Jump to ContentJump to Main Navigation
Social Determinants of Health$

Michael Marmot and Richard Wilkinson

Print publication date: 2005

Print ISBN-13: 9780198565895

Published to Oxford Scholarship Online: September 2009

DOI: 10.1093/acprof:oso/9780198565895.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 23 May 2017

Social patterning of individual health behaviours: the case of cigarette smoking

Social patterning of individual health behaviours: the case of cigarette smoking

(p.224) Chapter 11 Social patterning of individual health behaviours: the case of cigarette smoking
Social Determinants of Health

Michael Marmot

Richard G. Wilkinson

Oxford University Press

Abstract and Keywords

This chapter focuses on cigarette smoking as the individual health behaviour with the single largest impact on health inequalities. Drawing on data from the UK, a country where the smoking epidemic is now mature, the chapter first considers the nature and extent of the association of smoking with indicators of disadvantage, and trends in rates of current and ex-smoking by deprivation. It then estimates the contribution of smoking to death rates in different social groups, before addressing the question of why poor people are more likely to smoke and why they find it harder to give up. The final section outlines possible policy options to reduce smoking-induced inequalities.

Keywords:   smoking, health inequality, smoking epidemic, poverty, deprivation, mortality

11.1 Introduction

Poverty is intimately linked to a variety of behaviours which impact on health. As illustrated in Table 11.1, poor people in a country such as the UK are less likely than those who are well off to eat a good diet, more likely to have a sedentary lifestyle, more likely to be obese, and more likely to be regularly drunk. These associations may be due to a variety of factors, including poverty itself, as well as poorer access to education and information, rather than reflecting a single invariant cause, and not all of these associations may be found in all societies. Nevertheless their consistency is impressive.

Nowhere are the links between deprivation and health behaviours stronger than in the case of drug use, both legal and illegal. Alcohol abuse is frequently a marker for acute social breakdown and, through accidents and violence induced by drunkenness, has a significant impact on death rates (Makela et al. 1997). Binge drinking, which is more common in deprived groups in the population, may lead directly to sudden cardiac death (Kauhanen et al. 1997). Drug users are differentially recruited from groups with disturbed family backgrounds, low self-esteem, and impaired psychological functioning. The costs to society through drug-related crime and social disruption are immense. Cigarette smoking stands out as somewhat different: nicotine is not a drug whose acute effects lead directly to disturbed behavior, crime, or violence, but it imposes the greatest costs of all in terms of premature death. It too shows a strong association with indicators of social disadvantage.

Smoking, drinking, and drug use are individual behaviours which involve an element of personal choice. It is perhaps for this reason that they have frequently been seen not in a broad social context but as a matter of individual responsibility: if smokers wish to avoid the adverse effects of tobacco on their health, it is for them to change their behaviour and quit. If they don’t, they have brought ill health on themselves and it is no-one else’s fault. Persistence in unhealthy behaviours is seen as simply fecklessness rather than as a response to social circumstances. This victim-blaming approach is (p.225)

Table 11.1 Distribution of some health behaviours by level of socio-economic deprivation

Level of socio-economic deprivation





Least deprived

Most deprived

Men (%)

Eat fruit less often than weekly





Sedentary lifestyle





Body mass index >30





Drunk at least once per week





Women (%)

Eat fruit less often than weekly





Sedentary lifestyle





Body mass index >30





Drunk at least once per week





Source: Health Survey for England 1994.

unhelpful, in that it fails to address underlying questions of why disadvantaged people are drawn to these behaviours and the nature of the social and individual influences that maintain them. It has also been signally unsuccessful in leading to the development of effective interventions to achieve behaviour change in disadvantaged groups.

This chapter will focus on cigarette smoking, as the individual health behaviour with the single largest impact on health inequalities. As a legal, widespread and, until recently at least, little stigmatized behaviour, there is a wealth of detailed data available documenting its natural history, social patterning, and impact on health. Drawing on data from the UK, a country where the smoking epidemic is now mature, we consider first the nature and extent of the association of smoking with indicators of disadvantage, and trends in rates of current and ex-smoking by deprivation. We then briefly give estimates of the contribution of smoking to death rates in different social groups, before attempting to address the question of why poor people are more likely to smoke and why they find it harder to give up. The final section outlines possible policy options to reduce smoking-induced inequalities.

