Population tobacco control interventions and their effects on social inequalities in smoking: systematic review

Transcription

1 Additional material inluding a table on study suitability and quality is published online only at ontent/vol17/issue4. 1 MRC Soial and Publi Health Sienes Unit, Glasgow G12 8RZ, UK; 2 Centre for Reviews and Dissemination, University of York, York YO10 5DD, UK; 3 PEHRU, London Shool of Hygiene and Tropial Mediine, London WC1E 7HT, UK; 4 Division of Publi Health, University of Liverpool L69 3GB, UK; 5 MRC Epidemiology Unit, Cambridge CB2 0QQ, UK Correspondene to: Sian Thomas, /o MRC Soial and Publi Health Sienes Unit, Glasgow G12 8RZ, UK; sian.thomas@btinternet.om Reeived 20 Otober 2007 Aepted 7 April 2008 Published Online First 19 May 2008 This paper is freely available online under the BMJ Journals unloked sheme, see tobaoontrol.bmj.om/info/ unloked.dtl Population tobao ontrol interventions and their effets on soial inequalities in smoking: systemati review S Thomas, 1 D Fayter, 2 K Misso, 2 D Ogilvie, 1,5 M Pettirew, 3 A Sowden, 2 M Whitehead, 4 G Worthy 2 ABSTRACT Objetive: To assess the effets of population tobao ontrol interventions on soial inequalities in smoking. Data soures: Medial, nursing, psyhologial, soial siene and grey literature databases, bibliographies, hand-searhes and ontat with authors. Study seletion: Studies were inluded (n = 84) if they reported the effets of any population-level tobao ontrol intervention on smoking behaviour or attitudes in individuals or groups with different demographi or soioeonomi harateristis. Data extration: Data extration and quality assessment for eah study were onduted by one reviewer and heked by a seond. Data synthesis: Data were synthesised using graphial ( harvest plot ) and narrative methods. No strong evidene of differential effets was found for smoking restritions in workplaes and publi plaes, although those in higher oupational groups may be more likely to hange their attitudes or behaviour. Smoking restritions in shools may be more effetive in girls. Restritions on sales to minors may be more effetive in girls and younger hildren. Inreasing the prie of tobao produts may be more effetive in reduing smoking among lower-inome adults and those in manual oupations, although there was also some evidene to suggest that adults with higher levels of eduation may be more prie-sensitive. Young people aged under 25 are also affeted by prie inreases, with some evidene that boys and non-white young people may be more sensitive to prie. Conlusions: Population-level tobao ontrol interventions have the potential to benefit more disadvantaged groups and thereby ontribute to reduing health inequalities. Reduing soial inequalities in health is a priority for health poliy in many ountries. 1 Although the extent and auses of health inequalities have been extensively researhed, we know remarkably little about the atual effets of measures to redue suh inequalities, 2 and it is possible that a strategy that improved health in the population overall might atually widen inequalities between soial groups if its benefits were onentrated among the betteroff. 3 Smoking has been shown to be a major ontributor to soial inequalities in mortality and is the single greatest ontributor to preventable illness and premature death in the United Kingdom. 4 5 The importane of interventions to redue the assoiation of smoking with disadvantage is well reognised 6 and is refleted, for example, in the target set by the Department of Health to redue the prevalene of smoking in manual groups from 32% to 26% by Smokers from lower soioeonomi groups may be less likely than those from higher soioeonomi groups to quit as a result of partiipating in individually targeted approahes suh as smoking essation servies, although this soial gradient in quit rates may be offset by a greater penetration of smoking essation servies in disadvantaged areas. 8 The potential ontribution of population-level interventions, suh as restritions on tobao advertising and on smoking in publi plaes, to reduing soial inequalities in smoking has been less well researhed. 9 We arried out a systemati review of the differential effets of population-level tobao ontrol interventions by evaluating their effets in groups with different demographi and soioeonomi harateristis. Our overall aim was to identify whih interventions are most likely to be effetive in reduing smoking-related health inequalities. METHODS Searh strategy We identified primary studies in any language by searhing medial, nursing, psyhologial, soial siene and grey literature databases from their ineption dates to January We did not limit our searhes by study design. We also examined bibliographies and onferene abstrats, handsearhed key journals and ontated authors for additional information where neessary. Further details an be found in our full report at Study seletion and inlusion riteria Titles and abstrats were assessed for relevane independently by two reviewers. Potentially relevant studies were assessed for inlusion independently by two reviewers, with disagreements resolved through disussion and, where neessary, the involvement of a third reviewer. We inluded studies of any design that assessed the effets of a population-level tobao ontrol intervention (see box) in smokers, people at risk of taking up smoking, people at risk of exposure to environmental tobao smoke (ETS) or the general population. Studies had to report quantitative outomes for individuals or groups with different demographi or soioeonomi harateristis. 230 Tobao Control 2008;17: doi: /t

