Stopping smoking is the single most important thing most smokers can do to protect their health, and the earlier they do it the better.1 Most smokers recognise this and express a desire to stop2 3 and try to stop, often repeatedly.3 This paper aims to gain a better understanding of what factors influence smokers to try to stop and how this varies with major sociodemographic variables. This may help design better interventions to promote quit attempts.

Evidence from numerous surveys has shown that in many countries such as the UK the major motivation for wanting to quit is concern about health followed by cost.3 4 These are also the main reasons ex-smokers give for having stopped.3 4 However, it is possible in principle to differentiate reasons for wanting to quit from events that actually “trigger” quit attempts. There seems to be very little research focusing on these “triggers”. Thus, a smoker may cite a number of reasons for wanting to quit, including health and cost but one smoker may be finally prompted to try to quit by an acute illness while another may be prompted by advice from a General Practicioner (GP). We could not find any studies that asked smokers what finally triggered their quit attempts. It could be valuable to gain a better understanding of such triggers to complement what is already known about the reasons for wanting to stop.

Aside from current health problems, future health concerns and concern about the cost, it is possible to examine a range of factors that might act as triggers: health professional advice, pressure of other people, publicity campaigns, new treatments becoming available and other people quitting.

It seems likely that different triggers would be identified by different categories of smoker. For example, one might hypothesise that smokers from more deprived socioeconomic groups would be less likely to cite future health concerns than more affluent smokers. This arises from evidence that smokers from more economically deprived socioeconomic groups are less “future oriented”.5 By contrast, more economically deprived smokers would be expected to suffer more from health problems6 7 and so one might expect current health problems to be cited more often in this group than in more affluent smokers. The cost of smoking will necessarily impact more on those with less money8 and so this too might be expected to be cited more often as a trigger in more deprived smokers.

One might expect on common sense grounds that younger smokers would be more often motivated by future health concerns and older smokers by current health problems. One would also obviously expect pregnancy to be cited primarily by young women, though it is conceivable that some men might also cite it (with respect to their partners). Beyond this, there is no obvious basis for making predictions concerning which sociodemographic groups will cite particular triggers more often.

Recall bias is a limitation of self-report survey studies. This is a particular problem when it comes to recalling quit attempts.9 To examine to what extent forgetting may have led to differential reporting of different triggers, the relationship between the trigger given and time since quit attempt was explored.

It is of interest to know whether some triggers are more conducive to successful quitting than others. For example, with regards to stopping because of current health problems one could argue the case either way. It might be more likely to result in lasting abstinence because of the greater immediate need to quit. However, some health problems such as a respiratory infection may be a trigger to a quit attempt but not lead to lasting abstinence because once the condition has resolved the motivation to quit is reduced. Therefore, we considered it an empirical and open question concerning whether different triggers might result in lasting change.

The Smoking Toolkit Study (STS) involves a series of surveys of the adult population of England examining smoking and smoking cessation patterns (see http://www.smokinginengland.info). An important aim is to track attempts to stop, methods used in those attempts and success rates over time to assess the effects of events and policy initiatives on smoking cessation. It also seeks to provide a better understanding of what motivates quit attempts to help develop and target interventions to promote successful quitting. This paper uses data from the STS to address the following four questions:

• What do smokers report as the final trigger for their most recent quit attempt?

• How does the prevalence of these triggers differ by socioeconomic status, age and gender?

• To what extent is length of time since quit attempt associated with differential reporting of different triggers?

• To what extent are different triggers associated with success of the quit attempt up to the time of the survey?

METHODS

The STS 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.10 The study involves monthly household surveys of national samples in England, carried out by the British Market Research Bureau (BMRB).

The survey uses computer-assisted face-to-face interviews and a multistage quota sample 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, gender, social grade, region, working status and presence of children in the population.

A total of 20 535 respondents were interviewed from November 2006 through to September 2007. Demographic information and smoking status were assessed by using standard questions based on those used in other national surveys (eg, Office for National Statistics (ONS) annual survey).3 The data were weighted by age, gender 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 (eg, 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 (ie, smoked for a year or more). Cigarette smoking prevalence (using responses A and B) was calculated at 24.6% which is higher than the 2006 data from the General Household Survey (GHS) for England (22%)11 but similar to the 2006 ONS omnibus survey for Great Britain (24%).3 A total of 5611 had smoked some form of tobacco in the past year.

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 sample for this study comprised all those who had made at least one quit attempt in the past year (table 1).

Responders were asked: “What finally triggered your most recent quit attempt?”. Only one trigger was allowed. Respondents were shown the following options: advice from a GP/health professional; TV advert for a nicotine replacement product; government TV/radio/press advert; hearing about a new stop smoking treatment; a decision that smoking was too expensive; being faced with smoking restrictions; I knew someone else who was stopping; seeing a health warning on a cigarette packet; being contacted by my local NHS Stop Smoking Services; health problems I had at the time; a concern about future health problems; don’t know/can’t remember; or other (please specify). The list was constructed on the basis of prior questioning of smokers and analysis of possible triggering events.