11.2 Rates of cigarette smoking by material and cultural disadvantage

The gradient in cigarette smoking prevalence by occupational class is well known, as is also the high rate of smoking by lone parents (Marsh and McKay 1994), the unemployed (Lee et al. 1991), and the mentally ill (Hughes et al. 1986). But these links by no means fully characterize the extent of smoking’s association with disadvantage. Table 11.2, which draws on several years of recent data from the General Household Survey and closely replicates analyses of earlier data (Jarvis 1997), documents how (p.226) a whole range of circumstances independently predict current cigarette smoking. Thus, the odds of being a smoker are substantially increased in those in lower occupational class groups, those living in rented housing, without access to a car, who are unemployed, and in crowded accommodation. Above and beyond this, there is a substantial gradient by educational level, and an increased risk in those who are divorced or separated or lone parents. With the exception of lone parenthood, which is uncommon in men and appears not to carry a risk of smoking for them, the magnitude of the associations is very similar in men and women.

These independent associations imply extreme differences in smoking prevalence between groups with different constellations of circumstances. For example, by comparison with professional owner-occupiers with degree-level education and owning a car, the predicted odds of smoking for unemployed, unskilled, manual workers living in rented, crowded accommodation and with no car are 17.8, corresponding to smoking prevalences in these groups of about 15% and 75%. It should be noted that these variables do not provide an exhaustive list of factors influencing smoking prevalence, as other work shows that smoking is more common in people suffering from mental illness or who are heavy drinkers or who are homeless. Indeed, groups who have an extreme clustering of deprivation indicators (such as prisoners in gaol and

Table 11.2 Predictors of current cigarette smoking among men and women










Social class








































Rented tenure







No car




















Degree level




Higher < degree







A level







O level







CSE grade







No qualification







Lone parent







Divorced or separated







OR = odds ratio; CI—confidence interval.

Source: General Household Survey 2000–2003.

(p.227) homeless people sleeping rough) have been observed to have rates of smoking prevalence of 80–90% (Bridgwood and Malbon 1995; Gill et al. 1996).

The factors which predict smoking include material circumstances, cultural deprivation, and indicators of stressful marital, personal, and household circumstances. This illustrates what might be proposed as a general law of western industrialized society—namely, that any marker of disadvantage that can be envisaged and measured, whether personal, material, or cultural, is likely to have an independent association with cigarette smoking. Of course, this may not be true of all societies, such as, for example, Asian countries where there is an overriding cultural prohibition on women’s smoking and, in particular, may not be true of developing societies in which cigarette smoking is associated with images of glamour and western prosperity, rather than disadvantage and poverty. However, recent data suggest that smoking in India, as in Britain, is now associated with indicators of disadvantage (Sorensen et al. 2005).

11.3 Trends in cigarette smoking prevalence and rates of cessation by deprivation

Cigarette smoking prevalence has been on a declining trend in Britain for over 30 years, reducing overall from 53% in 1973 to 28% in 2003 in men, and from 42% to 25% in women. But over this same period, there has been a substantial widening of the gulf in prevalence between social groups. Figure 11.1 shows trends by a composite index of deprivation which incorporates several of the variables in Table 11.1. Respondents are assigned a score of 1 for each of the following: manual occupational class; rented housing; no car; unemployed; living in crowded conditions (one or more persons per room). The resulting index, with scores ranging from 0 among the affluent to 5 among the most deprived, is similar to the indices employed by Townsend (Townsend et al. 1988) and by Carstairs (Carstairs and Morris 1989), but is applied to

                   Social patterning of individual health behaviours: the case of cigarette smoking

Fig. 11.1 Cigarette smoking by deprivation in the UK. (Source: General Household Survey 1973 and 2003).

                   Social patterning of individual health behaviours: the case of cigarette smoking

Fig. 11.2 Smoking cessation by deprivation in the UK. (Source: General Household Survey 1973 and 2003).

individuals rather than areas. In both 1973 and 2003, for both men and women, there was an approximately linear increase in cigarette smoking with increasing deprivation. Among the most affluent, smoking rates more than halved over the years, reaching a figure of 16% in 2003. Among the most deprived, on the other hand, about 70% were smokers in 1973, and that still remained the case 30 years later.

