for high SES female adolescents.

Outcomes measuredSmoking initiation and transition

Intervention detailsNational longitudinal school-based survey ‘Add Health’ of individual adolescents about their health behaviours and life experience as young adults

Tobacco control policies have the biggest impact on reducing the likelihood of smoking initiation in low SES females, less of an impact on the likelihood of middle SES female group, and the least impact on high SES females. Stronger tobaccocontrol policies are positively related to lower likelihood of adverse transition in smoking, especially for the low SES female group

134

Details Method Results CommentsControls on access to tobacco productsAuthor, yearLipperman-Kreda 2012

Age (years)Underage tobacco sales – 4 confederate buyers (2 men and 2 women), who were over 18 years of age, but judged to appear younger by an independent panel, mean age 19 years.

Setting997 Tobacco outlets in 50 mid-sized California cities, USA

Study designCross-sectional

ObjectiveTo examine contextual and community-level characteristics associated with youth access to tobacco through commercial sources

Interventionunderage tobacco sales laws

SES variablesMedian family income, % population with college education (city level n=50)

Data sourcesAccess surveys

Participant selectionpurposive geographic sample

Participant characteristics997 Tobacco outlets in 50 mid-sized California cities, USA

Outcomes measuredRetailer compliance with underage tobacco sales laws

Intervention detailsPurchase attempts were made at 997 tobacco outlets in 50 mid-sized California cities by a team of two buyers. At each outlet a single buyer attempted to purchase a pack of Marlboro or Newport cigarettes, which are the most popular cigarette brands among high school-aged students. Each buyer asked for Marlboro in one outlet and Newport in the next one. If asked about their age they stated that they were over 18 years old, and if asked for an age ID they indicated they had none. If a sale was refused, the buyers left without attempting to pressure the clerk.

General population impactOverall rate of retailer non-compliance with underage tobacco sales laws in the 997 selected outlets was 14.3%. Buyer’s actual age, a male clerk and asking young buyers about their age were related to successful cigarette purchases. Buyer’s actual age and minimum age signs increased the likelihood that clerks will request identification (ID).

Impact by SES variableRetailer compliance with underage tobacco sales laws: at the community level, a greater percentage of residents with at least a college degree were associated with increased likelihood of non-compliance.Predictors of clerks requesting ID: at the community level, lower percentage of residents with at least a college degree was associated with retailers asking for an ID. Asking young buyers about their age was positively associated with successful purchases.Predictors of cigarette pack prices: higher cigarette prices of Marlboro but not Newport, were associated with higher median household income.

Author’s conclusion of SES impactYouth in communities with higher educational levels may have easier access

Internal validityThere were no significant differences between the sampled and the unsampled cities in relation to population size, ethnic diversity, household size and median household incomes.

External validityOnly 2 buyers conducted the surveys in each city which limits ability to consider characteristics of the buyers other than gender and age. Also limited to 2 brands.

Validity of author’s conclusionHigher education was a significant predictor of underage tobacco sales.So stricter enforcement of laws would not reduce gap between low and high SES in terms of smoking prevalence? Unclear how access to tobacco translates into smoking prevalence.

135

Details Method Results CommentsControls on access to tobacco products

Study analysisMultilevel logistic and linear regression

to cigarettes from retail stores. The relationships between community characteristics and cigarette prices varied by cigarette brand. Higher median household income was associated with higher prices of Marlboros.

136

Details Method Results CommentsControls on access to tobacco productsAuthor, yearMillett 2011

Age (years)13

SettingSecondary schools, England

Study designRepeat cross-sectional surveys of children in same schools before and after legislation

ObjectiveTo determine whether the law had a differential impact on the likelihood of regular smoking depending on FSM status among youth in England

InterventionLegislation in England, Scotland and Wales increasing minimum age for legal purchase of tobacco from 16 to 18 years, October 2007

SES variablesFree school meals (FSM) (eligibility assessed on basis of parental employment and status and income level)

Data sourcesSmoking, Drinking and Drug Use Among Young People in England (SDDU) annual survey of 11-15 year olds, by National Centre for Social Research and the National Foundation for Educational Research, National Foundation for Educational Research database

Participant selectionIn 2008, 264 schools agreed to take part (response rate 58%) and within these schools 7798 pupils aged between 11 and 15 years completed the survey (response rate 88%).

Participant characteristicsFSM group was significantly younger (mean age: 13.1 vs 13.2 years, p=0.002), more likely to be female (53% vs 49%,P=0.042) and contained significantly more pupils from ethnic minorities (22% vs 13% non-white, p<0.001) than the non-FSM group in 2008.

Outcomes measuredRegular smoking statusUsual source of tobaccoEase of tobacco purchase

Students receiving FSM were more likely to smoke (adjusted OR for FSM: 1.87, p<0.001).

General population impactIncreasing the minimum age for purchase was associated with a significant reduction in regular smoking among youth (adjusted OR 0.67; 95% CI 0.55 to 0.81,P=0.0005).

Impact by SES variableRegular smoking was not significantly different in pupils eligible for FSM compared with those that were not (adjusted OR 1.29; 95% CI 0.95 to 1.76, p=0.10).

Percentage of regular smokers who usually bought cigarettes from a vending machine decreased significantly in the non-FSM but not in the FSM group.

Percentage of regular smokers who usually bought cigarettes from friends and relatives or from other people increased significantly in the non-FSM but not the FSM group after the introduction of age restriction.

Regular smokers eligible for FSM were significantly more likely to be given cigarettes by their parents in 2006

Internal validityPupil records with missing values (e.g., not answering) for outcome variables and covariates were removed (10.4%).Baseline differences in age, gender and ethnicity but controlled for in analyses.Self-report smoking status but reported within schools rather than at home.Cross-sectional but response bias is likely to be low because the pupil response rate to the survey was very high (88% in 2008) in participating schools.Sample size was sufficient to detect a 10% relative reduction in smoking prevalence in the non-FSM group compared with the FMS group (at 80% power at the 5% level of significance).However, the sample size did not permit examination of whether the legislation reduced the volume of cigarettes smoked.

External validityAlthough the response rate for schools was only 58% in 2008, the sampling frame ensured

137

Details Method Results CommentsControls on access to tobacco products

Study analysisMultivariate logistic regression analysis adjusted for previous time trends, age, gender, ethnicity, alcohol and drug use

Intervention detailsData used from 2003 to 2008 excluding 2007.

(p<0.001) but this was no longer the case in 2008 (p=0.42).

Percentage of pupils who stated that they found it difficult to buy cigarettes from a shop did not increase in those eligible for FSM (25.2% to 33.3%; p=0.21) but did increase significantly in others (21.2% to 36.9%; p<0.01) between 2006 and 2008.

Percentage of regular smokers who were successful in buying cigarettes from a shop during their latest attempt decreased significantly in the non-FSM but not the FSM group between 2006 and 2008.

No differences in ease of purchase were found between pupils eligible for FSM and those not before or after the legislation (2006: p=0.34, 2008: p=0.55).

Author’s conclusion of SES impactIncreasing the minimum age for the purchase of tobacco in England was associated with a significant reduction in youth smoking and was neutral with regard to disparities.

that schools participating in the survey closely reflect the composition of schools in England generally.The survey did not include 16 and 17 year olds who weremost directly affected by the increase in age for the legal purchase of tobacco.

Validity of author’s conclusionSmokefree ban and alcohol restrictions also introduced during this time which may confound these results.

138

Details Method Results CommentsControls on access to tobacco productsAuthor , yearSchneider 2011

Age (years)17.6% aged 0-20 years

SettingCologne, Germany

Study designBefore and after observational study

ObjectiveTo compare number of vending machines and other commercial sources before and after new legislation and to examine association between commercial cigarette sources and area SES

InterventionElectronic locking devices on vending machines to prevent underage (<16 years) purchasing

SES variableIncomeUnemploymentSocial welfareLow-qualifying schools

Data sourcesGerman Sources of Tobacco for Pupils (STOP) study, observational

Participant selectionCologne selected because had existing sociogeographical data

Participant characteristics17.6% aged 0-20 years

OutcomesDensity of sources before and after legislation according to SES of each district

General population impactNumber of commercial sources declined by 12% from 2005 to 2009 resulting mainly from removal of 44% of outdoor cigarette vending machines (indoor machines decreased by 5%). Convenience cigarette sources reduced by only 0.9%, supermarket and drug stores +2.6%.

