Article Text

Download PDFPDF

Explaining the social gradient in smoking cessation: it’s not in the trying, but in the succeeding
  1. D Kotz1,
  2. R West2
  1. 1
    Department of General Practice, School for Public Health and Primary Care (CAPHRI), Maastricht University Medical Centre, Maastricht, The Netherlands
  2. 2
    Cancer Research UK Health Behaviour Unit, Department of Epidemiology and Public Health, University College London, London, UK
  1. Dr Robert West, Cancer Research UK Health Behaviour Unit, Department of Epidemiology and Public Health, University College London, London, UK; robert.west{at}


Objective: Smokers from more deprived socioeconomic groups are less likely to become ex-smokers. This study examined how far this is attributable in England to a social gradient in quit attempts, use of aids to cessation and/or the success of quit attempts.

Design: Cross-sectional household survey.

Setting: England.

Participants: 6950 respondents aged 16+ who had smoked in the past year, of whom 2983 had tried to stop in the past year and 469 reported having stopped at the time of the survey.

Main outcome measures: Reported attempts to stop smoking in the past year; the use of stop-smoking medications (nicotine replacement therapy, bupropion or varenicline); use of behavioural support offered by the National Health Service Stop Smoking Services; current smoking status.

Results: There was a large difference in success rates of quit attempts: 20.4% in social grade AB versus 11.4% in social grade E of those who made attempts were still not smoking by the time of the survey (p<0.001). However, there was no difference across social grades in the rate of attempts to stop (42.7% in social grade AB to 41.3% in E), use of medications to aid cessation (46.7% of those making quit attempts in AB to 50.9% in E) or use of NHS Stop Smoking Services (7.0% of those making quit attempts in AB to 4.8% in E).

Conclusions: Smokers in more deprived socioeconomic groups are just as likely as those in higher groups to try to stop and use aids to cessation but there is a strong gradient across socioeconomic groups in success, with those in the lowest group being half as likely to succeed compared with the highest. Determining the cause of, and counteracting, this gradient is paramount in reducing health inequalities.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

In England, cigarette smoking killed about 82 000 men and women in 2005 (17% of all deaths of adults aged 35 years and over).1 The rate of decline in smoking prevalence has slowed considerably in recent years2 and remains above 20%.3 There is a large socioeconomic gradient with 29% of adults in “routine and manual” occupational groups currently smoking cigarettes compared with 15% in “professional and managerial” groups.3 This gradient is estimated to be responsible for half of socioeconomic differences in mortality in men aged 35–69 years in developed countries.4 The present paper examines how far the gradient can be explained in terms of attempts to stop smoking and/or use of effective aids to cessation. This has implications for strategies aimed at narrowing the gap in smoking prevalence.

The UK government has set a reduction in smoking prevalence as a high priority.5 Its target is to reduce the prevalence of cigarette smoking among adults in England to 21% or less by 2010, with a reduction in prevalence among routine and manual groups to 26% or less.6 Reducing the gap in smoking prevalence requires a better understanding of what is causing it. There is evidence that much of the socioeconomic difference in smoking prevalence is attributable to differences in cessation rates. For example, a pseudo-birth-cohort analysis with combined data from the 1972–2004/5 general household surveys showed that smokers born more recently (1956–85) were less likely to give up smoking than earlier cohorts and that inequalities in smoking prevalence between the manual and non-manual group increased with age; the vast majority of manual men and women who started smoking remained smokers.7

What is less clear is what underlies the difference in cessation rates. The process of becoming an ex-smoker involves two potentially separable stages: initiating the quit attempt and then maintaining abstinence. Different factors may contribute to each of these.8 9 There is relatively little evidence on how far socioeconomic status (SES) contributes to each of these phases. One study from the United Kingdom found no difference in attempts to stop by SES but a difference in success rates9 while another found no difference in attempts but those with an intermediate level of educational attainment were less likely than those with the lowest level to succeed.8 However, both studies were small and the second study involved telephone interviews which may undermine representativeness.