Figure 11.2 shows rates of smoking cessation by deprivation in 1973 and 2003 (Jarvis 1997). Mirroring the observations in Fig. 11.1, it indicates that while rates of cessation more than doubled in affluent people (from 25% to 58%), among the poorest groups, there was little or no change, with 10% of ever-regular smokers giving up.

11.4 Contribution of smoking to differences in death rates by social group

It has long been acknowledged that smoking has a major bearing on observed differences in death rates by social class. One in two of those who smoke are ultimately killed by the habit if they persist and do not give up (Doll et al. 1994; Thun et al. 1995), and in view of higher prevalence and lower rates of cessation, it would be anticipated that smoking-related disease would bear more heavily on poorer groups. Poorer diet and factors such as earlier age of starting to smoke (leading to longer duration of smoking at any given age) and higher levels of smoke intake in poorer smokers (see below) would act to amplify smoking risks.

Unsurprisingly, observed rates of death from smoking-related diseases show a gradient that parallels the gradient in smoking prevalence. Standardized mortality ratios in unskilled male workers are three times higher than in professionals for heart disease, five times higher for lung cancer, six times higher for emphysema, and 14 times higher for chronic airways obstruction (Drever and Whitehead 1997). Similar, but somewhat (p.229) smaller, variations are seen in women. As would be expected from the preceding discussion, alternative indicators of socio-economic status such as housing tenure or access to a car are additionally predictive of death rates (Smith and Harding 1997; Goldblatt 1990).

Recently, the indirect methods developed by Peto and Lopez (Peto et al. 1992, 1994) to estimate deaths from smoking in different countries have been applied to deaths by socio-economic status within countries. For men aged 35–69, in England and Wales, estimates have been made of the proportion of deaths attributable to smoking by social class for the years since 1970. In 1970–1972, among men in social class I and II, the overall risk of dying in middle age was 36%, and just over one third of these deaths (13%) were estimated to be attributable to smoking. By 1996, the overall risk of dying had declined to 21%, and tobacco-attributable deaths were 4%. Thus, the reduction in deaths attributable to tobacco was responsible for over half of the overall reduction in risk of death. Men in social class V, by contrast, had an overall risk in 1970–1972 of dying in middle age of 47%, and over half of this (25%) was accounted for by tobacco. By 1996, the risk had declined only modestly to 43% and the estimated smoking-attributable element, to 19%.

Of course, the accuracy of these estimates depends on the adequacy of the data and of the assumptions underlying their calculation. The overall number of deaths attributed to tobacco is subject to considerable uncertainty, due to fairly approximate assumptions. For example, only half of the apparent excess of vascular deaths observed in smokers is attributed to tobacco. This somewhat arbitrary proportion, intended to be conservative, may either underestimate true smoking effects, or not be conservative enough. However, for the method to give seriously misleading results for differences in the proportions of deaths attributable to smoking in different social class groups, there would have to be major bias introduced in the calculations across different classes. It is difficult to see that any major bias could be present which would invalidate the estimates.

These findings carry a number of implications. First, the observed widening in overall risk of death between men in social classes I/II and V, that has been observed over the past 25 years, largely reflects changes in tobacco-attributable mortality. Deaths caused by tobacco have dropped far more steeply in social class I and II than in social class V, paralleling changes in smoking prevalence and cessation in these groups. Men in social class V have not experienced an absolute increase in risk of death: the widening of health inequalities that has occurred has been due to their failure to share, to the same extent, in the major overall reduction in risk of death consequent upon smoking cessation. This implies that further reductions in smoking prevalence, unless they are concentrated in poorer groups, may only serve to widen inequalities in death rates still further. The 1996 data indicate that somewhere around two thirds of the observed difference in risk of death across social class groups in middle age is caused by tobacco.

(p.230) 11.5 Why do poor people smoke?

The discussion so far has shown that disadvantaged groups in society are disproportionately likely to smoke and least likely to give up cigarettes. As a consequence, the burden of smoking-related disease also falls disproportionately on these groups. Those who can least afford to smoke smoke the most and suffer most from it. That nicotine is a powerful drug of addiction no doubt has much to do with this state of affairs, but we need to move beyond this to ask why it is that the poor are particularly drawn to this drug. The association could be mediated by higher rates of smoking initiation, stronger perceived rewarding effects (either positive or negative) leading to higher levels of dependence, or to greater difficulties in cessation through lower motivation, higher dependence, or fewer available coping resources. These influences are not mutually exclusive, and it could be that a variety of factors operate at each stage of the smoker’s career to accentuate the link with disadvantage.