2 Eligible outomes inluded hanges in smoking behaviour (suh as prevalene or onsumption), indiret measures of tobao onsumption (suh as illegal sales to minors or quantity of smuggled igarettes), exposure to ETS, intermediate outomes (suh as hanges in knowledge or attitudes), proess measures (suh as partiipation rates), implementation measures (suh as enforement of poliy hanges) and any health outomes (suh as mental health or wellbeing), as well as adverse or unintended effets. We also inluded qualitative data where these were linked to an inluded quantitative study. We exluded studies of interventions onduted exlusively within losed settings (suh as psyhiatri or addition treatment failities, detention entres or prisons) beause this review was onerned with effets in the wider population. We also exluded studies that assessed the effets of restritions on sales to minors (youths) by only reporting test purhases as outomes. This is beause we onsidered the minors undertaking the test purhases at retail outlets to be part of the intervention, their purhase attempts being a devie for evaluating the implementation and enforement of the intervention. Suh test purhases alone did not provide suffiient data for our purposes on the differential effets of an intervention between soial groups. We did, however, inlude studies that assessed the effets of restritions on sales to minors by reporting evaluation data from a larger population (suh as surveys of loal shoolhildren). Data extration and quality assessment Data were extrated and the quality of eah study was assessed independently by one reviewer and heked by a seond. We What is a population-level tobao ontrol intervention? We defined population-level tobao ontrol interventions as those applied to populations, groups, areas, jurisditions or institutions with the aim of hanging the soial, physial, eonomi or legislative environments to make them less onduive to smoking. These are approahes that mainly rely on state or institutional ontrol, either of a link in the supply hain or of smokers behaviour in the presene of others. Our definition was based on our pilot study 10 and soping searhes for the systemati review and inludes interventions suh as: Tobao rop substitution or diversifiation Removing subsidies on tobao prodution Restriting trade in tobao produts Measures to prevent smuggling Measures to redue illiit ross-border shopping Restriting advertising of tobao produts (Enforing) restritions on selling tobao produts to minors Mandatory health warning labels on tobao produts Inreasing the prie of tobao produts Restriting aess to igarette vending mahines Restriting smoking in the workplae Restriting smoking in publi plaes. Suh approahes ould also form part of wider, multifaeted interventions in shools, workplaes or ommunities. We did not inlude interventions whose main aim was to strengthen the apaity of individuals to stop smoking or to resist taking up smoking, even if these interventions were applied to whole groups or populations (for example, mass media health eduation ampaigns). These are approahes that mainly rely on individuals engaging voluntarily with measures intended to help them. summarised study quality using a sale of suitability of study design adapted from the riteria used for the Community Guide of the US Task Fore on Community Preventive Servies 11 and a six-item heklist of quality of exeution adapted from the riteria developed for the Effetive Publi Health Pratie Projet in Hamilton, Ontario 12 (see table on Tobao Control website). We extrated outome, proess and implementation data stratified by the soiodemographi harateristis speified in the PROGRESS riteria (plae of residene, rae or ethniity, oupation, gender, religion, eduational level, soioeonomi status (for example, represented by inome), and soial apital) 13 and also by age for interventions targeted at populations onsidered speifially at risk of smoking beause of their age (adolesents and young adults). For studies where it appeared that relevant data on differential effets may have been olleted but not reported, we ontated authors to request additional data. Data from qualitative studies were extrated using methods adapted from those developed by Britten et al 98 and their quality was assessed using published prompts for appraising qualitative researh. 99 Any disagreements at eah stage were resolved by disussion and, if neessary, the involvement of a third member of the review team. Data synthesis We adopted a hypothesis-testing approah to examine the balane of evidene about the differential effets of interventions and synthesised the data using a ombination of graphial and narrative methods, inluding a novel matrix or harvest plot (see fig 2). 100 For eah ategory of intervention and dimension of inequality, we populated the relevant row of this matrix by plaing a bar representing eah study in one of three olumns aording to whih of three ompeting hypotheses were most strongly supported by the results of that study: The null hypothesis that for any given demographi or soioeonomi harateristi there was no soial gradient in the effetiveness of the intervention The alternative hypothesis that there was a positive soial gradient in effetiveness, meaning that the intervention was more effetive in more advantaged groups (defined for this purpose as the more affluent, those with a higher level of eduation, those in more skilled oupational groups, males, older people or those in the majority or most advantaged raial or ethni group in the ontext of a partiular study) The alternative hypothesis that there was a negative soial gradient in effetiveness, meaning that the intervention was more effetive in more disadvantaged groups. RESULTS We sreened a total of referenes, identified 970 potentially eligible papers and finally inluded 84 studies (reported in 90 papers) (fig 1). We found only one qualitative study onduted in onjuntion with a quantitative study. 22 We approahed six authors for additional data but none was forthoming. We found relevant evidene for seven ategories of intervention: restritions on smoking in workplaes and publi plaes, restritions on smoking in shools, restritions on sales to minors, health warnings on tobao produts, restritions on advertising of tobao produts, prie of tobao produts and multifaeted interventions (see fig 2). Further details of the studies inluded in eah ategory an be found in our full report at a.uk/inst/rd/projets/tobao-ontrol.htm. Tobao Control 2008;17: doi: /t