Impact by SES variableThe lower the income level in a district, the higher the availability of cigarettes (Pearson’s r = .595; p = .009). The same occurred for the alternative indicators such as youth unemployment (Pearson’s r = .548; p = .019), the percentage of people receiving social welfare (Pearson’s r = .485; p = .041), and the percentage of pupils attending low-qualifying schools (Pearson’s r = .473; p = .048).In 2005 as well as in 2009, we found significantly fewer commercial cigarette sources in districts with above-average SES than in districts with below-average SES. This can be seen in terms of absolute as well as relative numbers. The density of commercial cigarette sources in 2005 in districts with above-average SES was 3.20 per 1,000 inhabitants and 4.84 per 1,000 inhabitants in the districts with below-average SES. In 2009, the numbers were 2.63 per 1,000 inhabitants and 4.44. per 1,000 inhabitants, respectively. The differences between socially advantaged and disadvantaged districts appeared to be

Internal validityPotential for limited interrater reliability between 3 geocoders

External validity‘natural experiment’ design is real life, but limited to one city so not representative of all German cities but appears comparable with Germany as a whole

Validity of author’s conclusionValid

Details Method Results CommentsControls on access to tobacco productsStudy analysisInventory of commercial cigarette sources in 2005 and 2007 and 2009 and mapped using Geographic Information System

significant in both years (2005: t(15) = 9.017, p < .001 and 2009: t(17) = 6.915, p < .001).

Author’s conclusion of SES impactIn districts with above-average SES, the supply density was lower than in districts with below-average SES, even at the beginning of the study. Decreases in the number of cigarette sources were reflected more sharply in regions of higher SES, which also emphasizes the social inequalities between these two areas.

140

Details Method Results CommentsControls on access to tobacco productsAuthor , yearWidome 2012

Age (years)Minors aged 15-18 years used for compliance checks

SettingMinnesota, US

Study designCross-sectional

ObjectiveTo test the association between point-of-sale advertising intensity and likelihood that a store would fail a compliance check

InterventionAge-of-sale tobacco checks

SES variablePoverty (below 150% poverty level)

Study analysisDescription of compliance check failure proportions for various types of stores and by census block group demographics. Failure proportions calculated. Chi-square tests to test

Data sources(1) Observations of the advertising environment in establishments (2) a record of age-of-sale tobacco checks where an undercover minor working with law enforcement attempts to purchase tobacco (3) Demographic data from the Year 2000 U.S. census.

Participant selection655 licensed tobacco vendors, both interior and exterior assessments were completed on 485 establishments (74.0%). Analyses conducted on 467 establishments that had complete assessments.

Participant characteristics

OutcomesCompliance -failure defined as the sale of tobacco to a youth, regardless of whether the store clerk examined the minor’s ID.

General population impactNo association found between tobacco point-of-sale marketing and compliance check failure.Of a total of 467 stores, 48 failed compliance check. Tobacco shops were most likely to fail compliance checks (44%). Supermarkets were least likely to fail (3%).

Impact by SES variablePoverty of stores block group was not associated with compliance failure of stores. Stores in block groups with greater percentage of people living in poverty were not more likely to fail compliance check.

Author’s conclusion of SES impactThere was no association between store advertising characteristics or poverty and stores’ compliance check failure. The relationship between advertising and real youth sales may be more nuanced as compliance checks do not perfectly simulate the way youth attempt to purchase cigarettes.

Internal validityCompletion of a full store assessment was not significantly associated with whether stores passed their compliance check (p = .931).Each store assessed by one assessor.

External validityAuthors report that Minnesota has less racial/ethnic diversity compared to other urban centres.Compliance checks may not be a very valid measure of commercial tobacco accessibility for minors.Only vendors with a current license can sell tobacco in state of Minnesota but this is not the case across all US states.Also stores who repeatedly violate youth access laws have license rescinded.Study was cross-sectional so cannot assess whether change in advertising leads to change in compliance check failure.

Validity of author’s conclusionValid as considers weakness of

141

Details Method Results CommentsControls on access to tobacco productswhether failure was associated with a store being situated in a block group that was in the top decile of each demographic item. Top decile used as a cut-off to examine more extreme examples of census block groups that had relatively high proportions of certain demographics, and t tests to examine whether the percentage of people in a block group for each demographic item was associated with compliance check failure. X2

tests to examine whether specific advertising practices were associated with failure.

the compliance check measure.

142

Details Method Results CommentsSchool-based preventionAuthor, yearBacon 2001

Age (years)11 (6th grade)

Setting6 middle schools in Florida, US

Study designCluster randomised controlled trial

ObjectiveTo examine effectiveness of ‘Too Good for Drugs II’ (TGFD II) programme

Intervention9 week prevention curriculum and follow-up after further 20 weeks, school-based curriculum and also involves community partners and parents; theoretical basis includes Social Learning Theory, Problem Behaviour Theory and Social Development Theory.

SES variables usedfree/reduced lunch status

Study analysis - RCT

Data sources

Participant selectionRandomly selected

Participant characteristics1318 sixth grade students, 52% female, 48% white, 33% African-American, 13% Hispanic, 6% Asian; 51% in receipt of free or reduced school lunches; 84% participated in 20-week follow-up

Outcomes measuredIntentions, attitudes and perceptions towards tobacco use

Intervention details9 lesson units (40 minutes each) at each grade by trained classroom teacher or TGFD II instructor; social and emotional competencies, reducing risk factors and building protective factors; emphasise cooperative learning activities, role-play and skills building methods;

General population impact8% 48/588 in intervention group indicated greater likelihood of actual tobacco use at end of programme compared to 12% (45/375) in control, p=0.04. No statistically significant difference between groups at 20 weeks follow-up.

Impact by SES variableThe overall findings of the comparison of change scores for treatment students indicates the programme was similarly effective in impacting students risk and protective factors regardless of economic status (perception of peer resistance skills; positive attitudes toward non-drug use, perceptions of peer normative substance use, perceptions of peer disapproval of substance use, association with prosocial peers, perceptions of locus of control self-efficacy)Student scores at end of programme and at 20 weeks follow-up showed significant multivariate overall effects for SES (before and after intervention).

Author’s conclusion of SES impactThe findings suggest the programme was equally effective for students regardless of SES

Internal validityConfounding influences of other interventions were not observed; 16-17% attrition; test of equivalence of attrition rates by treatment condition did not show attrition bias for students predisposition towards future tobacco use behaviours

External validityImpact by SES only relates to scores for substance use not just tobacco use which limits comparability

Validity of author’s conclusionEquity impact unclear

143

Details Method Results CommentsSchool-based preventionMercken 2012Author, yearCampbell 2008 (from Mercken 2012)

Age (years)12-13

Setting59 secondary schools in England & Wales

Study designRCT

ObjectiveTo assess the effectiveness of a peer-led intervention that aimed to prevent smoking uptake in secondary schools

Interventionschool-based, peer-led – influential students trained to act as peer supporters outside of the classroom

SES variables usedFAS and FSM (above 19% andequal or below 19%), area deprivation

Data sourcesRCT A Stop Smoking in Schools Trial (ASSIST)

Participant selectionSchools were randomly assigned to the control group to continue their usual smoking education (29 schools with 5372 adolescents) and intervention group (30 schools with 5358 adolescents) by stratified block randomization.

