Start2quit: a randomised clinical controlled trial to evaluate the effectiveness and cost-effectiveness of using personal tailored risk information and taster sessions to increase the uptake of the NHS Stop Smoking Services

Mass mailing and proactive strategy

Studies suggest that the direct marketing approach has potential as a population-based strategy for recruiting smokers into support services, and could provide treatment access to individuals who might not otherwise seek cessation care. Paul et al. 20 explored the acceptability of direct marketing and proactive contact offering cessation services to smokers. The authors reported that 92.8% of the sample found it acceptable for the health service to contact people to offer assistance and 55.7% said they were likely to take up the offer of individual counselling. This could be an overestimation of actual take-up of the service, but suggests that proactive contact is acceptable and that smokers are open to the idea of intensive counselling. The importance of proactively encouraging smokers to quit has also been demonstrated in studies exploring recruitment to telephone quit lines, lending support to the ‘cold call’ telephone approach. These studies suggested that demand and interest in using services or receiving information about quitting may be greater than is reflected in current usage rates, and that proactively offering services could result in an increase in uptake. 21–24 A recent systematic review of recruitment methods for smoking cessation programmes suggested that personal tailored messages and proactive and intensive recruitment strategies can enhance recruitment. 25 This review confirmed the conclusion reached by McDonald26 that interpersonal strategies have a positive effect on recruitment into smoking cessation programmes.

Lichtenstein and Hollis27 employed a more proactive recruitment method. They invited smokers attending a medical appointment to an immediate intervention where they were offered information about what attendance at the service would involve, and a strong referral message to the service. Attendance at the first session of the cessation programme increased to 11.3%, compared with 0.006% in a control group who received brief advice only. Fiore et al. 28 also showed that many primary care patients identified as smokers will accept treatment ‘if it is free, appropriately incorporated into the health-care delivery system to ensure convenience, and encouraged through proactive recruitment’.

In line with these findings, a major UK study used a proactive strategy to identify individual smokers and inform them about available cessation services. In a cluster randomised controlled trial, Murray et al. 29 identified all patients in general practices recorded as current smokers or with no status recorded. These patients were proactively informed by letter about the SSSs and given the option of being contacted by an advisor. Smokers in practices allocated to the intervention group indicating that they would like to speak to an advisor were contacted within 8 weeks by a researcher trained as an advisor and offered advice and an appointment. Smokers in control group practices received no further contact. Overall, the proportion of current smokers expressing interest was 13.8%, suggesting that more than the current 5% of the smoking population setting quit dates within the NHS were interested in receiving help. Furthermore, Murray et al. 29 reported a 7.7% absolute increase in smokers using the SSSs in the intervention group over the control group at the 6-month follow-up, and an increase of 1.8% in validated abstinence in those smokers requesting contact over the control group (4% vs. 2.2%).

This study by Murray et al. 29 was the first in the UK to assess a proactive method of recruitment to attract smokers into the SSSs. It demonstrated the potential to increase attendance, and also indicated that novel methods of marketing are needed in order to engage interested smokers to encourage use of the SSSs.

Individual computer-tailoring and risk information

One possible way of enhancing recruitment is to use individual characteristics to personalise and tailor communications. Computer-based systems can generate highly tailored materials, defined as ‘any combination of information or change strategies intended to reach one specific person and based on characteristics unique to that person’. 30 This technology offers a method for personalisation of communications to patients and can include an individualised risk communication element based on an individual’s own risk factors, more personally relevant to the consumer than information about population ‘average’ risks. 31

The use of fear in health promotion has been the subject of debate and is somewhat controversial, with claims that ‘shock tactics’ do not work, are too frightening, or can backfire and prove to be counterproductive by prompting a maladaptive behavioural response. There is also a general notion that healthy lifestyle campaigns and anti-smoking messages should be positive and reflect non-smoking role models rather than dealing with the ‘scary’ health consequences of smoking. 32 However, in a review of studies on fear appeals, Sutton33 concluded that increases in fear in communications are associated with increases in acceptance of the recommended action, in a linear relationship. Providing recipients with a reassuring message that adopting the recommended action would be effective, together with clear advice on how to go about it strengthens intentions to follow the advised course of action. 34

Fear messages about smoking can indeed push people to attempt to quit. The fear induced by such messages can be dealt with adaptively by a behavioural response that removes the reason to be fearful, such as quitting smoking, or maladaptively by, for example, denying the truth or personal relevance of the threat. 32 The likelihood of eliciting the desired response can be maximised by empowering the recipient and giving reassurance that it is possible, and also providing a ‘helping relationship’ that is needed to succeed. 35 This has been demonstrated to good effect in mass media campaigns, particularly that of the Australian national anti-smoking advertising campaign, which used graphic fear-based messages as a dominant part of the communication strategy, but tagged with the national quit helpline number and an additional advertisement encouraging calls to that number. 36 This strategy is also consistent with social cognitive models such as the health belief model, which posits that a greater perceived risk of a disease and perceived efficacy of the action in preventing it is associated with increased participation in a recommended behaviour. The health belief model also highlights the importance of providing a specific cue to action, which can act as a trigger and increase compliance with the recommended behaviour. 37,38

Additional justification for this approach lies in the evidence that smokers do not fully acknowledge their own personal vulnerability. Data show unequivocally that smokers acknowledge that their risks of health problems are higher than those of non-smokers. However, studies indicate that they substantially underestimate their own personal risk and tend to conclude that they are less likely to suffer health effects than other smokers. 39 A large literature demonstrates this ‘optimistic bias’ or ‘unrealistic optimism’. Weinstein et al. 40 provided further clear evidence that smokers engage in risk minimisation by convincing themselves that they are not as much at risk as other smokers. Moreover, smokers do not perceive the relationship between the amount smoked and their perceived risk. Thus, even if people are aware of the well-publicised risks, they resist the idea that the risks apply to them, and a key factor is getting participants to acknowledge that these risks are personally relevant.

Computer-tailoring can be used to customise an individual’s risk factors, enhancing perceived personal relevance and helping to overcome the tendency to deny that the information in the tailored messages applies to the recipient. These personal communications can both inform the smoker of their own personal risk, while at the same time promoting confidence and providing the helping relationship that is essential to encouraging acceptance of the advice and following the recommended course of action. These basic tenets were put together in the 3Ts (Tension, Trigger, Treatment) model proposed by West and Sohal. 41 They proposed that triggers can lead to sudden changes and that creating motivational tension can trigger action in smokers who are predisposed, or motivated, to change. The immediate availability of treatment can then prompt action.

Research has shown that individually tailored self-help materials have a small but useful effect over generic materials on smoking cessation. 42 The addition of personalised risk communication that is more personally relevant to the consumer has also been found to increase uptake of screening. 31 Computer technology can be used to produce a communication that combines the tenets of this model, and can also be combined with proactive recruitment methods with the potential to engage with and recruit a larger proportion of the smoking population in a relatively inexpensive way.

Opportunity to experience a support service without commitment

In addition to the factors noted as barriers to the use of support services, the literature also suggests that many smokers are unaware of, or have insufficient knowledge of or inadequate information about, the services available. 20,43 This lack of knowledge can also lead to the belief that ‘it wouldn’t help me anyway’. 20 The combination of ‘why quit’ messages, hard-hitting messages about the consequences of tobacco use and ‘how to quit’ messages, supportive and positive and emphasising quitting resources, was recommended in the Global Dialogue for Effective Stop Smoking Campaigns,44 an international review of the literature. This report also recommended that promotional efforts need to both build awareness that getting help will increase the chances of success and build awareness of and comfort with the quitting services.

Lichtenstein and Hollis27 demonstrated how a proactive and personal approach can be combined with an opportunity to gain more information about the service and what it involves at a no-commitment introductory session. Their intervention included an assessment, measurement of expired-air carbon monoxide (CO) level with an interpretation, video testimonials and the opportunity to ask questions, a voucher fee waiver and immediate scheduling of the smoker for the group. Thus, including a personal invitation with an appointment to a no-commitment introductory session offers the opportunity to gain an insight into what the service can offer, and also has the potential to increase service use.