A particularly important distinction to be drawn is between ever becoming a regular cigarette smoker and persisting with the habit. Among those who take up smoking but give up before their early 30s, there is no detectable increase in risk of premature death in comparison with those who never smoke (Doll et al. 1994, 2004). As shown in Fig. 11.3, there exists a gradient between ever-regular cigarette smoking and deprivation in both men and women, but among those aged 35–64, the gradient with current smoking is far steeper. Thus, although poor people are somewhat more likely to become smokers, the strongest association is with persisting smoking. What we need to explain above all is not so much why poor people start smoking, but why they do not give it up.

11.5.1 Disadvantage and smoking uptake

A gradient in uptake of smoking by level of deprivation is not hard to explain. Children growing up in poverty experience social environments outside the home where most adults are smokers, and the vast majority, 80% or more, will live in households where one or both parents smoke. Thus, cigarette smoking is modelled as normal adult behaviour and cigarettes are readily available to experiment with. But in addition to this, there is evidence that smoking is a measure of smoking trajectory, with prevalence being more closely related to people’s social destination than to their circumstances of origin (Glendinning et al. 1994). In the national cohort of all the babies born in one week in 1958, who have been followed up at regular intervals ever since (Ferri 1993), cigarette smoking at age 16 increased from 24% among those from the most affluent homes to 48% among the most deprived. But the gradient at age 16 was much sharper (80% among the most deprived) when cohort members were characterized by deprivation measured seven years later at age 23 by their own achieved social position rather than by the characteristics of the parental household. This implies that factors conferring an increased risk of smoking at age 16 (such as poorer (p.231)

                   Social patterning of individual health behaviours: the case of cigarette smoking

Fig. 11.3 Rates of current and ever-regular cigarette smoking among men and women by deprivation. (Source: General Household Survey 2000–2003).

school achievement and lower levels of self-esteem) also have a bearing on subsequent downward social mobility.

11.5.2 Motivation to give up smoking

A superficially attractive hypothesis to explain poor people’s lower likelihood of quitting smoking is that they are less well informed and concerned about effects on health, leading to a lower probability of attempts at quitting. Taking this view, the problem is (p.232) that the health education message simply isn’t getting through. Motivation to quit is not an easy construct to measure reliably, as it can fluctuate greatly over time and in different situations (raised in the doctor’s surgery, for example, but much lower when drinking in the pub with friends). At present, we are not able to do much better than ask people the question: how much they want to give up smoking altogether. Responses to this have been shown to have some validity by predicting future attempts to quit. In the General Household Survey, levels of expressed motivation to quit are essentially flat across deprivation categories—just over two thirds of cigarette smokers in each group saying that they want to give up. To the extent, therefore, that this measure can be taken at face value, there is no evidence that disadvantaged groups are any less likely than the more affluent to want to give up.

11.5.3 Nicotine dependence and deprivation

By contrast with the lack of variation in motivation to quit by disadvantage, there is strong emerging evidence that level of nicotine dependence increases systematically with deprivation. This is evident both from questionnaire indicators of dependence in the General Household Survey (for example, time to first cigarette of the day; perceived difficulty of going for a whole day without smoking) and from quantitative measures of smoke intake. Figure 11.4 shows levels of saliva cotinine (a measure of total nicotine intake) among smokers in the Health Survey for England (Bajekal et al. 2002). Increasingly high levels of nicotine intake are seen with increasing deprivation, with average intake being 30% higher in the most deprived than in the most affluent smokers. Poor people achieve their higher intakes both by choosing to smoke more cigarettes and by smoking each cigarette more intensively. There are indications that this may turn out to be a phenomenon of wide generality, as similar observations have been made in the USA comparing black with white smokers (Caraballo et al. 1998; PerezStable et al. 1998; Wagenknecht et al. 1990; English et al. 1994) and a gradient in intake by level of education has been observed in Czech smokers (Bobak et al. 2000).

Since nicotine dependence is an important determinant of ease of quitting, these findings suggest one reason for lower rates of cessation in those who are disadvantaged. Since smoking- related disease bears a dose–response relation to intake, they have implications for higher risk of disease in poor rather than affluent smokers. They also raise the question of just why it should be that poor smokers seek higher nicotine doses.