Participant characteristicsNot reported

Outcomes measuredsmoking behaviour in the past week

Intervention detailsTraining influential students to act as peer supporters during informal interactions outside the classroom to encourage peers not to smoke. During the 10-week intervention period, peer supporters undertook informal conversations about smoking with their peers when travelling to and from school, in

General population impactAt 1-year follow-up, the odds ratio of being a smoker in intervention compared with control group was 0·77 (95% CI 0·59–0·99). At2-year follow-up, the corresponding odds ratio of 0·85 (0·72–1·01) was not significant (p=0·067) which suggests an attenuation of this intervention effect over time. For the high-risk group (occasional, experimental, or ex-smokers at baseline), the odds ratios at 1-year follow-up of 0·75 (0·56–0·99) and at 2-year follow-up of 0·85 (0·70–1·02).In a three-tier multi-level model using data from all three follow-ups (immediately after the intervention (N = 10047), after 1 year (N = 9909) and after 2 years (N = 9666)) the odds of being a smoker in the intervention group compared with the control group was 0.78 (95% CI = 0.64–0.96)

Impact by SES variableReported in primary study: subgroup analyses showed no evidence of intervention having differential effect according to deprivation measured by FSM (0·99 [0·65–1·51]). However, the intervention does seem to have had a more pronounced effect in schools located in

Internal validityA slightly larger proportion of students in control schools came from less affluent backgrounds and did not have a family car than did those in intervention schools Saliva cotinine levels obtained which minimised reporting bias.Results may depend on the SES indicator used.

External validityResults are specific to study interventions

Validity of author’s conclusionValid

144

Details Method Results CommentsSchool-based preventionMercken 2012

Study analysisData of the assist trial were reanalysed according to methods reported in the ASSIST study. Multilevel modelling was used to explore intervention effects on adolescent smoking in different SES categories. Data from the three follow-up periods were modelled using a using a three-level multilevel model with schools at Level 3, students at Level 2 and follow-up measurements at Level 1. Models were estimated using the RIGLS estimation procedure combined with first-order penalized quasi-likelihood within MLWin 2.10 beta. Separate analyses were conducted for adolescents in the low, medium and high categories of the included SES indicators.

breaks, at lunchtime and after school in their free time. Peer supporters logged a record of all conversations in a diary. Trainers visited schools four times to meet with peer supporters to provide support, trouble shooting and monitoring of peer supporters’ diaries

south Wales valleys (0·58 [0·36–0·93].

Reported in secondary analyses: A significant main effect of intervention was found among adolescents scoring low (chi-square (df = 1) =5.97, P < 0.05, OR = 0.71, 95% CI = 0.54–0.93) and high (chi-square (df = 1) = 7.28, P < 0.05, OR = 0.68, 95% CI = 0.52–0.90) on the FAS. No significant main effects of the intervention on adolescent smoking behaviour were found in either group. However, a trend is visible among adolescents in schools with a low free school meal entitlement (chi-square (df = 1) = 3.56, P = 0.06, OR = 0.80, 95% CI = 0.63–1.01). The intervention was significant among adolescents in schools located in the valleys which can be considered to be a more deprived area (chi-square (df = 1) = 5.68, P < 0.05, OR = 0.53, 95% CI = 0.32–0.89) but not among adolescents in schools on other locations. Among adolescents in Valley schools, the intervention was also effective among those with low FAS scores (chi-square (df = 1) = 5.97, P < 0.05, OR = 0.71, 95% CI = 0.54–0.93). The additional analyses stratified by SES and gender showed that the ASSIST intervention was mostly effective among lower SES girls.

Author’s conclusion of SES impactThe results were mixed depending on the specific SES indicator used. The ASSIST

145

Details Method Results CommentsSchool-based preventionMercken 2012

study showed the strongest results for adolescents in the Valley schools, located in a deprived area. Social network approach allowing youngsters to deliver the intervention themselves seems promising in preventing the uptake of smoking in deprived adolescents.

146

Details Method Results CommentsSchool-based preventionMercken 2012Author, yearCrone 2003 (from Mercken 2012)

Age (years)13

Setting26 junior secondary education schools in the Netherlands

Study designRCT (an independent person tossed a coin)

ObjectiveTo investigate whether a peer group pressure and social influence intervention reduced the percentage of adolescents who start to smoke

Interventionschool-based social influence and peer group pressure to prevent smoking with a class-based competition

SES variables usedparental education

Study analysis

Data sourcesRCT

Participant selectionSchools were ranked by size, stratified by use of a national drug education programme and subsequently randomly assigned to the control and intervention group. At baseline, a sample of 2562 adolescents participated

Participant characteristicsNot reported

Outcomes measuredexperimenting with smoking or smoking daily or weekly

Intervention detailsThree lessons on knowledge, attitudes and social influence, followed by a class agreement not to start smoking or to stop smoking for the next 5 months. Video lessons on smoking and social influence were available as an optional extra during these 5 months. Classes having fewer than 10% smokers after 5 months were entered in the competition. The final activity of the

General population impact9.6% of the nonsmokers started to smoke in the intervention group, whereas 14.2% started to smoke in the control group (N = 1388, OR = 0.61, 95% CI = 0.41–0.90). After 1-year follow-up, the effect was no longer significant.

Impact by SES variableAt 5 months, smoking behaviour was significantly lower in adolescents who indicated that their parents had mid to high completed educations (chi-square (df = 1) = 4.21, P < 0.05, OR = 0.35, 95% CI = 0.13–0.95). The intervention did not result in smoking fewer cigarettes among adolescents who indicated that their parents had lower education (chi-square (df = 1) = 0.33, P > 0.05, OR = 0.80, 95% CI = 0.37–1.72). All significant intervention effects disappeared at 12 months follow-up. The additional analyses stratified by gender and SES furthermore showed that the intervention was only effective at 5 months follow-up among boys with higher parental educational levels (chi-square (df = 1) = 5.56, P < 0.05, OR = 0.24, 95% CI = 0.07–0.79).

Author’s conclusion of SES impactThe Dutch class competition study only had a significant effect among higher

Internal validityThe percentage of boys in the control group was higher than in the intervention group at baseline but this was adjusted for in analyses.Nonresponse was higher among smokers, especially in the control group but selective dropout was assessed using ITT under 3 different assumptions.Results may depend on the indicator used

External validityResults are specific to study interventions.

Validity of author’s conclusionValid.

147

Details Method Results CommentsSchool-based preventionMercken 2012Data from follow-up at 5 and 12 months were modelled using three-level multilevel models with school at Level 3, class at Level 2 and adolescent at Level 1. Models were estimated using the restricted iterative generalized least squares (RIGLS) estimation procedure combined with first-order penalized quasi-likelihood within MLWin 2.10 beta. The multilevel model was tested separately for adolescents in each of the categories of the two included SES indicators.

class was to make a photo expressing the idea of a non-smoking class. There were competition prizes for six classes with less than 10% smokers and a photo best expressing a non-smoking class

SES adolescents and appeared to widen the inequalities

148

Details Method Results CommentsSchool-based preventionMercken 2012Author, yearDe Vries 2006 (from Mercken 2012)

Age (years)13.5

Setting25 schools in 2 regions of Portugal

Study designRCT

ObjectiveNot stated

InterventionSocial influence intervention which was school-based with wider community. Interventions were developed for four levels: the individual adolescent level, the school level, the parental level and the out-of-school level.

SES variables usedspending money

Study analysisMultilevel modelling techniques were used to test for intervention effects

Data sourcesRCT - the European SmokingPrevention Framework (ESFA) study

Participant selectionTwo regions, consisting of 14 and 11 schools, respectively, were randomly assigned to the experimental and control condition. At baseline, 3102 adolescents participated in the intervention study in Portugal

Participant characteristicsNot reported

Outcomes measuredever smoking/never smoking

Intervention detailsLessons on effects of tobacco, reasons for (not) smoking, social influence processes, refusal skills and decision making and a smoke-free competition.Teachers received 48 hours of teacher training, manual and smoking cessation materials. Schools received the ESFA no-smoking policy manual and non-smoking posters. To the parents,

General population impactAt 24 months significantly fewer ever-smokers were found in the Portuguese experimental group (33.8%) than the control group (41.5%) (OR=0.73, 95% CI = 0.57–0.94).At 30 months 41.8% of the never smokers started to smoke 30 months later in the intervention group, whereas 53.8% of the never smokers in the control group (N = 1304, OR = 0.62, 95% CI = 0.48–0.80)In terms of non-smokers becoming weekly smokers in experimental vs control groups; 7.3% vs 9.1% respectively at 24 months (OR = 0.74, 95% CI = 0.41–1.34) and 7.9% vs 12.4% at 30 months (OR = 0.56, 95% CI = 0.37–0.84).