This research study brings together evidence on proactive and direct mail recruitment, on personalised computer-tailoring and risk information, and on offering the opportunity to experience a support service without commitment, to evaluate a complex intervention comprising all of these elements, in encouraging and increasing attendance at the English SSSs.

The intervention is further enhanced by the addition of a repeated personal letter, with a further invitation sent 3 months after the original to all participants who fail to attend a taster session. This is consistent with recommendations made by Lichtenstein and Hollis27 who proposed that, with repeated advice over time, a greater proportion will be likely to respond.

Principal research question

We hypothesised that smokers, identified from general practice records, sent brief personal tailored letters based on characteristics available in their primary care medical records and on a short screening questionnaire, and invited to a ‘Come and Try it’ taster session designed to inform them about the SSSs, were more likely to attend the services than those who received a standard generic letter advertising the service.

Primary objective

The primary objective of the study was to assess the relative effectiveness and cost-effectiveness of a complex intervention, consisting of proactive recruitment by a brief personal letter tailored to individual characteristics available in medical records, and an invitation to a ‘Come and Try it’ taster session to provide information about the SSSs, over a standard generic letter advertising the service, on attendance at the SSSs of at least one session.

Secondary objectives

The secondary objectives were to (1) assess the relative effectiveness of the intervention on biochemically validated 7-day point prevalent abstinence rates at the 6-month follow-up; (2) compare the cost-effectiveness of the intervention; (3) assess the relative effectiveness on additional periods of abstinence measured by self-report of not smoking for periods of 24 hours to 3 months at the 6-month follow-up; (4) assess the number of smokers attending the taster session and the number of smokers completing the 6-week NHS smoking cessation course; (5) assess the number of quit attempts made and any reduction in daily cigarette consumption; (6) explore the effectiveness of the intervention by socioeconomic status and social deprivation; (7) explore reasons for non-attendance and barriers to attendance at the SSS; and (8) determine predictors of attendance at the services and at the taster sessions (in the intervention group).

Target population

The target group was smokers motivated to quit who had not attended SSSs in the previous 12 months. We aimed to target smokers in areas of high deprivation and ethnic minorities, where smoking prevalence is high.

Inclusion/exclusion criteria

All current smokers willing to participate and returning the signed consent form, aged ≥ 16 years, able to read English, motivated to quit and who had not attended SSSs in the previous 12 months were eligible for inclusion in the study. For the purposes of this research, motivation to quit was defined as answering ‘yes’ to either or both of the following questions:

1. Are you seriously thinking of quitting in the next 6 months?

2. Would you think of quitting if appropriate help was offered at a convenient time and place?

Exclusion criteria were minimal because the aim was to recruit all smokers into the services. However, smokers aged < 16 years were excluded because of the need for parental consent to participate for this age group, and any patients identified considered by the GP to be unsuitable for the project, for example the severely or terminally ill, were excluded.

Stop Smoking Service and practice recruitment

We worked with the National Institute for Health Research Primary Care Research Network (PCRN) to identify SSSs willing to participate in the trial, and to then identify and recruit practices in the selected SSS areas. We aimed to select more practices in areas of high deprivation and of large ethnic communities to ensure full representation of smokers most in need of help, and to maximise the generalisability of the results. The SSSs and PCRNs assisted in identifying sufficient practices serving these populations.

The trial was originally planned to cover 10 different areas served by a SSS, and we initially recruited Camden (North London) and Oxfordshire SSSs and practices in those areas for the pilot phase of the trial. The extension of the trial and the increase in the size of the sample meant that the number of SSSs was increased to 18 in areas representative of the English SSS.

Participant recruitment

All smokers aged ≥ 16 years were identified from their medical records in participating practices. GPs screened the list to exclude anyone they deemed to be unsuitable for the research, for example severely or terminally ill, and the list was also screened to ensure that only one person from the same address was selected. All remaining persons on the list were sent a letter from their GP inviting them to participate in the trial, together with a participant information sheet describing the research, a consent form and a screening questionnaire. Participants were asked to consent to the use of information from their medical records and information given in the screening questionnaire to send them information about quitting, and for the researchers to access relevant data from their attendance at the SSSs. Data generated from the screening questionnaire were used both to assess the criteria for inclusion in the trial, to provide information for the computer-generated tailored intervention letter and to provide baseline characteristics. Participants returned these questionnaires to the practice using a Freepost envelope. Non-responders were sent a reminder and duplicate questionnaire after 3 weeks. All smokers returning the signed consent form and who were eligible to participate were randomised to the intervention or the control group. Patients had the opportunity to decline to participate but to return the questionnaire with basic information to update their smoking status in their GP practice records.

All recruitment materials can be found in Appendix 3. The baseline questionnaire is included in Appendix 4.

Control group

Participants allocated to the control group were sent a standard generic letter from the GP practice advertising the local SSS and the therapies available, and asking the smoker to contact the service to make an appointment to see an advisor.

Intervention group

Participants allocated to the intervention group received the following:

• a brief personalised and tailored letter sent from the GP that included information specific to the patient, using information obtained from the screening questionnaire and from their medical records

• a personal invitation and appointment to attend a ‘Come and Try it’ taster session to find out more about the services, run by advisors from the local SSS

• a repeated personal letter with a further invitation sent 3 months after the original to all participants who failed to attend a taster session following the first letter and invitation.

Each of these components is described in the following sections in detail.

Development of the tailored intervention letter

The overall objectives of the letter were to communicate personal risk level if the person continues to smoke, using individualised information on the risk of serious illness, and to encourage attendance at the SSS.

Development of the taster session and training

The goal of the taster session was to offer information about the SSS, promote the service, address any concerns or queries smokers may have about the service provided and encourage sign-up to a course. It was not intended to replicate the first session of a course.

A draft of the content of the taster sessions was prepared by one of the co-investigators (SG), and was developed into a standard protocol in consultation with the research team. The final standardised protocol also included items from the NHS Centre for Smoking Cessation and Training’s Standard Treatment Programme46 to ensure conformity with national guidelines, and a detailed manual was produced. The standard protocol for the taster session included:

• a motivational element, congratulating attendees on coming to the session

• an introduction to the SSS, emphasising that it is a free service and based on well-researched evidence

• emphasising the importance of stopping smoking and outlining the benefits of quitting, both health and financial/lifestyle

• information about the services offered, outlining the structure the treatment programme in one-to-one or group sessions, the length of sessions and of the course

• information about what to expect when they attend and the content of advice, for example emphasising that no-one will be forced to quit, but will be helped to explore the reasons for and against wanting to give up smoking, and helped to develop strategies to resist smoking after their quit date

• interaction between attendees, discussing, among other things, reasons for stopping

• discussion about withdrawal symptoms, with information about the available medications and range of nicotine replacement therapy (NRT) products available

• the opportunity to have a measurement of CO level taken, with an interpretation

• a 5-minute digital versatile disk showing group and one-to-one sessions in progress, and testimonials from previous successful attendees, produced by University College London (UCL) Media Services in collaboration with Camden SSS

• the opportunity to ask questions about the service

• details of how to contact the service and a clear and persuasive invitation to sign up for a group or individual session.

Between two and seven advisors in each SSS, already trained to give smoking cessation advice in group and one-to-one sessions, attended a 2-hour training session to enable them to facilitate the taster sessions according to the standardised protocol and manual. The training sessions took place within each SSS and were led by two members of the team (either HG or SG), and included an explanation and clarification of the study protocol and procedures, and specified the exact information to be delivered in the session. Only trained advisors led the taster sessions, and each session was run either by one advisor with additional administrative support provided by one other, or the presentation was divided between two advisors. They were encouraged to introduce themselves and describe their background and expertise to reassure attendees of their credibility and expertise.