3 Figure 1 Proess of study seletion. The inluded studies reported outomes by rae or ethniity, oupation, gender, eduational level, inome or age. As no studies reported outomes by plae of residene, religion or level of soial apital these harateristis were exluded from our analysis. Stronger designs tended to have been used for studies of the effets of restritions on smoking in workplaes, publi plaes and shools and restritions on sales to minors, of whih three were luster randomised ontrolled trials Studies of other types of intervention were predominantly ross-setional or retrospetive. Studies of restritions on sales to minors were the most likely to fulfil the riteria for quality of exeution, with one study meeting all six riteria 31 and two studies meeting five Two studies of restritions on smoking in shools met four riteria The remaining studies in this review met between zero and three of the riteria. Restritions on smoking in workplaes and publi plaes Fourteen studies, nine published between 1981 and 1999 and five published more reently, evaluated smoking restritions or bans in the workplae or in publi plaes in the United States, Australia, 15 New Zealand, 27 Israel, 17 Finland, 18 Sotland 22 and Wales. 19 The interventions onsisted of a total ban on indoor smoking, a smoking ban with exeptions, 22 restriting smoking to designated rooms or areas or displaying no-smoking signs in a hospital lobby. 16 The nature of the smoking ban was unlear in two studies. The balane of evidene from five omparatively weak studies suggested that, if anything, restritions on smoking in workplaes may be more effetive for staff in higher oupational grades. We found insuffiient evidene of differential effets by inome, 26 eduational level or ethniity, 27 inonsistent evidene of differential effets by age, and no evidene of differential effets by gender. Restritions on smoking in shools Three studies assessed the effets of restritions on smoking in shools, one published in and two published in These examined the effets of a smoking poliy in a UK shool, 29 student beliefs and support for a shool smoking ban in a mostly non-white population in California 30 and the effets of enforement ation on student smoking behaviour and attitudes in another US population. 28 These studies suggested that restritions on smoking in shools may be more effetive in girls than in boys 29 and in middle-shool than in high-shool students, 28 and that attitudes were more favourable in non- Hispani students than in Hispani students. 30 No studies provided evidene about possible differential effets by parental inome, oupation or eduational level. Restritions on sales to minors Thirteen studies, most published between 2000 and 2005, evaluated restritions on sales to minors in the United States, Sweden, 41 Finland, 37 Australia and New Zealand 35 in populations aged between 13 and 18 years of age. The interventions inluded eduation of retailers and the ommunity, enforement of legislation, or both. The evidene from two studies (one of an eduational intervention and one of ombined eduation and enforement) suggested that girls may be less likely to use tobao as a result of the intervention than boys. The evidene from six other studies (four of an enforement intervention and two of ombined eduation and enforement) on differential effets by gender was inonsistent One study of ombined eduation and enforement found that the intervention was less effetive in non-white students than in white students. 34 A seond weaker study of an enforement intervention found no evidene of differential effets by ethniity. 35 Three studies (two of an enforement intervention and one of ombined eduation and enforement) found larger effets in younger students than in older students Four other studies (one of an enforement intervention and three of ombined eduation and enforement) found inonsistenies in effets by age. No studies provided evidene about possible differential effets by parental inome, oupation or eduational level. Health warnings on tobao produts Five studies assessed the effets of health warnings and labelling of ontents on tobao produts in the general population, young adults 48 or shoolhildren. 49 Studies were published between 1997 and 2005 and were onduted in Australia, 46 Canada, the United States 49 and The Netherlands. 50 We found no onsistent evidene of differential effets on smoking behaviour by eduation for smoking behaviour or on smoking attitudes or behaviour by gender In three studies of young people, health warnings did not appear to hange attitudes or smoking behaviour No studies provided evidene about possible differential effets by inome, oupation or ethniity. Restritions on advertising of tobao produts Two studies assessed the effets of advertising restritions on hildren and young people. One study was set in Hong Kong 232 Tobao Control 2008;17: doi: /t

4 Figure 2 Evidene for soial gradients in effets of interventions. A supermatrix overing all ategories of intervention onsisting of six rows (one for eah dimension of inequality) and three olumns (one for eah of the three ompeting hypotheses about the differential effets of eah ategory of intervention). Eah study is represented by a mark in eah row for whih that study had reported relevant results. Studies with hard behavioural outome measures are indiated with full-tone (blak) bars, and studies with intermediate outome measures with half-tone (grey) bars. The suitability of study design is indiated by the height of the bar, where the highest bars represent the most suitable study designs (ategories A and B) and the lowest bars represent the least suitable (ategory D). Eah bar is annotated with the number of other methodologial riteria (maximum six) met by that study. and published in The other used national statistis from 1992 to assess smoking prevalene among adolesents in Norway, Finland, New Zealand and Frane. 45 We found no evidene of differential effets by gender or age. No studies provided evidene about possible differential effets by parental inome, oupation, eduational level or ethniity. Prie of tobao produts Forty-two studies provided information about the effets of the prie of tobao produts on smoking behaviour. Most were eonometri analyses applying statistial models to rosssetional or longitudinal survey data from various time periods between 1961 and These studies modelled the relation between the deision to smoke or the quantity of igarettes smoked and hanges in prie or tax. Most used survey data from the United States with 20 studies reporting data for adolesents or ollege students only and 13 reporting data for adults only or for young people and adults ombined Three studies were onduted in the United Kingdom while others were from Frane, 75 Spain, 73 Canada, 90 South Afria and Taiwan. Effets on adults Four studies found that igarette prie inreases had a greater effet in those on lower inomes Two UK studies found that effets on smoking were greater among those in manual oupations than those in professional oupations but a later UK study found no evidene of differential effets by oupation. 