Impact by SES variableThe intervention was significant in reducing smoking uptake among adolescents who indicated to have no to only a low amount of spending money (chi-square (df = 1) = 9.85, P < 0.01, OR = 0.62, 95% CI = 0.46–0.84). This effect was not seen among adolescents reporting to receive mid to high amounts of spending money (chi-square (df = 1) = 3.51, P > 0.05, OR = 0.57, 95% CI = 0.32–1.03). Additional analyses stratified by gender and SES showed that the intervention was mostly effective among girls.

Internal validityResponse rates differed between experimental and control groups; 41.7% vs 39.1% respectively.May not be a strong association between indicators such as adolescents’ pocket money and household income.‘mid to high’ spending money subgroup relatively small (n=182) which explains wide CI’s and probably why result not significant – so intervention might not decrease inequalities in smoking

External validityProcess evaluation included pupil report of exposure to each element of the intervention and showed it reasonably likely that the observed effects were attributable to the school-based elements of the intervention.

Interventions differed between countries and Portugal received the most intensive teacher training and pharmacists of smoking

149

Details Method Results CommentsSchool-based preventionMercken 2012on smoking behaviour in different SES categories. Data from follow-up at 30 months were modelled using three-level multilevel models with region at Level 3, school at Level 2 and adolescent at Level 1. Models were estimated using the RIGLS estimation procedure combined with first-order penalized quasi-likelihood within MLWin 2.10 beta. The multilevel model was tested separately for adolescents in each of the categories of the three included SES indicators.

information was offered on how to discuss non-smoking with their adolescents. Pharmacists furthermore offered cessation courses for 150 parents. At the community level, the Portuguese Health Minister and mayor of the community introduced the ESFA study on the national no smoking day

Author’s conclusion of SES impactThe results were mixed depending on the specific SES indicator used. When using spending money as a SES indicator, the intervention did appear to decrease inequalities in smoking.

cessation support for 150 parents; so results may only be generalisable to that type of intervention in that country.

Validity of author’s conclusionValid

CommentsThe ESFA study was a community-based intervention that took place in six European countries. In Finland, Denmark, UK and Portugal schools or regions were randomly assigned. In Spain and The Netherlands it was quasi-randomisation.Due to the fact that peer-led programmes were uncommon in the ESFA countries, programmeswere teacher-led.Since the strongest and significant long-term effects after 24 and 30 months were found in the Portuguese sample, only data of the ESFA study in Portugal were reanalysed on the impact by SES (Mercken 2012) and so only results for Portugal are

150

Details Method Results CommentsSchool-based preventionMercken 2012

extracted here.Details Method Results CommentsSchool-based preventionAuthor, yearMenrath 2012

Age (years)12

Setting53 public secondary general schools in Northern Germany (rural federal state of Schleswig-Holstein)

Study designQuasi randomised multicentre trial (6 schools included without randomisation to intervention group)

ObjectiveTo evaluate the effects of two validated school-based life skills programmes (Fit and Strong for Life and Lions Quest) in a high-risk sample of socially disadvantaged pupils.

SES variablesFamily Affluence Scale

Data sources

Participant selectionOversampled pupils with low SES by only including secondary general schools. 2/102 classes lost to follow-up at end of school year and 7 more lost to follow-up at six months

Participant characteristics102 classes with a total of1,561 pupils. 25% of thepupils had a low SES

Outcomes measuredSelf-report cigarettes smoked per week30-day smoking prevalence

Intervention details“Fit and Strong for Life” and “Lions Quest”. Both programmes foster life skills and self-efficacy and include the prevention of substance abuse (cigarettes, alcohol, and drug consumption). Fit and Strong for Life is a modular life skills programme for

General population impactIn the EGC analysis the effect of the intervention was observed with regard to smoking (cigarettes per week and 30-day smoking prevalence). In the ITT analysis only the effect on the 30-day smoking prevalence was significant.

Impact by SES variableANOVA with SES as a factor revealed no influence of SES on the effect of the intervention (SES* time*group).

Author’s conclusion of SES impactSchool-based life skills programmes have a positive effect on smoking prevention regardless of socioeconomic status. Socially disadvantaged children benefit from such programmes to a similar extent as other pupils.

Internal validityOver 50% of the schools eligible for the study initially agreed to participate.Authors report did ITT analysis and analysis of classes with 60% programme participation (EGC analysis).The intervention group included significantly more pupils who reported a higher SES.Loss to follow-up was 23%.Schools in the intervention group were asked to conduct one of the two programmes in classes 5 or 6.

External validityAnalyses do not appear to account for specific life skills programmes (“Fit and Strong for Life” and “Lions Quest”.)

Validity of author’s conclusionDoes not report data for results by SES and does not assess

151

Details Method Results CommentsSchool-based preventionMercken 2012

Study analysisRepeated standardised interviews before and after school year and at 6 months follow-up, repeated measures analyses of variance

primary and secondary schools. Each module covers the following six topics: (1) self-esteem and empathy, (2) coping with stress and negative emotions, (3) communication skills, (4) resistance skills and critical thinking, (5) problem solving and decision making, (6) health-related knowledge. The consecutive modules are not interdependent and may start at any grade regardless of whether the pupils have participated in the programme before or not. In this study we used the module for grades 5 and 6.Lions Quest has a comparable curriculum. It may be used in secondary schools from class 5 to 10. Based on the life skills approach it consists of seven major content areas: (1) behaviour in classes/groups, (2) self-esteem, (3) coping with emotions, (4) peer-relationships, (5) family-relationships, (6) decision making and (7) self-efficacy. In our study both programmes Fit and Strong for Life and Lions Quest were carried out by classroom teachers. Beforehand all teachers had to attend a one- or two-day training workshop.

effects of each life skills programme separately so difficult to assess equity impact of each specific programme.

152

Details Method Results CommentsMultiple policy interventionsAuthor, yearHelakorpi 2008

AgeRange for smoking initiation defined as 13 - 20 years

SettingFinland

Study designRepeat cross-sectional

ObjectiveTo assess the impact of the 1976 Tobacco Control Act (TCA) on smoking initiation across socioeconomic groups.

Intervention1976 Tobacco Control Act

SES variablesUpper white collar workers (upper level employees), lower white collar workers (lower level employees), blue collar workers (manual workers), farmers and entrepreneurs (other self-employed persons than farmers).

Analyses

Data sourcesNational Public Health Institute annual cross-sectional postal surveys from 1978 to 2002. Unique personal identification codes were used to link information on socioeconomic group from population censuses (every fifth year starting 1970 except 1985)

Participant selectionEach year an independent random sample (n=5000) of the population aged 15–64 years was drawn from the National Population Register.

Participant characteristics33,080 adults aged 25 to 64 years born between 1926 and 1975

InterventionThe 1976 TCA prohibited smoking in most public places, including public transport, and the sale of tobacco products to those below 16 years of age, and required obligatory health warnings on packages

Length of study14 years – 1978 to 2002

OutcomesSmoking prevalence (ever

General populationAmong men the secular cohort trend in smoking declined only in upper white collar workers, whereas in other socioeconomic groups the secular cohort trend was non-significant. A clear decline in the prevalence of male ever daily smokers concurrent with the TCA was found in all socioeconomic groups except farmers. The differences between the three largest socioeconomic groups in the effect of the TCA were statistically significant (p=0.007 for the interaction between SES and the TCA) among men. Smoking decline corresponding to the 1976 TCA was most marked among white collar employees. In the three largest socioeconomic groups, the secular cohort trend remained unchanged after TCA (p=0.60 for the cohort trend after TCA, controlling for the general secular cohort trend) and there was no difference between the three largest socioeconomic groups in this respect (p=0.64 for the interaction between SES and cohort trend after the TCA).Among women an increasing secular cohort trend in ever daily smoking was found in each socioeconomic group before the impact of the 1976 TCA (birth cohorts born in 1926–1962). A reversal of the female ever daily smoking trend concurrent with the introduction of the 1976 TCA was found in each examined socioeconomic group. The impact of the legislation was even in the three largest socioeconomic

Internal validityAverage response rate 1978–2002 was 70% among men and 79% among women.

External validityThe response rate has declined over the past 25 years in both genders and all age groups. The decline has been faster among men than women, and in younger than older age groups – this may have biased the results.