It was intended that each SSS should run between 4 and 12 taster sessions, depending on the number of participants recruited and the area covered by the SSS, and that up to 50 participants were invited to each session, which lasted approximately 1 hour. Sessions were held in the early evening normally beginning between 18.00 and 19.00. On arrival attendees were asked to sign an attendance sheet provided by the research team. An evaluation form was developed for immediate assessment of attendees views of the session, the form included space to indicate interest in signing up to the SSS. Advisors encouraged attendees to complete these evaluation forms before leaving the session and, when possible, participants were given a date and time of a group or one-to-one session before they left. The taster session protocol in included in Appendix 6.

Procedure

Recruitment, collection of baseline data and delivery of the intervention took place over 4 years, between January 2011 to June 2011 for the pilot phase, and between January 2012 and December 2014 for the main trial. The procedure in each practice took 12 weeks to complete, during which time the research assistants visited the practice on four occasions, assisting practice staff with mailing invitation letters and questionnaires to patients, processing returned questionnaires, and generating tailored and generic intervention letters. A series of purpose-written computer programs written in Visual Basic for Applications (1997–2003, Microsoft Corporation, Redmond, WA, USA) that read and write Microsoft Excel^® (Microsoft Corporation, Redmond, WA, USA) and Microsoft Word^® (Microsoft Corporation, Redmond, WA, USA) files facilitated these processes, and were installed on laptops for use in the practices.

The practice staff initially ran a search using the practice computer system to identify all patients recorded as smokers. The GPs then screened the list for exclusions. Research assistants visited the practice to first run the ‘Check for Duplicates’ computer program, which randomly selected one smoker from each address on the list, and then using the ‘Invite’ program-generated letters inviting smokers to participate, which were then mailed. Research assistants visited the practice a second time to process returns, using a third computer program ‘Risk’ which allocated identification numbers and randomised participants to the control or intervention group. This computer program also combined the data from the baseline questionnaire and medical records with the correct messages from a message library, written using Microsoft Word, to generate tailored letters and invitations to the taster session for those participants randomised to the intervention, and the generic letters for control participants. These were then mailed to the participants. The first taster session was held approximately 2 weeks after this mailing. Research assistants visited the practice on two further occasions to process responses from patients in the same manner as previously described. Further taster sessions were timed accordingly.

Three months later, all participants in the intervention group who had previously been invited to attend a taster session but had not attended were sent a second intervention letter (identical to the first) and a further invitation to a taster session (Figure 1).

FIGURE 1.

Study schedule showing the duration and timing of the procedure, intervention and follow-up. RA, research assisstant. Reproduced from Gilbert et al. 45 © The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY license.

Patients who returned the questionnaire with written consent but who were not eligible to take part in the study were sent a letter thanking them for responding and informing them that they did not fit the study criteria, and patients who returned questionnaires outside the time frame for processing were sent a similar letter informing them that the recruitment period had ended. Both of these letters were sent from the general practice and contained information about the local SSS, advising the smoker to contact the service for more information or to speak to an advisor.

Security and baseline data management

The patient-level data collected in this trial comprised information downloaded from practice records and information provided by participants on the consent form and baseline questionnaire. The information from the practice record was used to generate letters inviting patients to participate in the trial. It was also used, along with baseline questionnaire information, to generate the tailored and generic letters.

All data files and backup media remained in the practice until all eligible participants had been randomised and the tailored and generic letters generated. At this point data files were transferred to the study centre at UCL using proprietary encryption software (TrueCrypt, V61.1; TrueCrypt Foundation, Henderson, NV, USA).

At the end of recruitment in each practice anonymised average data on patients who were invited to participate in the study but did not respond were collected to establish the external validity of our results. These anonymised data comprised gender, date of birth and postcode. Date of birth was converted to age at the time of the invitation and the postcode converted to an Index of Multiple Deprivation (IMD) score via GeoConvert (2011, UK Data Service Census Support, University of Essex, University of Manchester). IMD is the government’s official measure of multiple deprivation at the small-area level, which provides a relative ranking of areas across England according to their level of deprivation. Names and all other identifying data were removed.

Baseline measures

Inclusion criteria (age, intention and motivation to quit, and previous SSS attendance), demographics (gender, marital status, qualifications, employment and ethnicity), self-reported health, dependence on nicotine (number of cigarettes per day and time from waking to first cigarette), smoking history (age started and previous quit attempts), determination and confidence to quit were assessed in the baseline screening questionnaire.

Outcome measures

Interim analyses and stopping guidelines

The study was initiated with a pilot phase conducted in seven practices recruited from two SSSs. This was intended to be approximately 20% of the original total sample. The criteria for judging the success of the pilot phase and proceeding to full trial was based on:

1. achieving a 7% response rate (i.e. a mean of 42 participants per practice giving consent and agreeing to randomisation) in the first seven practices

2. a preliminary analysis that suggested that the uptake of smoking cessation services in the intervention group was greater than in the control group (i.e. the difference in proportions, intervention minus control, was greater than zero).

No other stopping rules were applied.

Levels of significance

During the course of running the trial but prior to locking the database, based on further expert statistical advice, we devised an analysis plan for interpreting significance levels for analysis on multiple outcomes of interest. Hence, the interpretation of the results of the trial for the primary outcome, (1) engagement with SSS, and the main secondary outcome, (2) 7-day point prevalent abstinence, was governed by an alpha spending plan that preserved the study-wise alpha for (1) and (2). We hypothesised that these outcomes fall naturally into a hierarchy with (1) as a step prior to (2). We employed a hierarchical monitoring plan in which alpha was spent first on (1) and the remaining alpha was available for (2). The simple formula below describes alpha allocation in the hierarchy:

where subscript 2 = two sided; s = study-level critical alpha (0.05); e = engagement with SSS; and c = 7-day point prevalent abstinence. Thus, if the p-value for attendance at SSS was 0.02, there remains a p-value of 0.031 to spend on the second outcome of 7-day point prevalent abstinence and a p-value for that outcome of < 0.031 would be considered significant. If the p-value for the primary outcome (difference in smoking cessation service attendance) was > 0.05, then the overall study would be considered neutral and any finding on the second outcome considered exploratory with a nominal p-value.

Likewise, if there was a significant decrease in attendance within the intervention arm over the control arm, the second outcome would be considered exploratory with a nominal p-value.

Subgroup analyses

In order to assess whether or not the intervention was any more effective for any particular subgroup of smokers, we explored interactions between intervention and deprivation (defined in fifths), intervention and gender, and intervention and age (defined by categories 16–39 years, 40–64 years and ≥ 65 years), for the primary outcome (attendance) and 7-day point prevalent abstinence at the 6-month follow-up. We had planned at an early stage to analyse the interaction with deprivation,58 and with gender and age when drawing up our analysis plan,59 prior to the end of data collection.

Subsidiary analyses

We also explored any delayed effect of sending repeat reminders to smokers on the uptake of service, and any differences in attendance due to seasonal variations.

Practice recruitment

Eighteen SSSs spread across England, located in both high and low areas of deprivation and representing both large and small organisations, agreed to participate in the trial (Figure 2). Ninety-nine practices within the participating SSS areas, identified and approached by PCRNs, agreed to participate. The number of practices per SSS ranged from 3 to 10, and the practice list size ranged from 2205 to 26,000 (mean, 9723; median, 9725). Cumulative list sizes, shown in Table 2, ranged from 17,617 in Medway to 106,424 in Durham and Darlington. Current cigarette smokers aged 16–99 years were identified from computer records in participating practices (n = 141,488; 14.7% of the total list size). The proportion of smokers identified in each practice ranged between 4.8% and 39.7%, and within each SSS from 9.7% to 20.81%.

FIGURE 2.

Map showing level of socioeconomic deprivation in England and SSSs participating in Start2quit.

The study targeted higher-risk groups; therefore, practices in areas of high deprivation were preferentially selected as determined by the practice postcode converted to an IMD score (the government’s official measure of multiple deprivation at the small-area level). A majority (54.6%) of practices were located in areas of high deprivation (i.e. within the two highest quintiles of IMD scores, a score > 21.34) (Figure 3). This suggests that the catchment area for these practices were within areas of high deprivation or that a higher number of participants were living in highly deprived areas; however, the postcode of the practice does not always indicate the deprivation status of the catchment area. Nevertheless, a comparison of the mean IMD scores of all smokers identified for each practice with the IMD of the practice indicated a similar trend, and there was good correlation between smokers’ IMD scores and those of the practices (r = 0.80) (Figure 4).