Statistical analysis

The χ² test was used to assess associations between trigger type and each of the following characteristics: gender, age, social grade, number of cigarettes per day, time since quit attempt and current smoking status. Where significant associations were detected, χ² tests taking each individual trigger separately versus all other triggers combined were undertaken to explore the data further. Linear by linear associations are reported for age and social grade to examine whether a significant gradient was present. With the exception of gender, individual pairwise comparisons within sociodemographic variables were not analysed due to lack of power. Logistic regression analyses examining first order interactions between age, gender and social grade were conducted.

RESULTS

Of the 5611 responders who were smokers a year previously 43.5% (n = 2441) reported making at least one serious quit attempt in the last 12 months.

Of responders who reported making a quit attempt, 89% (n = 2174) identified a trigger prompting their quit attempt, 4.6% (n = 111) reported that just decided to quit and 6.4% (n = 156) did not know or could not remember. The triggers identified by at least 2% of respondents are shown in fig 1. Concern for future health problems was the trigger given for more than a quarter of quit attempts, followed by current health problems and then the decision that smoking was too expensive. Specific factors such as advice from a health care professional, knowing someone else who was stopping, smoking restrictions, family pressure and concern for children were reported less often as the final trigger with each prompting between 5.6% and 2% of quit attempts.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Triggers reported as prompting the most recent quit attempt. HP, health professional.

The χ² test revealed differences between the prevalence of triggers overall by age (χ² = 270.82, degrees of freedom = 85, p<0.001), gender (χ² = 65.31, degrees of freedom = 17, p<0.001) and social grade (χ² = 120.061, degrees of freedom = 68, p<0.001). No overall association was found between triggers and the number of cigarettes smoked per day among those who were currently smoking (χ² = 31.12, degrees of freedom = 30, p>0.05).

To examine these associations further, individual χ² analyses were conducted for each trigger versus all the other combined. With age and social grade, linear associations are reported and where present non-linear associations also reported (tables 2 and 3).

Being triggered by any health reason (ie, future concern, current health problem or general health) showed a significant relationship across social grade with those in the highest grades being more likely to be triggered by a health reason than those in the lowest social grades. However, a different pattern emerged between concern about future health and current health problems. Concern about future health problems was more frequently cited as triggering quit attempts in men and those in the higher socioeconomic status (SES) groups. Responders in the highest SES groups were nearly twice as likely to be triggered compared to the lowest (38.3% vs 21.7%). The inverse relationship, though weaker, was seen for health problems had at the time with those in the lowest social grades being a third more likely to be triggered by this than those in the highest two grades (22.1% vs 16.7%).

A significant age gradient was seen for the citing of any health reason as a trigger with the oldest reporting this approximately 1.5 times more than the youngest (53.2% vs 35.3%). Again when future health and current health were differentiated different relationships were observed. A non-linear association was found for age and future health concern with around 20% of responders in the lowest (16 to 24 years) and highest (65+ years) age groups citing being triggered by this and peaking at 30.6% in the middle age group (45 to 54 years). Health problems at the time were given as a trigger more frequently by older respondents.

Smokers from lower social grades cited cost more often than those from higher grades with the highest proportion of 16.1% being in grade D (the semiskilled and unskilled manual workers).

Advice from a GP or health professional was more frequently cited by women (7.0%) than men (4.3%) and increasing age, those in the oldest two age groups being more than twice as likely to cite being triggered by this advice then their younger counterparts.

Knowing someone else who was stopping was cited less often with increasing age, attempts prompted by this trigger being three times more common for those in the youngest age group compared to the oldest age group (6.4% vs 2.2%).

One would expect pregnancy to be cited by young women and indeed age and gender were found to be factors. There was no evidence of a difference by social grade.

A TV advertisement for a nicotine replacement product was cited less often with increasing age. 3.8% of those in the youngest age group (16 to 24 years) responded that they were triggered by a TV advertisement for NRT, reducing to 0.7% or less in those aged 35 years or over. Smokers with lower SES were more likely to cite health warnings on packets as a trigger.

The only significant interpretable interaction was between age and gender for the trigger “advice from a health professional” such that the effect of age was stronger in men than women (p<0.05 by logistic regression).

As shown in table 4, time since the quit attempt was significantly associated with prevalence of some of the triggers. Hearing about a new treatment, cost and smoking restrictions were somewhat less likely to be reported for attempts made longer ago. Family pressure and pregnancy were somewhat more likely to be reported as triggering attempts made longer ago. However the differences were small and did not affect the overall rank order of triggers cited. Neither were there interactions between time since quitting and the sociodemographic variables of interest.

The only association between current smoking status and trigger type concerned smoking restrictions. Those whose attempts were successful at the time of responding were twice as likely to report being triggered by smoking restrictions than those who had failed (χ² = 9.897, degrees of freedom = 1, p<0.01).

DISCUSSION

The most commonly cited triggers for specific quit attempts were similar to the “reasons” given for wanting to stop smoking in previous studies, namely concern about future health, current health problems and cost.3 4 Of the specific events that were cited, health professional advice was the most common but it was cited by a small minority of smokers.