Validity of author’s conclusionThere have been two major steps in Finnish tobacco control policy: the 1976 TCA, supplemented by a total tobacco advertising ban in 1978, and the environmental tobacco smoke amendment of the TCA in 1995. A significant rise in the price of tobacco products almost coincided with the 1976 TCA; tobacco prices rose substantially (real price increase 27%) in 1975–1976, (but since then annual increases have been either modest or negligible) and could explain some of the variability

153

Details Method Results CommentsMultiple policy interventionsLogistic regression smoked daily for at least a year) groups (p=0.14 for the interaction between

SES and the TCA). Moreover, the general cohort trend after the TCA differed from the secular cohort trend before TCA (p<0.001) and there were differences between the three largest SES groups in this respect (p=0.002 for the interaction between SES and the cohort trend after the TCA).

SESIn cohorts reaching the smoking initiation age after the TCA, the prevalence of ever smoking remained relatively stable among white collar female workers but tended to decline among blue collar female workers (odds ratio=0.88, 95% confidence interval 0.72 to 1.02), in contrast to the sharply increasing trend in older cohorts.

Author’s conclusion of SES impactAmong men, whose prevalence of ever smoking was potentially influenced by the 1976 TCA (those born in 1956 or later) the 1976 TCA appears to have had the greatest impact on male white collar employees. Among women, the apparent effect was very pronounced in all socioeconomic groups and among blue collar female workers the cohort trend tended to decline.

in results by SES.

154

Details Method Results CommentsMultiple policy interventionsAuthor , yearPabayo 2012

Age (years)12.7

SettingMontreal, Canada

Study designProspective cohort study

ObjectiveTo describe the association between smoking intolerance in schools, restaurants and corner stores near schools and the initiation of smoking in adolescents

SES variableParental educationSchools classified as low/medium/high SES based on mean household income

Study analysisCox proportional hazards modelling, limited to n=868 never smokers at baseline

Data sources‘The Natural History of Nicotine Dependence in Teens Study’. Self-report questionnaires administered in classroom, every 3 months from 1999 to 2005 in 1, 293 grade 7, age 12-13, students in 10 secondary schools. 7 English and 3 French language secondary public schools,

Participant selectionConvenience sample, 54.5% student response

Participant characteristicsMean age 12.7 (SD 0.5), range 11-16; 51% male

OutcomesSmoking initiation

Intervention detailsLongitudinal cohort and direct observation

General population impactStudents in smoking-intolerant schools (access and restrictions) were less likely to initiate smoking than students in smoking-tolerant schools (Hazard ratio [HR] = 0.83, 0.68, 1.01); attending schools located in neighbourhoods with smoking intolerant restaurants, HR=0.85 (0.68, 1.07). There was no association between corner store smoking intolerance and smoking initiation

Impact by SES variableHR for cigarette use initiation for low SES school, in schools=1.11 (0.88, 1.36), p=0.40; in restaurants=1.04 (0.83,1.31)p=0.74; in corner stores=1.10 (0.88, 1.37) p=0.59

Author’s conclusion of SES impactNot stated

Internal validity25% (219/868) lost to follow-up over 5 years, more likely to attend a low SES school (OR 1.7, 95% CI 1.2, 2.4; p<0.01)

External validityConvenience sample limits generalisability of this study

Validity of author’s conclusionStudy did not aim to assess differential impact by SES.

155

Details Method Results CommentsMultiple policy interventionsAuthor, yearWhite 2008

Age (years)12-17

SettingAustralia

Study designCross-sectional study

ObjectiveTo examine whether SES was associated with changes in smoking prevalence among Australian adolescents during 3 phases of tobacco control activity between 1987 and 2005

Intervention3 periods of tobacco control activity: low tobacco-control funding (1992-1996), high tobacco-control activity (1984-1991, and 1997-2005) which included smoking restrictions and increased tax

SES variableIndex of relative socio-economic disadvantage (IRSD) associated with residential postcode

Data sourcesRandom sample students aged 12-17 years from each Australian state and territory and three main education sectors, questions on smoking were part of a larger survey assessing use of alcohol and illicit drugs, 1987-2005, 19,000-22,000 students sampled each year

Participant selectionSchool acceptance rate has decreased over time but has stayed around 65% since 1999, Variation in school participation rates did not systematically co-vary with smoking prevalence

Participant characteristicsstudents aged 12-17 years

OutcomesSelf-reported smoking prevalence

Intervention detailsSelf-report anonymous surveys of cigarette use administered at school

General population impactThere was a significant and substantial reduction in the likelihood of smoking among all SES groups for older (16-17) and younger students (12-15) between 1987 and 2005 (all p <0.01).

Impact by SES variableFor younger students the reductions differed by SES (interactions p <0.01), with reductions in all smoking behaviours, greater for students from higher SES groups. Among older students, only the reductions in committed smoking differed across SES groups (interaction p < 0.01), and again reductions were greater among students from higher SES groups.

Between 1990 and 1996 the proportion of younger and older students involved with smoking increased significantly. Among younger students, the increase in monthly and weekly smoking was greater among lower SES students (interactions p < 0.05). Between 1996 and 2005 the prevalence of monthly and weekly smoking decreased significantly among both younger and older students, and these decreases were consistent across SES groups. For committed smoking, the interaction between year and SES was of borderline significance for students from both age groups, suggesting that the decrease may not be consistent across SES groups.

Internal validityOver the study period, Year 12 retention rates increased from 53% in 1987 to 75% in 2002 and 2005 so the characteristics of the student sample in Years 11 and 12 are likely to differ systematically across survey years.Individual students SES may not match the area IRSD.Self-reported smoking status so potential for bias

External validityCo-operation rate of schools was 85% in 1987 and 63% in 2005

Validity of author’s conclusionUnclear because changing prevalence estimates may be the result of different survey samples

156

Details Method Results CommentsMultiple policy interventionsStudy analysisLogistic regression analysis, controlled for sex, age and state and weighted to reduce the influence of under- or over-sampling of any state, education sector, age, or sex grouping

Author’s conclusion of SES impactThe magnitude of the decreases in smoking prevalence between 1996 and 2005 did not differ significantly between SES groups for most indicators of tobacco involvement. These findings suggest that the tobacco-control policies adopted in the late 1990s and early 2000s were effective in reducing smoking among Australian secondary students from all SES groups.

157

Details Method Results CommentsIndividual smoking cessation supportAuthor, yearRodgers 2005

Age (years)Mean=25, median = 22 (IQR 19-30)

Included persons from 16 years

SettingNew Zealand

Study designRCT

ObjectiveTo determine the effectiveness of a mobile phone text messaging smoking cessation programme

InterventionRegular personalised text messages providing smoking cessation advice, support, distraction

SES variableIncome

Study analysisLogistic regression

Data sourcesRCT

Participant selectionRecruited from adverts on websites, media, email and text messaging mailing lists and posters at tertiary education institutions

Participant characteristics1705 smokers who wanted to quit, 58% female, mean number cigarettes smoked per day = 15 average previous quit attempts=2 per person;

OutcomesSelf-reported smoking statusBiochemically verified abstinence on random selection

Intervention detailsFree five text messages per day for week prior to negotiated quit date and for four weeks after quit date.Control group received free month of text messaging if participated until 26 weeks.