FIGURE 3.

Percentage of practices (%) within each IMD quintile.

FIGURE 4.

The IMD scores of practices compared with the mean IMD scores of all smokers living in the practice catchment area.

One would also expect to find a high proportion of smokers identified in areas of high deprivation, indicated by a high IMD score, and in many cases there appears to be a close match. However, in some areas the percentage was lower than would be expected, and this may be accounted for by a high ethnic population, in which smoking prevalence may be lower in women, for example in Derby and Brent (Figure 5).

FIGURE 5.

The IMD scores of practices and the proportion of smokers identified in the practice. Derby and Brent are highlighted in blue.

Participant recruitment

The recruitment of participants for the pilot phase of the trial was conducted between January and March 2011. Recruitment for the main phase began in January 2012 and was completed in October 2013.

General practitioners excluded 4186 patients considered to be unsuitable to take part in the study. Reasons for exclusion included patients unable to understand English sufficiently, serious pre-existing condition or terminal illness, and severe cognitive, mental or psychological impairment. However, most practices did not give the reasons for exclusion and we are unable to estimate numbers in each category. A further 25,086 patients were excluded because of duplicate addresses, ensuring that only one person from the same address was selected. All remaining persons on the list (n = 112,216) were invited to participate in the trial. Of these, 21,971 (19.6%) replied. However, 5333 replied to say they were non-smokers and 420 were returned to the practices unopened as a result of incorrect addresses or deceased. Records are not always accurate, and those returned from non-smokers are likely to represent only a portion of those recorded as smokers but are actually non-smokers. Thus, our best estimate of total potentially eligible is 106,463 (see Figure 6).

The total number of questionnaires returned from potentially eligible participants was 16,638 and, of these, 10,380 declined to take part in the study but returned the questionnaire with basic information only to update their smoking status in their practice records. A further 1874 were willing to take part in the study but did not fit the inclusion criteria, leaving 4384 participants enrolled in the trial, representing a response rate of 4.1%. Of these, 2636 were allocated to the intervention group and 1748 to the control group. One participant from the intervention group withdrew from the study before follow-up commenced and 4383 were analysed. See the Consolidated Standards of Reporting Trials (CONSORT) flow chart (Figure 6) for details.

FIGURE 6.

The CONSORT diagram of recruitment and flow of participants through the trial. a, Percentage of total list size; b, total invitations sent minus non-smokers and wrong address/deceased; c, of total potentially eligible. Adapted from Gilbert et al. 45 © The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY license.

Follow-up

Follow-up data collection took place 6 months post randomisation and was conducted between August and November 2011 for the pilot phase and between August 2012 and July 2014 for the main study.

Complete validation data of attendance at the SSS were obtained for each participant from SSSs at the end of the quarter following the end of the 6-month follow-up period in each area. Additional data were obtained by telephone interview or postal questionnaire. In total, 2910 (66.4%) completed the full telephone interview, 302 (6.9%) completed a shorter paper version of the follow-up questionnaire returned by post and an additional 160 (3.7%) completed the four basic questions related to the primary outcome, giving a total response rate of 3372 (76.9%). There was no difference in follow-up response between the treatment groups: 76.7% and 77.3% in the intervention and control groups, respectively. The reasons for loss to follow-up were declined to complete the interview (n = 150), not able to be contacted (n = 857), and died within the 6-month follow-up period (n = 4) (see Figure 6). The timing of completing the follow-up ranged from 20 days prior to the due date (180 days after randomisation) to 194 days after the due date (mean, 27.53 days; median, 22 days). This was not statistically different between the intervention and the control groups.

There were large differences in recruitment between SSSs, ranging from 2.3% in Brent to 6.7% in Oxfordshire, and also in follow-up response. There were also large variations between practices within SSSs. We deliberately included some practices in areas with high ethnic minority populations, for example in Derby and Brent, but in these practices recruitment was especially low (Table 3).

Biochemical validation of 7-day abstinence

Of the 630 participants who answered ‘not at all’ in response to the question ‘How often do you currently smoke cigarettes or rollups?’ at the follow-up, and who were asked to provide a salivary cotinine sample to validate abstinence, 595 (94.4%) agreed to send a saliva sample for analysis and 443 (70.3%) returned a sample; 399 (63.3%) samples were sent for analysis. Samples from 44 participants who reported that they had resumed smoking between follow-up and returning the sample were not analysed.

Of the samples analysed, 345 (54.8%) were validated (249 had a cotinine content of < 12 ng/ml, 36 were from participants using NRT and 60 from participants were using e-cigarettes). Of the 54 samples not validated, 30 had a cotinine content of > 12 ng/ml and were from participants not using NRT or e-cigarettes, five samples came from participants who had smoked in the previous 6 days and 19 samples were of insufficient volume to be analysed. There was no difference between the intervention and control groups.

Baseline

Table 4 shows the demographic and smoking characteristics by intervention and control group.

The sample was 50.9% male and had a mean age of 49.3 years; 54.5% lived with a spouse or partner and 95.6% were of a white ethnic background. Half of the sample (50.7%) were living in areas of high deprivation, defined as IMD quintiles 4 and 5, and 32% of the sample were living in a household with another smoker. In terms of nicotine dependence, approximately one-quarter were highly dependent; however, a high proportion fell into the low-dependence category, probably accounted for by the non-daily smokers (7.8%), who were not excluded from this study. The mean age at which participants started to smoke was 16.5 years. Regarding intention to quit and motivation, 42.5% were planning to quit some time in the next 6 months, 55.7% had previously quit for > 1 month and 33.9% had previously attended the SSS (but not in the previous 12 months). Motivation and determination to quit were relatively high (mean 3.76 and 3.74, respectively, scored on a 1–5 scale), but confidence in this ability to quit was lower (mean 2.71). Although 74.1% reported having no health problems linked to smoking, 45.7% were recorded as having at least one QOF disease.

Characteristics associated with study attrition

There were differences between those who completed the follow-up interview, either by telephone or by post, and those who did not complete. Participants who did not complete were younger (mean age 45.2 years vs. 50.5 years; p < 0.001), more likely to be living in a highly deprived area [IMD quintile 5 (27.7%) vs. IMD quintile 1 (16.4%); p < 0.001] and were more likely to be single (30.1% vs. 20.6%; p < 0.001). Non-completers had a higher dependence score (mean score 2.73 vs. 2.57; p = 0.004), were more likely to have not previously quit for > 24 hours [longest quit attempt < 24 hours (29.6%) vs. > 1 month (21.7%); p = 0.003] and were more likely to intend to quit in the next 2 weeks than to intend to quit in the next 6 months (28% vs. 19.2%, respectively; p = 0.001). Non-completers also scored higher on wanting and determination to quit (3.86 vs. 3.73 and 3.85 vs. 3.71, respectively; both p < 0.001) and were more confident in their ability to quit (2.79 vs. 2.69; p = 0.005). Participants who completed the follow-up were more likely to have previously attended the SSS (20% vs. 24.6%; p = 0.001). Completion was not related to self-reported or recorded health status (Table 5).

External validity

Anonymised data of patients who were invited to participate in the study but did not accept were compared with the data of the participants to explore whether or not the sample was representative of the whole population of smokers. Numbers in the anonymised data set are higher than those stated in Figure 6 because of errors in the initial search identifying smokers, leading to a difference in the electronic anonymised data retrieved from practices.