We are aware of no other studies that have separated out attempts and success of attempts in this way. However, there is a large body of research on motivation to quit. One study found that smokers with lower education and with low income were less likely to report the intention to quit than smokers with higher education and income.10 However, other much larger studies of representative samples in the UK have consistently shown that smokers with lower SES report wanting to stop to the same degree as higher SES smokers.3 11 Social deprivation has been associated with “hard core” smoking; a syndrome which includes the absence of the desire to give up smoking12; however, this construct also includes indices of dependence and it is not clear how far the latter might account for the SES gradient. Taken together, these data suggest that there would be little or no difference in attempts to stop by SES. However, there may be many factors other than motivation that contribute to this behaviour.

It is also possible that SES differences in rates of becoming an ex-smoker are the result of more deprived smokers being less likely to use effective aids to cessation such as nicotine replacement therapy (NRT) or the National Health Service Stop Smoking Services. This is what would be predicted by the ”inverse care law”,13 which states that individuals in lower strata of society who have the greatest need of healthcare services are least likely to gain benefit from them. We are aware of one study carried out in the United States examining this in the case of smoking cessation.14 This study found that use of behavioural and pharmacological aids to cessation was higher in more highly educated and more affluent smokers. The healthcare system in the United States is different from that in the UK and it remains to be seen whether a similar pattern would be observed in the UK. A study on utilisation of behavioural support, provided free of charge via the NHS in the United Kingdom, reported that smokers from more deprived geographical areas were more, rather than less, likely to use this service.15 16 However, we are not aware of good evidence from the UK on the social gradient in use of other aids such as NRT. A survey of utilisation of medications in the UK found no evidence for differences by SES but the sample was small and would have had limited power to detect a gradient.17

The present paper uses data from the “Smoking Toolkit Study” (a series of monthly national surveys focusing on smoking and smoking cessation patterns), to examine socioeconomic differences in: (1) attempts to stop smoking in the past year, (2) use of aids to cessation of those attempts, and (3) short-term “success” rates.


The Smoking Toolkit Study has been designed to provide ongoing, up-to-date national statistics on key performance indicators relating to smoking cessation to guide policy and clinical practice.18 The study involves monthly household surveys of national samples in England, carried out by the British Market Research Bureau.

The survey uses computer-assisted face-to-face interviews and a two-stage mixed probability and quota sampling method designed to maximise representativeness within the age range 16 and over. In the first stage, grouped output areas (containing 300 households) have an equal chance of being selected. The interviewers then go to the selected areas and attempt to secure interviews with members of households—one member per household, according to quotas based on known percentages for age, sex, social grade, region, working status and presence of children in the population.

A total of 25 591 respondents were interviewed from November 2006 through to January 2008. Demographic information and smoking status were assessed by using standard questions based on those used in other national surveys.17 The data were weighted by age, sex and socioeconomic group to match the 2001 census.

People who had smoked in the past year were identified by asking respondents to indicate which of a series of statements applied to them: (a) I smoke cigarettes (including hand-rolled) every day; (b) I smoke cigarettes (including hand-rolled), but not every day; (c) I do not smoke cigarettes at all, but I do smoke tobacco of some kind (for example, pipe or cigar); (d) I have stopped smoking completely in the last year; (e) I stopped smoking completely more than a year ago; (f) I have never been a smoker (that is, smoked for a year or more). Cigarette smoking prevalence (using responses (a) and (b)) was calculated at 24%, which is slightly higher than the 2006 data from the General Household Survey for England (22%)3 and the 2007 Office for National Statistics omnibus survey for Great Britain (22%).17 Those who endorsed (a) to (d) (n = 6950) comprised the sample.

The number of quit attempts in the last year was assessed by asking: “How many serious attempts to stop smoking have you made in the last 12 months? By serious attempt I mean you decided that you would try to make sure you never smoked again. Please include any attempt that you are currently making and please include any successful attempt made within the last year”.