General population impactNot smoking in past week:RR at 6 weeks = 2.20 (1.79 to 2.70)RR at 12 weeks = 1.55 (1.30 to 1.84)RR at 26 weeks = 1.07 (0.91 to 1.26)All participants with missing status are assumed to be smoking.Current non-smoking at 6 weeks = 28.1% vs 12.8%;assuming rate of true quitters is same as sample assessed for cotinine then current non-smoking at 6 weeks = 13.9% vs 6.2%; absolute difference in quit rates at 6 weeks is reduced to 7.7% from 15.3%;

Impact by SES variableEffect was consistent across income level

Author’s conclusion of SES impactText messaging can double quit rates and this effect was consistent across major subgroups including income level

Internal validityRandom sampling for salivary cotinine showed over-reporting of quit rates but not different between active and control group; 74% (n=1265) follow-up rate at 26 weeks which was different between groups (69% in intervention group vs 79% in control group), meant there was some uncertainty about between group differences at 26 weeks; for example reported quit rates increased amongst control group from 13% at 6 weeks to 24% at 26 weeks (this would have led to underestimating of treatment effects);

External validityParticipants had to be in the contemplative stage of change to be included; participants could use other smoking cessation strategies and were informed of quitline and government subsidy for NRT at baseline; participants had to own a mobile phone

Validity of author’s conclusionvalid

158

Details Method Results CommentsIndividual smoking cessation supportAuthor, yearYbarra 2013

Age (years)18-25

SettingNational, text-based, USA

Study designPilot Quasi RCT

ObjectiveTo address the lack of smoking cessation programs available to young adults, Stop My Smoking (SMS) USA, a text messaging–based smoking cessation program, was developed and pilot tested

Intervention6-week text messaging intervention

SES variableIncome (<$35,000 vs. higher);Education (enrolled/not enrolled in higher education)

Study analysis

Data sourcesRCT

Participant selectionPurposefully targeted a diversity of communities. Recruited nationally through online advertisements (e.g., Craigslist) between May 3, 2011 and August 4, 2011. Eligibility criteria included the following: being between the ages of 18–25, able to read and write in English, owning cell phone, being cognizant of how to send and receive text messages, being currently enrolled or intending to enrol in an unlimited text messaging plan, smoking 24 cigarettes or more per week (at least four per day on at least 6 days/week), seriously thinking about quitting in the next 30 days, and agreeing to smoking cessation status verification by a significant other (e.g., family member, friend).

Participant characteristics1,916 people expressed interest in participating, 585 (31%) of whom appeared eligible based upon the online screener form. Of these 585, contact was not made with 49% (n = 284 ‘passive refusals’). Fifteen percent (n = 90) declined to

General population impact40% of the participants in the intervention arm had a verified quit status compared with 30% in the control arm at 3 months post quit. The observed difference was not statistically significant (OR = 1.62, 95% CI:0.82, 3.21).Participants in the intervention were significantly more likely to have quit at 4 weeks post quit (39%) than those in the control group (21%; aOR = 3.33, 95% CI: 1.48, 7.45); this was true also for 7-day point prevalence (44% vs. 27%; aOR = 2.55, 95% CI: 1.22, 5.30).Cessation rate among intervention participants was stable between 4 weeks and 12 weeks, but increased among control participants

Impact by SES variableThe intervention appeared to be helpful for young adults not currently enrolled in higher education settings (45% vs. 26% control had quit at 3 months; p = .07).Enrolment in higher education settings was an effect modifier within the context of other potentially influential characteristics (arm assignment × school status; aOR = 4.7, 95% CI: 1.01, 22.3).

Author’s conclusion of SES impact

Internal validityFeasibility sample – small sample size so not sufficiently powered particularly for subgroup results.Imbalance in the minimum number of participants intended for each study arm within each subgroup (e.g., male heavy smokers). As a result, allocation concealment was broken for the last eight participants enrolled. To rectify the imbalance, these participants were manually assigned to the arm subgroup that required additional participants to become balanced.Eighty-seven percent of participants responded at 4 weeks post quit and 80% at 3 months post quit. Differential follow-up between the intervention and control groups was not observed at either follow-up time.Employment status differed between groups at baseline – does not add up to 100% in control group in Table 2.Study reports that allocation

159

Details Method Results CommentsIndividual smoking cessation supportLogistic regression participate.

n = 211 consented to participate and were randomized into the study. Final sample n = 164 (47 did not complete online baseline survey following randomisation): 101 in the intervention and 63 in the control groups.Mean age 22, daily smokers, 44% female, 84-90% low income (<$35,000)

Outcomes3-month continuous abstinence (reported smoking five or fewer cigarettes since their quit date and verified by phone with significant other),smoking five or fewer cigarettes since quit day at 4 weeks post quit (verified by a significant other);7-day point prevalence abstinence at 4 weeks;Acceptability

Intervention detailsTailored to young adult smokers based on quitting stage.2 weeks of Pre-Quit messages aimed at encouraging them to clarify reasons for quitting and to understand their smoking patterns and tempting

The intervention appeared to be more influential for intervention participants not enrolled in higher education compared with control participants not enrolled in higher education aOR of verified quit at 3 months =2.7, 95% CI: (1.0, 7.4).

‘unknown’ for 47 participants who did not complete baseline survey but looking at numbers in each group it is possible that majority of these participants were from control group although all participants were blinded to treatment

External validityParticipants had to be seriously thinking about quitting in next 30 days – motivated sample. Financial incentives for completing follow-up surveys.Sample classed as ‘low income’ and low income defined as <$35,000! However, 43% report an annual household income of less than $15,000. Majority were male which is unusual.Having this type of control group suggests content of text messages is important.

Validity of author’s conclusionOverall significant increased quit rates in intervention group vs control group at 6-weeks were not sustained at 3 months. Tailored text messaging appears to benefit

160

Details Method Results CommentsIndividual smoking cessation support

situations/triggers/urges. Early Quit messages, sent on Quit Day and through the first week post quit, talked about common difficulties and discomforts associated with quitting and emphasized the use of coping strategies. Late Quit messages encouraged participants to recognize relapse in a different way (e.g., situations, confidence) and provided actionable information about how to deal with issues that arise as a non-smoker (e.g., stress, moods).Text message at Post-Quit Day 2 and 7 that asked their smoking status. At either time point, if participants reported smoking, they were pathed to Relapse messages that focused on helping them get back on track and to recommit to quitting. If participants were smoking at both days, they were pathed to an Encouragement arm that focused on norms for quitting and suggested that participants try quitting again at later time.Participants received four messages per day during the 2-week Pre-Quit stage, with the exception of Day 1 and Day 14 when they received five and six messages, respectively. In the Early Quit stage, participants received nine messages on both Quit Day and Post-Quit Day 2, eight messages on the third day, and then

youth not enrolled in higher education.

161

Details Method Results CommentsIndividual smoking cessation support

one fewer message each day until the last day of the week when four messages were received. In Late Quit, participants received two messages per day for 2 weeks and then one message per day during the final week. Participants in Relapse received two messages per day; those in Encouragement received one message per day for 4 days.Intervention group participants had access to two program components first used in the STOMP NZ program (Rodgers et al., 2005): (a) Text Buddy (another person in the program that a participant was assigned to so they could text one another for support anonymously during the program; assignment was sequential so that buddies would be in similar stages during the quitting process); (b) Text Crave (immediate, on-demand messages aimed at helping the participant through a craving). A project Web site (StopMySmoking.com) provided additional quitting resources, technical support, and a discussion forum.Control group received similar number of text messages, message content was aimed at improving sleep and exercise habits within the context of how it would help the participant quit smoking. Messages

162

Details Method Results CommentsIndividual smoking cessation support

were not tailored based on quitting stage nor were Text Buddy and Text Crave components available

163

7.7 Appendix G Quality assessment

Study Study design#

Quality of execution##

Gen

eral

isab

ility

+

Rep

rese

ntat

iven

ess*

Ran

dom

isat

ion*

*

Com

para

bilit

y***

Cre

dibi

lity

of

da

ta

colle

ctio

n in

stru

men

ts†

Attr

ition

rate

††

Attr

ibut

abili

ty

to

inte

rven

tion†

††

Smoking restriction in schools, workplaces, and other public placesAkhtar 2010 1.2 yes n/a n/a yes yes yes nationalGalan 2012 1.1 yes n/a n/a yes n/a regionalMacKay 2010 1.2 yes n/a n/a yes yes nationalMillett 2013 1.2 yes n/a n/a yes yes nationalMoore 2011 1.2 yes n/a n/a yes yes yes nationalMoore 2012 1.2 yes n/a n/a yes yes yes nationalNabi-Burza 2012 1.1 n/a n/a yes n/aNoach 2012 1.1 n/a n/a yes n/aWoodruff 2000 2.1 yes n/a n/a yes nationalControls on advertising, promotion and marketing of tobaccoGilpin & Pierce 1997 1.1 yes n/a n/a yes n/a nationalHammond 2011 3.1 yes yes n/a yesPucci 1998 1.1 n/a n/a yes n/aMass media campaignsVallone 2009 1.2 n/a n/a yes yesIncreases in price/tax of tobacco productsBiener 1998 1.1 yes n/a n/a n/a regionalGilpin & Pierce 1997 1.1 yes n/a n/a yes n/a nationalGlied 2002 1.4 n/a n/a yes yesGruber 2000 1.4 yes n/a n/a yes yes nationalMadden 2007 1.3 n/a n/a yesPerretti-Watel 2010 1.1 yes n/a n/a n/a regionalControls on access to tobacco productsKim 2006 1.3 yes n/a n/a yes yes nationalLipperman-Kreda 2012