Overall, differences were found between participants and non-participants in gender, age and deprivation (Table 6). Males were under represented in the sample (50.9% vs. 54.3%); however, the difference was small, and may only be significant as a result of the large sample size. The difference was more pronounced in some SSSs than others (Figure 7). Participants were significantly older than non-participants (mean age 49.3 years vs. 43.3 years) and this was reproduced consistently in each SSS. The IMD score was significantly different between participants and non-participants, but the difference was small (1.18); however, there was large variation in IMD scores between individual SSSs. In areas where the population had higher IMD scores, for example Derby SSS and Durham and Darlington SSS, the mean IMD score of participants was lower than that of non-participants, whereas in areas of lower deprivation (Buckinghamshire and Swindon SSSs) the sample recruited tended to be more deprived than the average (Figure 8).

FIGURE 7.

Percentage male participants and non-participants by SSS area.

FIGURE 8.

The IMD scores of participants and non-participants by SSS area.

Overall, although the sample is consistently older than the average smoker in all areas, in terms of gender and IMD score the differences are small; therefore, we can probably consider the sample to be representative and have reasonable external validity.

Primary outcome

The proportion of people entering the smoking cessation service (i.e. attending the first session of a 6-week course) over a period of 6 months from the receipt of the invitation letter, as measured by records of attendance at the SSSs, was significantly higher in the intervention group than in the control group (17.4% vs. 9.0%; unadjusted OR 2.12, 95% CI 1.75 to 2.57; p < 0.001) (Table 7).

Secondary outcomes

The main secondary outcome of 7-day point prevalent abstinence at the 6-month follow-up, validated by salivary cotinine, was significantly higher in the intervention than in the control group (9.0% vs. 5.6%; unadjusted OR 1.68, 95% CI 1.32 to 2.15; p < 0.001). All other periods of abstinence measured by self-report (24-hour and 7-day point prevalent abstinence, 1- and 3-month prolonged abstinence), 3-month-validated prolonged abstinence and the number completing the 6-week SSS course were significantly higher in the intervention group than in the control group (see Table 7).

There was a slight reduction in daily cigarette consumption in those continuing to smoke in both the intervention and the control groups, and the proportion of these who had made a quit attempt (approximately one-quarter) was similar in both groups. Intention and motivation to quit changed little in these continuing smokers, and there was no difference between the groups (Table 8).

Subgroups

Table 9 shows subgroup interactions in outcomes.

The effect of the intervention on attendance at the SSS was greater for males (19% vs. 8%; OR 2.70, 95% CI 2.04 to 3.57) than for females (15.7 vs. 10.1%; OR 1.67, 95% CI 1.28 to 2.19), and the interaction term was significant (p-for-interaction = 0.01). There was also a differential effect for males and females for validated 7-day point prevalent abstinence (p-for-interaction = 0.01). The intervention was more effective for males (OR 2.37, 95% CI 1.63 to 3.42) than for females (OR 1.23, 95% CI 0.88 to 1.72). Although attendance at the SSS in the intervention group was similar in all IMD quintiles, it was lower in the control group for participants in quintiles 2–4 (medium deprivation) than for those in quintiles 1 or 5, thus the differential was significantly greater (p-for-interaction = 0.001). Abstinence was lower overall in the higher-deprivation groups, but with no significant interaction. There were no significant interactions by age category. Both attendance and abstinence increased with age in both groups.

Effect of repeat reminders

We also explored any effect on the uptake of service of sending repeat reminders to smokers approximately 9–15 weeks (63–105 days) after randomisation. Mean time in days from randomisation to attendance in valid cases (i.e. dates reported by SSSs falling within the 6-month time frame) was 47.35 days (SD 37.91 days) in the intervention group and 55.99 days (SD 49.01 days) in the control group. The most frequent time for attendance in the intervention group was between 19 and 29 days (mode 21 days), with no noticeable increase between 63 and 105 days. In the control group, multiple modes were 15, 40 and 62 days, but there was no clear pattern in the peaks (Figure 9).

FIGURE 9.

Time from randomisation to attendance at the SSS in the (a) intervention group (n = 445) and (b) the control group n = 147. Reproduced from Gilbert et al. 45 © The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY license.

Seasonal variation

To assess any seasonal variation in attendance and any differences between the groups, we examined attendance by quarter and by group. Attendance was significantly higher overall in earlier months (17.4% in January to March vs. 11.4% in October to December; p < 0.001), but the relative effect between groups was not significantly different (Table 10).

Variation in outcome by Stop Smoking Service

To assess any variation in the outcome by SSS we examined attendance and 7-day-validated abstinence in each SSS area by treatment group. Overall attendance at the SSSs varied from 2.1% to 23.1%, and attendance varied from 3.3% to 28.3% in the intervention group and from 0% to 16.7% in the control group. In all SSSs, the number attending was higher in the intervention group than in the control group (Figure 10). The ICC for the overall attendance was 0.031 (95% CI 0.01 to 0.09), suggesting that around 3% of participants’ tendency to attend the SSS was explained by the SSS in which they were located.

FIGURE 10.

Percentage of participants attending SSSs by treatment group and by SSS.

The proportion of participants achieving validated 7-day abstinence varied from 2.1% to 13.4% overall, and from 3.3% to 16% and from 0% to 10.5% in the intervention and control groups, respectively (Figure 11). The ICC for 7-day abstinence was 0.034 (95% CI 0.011 to 0.096), again suggesting that around 3% of participants’ tendency to quit smoking was explained by the SSS in which they were located.

FIGURE 11.

Percentage of all participants with validated 7-day abstinence by treatment group and by SSS.

Background

Adherence can be defined as the degree to which behaviour coincides with the recommendations of health-care providers. 62 In trials this can be applied to the intervention as a recommended strategy and accurate measurement of adherence behaviour is necessary to be able to assess whether or not the outcomes can be attributed to the recommended strategy. Any decisions or recommendations based on the outcome depend on valid and reliable measurement of adherence to the treatment,63 and inadequate adherence can adversely impact the effectiveness of an intervention.

One component of this intervention consisted of the opportunity to gain more information about the SSS and what it involves at a no-commitment taster session. This session was part of the strategy to encourage attendance at SSSs and, therefore, attendance at the session was hypothesised to be a major influence on the decision to attend the SSS. Not all participants took up the offer of a taster session. In order to assess the impact of the taster sessions on the primary outcome of the study, details of the provision and attendance at the taster sessions are reported. In addition to the original primary outcome (attendance at the SSS), the trial was powered to assess 7-day point prevalent abstinence at the 6-month follow-up, validated by salivary cotinine analysis. This allowed assessment of whether or not this intervention, and subsequent attendance at the SSS, also translated to increased quit rates. Thus, the first aim of this part of the analysis was to examine the flow of participants from attendance at the taster session, through attendance at the SSS, to achieving validated 7-day abstinence, and the associations between attendance and abstinence.

One approach to measuring adherence behaviour is to ask participants for their subjective ratings of attendance. However, literature on adherence to both medical and behavioural regimes has shown that the subjective reports of patients are problematic. Inaccuracies can result and adherence can be either over- or underestimated. 63 More reliable objective strategies were used in this study to measure attendance both at the taster session and at the SSS. However, self-reported adherence can be revealing, and the second aim of this section was to describe the self-reported attendance, and the amount of agreement and disagreement with the objective measures of attendance.

Methods

Results

Taster session and Stop Smoking Service attendance and smoking status outcome

Of the 2635 participants in the intervention group invited to attend a taster session, 739 (28%) attended.

In the intervention group more participants who had attended a taster session attended the SSS than those who had not (45.7% vs. 6.3%), and more participants who attended a taster session and the SSS achieved 7-day abstinence (28.7%) than those who only attended the taster session (10%) or only attended the SSS (17.5%). Participants who did not attend a taster session nor the SSS had the lowest rates of validated 7-day abstinence (4.4% and 4.7% in the intervention and control groups, respectively) (Figure 12).

FIGURE 12.

Flow chart showing numbers attending a taster session, the SSS and achieving a validated 7-day point prevalent abstinence in the intervention and the control groups.

Summary

The opportunity to find out more about the SSS by attending a taster session was taken up by less than one-third of participants. The large difference of 40% in SSS attendance between attendees and non-attendees at taster sessions suggests that the session was successful in encouraging uptake of the service, although it is possible that the more motivated smokers were more likely to attend the taster session and also more likely to attend the SSS.