11.5.4 Functional aspects of nicotine use: positive and negative rewards

Nicotine has a number of positively rewarding effects which could serve as the basis for cigarette use. Although euphoriant effects are not as prominent as with many drugs, they are reported for at least some cigarettes by smokers (Pomerleau and Pomerleau 1992) and could achieve a greater valence for people whose lives are generally deficient in rewards. It is difficult to think of evidence which would strongly support this hypothesis but, equally, it should not be ruled out either.


                   Social patterning of individual health behaviours: the case of cigarette smoking

Fig. 11.4 Saliva cotinine by deprivation in adult smokers in UK. (Source: Health Survey for England 2001).

An alternate functional view of smoking is that it is self-medication. Cigarette smoking is seen as a means of regulating mood, of managing stress, and of coping with all the hassles and strain resulting from material deprivation (Graham 1987; Smith and Morris 1994). This account chimes with smokers’ self-reports of the calming effects from cigarettes and with poor women’s observation that smoking is the one thing they do for themselves, that gives them space from the difficult task of caring for children in poverty (Graham 1994). It also recalls soldiers’ demands for cigarettes in the First World War to help them cope with the rigours of life in the trenches.

Attractive though the self-medication view of smoking is, it faces several major objections. The most serious of these is the nature of nicotine as a drug. Pharmacologically, nicotine is a stimulant, similar to drugs such as amphetamine. Sedative or anxiolytic effects, if they exist, are very hard to find either in animal models or in humans. Smoking is closely associated with adverse mood states, but there is no good evidence in humans that it ameliorates them other than through withdrawal relief (Schoenborn and Horm 1993; Anda et al. 1990). Smokers’ self-reports of the calming effects from cigarettes could refer to relief of nicotine withdrawal by smoking. Onset of withdrawal symptoms is rapid and certainly occurs with overnight abstinence during sleep, and may, in a more subtle way, begin within an hour or two of the last cigarette. Stress modulation over the course of the day suggests that mood-elevating effects of cigarettes are attributable to relief of adverse mood from incipient withdrawal rather than to any absolute benefits (Parrott 1995).

Studies of the process of smoking cessation have found that successful quitting leads to lower, rather than to higher, levels of perceived stress, consistent with the idea that smoking may actually be a stressor rather than relieve stress (Cohen and Lichtenstein 1990).

11.5.5 Giving up smoking could be particularly difficult for poor people

If it is difficult to find unequivocal reasons why smoking should be more rewarding for the poor and disadvantaged, there are a number of identifiable factors which would (p.234) tend to make it harder for them to give up. Even if poor people are as likely to make an attempt at quitting, higher levels of nicotine dependence would place a barrier between the attempt and success. Poor people’s generally smokier environment and their much greater likelihood of having a smoking partner would further reduce their chances of succeeding (Jarvis 1997).

It is important to appreciate how hard it is for smokers to give up. Estimates of the chances of succeeding for at least a year in a serious unaided attempt to quite are no better than about 1 in 100 (Jarvis 1997; Cohen et al. 1989). When preparing themselves for an attempt at giving up smoking, smokers need to take a medium- to long-term view and be prepared to tolerate the discomfort of nicotine withdrawal and cravings for several weeks at least before things start to get easier. If this is difficult for everybody, it may be especially difficult for those whose lives are stressful and full of hassles to forego the certainty of short-term craving relief and elimination of withdrawal for longer-term gains in disposable income, health, and well-being. The logic of addiction, if not of economics, may win out and dictate buying the next pack of cigarettes.

11.6 Implications for policies to promote cessation

Two general kinds of approach can be identified which might reduce the association of smoking with disadvantage. Improvements in housing, education, and employment would target the underlying social conditions which foster high levels of smoking. There is little doubt that substantial progress in this direction would greatly facilitate reductions in smoking, as well as contributing to the wider adoption of other desirable health behaviours, such as an improved diet. But such change is not easy to achieve, and most government policy has sought instead to target the downstream factors which more proximally determine smoking behaviour.