The use of smoking cessation aids was assessed for the three most recent quit attempts and included NRT over the counter (for example, patches, gum, inhaler), NRT on prescription or given by a health professional, bupropion, varenicline, attending an NHS Stop Smoking Service group session or an NHS Stop Smoking Service individual counselling session.

Short-term success at stopping smoking was assessed by asking respondents who had made a quit attempt during the past 12 months whether they were still not smoking. This was checked against their response to the smoking status question (above).

Nicotine dependence in current smokers was assessed by the number of cigarettes smoked per day and the Fagerström test for nicotine dependence (FTND).19 The sum score on the six items of the FTND has a range from 0–10, with higher scores indicating higher levels of nicotine dependence.

Social grade was defined according to the National Statistics Socio-Economic Classification as: AB (managerial and professional occupations), C1 (intermediate occupations), C2 (small employers and own account workers), D (lower supervisory and technical occupations) and E (semi-routine and routine occupations, never workers, and long-term unemployed).

The statistical significance of differences in proportions across social grade was calculated using χ2 tests. The significance in cigarettes per day and FTND scores were assessed by means of analyses of variance. To examine social grade differences in proportions adjusting for age and sex we used multiple logistic regression analyses with all these predictors entered together.


The prevalence of smoking, including other forms of tobacco than cigarettes, was 25% with a large social gradient from 16% in AB to 37% in E (p<0.001 by χ2 test). A similar gradient was found among smokers who had smoked in the past year for cigarettes per day and nicotine dependence. The mean number of cigarettes smoked per day ranged from 12.1 (SD 8.7) in AB to 15.8 (SD 10.0) in E, and the mean level of nicotine dependence (according to the FTND) ranged from 2.3 (SD 2.3) in AB to 3.8 (SD 2.6) in E (p<0.001 by analysis of variance in both cases) (table 1).

Table 1 Demographic and smoking characteristics of last year smokers and use of smoking cessation aids among smokers trying to quit last year

There was no social gradient in attempts to stop smoking. Neither did we find a significantly lower rate of use of NRT bought over the counter or the NHS Stop Smoking Services in lower social grades (table 1). In fact, smokers from lower social grades were more likely to use NRT obtained on prescription than those in higher grades (p = 0.016 by χ2 test). There was also a small and non-significant gradient in use of NHS Stop Smoking Services. By contrast there was a large socioeconomic gradient in the success rates of those who attempted to stop with those in the lowest social grade having just over half the success rate of those in the highest (p<0.001) (table 1).

A multiple logistic regression analysis showed that the social gradient in cessation rate remained after adjusting for age and sex; compared with the highest social grade AB, the odds of having stopped smoking were significantly lower for the lower social grades: odds ratio (OR) 0.79 (95% CI 0.59 to 1.05) for social grade C1, OR 0.67 (0.49 to 0.90) for social grade C2, OR 0.68 (0.50 to 0.93) for social grade D and OR 0.50 (0.33 to 0.74) for the lowest social grade E. The reverse social gradient in use of NRT obtained on prescription remained marginally significant in the lowest social grade after the same adjustment; compared with the highest social grade AB, the odds of having used NRT on prescription were OR 0.99 (0.67 to 1.45) for social grade C1, OR 1.39 (0.94 to 2.05) for social grade C2, OR 1.22 (0.81 to 1.83) for social grade D and OR 1.53 (0.98 to 2.39) for social grade E.


Our results indicate that socioeconomic differences in smoking cessation rates cannot be attributed to a lower likelihood of making quit attempts or using aids for smoking cessation. More deprived smokers appear to be engaged with public health messages about smoking, but their quit attempts are less likely to succeed.