Participants in both the intervention and control groups were more likely to be abstinent if they had attended the SSS. This increase in the proportion abstinent suggests that attending the SSS is more likely to result in abstinence than quitting alone, and confirms the consistent evidence that individual counselling increases the likelihood of cessation. 64,65 Participants who received the intervention and attended the SSS were most likely to be abstinent, whereas participants who received the intervention but did not attend the SSS were not more likely to quit than those in the control group who had not attended the SSS. Attending either the taster session or the SSS (i.e. having some contact with the support services or an advisor) was beneficial, although clearly attending the SSS was most beneficial, and the abstinence figures for the intervention group suggest a possible added benefit of receiving a personal risk letter.

The self-reported SSS attendance figures reflect the effectiveness of the intervention. However, the discrepancy between self-reported attendance at the SSS and the objective measures confirmed the inadequacy of subjective measurement data. Recall or retrospective bias can account for some of this discrepancy, but there is also an issue of interpretation or misunderstanding. Although every effort was made to word the questions clearly, there may have been a lack of differentiation in the minds of the respondents between the taster session and the SSS course, which could account for the slightly better agreement in the control group than in the intervention group. Alternatively, participants may have received advice from elsewhere and not have signed up to a SSS course. In addition, social desirability, the desire to give the impression of compliance, could explain some of the discrepancy. Overall, this demonstrates the importance of collecting objective data, when possible.

Background

Thorough evaluation of an intervention should include an assessment of how acceptable and appropriate participants found the intervention. Participant satisfaction is an important determinant of the potential impact, as it is associated with adherence to the recommended advice and compliance to medical regimes in primary health care. 66,67 This is particularly important when evaluating personal health communication. The superiority of tailored health communications depends on how well received the material was by the target population. 68 Furthermore, assessing the perception of the tailored letter used in this study is crucial because of the use of personal information related to diseases. Issues related to health typically tend to be anxiety-laden and associated with negative rather than positive feelings,69 and a fear-driven appeal message presenting individuals with information about themselves can prompt a maladaptive behavioural response by, for example, denying the truth or personal relevance of the threat. 32

The focus of this section is to evaluate the perception of both the personal risk letters and of the taster session, assessed using measures from previous trials of tailored feedback,70,71 and included in the telephone interview at the 6-month follow–up, using both closed and open questions. Perception of the taster session was also assessed by an evaluation form completed immediately following each session.

Methods

Results

Summary

This analysis of the perception of the intervention suggested that both parts were well received by the target population. The personal letter was well accepted, and participants who received this personal letter as opposed to the generic letter were more likely to read it and to discuss it with others. They saw it as personally relevant and it appeared to increase their confidence and determination to quit. With regard to whether or not the intervention group found that the letter may have caused upset, few respondents reported feeling very or extremely angry, anxious or depressed.

The taster sessions were also viewed positively by the majority of attendees. However, actual attendance at the SSS resulting from the session was lower than the stated intention to sign up immediately following attendance at the taster session. The comments left by attendees immediately after the session also endorsed the positive view of the sessions, suggested that the taster sessions were reassuring and built the intended awareness and comfort with the services. They also offered some insight into the reasons smokers might not attend, such as the fear of criticism and victimisation.

Questions asking respondents to attribute their decision to attend the SSS or to quit might suggest that, for many smokers, both the letter and the taster session worked in synergy to encourage attendance and attempting to quit. However, in view of the response rate to these questions and possible problems of interpretation and recall, there is insufficient evidence to corroborate this conjecture and further research is necessary.

Additional reasons given for making an appointment with the SSS and for attending the taster session indicated that smokers who attended were thinking of quitting and were prompted by the communication to act on their plans. Similarly, reasons for not attending the taster session suggested that quitting was not a priority and other commitments were more important at this time.

Finally, the additional question included to assess whether or not reminders would encourage attendance suggested that it would not be worthwhile. The control question did not suggest contamination bias as a result of a large number of control participants having access to information from the taster sessions.

Background

Treatment fidelity refers to the extent to which an intervention that is delivered matches that described in the intervention manual. Knowing the fidelity compared with the manual is important for interpretation of results, as without this knowledge it is impossible to determine how much the intervention in question is the primary mechanism in any changes observed. 72

Assessment of the fidelity of delivery in complex behavioural interventions requires authors to report the content of the intervention, the characteristics of those delivering the intervention, the setting, the length and adherence of the delivery to the protocol. However, many studies fail to report the content of the intervention or the content and duration of training, and a recent systematic review found that few studies of psychosocial treatments evaluated treatment fidelity. 73 For behavioural interventions, in particular, variation in delivery quality can strongly mediate intervention effectiveness. This failure to account for fidelity may lead to underestimation of the intervention effects and faulty conclusions that interventions are not effective, when in reality they were not implemented as intended, or to the acceptance of statistically effective interventions, which differ greatly from their initial design. 72

The use of present/absent checklists has been identified as the most reliable method to monitor the fidelity of delivery of study interventions. 72,74,75 Key to this method is the prespecified intervention content or ‘treatment manual’, a manual containing explicit guidelines about the content and method of delivery of an intervention to ensure that all providers receive the same training and information. In addition to identifying the extent of fidelity of delivery, it is also important to understand the ‘essential’ components of an intervention. 76,77 A taxonomy of behaviour change techniques (BCTs) has been developed and modified to apply to behavioural support for smoking cessation. 78

Procedures to assess the fidelity of delivery of the taster sessions were embedded into the trial. We aimed to answer the following research questions:

1. To what extent did advisors adhere to protocol-specified content and BCTs?

2. Were advisor or session characteristics related to adherence to protocol-specified content?

3. Was adherence to protocol-specified content related to participants’ attendance at the SSS or to validated 7-day point prevalent abstinence?

Method

Results

Adherence

A median of 71.23% (mean 68.53%) of protocol-specified behaviours was delivered across sessions, ranging from 29% to 96%. As can be seen in Figure 13, low-adherence sessions were not concentrated in particular SSS areas.

FIGURE 13.

Percentage adherence to manual-specified content in each analysed taster session.

Median fidelity to specific BCTs across sessions varied from 50% (‘Summarise information’ and ‘Give options for support’) to 100% in five of the BCTs. However, two of the five consisted of only one behaviour, and another two consisted of only two behaviours. The fifth one, ‘provide information on the consequences of smoking and smoking cessation’, consisted of six behaviours and had the highest adherence (Figure 14).

FIGURE 14.

Median fidelity to BCTs across sessions.

Discussion

The observed median of 71.2% adherence to specified behaviours is higher than that reported by many other similar studies. 79–81 The standardised training in helping smokers to quit that all advisors had received, in addition to having previous experience of running individual or group stop smoking courses, could account for this. However, although the adherence of many was reasonably high, one-quarter fell below 50% and one as low as 29%.

Overall, adherence was greater in sessions that were run jointly by two advisors. The second advisor possibly acted as a safety net, delivering key items that the first advisor failed to present. In addition, presenting in pairs could have mitigated feelings of nervousness for some advisors who had less experience of facilitating groups. Shorter sessions were associated with higher levels of adherence, and being female and in the mid-range age category were significantly associated with adherence to the protocol. Of the BCTs, ‘provide information on consequences of smoking and smoking cessation’ had the highest median percentage fidelity. Information on the consequences of smoking is likely to be a common area of focus for SSS advisors as the health impact of smoking is the most common reason for smokers to attempt to quit.

Adherence to the protocol was not related to the main outcome of attendance at the SSS or to abstinence from smoking. Responding to an invitation to a smoking cessation study and then attending a session to find out more about the SSS indicates a high level of interest in both quitting smoking and getting support to do so, and indicates generally high levels of motivation of attendees. This enthusiasm could indicate that the presentation of protocol-specified content was less important than theorised, as attendees were already feeling inclined to sign up for a SSS course.