1.1 n/a n/a yes n/a yes

Millett 2011 1.2 yes n/a n/a yes yes nationalSchneider 2011 1.2 n/a n/a yes n/a yesWidome 2012 1.1 yes n/a n/a yes n/a regionalSchool-based preventionBacon 2001 3.1 yes yes yes yesCampbell 2008++ 3.1 yes yes yes yes yesCrone 2003++ 3.1 yes yes yes

Study Study design#

Quality of execution##

Gen

eral

isab

ility

+

Rep

rese

ntat

iven

ess*

Ran

dom

isat

ion*

*

Com

para

bilit

y***

Cre

dibi

lity

of

da

ta

colle

ctio

n in

stru

men

ts†

Attr

ition

rate

††

Attr

ibut

abili

ty

to

inte

rven

tion†

††

De Vries 2006++ 3.1 yes yes yes yesMenrath 2012 3.3 yes yes yesMultiple policy interventionsHelakorpi 2008 1.2 yes n/a n/a yes yes nationalPabayo 2012 1.3 n/a n/a yes yesWhite 2008 1.2 yes n/a n/a yes nationalIndividual smoking cessation supportRodgers 2005 3.1 yes yes yes yesYbarra 2013 3.3 yes yes yes++study identified in Mercken 2012

#Study designs see Table 1

## Quality of execution*Representativeness: Were the study samples randomly recruited from the study population with a response rate of at least 60% or were they otherwise shown to be representative of the study population?**Randomisation: Were participants, groups or areas randomly allocated to receive the intervention or control condition?***Comparability: Were the baseline characteristics of the comparison groups comparable or if there were important differences in potential confounders were these appropriately adjusted for in the analysis? If there is no comparison group this criterion cannot be met.†Credibility of data collection instruments: Were data collection tools shown to be credible, e.g. shown to be valid and reliable in published research or in a pilot study, or taken from a published national survey, or recognized as an acceptable measure (such as biochemical measures of smoking).††Attrition Rate: Were outcomes studied in a panel of respondents with an attrition rate of less than 30% or were results based on a cross-sectional design with at least 200 participants included in analysis in each wave?†††Attributability to intervention: Is it reasonably likely that the observed effects were attributable to the intervention under investigation? This criterion cannot be met if there is evidence of contamination of a control group in a controlled study. Equally, in all types of study, if there is evidence of a concurrent intervention that could also have explained the observed effects and was not adjusted for in analysis, this criterion cannot be met.+ Generalisability: Is the study generalisable at National, State/Regional, or Local level?Randomisation and comparability are not applicable (N/A) for all study designs except controlled trials coded 3.1, 3.2 or 3.3. Attrition rate is N/A to cross-sectional studies coded 1.1.

165

7.8 Appendix H Summary of equity impact of youth polices/interventionsAuthor, year

Age Study design Setting, country

SES variable policy/intervention

outcome Equity impact

Smoking restriction in schools, workplaces and other public placesAkhtar 2010

11 Repeat cross-sectional

Primary schools, Scotland

SEC, FAS Smokefreenational legislation

SHS exposure Greatest absolute reduction for low SES, but relative inequalities may have widened

Galan2012

15-16 Cross-sectional

Secondary schools,Spain

Census tractof school, parental education

Voluntary compliance

Smoking on school premises

A higher probability of smoking on school premises among adolescents whose fathers had a lower level of educational attainment. However, at the school level there was no significant impact

MacKay 2010

0-14 Repeat cross-sectional

Hospitals, Scotland

Area deprivation score (IMD)

Smokefree national legislation

Admission rates There were no significant interactions between hospital admissions for asthma and quintile of SES. All SES subgroups associated with significant reduction in admissions

Millett 2013 0-14 Interrupted time series

Hospitals, England

Area deprivation score (IMD)

Smokefree national legislation

Admission rates Significant and similar reductions in asthma admission rates among children from different SES groups

Moore 2011

11 Repeat cross-sectional

Primary schools, Wales

FAS Smokefree national legislation

SHS exposure Reductions limited to children from more affluent households, whose exposure was already significantly lower prior to legislation, leading to increased socioeconomic disparity

Moore 2012

11.2 Repeat cross-sectional

Primary schools, Scotland, Northern Ireland, Wales

FAS Smokefree national legislation

SHS exposureSmoking restrictions in the home and car

Declines in exposure occurred predominantly among children with low exposure before legislation, and from more affluent families. Substantial socioeconomic gradients in proportions of children

166

Author, year

Age Study design Setting, country

SES variable policy/intervention

outcome Equity impact

with higher SHS exposure levels remained unchanged.No change in inequality following legislation for home and car-based smoking restrictions (socioeconomic patterning remained stable).

Nabi-Burza 2012

0-18? Single cross-sectional

Paediatric practices, USA

Parental education

Voluntary smokefree car policy

smoking behaviour in cars and home

Parental education level was not significantly associated with strictly enforced smokefree car policy on its own, only significant in interaction with child age and amount smoked. College educated parents of children aged <1 year were more likely to have strict smoke-free car policies.

Noach 2012

15 Cross-sectional

Secondary schools, Israel

Parental education

Voluntary compliance

SHS exposure Parental education was a significant predictor of smoking in the home but not at school, exposure was significantly greater amongst adolescents whose parents had less education

Woodruff 2000

19 Before and after experimental study

US Navy recruitment centre, females only

Education 8-week 24-hour smoking ban

Smoking relapse

Education did not predict smoking relapse

Controls on advertising, promotion and marketing of tobaccoGilpin & Pierce 1997

14-21 in 1979-89

Cross-sectional

US population surveys

Education Tobacco marketing

Smoking initiation

Level of education impacted on initiation rates with initiation rates highest among high school dropouts and lowest amongst those who eventually attended college

Hammond 18-19 RCT, US online Education Cigarette Brand appeal Reactions to/perceptions of different

167

Author, year

Age Study design Setting, country

SES variable policy/intervention

outcome Equity impact

2011 convenience sample

survey, females only

packaging types of packs was the same by SES for nearly all the measures

Pucci 1998 5-1415-19

Cross-sectional

US field observation

Median household income

Advertising buffer zones around schools

Advertising density

Neighbourhoods with the lowest median household incomes showed highest advertising density inside school buffer zones

Mass media campaignsVallone 2009

12-17 Cross-sectional

US population survey

Median household income, median household education at zip code level

American Legacy Foundation’s truth® campaign

Awareness,receptivity to the campaign

Youth who lived in zip codes in which the median household income was less than or equal to US$ 35,000 had a lower level of confirmed awareness than respondents in other income categories. Zip code level median household income was not associated with confirmed awareness and there were no differences in receptivity by zip code level income or education

Increases in price/tax of tobacco productsBiener 1998

12-17 Cross-sectional

US statewide survey

Household income

Cigarette tax increase

Smoking behaviour

Low-income teenagers more likely than more affluent teens to cut costs by cutting down on smoking or (less often) by switching to cheaper brands. Young low-income smokers were not more likely than wealthier teenagers to consider quitting

Gilpin & Pierce 1997

14-21 in 1979-1989

Cross-sectional

US population surveys

Education Cigarette tax increase

Smoking initiation

Level of education impacted on initiation rates with initiation rates highest among high school dropouts and lowest amongst those who eventually attended college

Glied 2002 14-23 Cohort with US Family income Cigarette tax Smoking Tax at age 14 had a statistically

168

Author, year

Age Study design Setting, country

SES variable policy/intervention

outcome Equity impact

in 1979 longitudinal and cross-sectional data

population survey

increase behaviour significant negative effect on current smoking including initiation for low income people. Elasticities declined over time for low income people. By age 39 the effect of taxes at age 14 has largely disappeared.