The policies to reduce smoking that have been followed over the past decade or more have been successful in reducing overall prevalence, but have had the paradoxical effect of increasing health inequalities. The main planks of policy have been interventions designed to increase people’s motivation to quit (for example, price and restrictions on smoking in public places) and health education campaigns. Addressing dependence has received less attention. While price increases are effective in reducing consumption, there is some uncertainty about their impact on the poor, for whom tobacco expenditure amounts to a high proportion of disposable income (Marsh and McKay 1994). Some economists have argued that poor smokers respond equally as much or more to price than do affluent smokers (Townsend et al. 1994), while others are less certain. The very low rates of cessation seen in disadvantaged groups are inconsistent with the idea that, for them, price is an effective means of promoting cessation. Price increases may influence poor people to switch to cheaper and higher yielding brands (Jarvis 1998), to roll their own cigarettes, and to cut down on the number of cigarettes they smoke, rather than to quit altogether. Because of the phenomenon of nicotine compensation, lowering cigarette consumption is unlikely to confer any benefit in lowering risk of smoking-related disease.

(p.235) Restrictions on smoking in public places and the workplace carry an effective message about the social acceptability of cigarettes, and, as well as protecting non-smokers from other people’s smoke, may enhance motivation and attempts to give up smoking. Legislative bans have recently been introduced in Ireland, Italy, Norway, and New Zealand with good acceptance and success. A complete ban is under consideration in Scotland but, in England, the government is planning a partial ban that will exempt pubs that do not serve food as well as private clubs (White Paper 2004). Since drink-only pubs are found disproportionately in poorer areas, a partial ban risks widening health inequalities by protecting affluent people from exposure to tobacco smoke, while leaving the culture of smoking in poor neighbourhoods unchallenged.

One key factor amenable to intervention would appear to be nicotine dependence. Nicotine replacement (NRT) and bupropion (Zyban) have been shown to be uesful and cost-effective aids to cessation (Cromwell et al. 1997; Buck et al. 1997). NRT approximately doubles success rates from both brief and intensive treatments (Silagy et al. 2000) and there is evidence that its success is maintained in real-world settings (Shiffman et al. 1998). NRT and bupropion specifically target the dependence problem. The National Institute for Clinical Excellence reviewed these pharmacotherapies in 2002 and recommended that both should be reimbursed by the NHS as first-line treatments (National Institute for Clinical Excellence 2002). Now that smoking cessation services (including behavioural support as well as pharmacological aids) are provided by the NHS throughout the country, disadvantaged smokers have access to free and effective treatment. The challenge remains to ensure good uptake of these services by poor as well as by affluent smokers.

If the chances of achieving high rates of cessation in poor smokers are seen as questionable, an alternative, or complementary, approach, would be to take new initiatives in the area of product modification to make smoking less harmful. The advent of novel forms of nicotine delivery, as in NRT and some innovative products from the tobacco industry (Sutherland et al. 1993), is focusing attention on harm reduction as a potentially important new arm of policy (Warner et al. 1997; Bates et al. 1999; Royal College of Physicians Tobacco Advisory Group 2002). There is compelling evidence that people smoke for nicotine, but much of the burden of smoking-related disease is attributable to other smoke components, particularly the tar fraction. Smoking cigarettes has been likened to injecting drugs through a dirty syringe. The potential benefit of shifting the market toward safer forms of nicotine delivery is illustrated by the case of Sweden, a country which has the lowest male rate of cigarette smoking in Europe (below 20%) and also the lowest rate of lung cancer. But its rate of tobacco dependence is not low, as 20% of adult males use moist oral snuff, a non-combustible form of nicotine delivery which carries considerably less risk than cigarettes (Bolinder 1997; Bates et al. 2003).

The risk is that an intensification of current smoking-control policies, without fresh thinking, may well succeed in further reducing prevalence, but only at the cost of still wider health inequalities.

(p.236) References

Bibliography references:

Anda, R.F., Williamson, D.F., Escobedo, L.G., Mast, E.E., Giovino, G.A., and Remington, P.L. (1990). Depression and the dynamics of smoking: a national perspective. Journal of the American Medical Association, 264, 1541–5.

Bajekal, M., Primatesta, P., and Prior, G. (2002) Health Survey for England 2001. The Stationery Office, London.

Bates, C., Fagerstrom, K., Jarvis, M.J., Kunze, M., McNeill, A., and Ramstrom, L. (2003). European Union policy on smokeless tobacco: a statement in favour of evidence based regulation for public health. Tobacco Control, 12, 360–7.

Bates, C., McNeill, A., Jarvis, M., and Gray, N. (1999). The future of tobacco product regulation and labelling in Europe: implications for the forthcoming European Union directive. Tobacco Control, 8, 225–35.