The study is limited by focusing solely on adherence to the protocol and neglecting more subtle competency-related variables and non-specific effects, such as communication characteristics (e.g. providing examples to clarify points, and empathetic tone). In addition, a number of advisors provided extra information, both relevant and irrelevant, which was not analysed.

The results of this evaluation indicate that the study intervention is not likely to have been significantly impacted by issues of fidelity, and we can have greater confidence that variability in the main outcome of the study is not a result of variability in SSS advisor adherence to the protocol of the taster sessions.

Background

Until recently, little research had been devoted to identifying and removing barriers to entering stop smoking programmes, prompting Fiore et al. 65 in the US Department of Health and Human Services guidelines to stress the importance of further investigation in order to identify and remove barriers to treatment. Research both before and since this call has identified that major barriers include a lack of awareness, low expectations of effectiveness, most smokers underestimating the benefits of treatment and a belief that motivation and willpower are sufficient. 10,11,16,43,82–86 The belief that ‘if you really want to quit smoking you will succeed on your own as well as you would with help’ is prevalent. 16 Concern about lack of empathy and notions that treatments are unavailable and hard to access also permeate the literature. 43,82,83,85,86

Copeland et al. 49 adopted a more structured approach to identifying the most common reasons for not seeking treatment and developed a validated 40-item measure of reasons for not entering smoking cessation programmes. Copeland et al. 49 developed and used the TBQ, specifically with low socioeconomic smokers in the USA, to assess barriers to entering smoking cessation programmes and to relate the barriers to demographics, stage of readiness to quit smoking and dependence.

As part of the Start2quit study we adapted the TBQ and distributed it to a sample of English smokers in order to explore the barriers to seeking help and support to quit in the UK, specifically to attending the English SSSs. The aims of the study were:

1. to identify perceived factors that prevent smokers from seeking help in a UK population

2. to examine the factor structure of the adapted TBQ and investigate whether or not it shows the same dimensions across different groups.

In a more detailed analysis, logistic regressions were conducted to determine the most important barriers and whether or not they differ according to demographic and smoking-related characteristics. However, this work was not within the remit of this report and a more extensive analysis and interpretation will be reported in a future output.

Method

Results

Discussion

Exploratory PCA to identify barriers to entering the SSS yielded a 30-item, seven-factor solution, accounting for 53.06% of the variance. Work and time constraints accounted for the highest proportion of variance (14.81%); the notion that smokers should quit on their own and that nothing can help in quitting smoking were the next highest (9.93% and 8.19%, respectively). Other barriers emerging were lack of interest in quitting, lack of social support to attend, lack of privacy at SSS programmes and a lack of information about the SSS.

The results of the PCA were similar to those reported by the developers of the original US questionnaire. 49 Both analyses extracted a seven-factor solution comprising very similar scales. ‘Work and time constraints’ and ‘smokers should quit on their own’ correspond to those found by Copeland et al. 49 ‘Opinions about professional assistance’ and ‘misinformation about professional assistance’ correspond to the combined and renamed ‘nothing can help in quitting smoking’, and this factor is more specific in representing the belief that support offered is ineffective. Although ‘preparedness to quit’ and ‘lack of interest in quitting’ equate and represent smokers not ready to quit, in our study lack of interest in quitting accounted for only 6.54%. This might be because all participants in the present study had already expressed an interest in quitting prior to taking part in the study. The remaining three factors were also similar but were renamed to be more appropriate for each scale. Responses to the open questions also support this interpretation. The most frequently given answers relate to the factors accounting for most of the variance, with the only difference being that ‘lack of interest in quitting’ has the highest frequency of responses.

The results were also consistent with previous research suggesting that time constraints, disbelief in the efficacy of programmes, lack of childcare and transportation, and lack of knowledge about smoking cessation and support programmes are the main perceived barriers to entering smoking cessation programmes. 16,43,82–86 Although other commitments are often put forward as a reason for not attending, the notion that motivation is up to them and support can only help them stay abstinent88 is prevalent. There is a need to challenge the commonly held perception that really wanting to quit is sufficient. 88 A culture change is necessary and health-care professionals need to become more proactive84 if use of support services globally is to be increased.

The size of the sample, more than adequate to meet the requirement of a factor analysis and large enough to allow cross-validation of the questionnaire,89 is a strength of this study. However, the smokers that participated in the present study were recruited for the larger study and were somewhat motivated to quit; therefore, the sample may not be fully representative of the wider population of smokers in England.

This analysis has shown the validity of the TBQ on an English population of smokers motivated to quit smoking following an explicit invitation to attend the SSS (either with risk information and taster session invitation or without). This validation of the TBQ will also enable the comparison of data across studies and populations. Delineating the most common reasons for not seeking treatment could lead to changes to the advertising of services or in information given to smokers by clinicians and could increase treatment seeking. 82

Assessment of costs

A micro-costing approach was applied to estimate the costs of the two trial interventions. The costing methods involved three steps: (1) identifying the relevant cost items (identification), (2) measuring the use of the identified cost items (measurement) and (3) placing a value on these cost items (valuation). 91

Measurement of resource use

Health-care utilisation was systematically collected for each participant alongside the trial. Two main components of health-care resources were used in the trial.

First, the use of resources associated with the delivery of the trial interventions was recorded. This included staff time, consumables (such as postage and printing) and resources required for the training of advisors in the delivery of the taster sessions. Second, patients’ use of health and social care services was recorded using comprehensive service-use questionnaires, as used previously in a number of trials. Table 23 lists the resource items for the two trial interventions. Only resources directly related to the intervention in practice were included in the costing exercise. Resources that were used solely for the purpose of research were excluded.

Smokers aged ≥ 16 years were identified from their medical records by practice staff from participating practices. GPs from these practices then carried out further screening to exclude smokers deemed to be unsuitable for the research, such as those who were severely or terminally ill. GPs spent approximately 2.5 hours screening each practice’s smokers list for eligibility, but GP time was spent for research purposes and was not considered part of the intervention cost.

All eligible smokers were then sent a baseline questionnaire with a cover letter from their GP. A Freepost envelope was included for the return of the questionnaire to the practice. For non-respondents, a reminder and duplicate questionnaire was sent to the smoker 3 weeks after the first mailing. The cost components of the baseline questionnaire consisted of consumables such as printing, envelopes and postage, and staff time used to prepare and post the letters. It is noted that, although baseline questionnaires were sent to both trial groups, the cost was included only for the intervention group because the questionnaire was needed to gather information for generating the tailored letter. For the control group, generic letters would be sent to smokers directly after screening if implemented in the practice. Therefore, the cost of the baseline questionnaire was considered a research cost and excluded from the analysis.

Smokers who returned a completed baseline questionnaire, signed the consent form and were eligible to participate were randomised in a ratio of 3 : 2 to the intervention or the control group. The control group received a standard generic letter advertising the SSS and asking the smoker to contact the service to make an appointment to see an advisor. The relevant resource use includes stationery, mailing and staff time to prepare and post the letter.

In the intervention group, participants were sent a brief personalised and tailored letter based on their medical records and information provided in the returned baseline questionnaire. The resource used for generating the tailored letter includes consumables (printing, envelopes and postage), and staff time used to prepare and post the letters. The intervention letter also included an invitation and an appointment to attend a ‘Come and Try it’ taster session. Resource use associated with the taster session includes training and time of facilitators, equipment and venue overheads.

In addition to the intervention-related resource use, data on health resource use beyond the trial interventions were collected alongside the trial. These resources include the use of the SSSs, other smoking cessation aids and wider health-care resource use.

Numerical information on attendance at SSSs and the intervention type for each participant was provided by the SSS. Participants were asked at follow-up about the smoking cessation aids and other health-care services they had used in the previous 6 months. Smoking cessation aids included pharmacological (NRT, bupropion and varenicline) and non-pharmacological interventions (GP, nurse, pharmacist visits, smoking helplines). Wider health resource use consisted of primary health-care services (visits to primary care professionals) and secondary health-care services (hospital admissions and outpatient attendances).