53 There was also some evidene to suggest that those with higher levels of eduation may be more sensitive to prie We found no lear evidene for differential effets by gender or ethniity. Effets on young people All 20 studies restrited to adolesents or ollege students found that these groups were sensitive to prie and onluded that inreasing the prie of tobao produts would redue youth smoking. The only study omparing hildren within different age groups found that those aged 17 or 18-years-old were more sensitive to prie inreases than those aged between 13 and 16-years-old. 68 Four studies found that boys aged were more sensitive to prie than girls All three studies whih examined effets by ethniity found that blak or Hispani adolesents were more affeted by prie inreases than their white ounterparts. No studies provided evidene about possible differential effets by parental inome, oupation or eduational level. Multifaeted interventions Five studies assessed the effets of ombinations of interventions, mainly the ombined effets of different anti-tobao laws Studies were published between 1997 and Two Tobao Control 2008;17: doi: /t

5 studies examined the impat of the 1976 National Tobao Control At in Finland. One study assessed the impat of Frenh legislation inluding restritions on smoking in the workplae, advertising restritions, health warnings on tobao produts and restritions on sales to minors. This study involved a survey of hospital employees, mainly female nurses and healthare workers. 93 One study assessed smoking restritions in Californian shools as part of an independent evaluation of the Californian Tobao Control Prevention and Eduation Program. 97 The fifth study assessed the effets of prie inreases and tobao ontrol legislation in Canada. 96 The effets of the omponents of these interventions were not assessed separately within the studies and we therefore lassified them as multifaeted interventions in our analysis. One study found that the introdution of a tobao ontrol at in Finland redued the rate of smoking initiation among young people. 94 We found no evidene of differential effets by gender (interventions in all four studies were effetive for both men and women) or ethniity (one study). 97 No studies provided evidene about possible differential effets by inome, oupation or eduational level. DISCUSSION Prinipal findings This review has systematially and omprehensively applied an equity lens to tobao ontrol interventions, re-examining the available evidene about the impat of poliy measures and other population-level interventions in order to assess their role in takling health inequalities. 101 The literature is international, with over half of the studies having been onduted in the United States and just six in the United Kingdom, and is dominated by eonometri analyses (half of the inluded studies) modelling the effets of the pries of tobao produts. Overall, we found no strong evidene that restritions in workplaes and publi plaes are more effetive in reduing smoking in more advantaged groups, although smoking behaviour and attitudes may be more favourably affeted among those in higher oupational grades. We found evidene from single studies that smoking restritions in shools may be more effetive in girls and in younger shoolhildren, but there was an absene of evidene with respet to other possible differential effets. We found more, better-quality evidene on the differential effets of restritions on sales to minors: restritions seem to be more effetive in girls and in younger shoolhildren, and one study of a ombined eduation and enforement intervention found restritions on sales to minors to be more effetive in white than non-white groups. For health warnings on tobao produts and restritions on tobao advertising, the lak of robust studies makes firm onlusions diffiult. The effets of health warnings do not appear to be subjet to a soiodemographi gradient, but their effets have not been examined with respet to inome, oupation or ethniity and the evidene with respet to eduational level, gender and age is not onvining. The effets of advertising bans also show no differential by gender or age, but the evidene is not strong and other potential gradients have not been examined in primary studies. The balane of eonometri evidene suggests that inreasing the prie of tobao is more effetive in reduing smoking in lower-inome adults and those in manual oupations. There was also some evidene to suggest that smokers with higher levels of eduation may be more responsive to prie, although this evidene was limited to somewhat speifi study populations (men in Taiwan and pregnant women in the United States, whose response to priing may be onfounded by knowledge of the risks of smoking during pregnany). The evidene with respet to differential effets by gender, ethniity or age is not onsistent. Although we found fewer studies assessing the effets of priing in hildren, it appears that boys, non-white hildren and perhaps also older hildren may be more prie-sensitive. We found no evidene as to how the effets on hildren varied by household inome. Strengths and weaknesses of the review We made extensive attempts to obtain both published and unpublished studies and to inlude a wide range of study designs in order to avoid overlooking evidene from weaker studies whih to date have mainly been exluded from systemati reviews. However, it remains possible that we have not identified all relevant tobao ontrol intervention programmes or poliies, given that some may not have been formally evaluated or reported. One diffiulty in dealing with a diverse publi health evidene base is the need to inorporate onsiderable heterogeneity in intervention, study design and appropriateness of that design, study quality and study outomes (in this ase, hard behavioural and softer attitudinal outomes). The stratifiation of outomes by soial group adds another level of omplexity. To manage this we developed a novel graphial method, the harvest plot, to synthesise and display the balane of evidene to support ompeting hypotheses about possible soial gradients in the effets of the interventions. This methodologial development is a onsiderable strength of the review and may be of use to others reviewing the publi health literature; the rationale for this method and its advantages and disadvantages are disussed in a separate methodologial paper. 