Bobak, M., Jarvis, M., Skpodova, Z., and Marmot, M. (2000). Smoke intake among smokers is higher in lower socioeconomic groups. Tobacco Control, 9, 310–12.

Bolinder, G. (1997). Smokeless tobacco a less harmful alternative? In: The tobacco epidemic, Vol. 28 (ed. C.T. Bolliger and K.O. Fagerstrom). Karger, Basel.

Bridgwood, A. and Malbon, G. (1995) Survey of the physical health of prisoners. HMSO, London.

Buck, D., Godfrey, C., Parrott, S., and Raw, M. (1997) Cost effectiveness of smoking cessation interventions. Centre for Health Economics and Health Education Authority, York.

Caraballo, R.S., Giovino, G.A., Pechacek, T.F., Mowery, P.D., Richter, P.A., Strauss, W.J., et al. (1998). Racial and ethnic differences in serum cotinine levels of cigarette smokers. Journal of the American Medical Association, 280, 135–9.

Carstairs, V. and Morris, R. (1989). Deprivation and mortality: An alternative to social class? Community Medicine, 11, 210–19.

Cohen, S. and Lichtenstein, E. (1990). Perceived stress, quitting smoking, and smoking relapse. Health Psychology, 9, 466–78.

Cohen, S., Lichtenstein, E., Prochaska, J.O., Rossi, J.S., Gritz, E.R., Carr, C.R., et al. (1989). Debunking myths about self-quitting. Evidence from 10 prospective studies of persons who attempt to quit smoking by themselves. American Psychologist, 44, 1355–65.

Cromwell, J., Bartosch, W.J., Fiore, M.C., Hasselblad, V., and Baker, T. (1997). Cost-effectiveness of the clinical practice recommendations in the AHCPR guideline for smoking cessation. Journal of the American Medical Association, 278, 1759–66.

Doll, R., Peto, R., Boreham, J., and Sutherland, I. (2004). Mortality in relation to smoking: 50 years’ observations on male British doctors. British Medical Journal, 328, 1519–20.

Doll, R., Peto, R., Wheatley, K., Gray, R., and Sutherland, I. (1994). Mortality in relation to smoking: 40 years’ observations on male British doctors. British Medical Journal, 309, 901–911.

Drever, F. and Whitehead, M. (eds.) (1997) Health inequalities. The Stationery Office, London.

English, P.B., Eskenazi, B., and Christianson, R.E. (1994). Black-White differences in serum cotinine levels among pregnant women and subsequent effects on infant birthweight. American Journal of Public Health, 84, 1439–43.

Ferri, E. (ed.) (1993) Life at 33: the fifth follow-up of the National Child Development Study. National Children’s Bureau, London.

Gill, B., Meltzer, H., Hinds, K., and Pettigrew, M. (1996) Psychiatric morbidity among homeless people. HMSO, London.

Glendinning, A., Shucksmith, J., and Hendry, L. (1994). Social class and adolescent smoking behaviour. Social Science and Medicine, 38, 1449–60.

Goldblatt, P. (ed.) (1990) Longitudinal study: mortality and social organization. HMSO, London.

Graham, H. (1987). Women’s Smoking and Family Health. Social Science and Medicine, 25, 47–56.

Graham, H. (1994). Gender and class as dimensions of smoking-behavior in Britain—insights from a survey of mothers. Social Science and Medicine, 38, 691–8.

Hughes, J.R., Hatsukami, D.K., Mitchell, J.E., and Dahlgren, L.A. (1986). Prevalence of smoking among psychiatric outpatients. American Journal of Psychiatry, 143, 993–7.

Jarvis, M.J. (1997). Patterns and predictors of unaided smoking cessation in the general population. In: The Tobacco Epidemic, Vol. 28 (ed. C.T. Bolliger and K.O. Fagerstrom), pp. 151–64. Karger, Basel.

(p.237) Jarvis, M.J. (1998). Supermarket cigarettes: the brands that dare not speak their name. British Medical Journal, 316, 929–31.

Kauhanen, J., Kaplan, G.A., Goldberg, D.E., and Salonen, J.T. (1997). Beer binging and mortality: Results from the Kuopio ischaemic heart disease risk factor study, a prospective population based study. British Medical Journal, 315, 846–51.