Our finding of no social gradient in attempts to stop supports those of West et al9 and Hyland et al.8 The finding of a social gradient in success of attempts is in accordance with West et al9 but conflicts with Hyland et al.8 The size of the present sample and the strength of the association between social grade and success leads to the view that the association found here is genuine.

Our finding that lower SES smokers were at least as likely to use aids to cessation conflicts with those of Shiffman et al.14 One possible explanation for this may be that treatments for smoking cessation are reimbursed in England but not in the United States. As a consequence, increased use of pharmacological treatment among US smokers with higher incomes may be due to greater insurance coverage or greater ability to pay for non-covered medications.14 However, we did not find a social gradient in NRT purchased over the counter either. This suggests that another factor must be in play. The UK government has adopted a strategy of maximising NRT use in the smoking population and this may have played a part. However, there may be other cultural and economic factors involved as well, such as the fact that the US has a wider income gradient than the UK.20

The question remains as to what factors may account for differences in quitting success by SES. Nicotine dependence is one possibility. Our data confirm previous reports of higher nicotine dependence scores in smokers from more deprived socioeconomic groups,12 and nicotine dependence has been found in other studies to predict failure of attempts to stop smoking.21 Unfortunately, we could not adjust our analyses for nicotine dependence as this measurement may be subject to recall bias in data obtained from ex-smokers. Another possible explanation is that smokers in more deprived socioeconomic groups have more smokers in their immediate circle of family, co-workers and friends. Qualitative research has shown that smoking is often deeply embedded in the lives of disadvantaged smokers because they live, socialise, and/or work with other smokers, which makes it more difficult to refrain from smoking.22 It has also been shown that smoking cessation is influenced by unfavourable social trajectories such as childhood and educational disadvantages and early motherhood in female smokers.23 Smokers from more deprived socioeconomic groups also have higher levels of stress which can play a part in relapse because smokers often use smoking to cope with stressful aspects of their lives.22 24 Factors underlying differences in quitting success by SES is an important topic for future research.

One limitation of the current study is that it uses cross-sectional data. Many of the respondents who were not smoking at the time of the survey will have relapsed afterwards. Further research using longitudinal data is needed to establish the differential long-term success rates of smokers in different socioeconomic groups. Secondly, we only measured the smokers’ reports of having used or not used a certain treatment for smoking cessation and not how compliant they were with the use of that treatment (for example, if they used the prescribed dose of NRT or to what extent they attended behavioural counselling sessions). It cannot be ruled out that compliance with treatment varies across socioeconomic groups. A third potential limitation is the fact that our study did not involve a pure probability sample. However, the present method has been found to yield results very similar to those of pure probability samples and neither of these can eliminate the problem of non-response bias. The major limitation of the study concerns the inability to test other hypotheses to explain the social gradient in success rates. Because the data were cross-sectional we could not reliably assess the role of factors such as nicotine dependence and social milieu that might be subject to recall bias in those who had attained abstinence. This should be a priority for future studies.

We conclude that smokers from more deprived socioeconomic groups are less successful in quitting smoking but try to quit as often and use aids to cessation at least as often as less deprived smokers. Research now needs to focus on possible explanations for this so that this important source of health inequality can be addressed.

What is already known

There is a strong socioeconomic gradient in the rate at which smokers become ex-smokers but it is not clear how far this relates to differences in the rate at which they make quit attempts and/or use effective aids to cessation.

What this study adds

The difference in cessation rates is not attributable to differences in rate of attempts to stop but to differences in the rates of success of quit attempts. This difference in success rate is not attributable to differences in adoption of effective smoking cessation aids.


View Abstract


  • Competing interests: RW undertakes research and consultancy for, and has received travel expenses and hospitality from, companies that develop and market smoking cessation medications. He has a share on a patent for a novel nicotine delivery device.

  • Funding: The Smoking Toolkit Study is funded by Cancer Research UK, Pfizer, J&J and GlaxoSmithKline. DK is the recipient of a European Respiratory Society Fellowship (Number 351).