100 Strengths and weaknesses of the available evidene There are undoubted limitations in the evidene base, most notably a lak of prospetive evaluations. A partiular hallenge is the diffiulty of attributing outomes solely to the intervention in question. Authors often did not report o-interventions or desribe other ontextual fators that might have influened the suess of the intervention. Although we exluded studies fousing solely on individual-level interventions, population tobao ontrol poliies rarely exist in isolation and several studies inluded individual-level interventions suh as smoking essation lasses alongside workplae smoking bans. A deision to intervene at one level (poliy) ould be adversely affeted by ations at other levels; alternatively, there ould be a synergisti effet. 102 Contextual information would also help poliy-makers and pratitioners better understand how suessful interventions ould be implemented. 103 The ompleteness and larity of reporting in primary studies in this field would also be improved by the inlusion of more methodologial details (suh as study design, sampling, population harateristis, data olletion tools, methods of analysis and attrition rates), by assessing the differential impat of interventions aross different soiodemographi groups and by reporting data on hanges in smoking behaviour rather than relying on hanges in attitudes whih may be a poor proxy for behaviour hange. One of the more obvious limitations is the absene of qualitative researh on population-level tobao interventions and their effets on soial inequalities in smoking. Although we sought suh studies, we found only one. New 234 Tobao Control 2008;17: doi: /t

6 qualitative researh will also have an important part to play in identifying intended and unintended effets of poliy interventions and barriers to hange before implementation. 102 Impliations for poliy and pratie The urrent EU green paper on poliy options for progressing towards a smoke-free Europe notes that smoke-free poliies may redue soioeonomi inequalities in health and alls for qualitative and quantitative evidene on the impats of suh poliies. 104 Our systemati review addresses this all, ontributes a step towards understanding the interventions that are effetive for different soial groups and may inform deisions about takling soial inequalities in smoking. The most ompelling evidene of a soial gradient in effetiveness whih favours the least well off is for the prie of tobao produts; although we also found some evidene to suggest an apparently greater effet of prie on those with higher levels of eduation, suh evidene is limited and requires further investigation. Inreasing the prie of tobao is therefore the population-level intervention for whih we found the strongest evidene as a measure for reduing smoking-related inequalities in health. However, the effets of inreasing tobao taxation may be undermined by tax-evasion or tax-avoidane measures suh as smuggling and ross-border shopping. 105 The Aheson inquiry 106 and other ommentators have also raised onern about the long-term effet of prie rises on disadvantaged households, where smokers are more likely to be niotine-dependent and for whom living in hardship is the primary deterrent to quitting. Any further inrease in tobao taxation would therefore require extra measures to support essation among low-inome households. None the less, we found more evidene to support inreasing the prie of tobao produts than to support other more visible interventions suh as health warnings and advertising restritions, whose differential effets appear under-explored. However, although interventions suh as health warnings and advertising restritions may not in themselves affet inequalities, they may be important as part of a wider tobao ontrol strategy, if they help to eliit publi support for other measures. 109 The evidene on restritions on sales to minors suggests that these may be effetive in deterring younger smokers, though their effetiveness depends on enforement as unenfored voluntary agreements with retailers are less effetive in reduing sales. 105 Priing may be less effetive among some groups of younger smokers, perhaps beause they may obtain their igarettes from non-ommerial soures. 105 Among younger smokers restritions in shools (whih affet onsumption) and health warnings (whih affet attitudes to smoking) may therefore be more produtive. Appropriately enfored restritions on sales to minors may offer the greatest promise as part of a strategy for takling inequalities. While ombinations of interventions are also likely to be an important part of the poliy armoury inluding restritions in shools (whih affet onsumption) and health warnings (whih affet attitudes to smoking) the differential effets of suh ombinations largely remain an area for further researh. It is also important to identify poliies that have the potential to inrease inequalities. Our findings are enouraging, as we found little evidene of adverse effets in this regard. One exeption was workplae restritions, whih may be more effetive among higher oupational grades. This suggests that the implementation of suh poliies should be aompanied by measures to promote adherene aross all oupational grades. What is already known on this subjet Reduing soial inequalities in smoking and its health onsequenes is a publi-health and politial priority. Little is known about the atual effets of measures to redue health inequalities in general or about the differential impats of tobao ontrol measures in partiular. It is possible that a strategy whih suessfully redued smoking in the population overall might widen inequalities if its benefits were onentrated among the better-off. What this study adds This is the most omprehensive review to date of the potential effets on heath inequalities of population-level tobao ontrol interventions and makes an important ontribution towards understanding the effets of interventions in