A number of associations were found between triggers cited and sociodemographic variables, most notably smokers from higher social grades were more likely to cite a concern about future health whereas lower social grade smokers were more likely to cite immediate influences and especially cost and current health problems. This may be due in part to those in the lower social grades being more likely to have poorer health status.6 7 The lowest social grades were also more likely to be triggered to make an attempt by knowing someone else who was ill or had died from smoking, which may also be a reflection of this.

The future orientation of higher social grade smokers was predicted from the previous literature. The fact that it was evident in triggers for actual quit attempts and not just a reason for wanting to quit emphasises the importance of this as a potential factor to consider in smoking cessation communications and counselling. However, before this information can be used, it is important to know how far this factor is malleable. If it proves to be malleable then it would make sense to place greater emphasis on this in publicity campaigns and counselling in lower income smokers to compensate for their relatively low experience of this motivation in the normal course of events. This could be tested in studies specifically examining the effectiveness of smoking cessation messages focusing on future health in lower social grade smokers. If it turns out that this motivation is hard to inculcate in those not already experiencing it, it suggests that communication campaigns and counselling targeted at lower social grade smokers should focus on reinforcing the more immediate motivations to which this group is susceptible. This is an important future area of research.

The social gradient in cost as a trigger reinforces econometric research suggesting that lower income smokers are more responsive to price fluctuations.12 However, the econometric data relate primarily to cigarette consumption and it is noteworthy that even in the UK where the cost of cigarettes is high, smoking is more prevalent among those of lower SES and, among smokers, those of lower SES smoke more cigarettes per day.13 Therefore, while cost may be cited more often as a trigger for quitting in lower SES smokers this may not translate into actual long-term cessation. This is an area that requires further research.

Citation of future health concern showed a curvilinear relationship with those aged 16 to 25 years or over 65 years citing it least often. It is tempting to conclude that young people are less responsive to concerns about health problems that may arise many years into the future, but it should be remembered that the rate of attempts to stop smoking are higher in younger than older smokers and so quitting because of future health concerns is a very important motivator even for the youngest age group. It is also important to note that if the proportion of other triggers is greater, then the proportion of quit attempts driven by future health concerns must be less. It is apparent that young smokers more often cite specific events such as TV campaigns as triggers, and these may well feed into health concerns.

Women were more likely to be triggered by advice to stop by health professionals than men. Other research indicates that women are more likely to report receiving advice on smoking from a health professional compared with men.3 One would expect this to occur in younger women because of the need to stop during pregnancy. In fact, in the present study if one limits the analysis to women aged 40 or over the difference disappears, so it seems likely that the difference is due to pregnancy.

Older respondents were more likely to cite advice from a health professional as a trigger to quitting. It is not clear whether this is because they are more responsive to such advice or they are more likely to receive the advice. This is a potentially useful area for future research.

There is some evidence for recall bias going in different directions for different triggers although the effects of these are small. Probably the most interesting is pregnancy where there is a greater likelihood of this being reported as a trigger for attempts made longer ago. This might be expected given that pregnancy is a long-lasting salient event so therefore more likely to be confabulated as the final trigger. This indicates that studies examining pregnancy as a trigger should focus on quit attempts made within the last 6 months.

The fact that smokers who reported having been prompted to quit by smoking restrictions were more likely to still be abstinent suggests that this policy intervention may do more than prompt quit attempts; it may increase the chances that such attempts succeed. This could be because of the greater effort required to relapse when in social situations.

This study had a number of limitations. First of all, it is not clear how well smokers can remember the triggers to their quit attempts. It is possible that some are remembered better than others and also that there are distortions in their recall. Thus, it could be that GP advice or media campaigns were more commonly a trigger than has been reported. For these reasons, surveys can only provide a broad indication of factors that may be important and research is needed to examine the reliability and validity of self-reports.

A second limitation is that respondents were asked to choose from a list of possible triggers. It is possible that triggers higher up in the list presented on the computer screen were more likely to be cited but this seems unlikely to have had a major impact as the items cited most often were towards the bottom of the screen (ie, current health problems and future health concern). It is also possible that different results would have been obtained had we allowed a free response. In fact this was performed with 387 smokers and recent ex-smokers from two subsequent waves of the STS and a similar pattern of result emerged with future health concern being cited by 24% of respondents, current health problems by 21%, cost by 14% and advice from a health professional by 4%.

A third limitation is that some of these were mental events, such as a concern about future health, while others were external events such as health professional advice. It is possible that, when choosing from the list, respondents had to make arbitrary judgements in cases where more than one of these might apply. Future research should adopt a more structured approach, first of all asking respondents whether they could remember any specific event that made them try to stop on this occasion and then, once they have remembered this event, asking them to say what it was. If there was no specific event one would then ask whether they could remember what they were thinking and feeling at the time.

This study provided some potentially useful insights into motivation to make quit attempts and identified areas for future research. Of particular note were the social class gradients in current versus future health issues.