Lee, A.J., Crombie, I.K., Smith, W.C.S., and Tunstall–Pedoe, H.D. (1991). Cigarette smoking and employment status. Social Science and Medicine, 33, 1309–12.

Makela, P., Valkonen, T., and Martelin, T. (1997). Contribution of deaths related to alcohol use of socioeconomic variation in mortality: register based follow up study. British Medical Journal, 315, 211–6.

Marsh, A. and McKay, S. (1994) Poor smokers. Policy Studies Institute, London.

National Institute for Clinical Excellence (2002) Guidance on the use of nicotine replacement therapy (NRT) and bupropion for smoking cessation. Technology Appraisal Guidance No. 39. National Institute for Clinical Excellence, London.

Parrott, A.C. (1995). Stress modulation over the day in cigarette smokers. Addiction, 90, 233–44.

Perez Stable, E.J., Herrera, B., Jacob, P., and Benowitz, N.L. (1998). Nicotine metabolism and intake in black and white smokers. Journal of the American Medical Association, 280, 152–6.

Peto, R., Lopez, A.D., Boreham, J., Thun, M., and Heath C., Jr. (1992). Mortality from tobacco in developed countries: indirect estimation from national vital statistics. Lancet, 339, 1268–78.

Peto, R., Lopez, A.D., Boreham, J., Thun, M., and Heath, C. (1994) Mortality from smoking in developed countries 1950–2000: indirect estimates from national vital statistics. Oxford University Press, Oxford.

Pomerleau, C.S. and Pomerleau, O.F. (1992). Euphoriant effects of nicotine in smokers. Psychopharmacology, 108, 460–5.

Royal College of Physicians Tobacco Advisory Group (2002). Protecting Smokers, Saving Lives. Royal College of Physicians, London.

Schoenborn, C.A. and Horm, J. (1993) Negative moods as correlates of smoking and heavier drinking: implications for health promotion. National Center for Health Statistics, Hyattsville, Maryland.

Shiffman, S., Gitchell, J., Pinney, J.M., Burton, S.L., Kemper, K.E., and Lara, E.A. (1998). Public health benefit of over-the-counter nicotine medications. Tobacco Control, 6.

Silagy, C., Lancaster, T., Stead, L., Mant, D., and Fowler, G. (2000). Nicotine replacement therapy for smoking cessation (Cochrane Review). In: The Cochrane Library, Issue 1, 2003 (ed. T. Lancaster, C. Silagy, and D. Fullerton). Update Software, Oxford.

Smith, G.D. and Morris, J. (1994). Increasing inequalities in the health of the nation. British Medical Journal, 309, 1453–4.

Smith, J. and Harding, S. (1997). Mortality of women and men using alternative social classifications. In: Health inequalities: decennial supplement (ed. F. Drever and M. Whitehead). The Stationery Office, London.

Sorensen, G., Gupta, P.C., and Pednekar, M.S. (2005). Social disparities in tobacco use in Mumbai, India: the roles of occupation, education, and gender. American Journal of Public Health, 95, 1003–8.

Sutherland, G., Russell, M.A.H., Stapleton, J.A., and Feyerabend, C. (1993). Glycerol particle cigarettes: a less harmful option for chronic smokers. Thorax, 48, 385–7.

Thun, M.J., Day–Lally, C.A., Calle, E.E., Flanders, W.D., and Heath, C.W. (1995). Excess mortality among cigarette smokers: changes in a 20-year interval. American Journal of Public Health, 85, 1223–30.

Townsend, J., Roderick, P., and Cooper, J. (1994). Cigarette smoking by socioeconomic group, sex, and age: Effects of price, income, and health publicity. British Medical Journal, 309, 923–7.

Townsend, P., Phillimore, P., and Beattie, A. (1988). Health and deprivation: inequality and the North. Croom Helm, London.

Wagenknecht, L.E., Cutter, G.R., Haley, N.J., Sidney, S., Manolio, T.A., Hughes, G.H. et al. (1990). Racial differences in serum cotinine levels among smokers in the Coronary Artery Risk Development in (Young) Adults Study. American Journal of Public Health, 80, 1053–6.

Warner, K., Slade, J., and Sweanor, D.T. (1997). The emerging market for long-term nicotine maintenance. Journal of the American Medical Association, 278, 1087–92.

White Paper (2004) Choosing health—making healthy choices easier. The Stationery Office, London.