different soial groups. In terms of reduing soial inequalities in smoking, we found better evidene to support inreasing the prie of tobao produts than to support more visible interventions suh as health warnings and advertising restritions. We found little evidene of poliies that have the potential to inrease inequalities. In partiular, we found no strong evidene that smoking restritions in workplaes and publi plaes are more effetive among more advantaged groups. This supports the ase for legislating for mandatory workplae bans, rather than relying on willing employers to introdue voluntary bans. Unanswered questions and future researh We have identified many gaps in the evidene base on interventions to redue soial inequalities in smoking. In partiular, we know little about the differential effets of most ategories of intervention by inome, gender or ethniity. For tobao priing a relatively well researhed field we also need to know more about effets on adolesents from lower-inome households and on young people in general, and on lowerinome adults who are likely to be niotine-dependent. For restritions on sales to minors another relatively well researhed field it is unlear whether differential effets vary between interventions that involve eduation, enforement or both. Where population-level studies are arried out there ould be greater use of pre-planned subgroup analyses, speifially to shed light on effets on inequalities, but there also remains a need for robust evaluations of targeted interventions (even aepting that these may not provide evidene about effets on inequalities). Perhaps the most important observation is that muh of the existing evidene derives from the United States. The greatest researh priority should therefore be to develop relevant evidene for other ountry ontexts with a fous on behavioural outomes. The introdution of new populationlevel tobao ontrol poliies suh as the restritions on smoking in publi plaes now introdued in all the ountries of the United Kingdom and elsewhere provides suh an opportunity. Aknowledgements: We thank Christine Godfrey, Hilary Graham, Gerard Hastings, Betsy Kristjansson, Johan Makenbah, Alan Marsh, Steve Platt, George Thomson and Tobao Control 2008;17: doi: /t

7 Peter Craig for their omments and suggestions on drafts of the study protool and reports; James Coates for the design and onstrution of the Aess database for the review; and Caroline Main for assistane with sreening searh results, assessment of studies for inlusion and design of the data extration form. Funding: This review was funded by the Department of Health Poliy Researh Programme (PRP) (referene number RDD/030/077). This work was undertaken by all the authors, who reeived funding from Department of Health Poliy Researh Programme. The views expressed in the publiation are those of the authors and not neessarily those of the Department of Health. MP was funded by the Chief Sientist Offie of the Sottish Exeutive Health Department. DO was funded by a Medial Researh Counil fellowship. The authors work was independent of the funders. Competing interests: None. Contributors: DO, AJS, MP and MW designed the study. DO designed and populated the harvest plot. KM onduted the literature searhes. ST, DF and GW sreened the searh results, assessed studies for inlusion, onduted data extration and quality assessment and synthesised the data. ST, DF, GW, MP and AJS heked data extration and quality assessment. All authors ontributed to the interpretation of findings for researh and poliy. ST wrote the first draft of the manusript and all other authors ontributed to its ritial revision and approved the final version. ST is guarantor. REFERENCES 1. Leon D, Walt G, Gilson L. International perspetives on health inequalities and poliy. BMJ 2001;322: Tugwell P, Kristjansson B. Moving from desription to ation: hallenges in researhing soio-eonomi inequalities in health. Can J Publi Health 2004;95: Maintyre S, Chalmers I, Horton R, et al. Using evidene to inform health poliy: ase study. BMJ 2001;322: Jha P, Peto R, Zatonski W, et al. Soial inequalities in male mortality, and in male mortality from smoking: indiret estimation from national death rates in England and Wales, Poland, and North Ameria. Lanet 2006;368: Department of Health. The NHS aner plan: a plan for investment, a plan for reform. London: Stationery Offie, Jarvis JD, Wardle J. Soial patterning of individual health behaviours: the ase of igarette smoking. In: Marmot M, Wilkinson RG, eds. Soial determinants of health. 2nd ed. Oxford: Oxford University Press, 2006: Department of Health. Delivering the NHS Caner Plan. Caner prevention: smoking. London: DH, Bauld L, Judge K, Platt S. Assessing the impat of smoking essation servies on reduing health inequalities in England: observational study. Tob Control 2007;16: Dahlgren G, Whitehead M. Levelling up. Part II: European strategies to takle soial inequities in health: a disussion paper on European strategies for takling soial inequities in health. Copenhagen: World Health Organization, Available from Ogilvie D, Pettirew M. Reduing soial inequalities in smoking: an evidene inform poliy? A pilot study. Tob Control 2004;13: Briss P, Zaza S, Pappaioanou M, et al. Developing an evidene-based guide to ommunity preventive servies methods. Am J Prev Med 2000;18(1S): Thomas H. Quality assessment tool for quantitative studies. Hamilton, Ontario: Effetive Publi Health Pratie Projet, Available from myhamilton.a/nr/rdonlyres/04a24ebe-2c46-411d-aeba-95a60fdef5ca/0/ QualityTool2003.pdf. 13. Evans T, Brown H. Road traffi rashes: operationalizing equity in the ontext of health setor reform. Inj Control Saf Promot 2003;10: Beker D, Conner H, Waranh H, et al. The impat of a total ban on smoking in the Johns Hopkins Children s Center. JAMA 1989;262: Borland R, Owen N, Hoking B. Changes in smoking behaviour after a total workplae smoking ban. Aust J Publi Health 1991;15: Dawley HH, Morrison J, Carrol S. The effet of differently worded no-smoking signs on smoking behavior. Int J Addit 1981;16: Donhin M, Baras M. A smoke-free hospital in Israel a possible mission. Prev Med 2004;39: Heloma A, Jaakkola MS. Four-year follow-up of smoke exposure, attitudes and smoking behaviour following enatment of Finland s national smoke-free work-plae law. Addition 