The UK's Stop Smoking Services have played an important role in efforts to reduce tobacco use in the UK for almost 15 years (Bauld et al., 2012; Flack et al., 2006; Godfrey et al., 2005
). The average impact of the services included in this analysis was 152 extra ex-smokers per 100,000 population between mid-2010 and mid-2011 (Murray et al., 2012
). In general, quit rates in this study followed those reported in previous research: more affluent smokers, older clients and clients taking varenicline or combination therapy were most likely to quit (Brose et al., 2011; Ferguson et al., 2005
). Overall, the effectiveness of different interventions offered by the services did not differ significantly for clients from more or less affluent groups. For most clients, the most effective form of behavioural support was open groups. In terms of advisor type, clients who saw nurses were generally less likely to quit than clients who were treated by other advisor types.
Examining in more detail any differences between intervention types in the study, the form of behavioural support with the most success was open, rolling groups. The exception was for the long-term unemployed and the very small number (n = 2000) of prisoners in the study, for whom open groups were less effective than one-to-one counselling. There are a number of possible explanations for this difference, including the fact that for the long-term unemployed, for example, engaging with interventions that require interpersonal skills may be challenging or uncomfortable (McQuaid, Green, & Danson, 2005; Niepel, 2010
The professional background of the advisor delivering support made little difference to outcomes, with the exception of nurses who were found to be less effective than smoking cessation specialists when treating most groups of clients. Interestingly, clients who saw health care assistants (HCAs) had higher success rates despite these assistants working in the same settings (primary care premises) as nurses (model not shown OR main effect 1.27 (1.10 to 1.46)). These HCAs have fewer formal qualifications but may have more time to spend with clients. They may also be able to build rapport more effectively because they are more likely to be recruited from the same local communities as smokers themselves (Bosley & Dale, 2008; Lizarondo, Kumar, Hyde, & Skidmore, 2010; NHS Scotland, 2010; Petrova, Vail, Bosley, & Dale, 2010
In addition, although clients supported by GPs can achieve high quit rates this association disappeared once the SSS where they worked was taken into account — only a very small number of SSS recorded having GPs as smoking cessation advisors. It is possible that the boundary was blurred between brief advice and more intense interventions in the reporting of these SSS.
Some differences in intervention effectiveness for SES groups were found when SES was measured by NS-SEC rather than free prescriptions. The advantage of NS-SEC is that it provides a more finely graded distinction between SES groups which may be necessary to tease out whether and where changes should be made to services. There was no evidence that disadvantaged smokers were less likely to quit because they were attending less effective services i.e. it was not the case that disadvantaged smokers were not accessing open groups or were predominantly receiving behavioural support from nurses.
5.1. Limitations of the study
This study only provides short-term outcomes because routine data, where follow up is conducted at four weeks, was analysed. The advantage of this short period was that less than 30% of cases were lost to follow up. In smoking cessation studies lost to follow up tends to be an indicator of relapse (West, 2006
) so these cases could be retained in the analysis as non quitters avoiding selection bias. A particular limitation of routine data is that quality varies resulting in a lower level of accuracy than would be the case in a research study. Data on level of tobacco dependence and social support was limited to a few cases and could not be used in the analysis. Data on the advisor type was also poorly collected (it is not a national requirement for reporting) so further research is needed to corroborate our results on the relationship between who provides support and outcomes.
The naturalistic design has the advantage of real world observation but it meant that there was no control group so we cannot be certain whether using the services significantly increased quitting. Background quit rates (the chances of a smoker quitting with no help from the services) have however been estimated to be between 2% and 3% for 2006 (West, 2006
) which is considerably lower than the 34% quit rate for this study.
We measured disadvantage by two proxies: NS-SEC and eligibility for free prescriptions. Both of these have weaknesses. Only personal NS-SEC was available which may not reflect the disadvantage of household members who are not the chief income earner. Secondly, NS-SEC can be difficult to classify. NS-SEC was recorded by advisors rather than researchers with expertise in classifying socio-economic status and the questions asked to ascertain NS-SEC were simpler than those used by government surveys. This may have caused some idiosyncrasies in the results — occasionally quit rates by intervention type for clients with routine and manual occupations were more similar to clients with professional and managerial occupations than to clients with intermediate occupations.
Treatment was free for all clients through the UK government funded National Health Service. In other countries more affluent clients might be asked to pay for treatment. Eligibility for free prescriptions can be determined by medical reasons as well as SES. The collection of data on medical conditions was not collected consistently or in enough detail for us to determine whether the free prescriptions were due to health or income reasons. We excluded pregnant women but data on whether women had a child under the age of one was not collected so we could not exclude such women. Eligibility for free prescriptions is only a measure of disadvantage for 19 to 59 year olds in England. Results for all age groups are presented elsewhere (Murray et al., 2012
). In general, results were similar.
SSS level disadvantage did not significantly predict quitting. This may be because the health administration areas were too large or SSS disadvantage does not necessarily correlate with the disadvantage of individual clients. Ethnic minorities varied by health administration area and to increase the robustness of the analysis all ethnic minorities were combined, as the numbers were very small. In future research it would be helpful to analyse individual ethnic groups separately.
It is likely that group dynamics affect the chances of successful smoking cessation for smokers who took part in groups (Paddock et al., 2011
). Group dynamics are extremely complex to model particularly in an open group setting where members change from session to session (Morgan-Lopez et al., 2011; Paddock et al., 2011
) and no attempt to assess these was included here. Thus standard errors may have been underestimated (Paddock et al., 2011
). Different group dynamics may exist in groups with all low SES members, all high SES members or mixed membership. In future research focussing on socio-economic status and group dynamics may provide more insight into how to increase quit rates amongst more disadvantaged smokers taking part in open and closed groups.
Another concern with intervention types is that there are similarities between one-to-one and drop-ins, drop-ins and open groups, and open and closed groups. Thus these interventions may in practice lie on a continuum rather than be distinct entities. The same issues apply with age and SSS disadvantage. These continuous variables were divided into quartiles and quintiles to aid interpretation and allow the possibility of non linear effects.
There was significant variation between advisors and between health administration areas in the full multivariate model. Thus there must be more characteristics that can explain the variation which we were not able to model, which may have implications for the quit rates of low SES and high SES groups, such as differences in health administration areas' priorities, leadership, enthusiasm and cultures (Amos et al., 2011
) and motivation, gender or rapport with clients for advisors (Hiscock, Moon, Pearce, Barnett, & Daley, 2012
This study provides new insight into the types of behavioural support and smoking cessation service characteristics that are associated with biochemically validated quit rates amongst smokers from different socio-economic groups. As other studies have found, smokers who were more disadvantaged were less likely to be successful in their quit attempt than more affluent groups, even after accessing cessation services. With the exception of the most disadvantaged clients (the long-term unemployed and prisoners), smokers who attended open group therapy for smoking cessation were more successful than those who received one-to-one support. Clients were equally likely to quit if they received support to stop from staff trained as smoking cessation specialists, health care assistants or pharmacists, but less likely to quit if they were treated by nurses. Further research is required to explore in more detail how the background and training of those involved in delivering smoking cessation services may interact with client characteristics to affect outcomes.