Gruber 2000

13-18 Cross-sectional, econometric

US surveys Parental education

Cigarette prices, clean air, access

Price elasticity Price is the most important determinant of smoking by teens aged 16-18 years but not younger teenagers. Sensitivity to prices increases for those with less educated parents, sensitivity to price intensity increased for those with more educated parents

Madden 2007

19 Retrospective longitudinal

Ireland, survey of women

Education Cigarette taxes Smoking initiation

Increased cigarette prices were associated with later initiation among those with an intermediate education, but not those with only a primary education.

Perretti-Watel 2010

19.5 Cross-sectional

France, regional survey, university students

Parental education

Tobacco price increase

Smoking behaviour

Smokers with a lower SES were less likely to react to the price increase

Controls on access to tobacco productsKim 2006 15 Cohort US school-

based survey, females only

Parental education,parental income

Statewide tobacco control policies, statewide cigarette excise tax

Smoking behaviour

Stronger state level tobacco policies on age of sale were associated with lower likelihood of smoking initiation and adverse transition among low SES girls, although the effect sizes were small.

Lipperman-Kreda-2012

19 Cross-sectional

Tobacco retailers, California,

percentage of population with a

Underage tobacco sales laws

Compliance Higher education was a significant predictor of underage tobacco sales.

169

Author, year

Age Study design Setting, country

SES variable policy/intervention

outcome Equity impact

USA college education,median household income

Millett 2011 13 Repeat cross-sectional

Secondary schools, England

Free school meals Legislation in England, Scotland and Wales increasing minimum age for legal purchase of tobacco from 16 to 18 years

Smoking behaviour

Significant reduction in regular smoking among youth, regular smoking was not significantly different in pupils eligible for FSM compared with those that were not. Higher access reported for other sources by FSM eligible pupils. Following increase in age of sale significant reduction in access in non-FSM but not FSM.

Schneider 2011

17.6% aged 0-20

Before & after City-wide (Cologne), Germany, observational

Income, unemployment, social welfare, low-qualifying schools

Electronic locking devices on vending machines to prevent underage (<16 years) purchasing

Density of vending machines

The lower the income level in a district, the higher the availability of cigarettes, significant difference both before and after locking devices

Widome 2012

15-18 Cross-sectional

Licensed tobacco vendors, Minnesota, US

Below 150% poverty level

Age-of-sale tobacco checks

Compliance There was no association between store advertising characteristics or poverty and stores’ compliance check failure.

School-based preventionBacon 2001

11 Cluster RCT Middle schools, Florida, US

Free/reduced lunch status

‘Too Good for Drugs II’

Intentions, attitudes and perceptions towards tobacco use

Programme was similarly effective in impacting students risk and protective factors regardless of SES

Campbell 2008*

12-13 RCT England & Wales

FAS, FSM Peer-led social network based smoking uptake

Smoking in past week

The results were mixed depending on the specific SES indicator used. The intervention was most effective

170

Author, year

Age Study design Setting, country

SES variable policy/intervention

outcome Equity impact

prevention for adolescents in the Valley schools, located in a deprived area, particularly low SES girls.

Crone 2003*

13 RCT Netherlands Parental education

Anti-smoking class-based intervention

Experimenting with smoking or smoking daily or weekly

The intervention had a significant effect among higher SES adolescents and appeared to widen the inequalities in the short-term. All significant intervention effects disappeared at 12 months follow-up.

De Vries 2006*

13.5 RCT Portugal Spending money Smoking prevention school policies, also parents and community

Ever/never smoking

The results were mixed depending on the SES indicator used. When using spending money as a SES indicator, the intervention appeared to decrease inequalities in smoking but results unclear due to small number in ‘mid to high’ spending money subgroup and use of ‘spending money’ as proxy measure of SES.

Menrath 2012

12 Quasi randomised

Public secondary general schools, Northern Germany

FAS Two validated life skills programmes including element of smoking prevention

Self-report cigarettes smoked per week,30-day smoking prevalence

The two school-based life skills programmes had a positive effect on smoking prevention and benefitted children of all SES equally.

Multiple policy interventionsHelakorpi 2008

13-20 Repeat cross-sectional

Finland, national postal survey

Occupation 1976 Tobacco Control Act (smokefree, age of sale, health warnings)

Smoking prevalence (ever smoked daily for at least one year)

1976 TCA appears to have had the greatest impact on male white collar employees. Among women, the apparent effect was very pronounced in all socioeconomic groups and among blue collar female workers the cohort trend

171

Author, year

Age Study design Setting, country

SES variable policy/intervention

outcome Equity impact

tended to decline.Pabayo 2012

12.7 Cohort, convenience sample

Canada, school-based observational study

Household income

Smoking intolerance in schools, restaurants and corner stores near schools

Smoking initiation

No significant impact on smoking initiation by SES for smoking intolerance in schools, restaurants or corner stores

White 2008 12-17 Cross-sectional

Australia, school-based survey

Area-based Index of Relative Socio-Economic Disadvantage (IRSD)

3 periods of tobacco control activity: low tobacco-control funding (1992-1996), high tobacco-control activity (1984-1991, and 1997-2005) which included smoking restrictions and increased tax

Smoking prevalence

The magnitude of the decreases in smoking prevalence between 1996 and 2005 did not differ significantly between SES groups for most indicators of smoking behaviour. Less impact on younger low SES in period of low tobacco control funding.

Individual smoking cessation supportRodgers 2005

25 (mean)

RCT New Zealand, any setting, any location

Income level Text-messaging Smoking cessation

Text messaging doubled quit rates and this effect was consistent across major subgroups including income level

Ybarra 2013

22 (mean)

RCT USA, national

Enrolment in higher education

Text-messaging Continuous abstinence,7-day point prevalence

Significant increase in quit rates in intervention group vs control group at 6-weeks were not sustained at 3 months. Tailored text messaging appeared to benefit youth not enrolled in higher education.

*study identified in Mercken 2012

172

173

7.9 Appendix I Equity impact model of youth policies/interventions by SES measure

First author Income Area-level deprivation Education Occupationpos neu neg pos neu neg pos neu neg pos neu neg

Smoking restriction in cars, schools, workplaces and other public placesAkhtar 2010 Galan 2012 MacKay 2010 Millett 2013 Moore 2011 Moore 2012 Nabi-Burza 2012 Noach 2012 Woodruff 2000 Controls on advertising, promotion and marketing of tobaccoGilpin 1997 Hammond 2011

Pucci 1998 Mass media campaignsVallone 2009

Increases in price/tax of tobacco productsBiener 1998 Gilpin 1997 Glied 2002 Gruber 2000 Madden 2007 Perretti-Watel 2010 Controls on access to tobacco productsKim 2006 Lipperman-Kreda 2012

Millett 2011 Schneider 2011 Widome 2012 School-based preventionBacon 2001

Campbell 2008** Crone 2003** DeVries 2006** Menrath 2012 Multiple policy interventionsHelakorpi 2008 Pabayo 2012 White 2008 Individual smoking cessation supportRodgers 2005

174

First author Income Area-level deprivation Education Occupationpos neu neg pos neu neg pos neu neg pos neu neg

Ybarra 2013 **Study identified in Mercken 2012

This equity impact model of youth studies should be read in conjunction with the text in

Section 2.2.5

This matrix is based upon a hypothesis-testing model adapted from a model used in the York

review16:

The null hypothesis of a neutral equity impact that for any given socio-economic characteristic related to education, occupation or income, there is no social gradient in the effectiveness of the intervention.

The hypothesis of a positive equity impact defined as evidence that groups such as lower occupational groups, those with a lower level of educational attainment, the less affluent, those living in more deprived areas, are more responsive to the intervention.

The hypothesis of a negative equity impact defined as evidence that groups such as higher occupational groups, those with a higher level of educational attainment, the more affluent, or those who live in more affluent areas are more responsive to the intervention.

Key to symbol colour= “hard outcome” such as smoking prevalence or consumption; = “intermediate outcome” such as beliefs and attitudes

Neu = evidence supports null hypothesis i.e. neutral equity impactPos = evidence supports hypothesis of positive equity impactNeg = evidence supports hypothesis of negative equity impact

175