2003;98: Kassab J, Morgan G, Williams E, et al. Smoking prevalene and attitudes of Gwynedd Health Authority staff towards passive smoking and the authority s nonsmoking poliy. Health Trends 1992;24: Offord KP, Hurt RD, Berge KG, et al. Effets of the implementation of a smoking free poliy in a medial enter. Chest 1992;102: Olive KE, Ballard JA. Changes in employee smoking behavior after implementation of restritive smoking poliies. South Med J 1996;89: Parry O, Platt S. Smokers at risk: Impliations of an institutionally bordered riskredued environment. Health Plae 2000;6: Sorensen G, Rigotti NA, Rosen A. Effets of a worksite non-smoking poliy: evidene for inreased essation. Am J Publi Health 1991;81: Sorensen G, Beder B, Prible CR, et al. Reduing smoking at the workplae: implementing a smoking ban and hypnotherapy. J Oup Environ Med 1995;37: Stillman FA, Beker DM, Swank RT, et al. Ending smoking at the Johns Hopkins Medial Institutions: an evaluation of smoking prevalene and indoor air pollution. JAMA 1990;264: Tang H, Cowling DW, Lloyd JC, et al. Changes of attitudes and patronage behaviors in response to a smoke-free bar law. Am J Publi Health 2003;93: Waa A, Gillespie J. Reduing exposure to seond hand smoke: hanges assoiated with the implementation of the amended New Zealand Smoke-free Environments At 1990: Wellington: HSC Researh and Evaluation Unit, 2005: Kumar R, O Malley PM, Johnston LD. Shool tobao ontrol poliies related to students smoking and attitudes toward smoking: national survey results, Health Edu Behav 2005;32: Thrush D, Fife-Shaw C, Breakwell G. Evaluations of interventions to redue smoking. Swiss J Psyhol 1999;58: Trinidad DR, Gilpin EA, Piere JP. Compliane and support for smoke-free shool poliies. Health Edu Res 2005;20: Altman DG, Wheelis AY, MFarlane M, et al. The relationship between tobao aess and use among adolesents: a four ommunity study. So Si Med 1999;48: Forster JL, Murray DM, Wolfson M, et al. The effets of ommunity poliies to redue youth aess to tobao. Am J Publi Health 1998;88: Hinds MW. Impat of a loal ordinane banning tobao sales to minors. Publi Health Rep 1992;107: Jason LA, Pokorny SB, Shoeny ME. Evaluating the effets of enforements and fines on youth smoking. Crit Publi Health 2003;13: Laugesen M, Sragg R. Changes in igarette purhasing by fourth form students in New Zealand N Z Med J 1999;112: Livingood WC, Woodhouse CD, Sayre JJ, et al. Impat study of tobao possession law enforement in Florida. Health Edu Behav 2001;28: Rimpela AH, Rainio SU. The effetiveness of tobao sales ban to minors: the ase of Finland. Tob Control 2004;13: Siegel M, Biener L, Rigotti NA. The effet of loal tobao sales laws on adolesent smoking initiation. Prev Med 1999;29: Staff M, Marh L, Brnabi A, et al. Can non-proseutory enforement of publi health legislation redue smoking among high shool students? Aust N Z J Publi Health 1998;22: Staff M, Bennett CM, Angel P. Is restriting tobao sales the answer to adolesent smoking? Prev Med 2003;37: Sundh M, Hagquist C. Effets of a minimum-age tobao law Swedish experiene. Drug Edu Prev Poliy 2005;12: Thomson CC, Gokhale M, Biener L, et al. Statewide evaluation of youth aess ordinanes in pratie: effets of the implementation of ommunity-level regulations in Massahusetts. J Publi Health Manag Prat 2004;10: Tutt D, Bauer L, Edwards C, et al. Reduing adolesent smoking rates. Maintaining high retail ompliane results in substantial improvements. Health Promot J Austr 2000;10: Fielding R, Chee YY, Choi KM, et al. Delines in tobao brand reognition and ever-smoking rates among young hildren following restritions on tobao advertisements in Hong Kong. J Publi Health 2004;26: Joossens L. The effetiveness of banning advertising for tobao produts. Brussels: Union Internationale Contre le Caner, Borland R, Hill D. Initial impat of the new Australian tobao health warnings on knowledge and beliefs. Tob Control 1997;6: Gospodinov N, Irvine IJ. Global health warnings on tobao pakaging: evidene from the Canadian experiment. Top Eon Anal Pol 2004;4: Koval JJ, Aubut JA, Pederson LL, et al. The potential effetiveness of warning labels on igarette pakages: the pereptions of young adult Canadians. Can J Publi Health 2005;96: Robinson TN, Killen JD. Do igarette warning labels redue smoking?: paradoxial effets among adolesents. Arh Pediatr Adoles Med 1997;151: Willemsen MC. The new EU igarette health warnings benefit smokers who want to quit the habit: results from the Duth Continuous Survey of Smoking Habits. Eur J Publi Health 2005;15: Berg GD, Kaempfer WH. Cigarette demand and tax poliy for rae groups in South Afria. Appl Eon 2001;33: Bishai DM, Merer D, Tapales A. Can government poliies help adolesents avoid risky behavior? Prev Med 2005;40: Borren P, Sutton M. Are inreases in igarette taxation regressive? Health Eon 1992;1: Chaloupka F. Clean indoor air laws, addition, and igarette smoking. Appl Eon 1992;24: Chaloupka FJ. Rational additive behavior and igarette smoking. J Polit Eon 1991;99: Chaloupka FJ, Tauras JA, Grossman M. Publi poliy and youth smokeless tobao use. South Eon J 1997;64: Chaloupka FJ, Grossman M. Prie, tobao ontrol poliies and youth smoking. Cambridge, MA: National Bureau of Eonomi Researh, Working Paper Chaloupka FJ, Wehsler H. Prie, tobao ontrol poliies and smoking among young adults. Cambridge, MA: National Bureau of Eonomi Researh, Working Paper Tobao Control 2008;17: doi: /t

8 59. Colman G, Remler DK. Vertial equity onsequenes of very high igarette tax inreases: if the poor are the ones smoking, how ould igarette tax Inreases be progressive? Cambridge, MA: National Bureau of Eonomi Researh, Working Papers Czart C, Paula RL, Chaloupka FJ, et al. The impat of pries and ontrol poliies on igarette smoking among ollege students. Contemp Eon Poliy 2001;19: DeCia P, Kenkel D, Mathios A. Putting out the fires: will higher taxes redue the onset of youth smoking? J Polit Eon 2002;110: Delnevo CD, Hrywna M, Foulds J, et al. Cigar use before and after a igarette exise tax inrease in New Jersey. Addi Behaviors 2004;29: Ding A. Youth are more sensitive to prie hanges in igarettes than adults. Yale J Bio Med 2003;76: Emery S, White MM, Piere JP. Does igarette prie influene adolesent experimentation? J Health Eon 2001;20: Evans W, Farrelly M. The ompensating behavior of smokers: taxes, tar, and niotine. Rand J Eon 1998;29: Farrelly MC, Bray JW, Pehaek T, et al. Response by adults to inreases in igarette pries by soiodemographi harateristis. South Eon J 2001;68: Goel RK, Nelson MA. Tobao poliy and tobao use: differenes aross tobao types, gender and age. Appl Eon 2005;37: Gruber J. Youth smoking in the US: pries and poliies. Cambridge, MA: National Bureau of Eonomi Researh, Working Paper Katzman B, Markowitz S, MGeary KA. The impat of lending, borrowing, and antismoking poliies on igarette onsumption by teens. Cambridge, MA: National Bureau of Eonomi Researh, Working Paper Lee JM, Hwang TC, Ye CY, et al. The effet of igarette prie inrease on the igarette onsumption in Taiwan: evidene from the National Health Interview Surveys on igarette onsumption. BMC Publi Health 2004;4: Lewit EM, Coate D. The potential for using exise taxes to redue smoking. J Health Eon 1982;1: Liang L, Chaloupka FJ. Differential effets of igarette prie on youth smoking intensity. Niotine Tob Res 2002;4: Lopez Niolas A. How important are tobao pries in the propensity to start and quit smoking? An analysis of smoking histories from the Spanish national health survey. Health Eon 2002;11: Ohsfeldt RL, Boyle RG, Capilouto EL. Tobao taxes, smoking restritions, and tobao use. Cambridge, MA: National Bureau of Eonomi Researh, Working Paper Peretti-Watel P. Priing poliy and some other preditors of smoking behaviours: an analysis of Frenh retrospetive data. Int J Drug Poliy 2005;16: Ringel JS, Wasserman J, Andreyeva T. Effets of publi poliy on adolesents igar use: evidene from the National Youth Tobao Survey. Am J Publi Health 2005;95: Ringel JS, Evans WN. Cigarette taxes and smoking during pregnany. Am J Publi Health 2001;91: Ross H, Chaloupka FJ. The effet of publi poliies and pries on youth smoking. South Eon Journal 2004;70: Tauras JA. Publi poliy and smoking essation among young adults in the United States. Health Poliy 2004;68: Tauras JA. Can publi poliy deter smoking esalation among young adults? J Poliy Anal Manage 2005;24: Tauras JA, Chaloupka FJ. Prie, lean indoor air laws, and igarette smoking: evidene from longitudinal data for young adults. Cambridge, MA: National Bureau of Eonomi Researh, Working Paper Tauras JA, O Malley PM, Johnston LD. Effets of prie and aess laws on teenage smoking initiation: a national longitudinal analysis. Cambridge, MA: National Bureau of Eonomi Researh, Working Paper Thomson CC, Fisher LB, Winikoff JP, et al. State tobao exise taxes and adolesent smoking behaviors in the United States. J Publi Health Manag Pratie 2004;10: Townsend J, Roderik P, Cooper J. Cigarette smoking by soioeonomi group, sex, and age: effets of prie, inome, and health publiity. BMJ 1994;309: Townsend JL. Cigarette tax, eonomi welfare, and soial lass patterns of smoking. Appl Eon 1987;19: Tsai YW, Yang CL, Chen CS, et al. The effet of Taiwan s tax-indued inreases in igarette pries on brand-swithing and the onsumption of igarettes. Health Eon 2005;14: Wasserman J, Manning WG, Newhouse JP, et al. The effets of exise taxes and regulations on igarette smoking. J Health Eon 1991;10: Chaloupka FJ, Paula RL. Sex and rae differenes in young people s responsiveness to prie and tobao ontrol poliies. Tob Control 1999;8: Glied S. Youth tobao ontrol: reoniling theory and empirial evidene. J Health Eon 2002;21: Gruber J, Sen A, Stabile M. Estimating prie elastiities when there is smuggling: the sensitivity of smoking to prie in Canada. Cambridge, MA: National Bureau of Eonomi Researh, Working Paper Lewit EM, Hyland A, Kerrebrok N, et al. Prie, publi poliy, and smoking in young people. Tob Control 1997;6:S17 S Nonnemaker JM. The impat of state exise taxes, shool smoking poliies, state tobao ontrol poliies and peers on adolesent smoking [dissertation]. Minneapolis, MI: University of Minnesota, Cooreman J, Mesbah H, Leynaert B, et al. Evaluation of the impat of a smoking ban in a large Paris hospital. Sem Hop 1997;73: Helakorpi S, Martelin T, Torppa J, et al. Did Finland s tobao ontrol at of 1976 have an impat on ever smoking? An examination based on male and female ohort trends. J Epidemiol Community Health 2004;58: Heloma A, Nurminen M, Reijula K, et al. Smoking prevalene, smoking-related lung diseases, and national tobao ontrol legislation. Chest 2004;126: Stephens T, Pederson LL, Koval JJ, et al. Comprehensive tobao ontrol poliies and the smoking behaviour of Canadian adults. Tob Control 2001;10: Unger JB, Rohrbah LA, Howard KA, et al. Attitudes toward anti-tobao poliy among California youth: assoiations with smoking status, psyhosoial variables and advoay ations. Health Edu Res 1999;14: Britten N, Campbell R, Pope C, et al. Using meta ethnography to synthesise qualitative researh: a worked example. J Health Serv Res Poliy 2002;7: Dixon-Woods M, Shaw RL, Agarwal A, et al. The problem of appraising qualitative researh. Qual Saf Health Care 2004;13: Ogilvie D, Fayter D, Pettirew M, et al. The harvest plot: a method for synthesising evidene about the differential effets of interventions. BMC Med Res Methodol 2008;8: Giskes K, Kunst A, Ariza C, et al. Applying an equity lens to tobao-ontrol poliies and their uptake in six western-european ountries. J Publi Health Poliy 2007;28: Campbell NC, Murray E, et al. Designing and evaluating omplex interventions to improve health are. BMJ 2007;334: Arai L, Roen K, Roberts H, et al. It might work in Oklahoma but will it work in Oakhampton? Context and implementation in the effetiveness literature on domesti smoke detetors. Inj Prev 2005;11: Health & Consumer Protetion Diretorate-General. Towards a Europe free from tobao smoke: poliy options at EU level. Brussels: European Commission, Ogilvie D, Gruer L, Haw S. Young people s aess to tobao, alohol, and other drugs. BMJ 2005;331: Aheson D. Independent inquiry into inequalities in health report. London: Stationery Offie, Graham H. Promoting health against inequality: using researh to identify targets for intervention a ase study of women and smoking. Health Edu J 1998;57: Marsh A. Tax and spend: a poliy to help poor smokers. Tob Control 1997;6: Kunst A, Giskes K, Makenbah J. Soio-eonomi inequalities in smoking in the European Union. Applying an equity lens to tobao ontrol poliies. Rotterdam: Erasmus University; Available from Tobao Control 2008;17: doi: /t