Unit costs for resource use

The total costs of the identified resources were calculated using national unit costs from a range of published sources. The use of national unit costs can eliminate the difference in local costs among different sites and reflect the actual cost after generating the intervention to the real practice. 50 Table 24 shows unit costs employed to calculate the total cost per participant in the trial. The main sources from which unit costs were obtained include the Personal Social Services Research’s Unit Costs of Health and Social Care 2013,90 the NHS Reference Costs 2012–13,95 Prescription Cost Analysis England 201396 and the 2008 Smoking Cessation Services: Costing Report. 94

Health outcome measures

In addition to the clinical outcome measures used in the statistical analysis (the uptake of SSS and quit rates), health benefits were measured in QALYs in the economic evaluation. Unlike clinical outcomes such as quit rates, which can only be compared between smoking interventions, QALYs are a standard and internationally recognised method enabling comparisons across different health-care programmes. QALYs take into account both quantity and health-related quality of life and are recommended by NICE technology appraisal guidance. 50

In this study, QALYs were calculated as the area under the curve, following the trapezium rule, using utility scores measured by EQ-5D questionnaires at baseline and at the 6-month follow-up. 51 An EQ-5D questionnaire is a self-complete questionnaire that evaluates participants’ quality of life on five dimensions: mobility, self-care, usual activities, pain/discomfort and anxiety/depression. Each dimension has three levels, indicating no problem, some or a moderate problem and a major problem. Based on their combined answers to the five questions, participants could be classified as being in 1 of 245 possible health states (the 243 states arising from the EQ-5D, plus unconscious and deceased). A preference-based index utility score was derived for each participant’s response to the EQ-5D questionnaire using the UK population tariff calculated by a time trade-off method. 97,98 Perfect health has a utility value of ‘1’ and death was assigned a value of ‘0’; some health states may be considered worse than death and have negative scores.

Cost-effectiveness analysis

To assess the relative cost-effectiveness of the two trial interventions, we employed an incremental CEA to compare the mean difference in costs with the mean difference in health benefits. The ICER was used to combine costs and health benefits in a single measure to which a decision rule for cost-effectiveness can be applied:99

where E represents the change in effects and C represents the costs of intervention, measured in monetary units, and subscripts ‘I’ and ‘C’ refer to intervention and control arm, respectively. In this study, three sets of ICER were conducted using the three outcome measures using the following formulas.

1. Cost per additional attendee to the SSS:

   (4)ICER1=ΔCΔE=(Average cost of the tailored letter +average cost of the taster session)−average cost of the generic letterSSS attendance rateI − SSS attendance rateC.

2. Cost per additional quitter at 6 months:

   (5)ICER2=ΔCΔE=(Average cost of the tailored letter +average cost of the taster session +average SSS costI + health-care costI)−(Average cost of the generic letter +average SSS costC + health-care costC)Quit rate​ at 6 monthsI−Quit rate at 6 monthsC.

3. Cost per ​additional QALY gained at 6 months:

   (6)ICER3=ΔCΔE=(Average cost of the tailored letter +average cost of the taster session +average SSS costI + health-care costI)−(Average cost of the generic letter +average SSS costC + health-care costC)QALY gain at 6 monthsI−QALY gain at 6 monthsC.

Uncertainty assessment

A feature of CEA is the uncertainty that exists around the ICER. In this study, uncertainty around the decision to adopt the intervention is assessed through a non-parametric bootstrap resampling technique. Non-parametric bootstrap is a method that resamples observations from the original samples. 93 Bootstrapping has been proposed as an efficient approach for calculating the confidence limits for the ICER, as its validity does not depend on any specific form of underlying distribution. 100–103 We performed the bootstrap replications 5000 times; for each bootstrapped resample, an estimate of the differential costs and QALYs was calculated and 95% CIs for the ICERs were constructed based on the bootstrapping results. 101,102 In addition, cost-effectiveness planes (CEPs) and CEACs were plotted based on the outcomes of the bootstrap iterations. 104

Handling missing data

Missing data are a common and considerable problem in most economic evaluations alongside clinical trials because of the high missing rate of the health-care resource-use questionnaires. In this study, missing data were handled using Rubin’s multiple imputation method. 99,105–107 We assumed any missing data to be missing at random when the probability of the data being missing can depend on the observed values of the individual but not on the missing variable values of the individual. A total of 50 imputed data sets were generated to ensure efficient and reproducible estimates.

Sensitivity analysis

Besides the multiple imputation analysis, a sensitivity analysis was undertaken to repeat the CEA using complete cases, whereby the results are analysed only for those participants who had both completed cost and outcome data at the same time. Because the intervention of interest in this study was a multicomponent intervention, the costs and effects of each intervention element were explored as the second part of the sensitivity analysis.

Long-term costs and outcomes predictions

The within-trial analysis shows the short-term cost-effectiveness of the interventions. However, in common with many preventative interventions, many studies on smoking cessation have demonstrated that the majority of benefits are gained from the reduced risk of developing smoking-related diseases, such as lung cancer, myocardial infarction, COPD and stroke, and the reduced health-care costs and improved health-related quality of life it brings over a longer time period. 108,109 These benefits may not be evident until later in life; hence, the 6-month follow-up period of the trial is not sufficient to capture the real value of the smoking interventions. 99,107,110 Therefore, in this study, besides the within-trial analysis, we included an estimate of lifetime health-care cost savings and QALY gains associated with the two interventions.

The method employed in this study’s long-term CEA was based on an economic model that was developed to estimate smoking-related costs and consequences in adults in England. 52 This is a Markov model that simulated the pathway between smoking and four main smoking-related conditions (lung cancer, myocardial infarction, COPD and stroke) and evaluated the related costs of current smokers, ex-smokers and never-smokers. However, this study only reported the cost savings from quitting smoking. In the absence of estimated lifetime health outcomes, we combined and utilised another English study by Vogl et al. ,111 which reported health-related quality of life by smoking status.

Both future costs and health outcomes were discounted at 3.5% per annum when necessary, in line with the NICE guidelines for the technology appraisal methods. 50 Similar to the short-term CEA, ICERs were calculated and the uncertainty of the calculated ICERs were tested using the bootstrapping method.

Costs

Intervention costs reflect the value of resources required to deliver the trial interventions. Table 25 summarises the average cost of each intervention component for both treatment groups. The cost of screening was £0.30 per person recruited into the trial. For the intervention group, the cost of the baseline questionnaire used to gather information for generating the tailored invitation letter was £32 per person. The average costs of the tailored and generic invitation letter were £13 and £0.60 per participant, respectively. The average cost of the taster session was £27 per person per session; this included £0.10 for the training manual and £26 to provide the session. Four participants in the control group attended the taster sessions; the costs of these sessions were included in the control group according to the intention-to-treat principle.

Table 26 and Figure 15 show the mean cost for the interventions and the subsequent use of health services over the 6-month follow-up period. The mean intervention cost in the intervention group was £54 (SD £12) per participant, whereas the corresponding average cost in the control group was only £0.90 (SD £2) per participant. Costs associated with smoking cessation, including use of SSSs, and pharmacological and non-pharmacological cessation aids, were significantly higher in the intervention group than in the control group. The wider health resource-use cost was also higher in the intervention group, but the difference here is insignificant.

FIGURE 15.

Cost difference between two intervention groups.

The estimated total mean costs over the 6-month follow-up period were £777 (SD £2176) in the intervention group and £679 (SD £1860) in the control group. Participants’ average cost relies heavily on their wider health resource use over the past 6 months. The cost difference between the two groups was £98 (95% CI –£26 to £222). After adjusting for baseline resource use, the adjusted cost difference was £92 (95% CI –£32 to £216). The 95% CIs indicate that the total cost differences were not statistically significant between the two treatment groups.

Outcomes: utility and quality-adjusted life-years

The primary health economic outcome was QALY gains over 6 months, estimated using the EQ-5D. Mean EQ-5D scores in each intervention group at both baseline and the 6-month follow-up are reported in Table 27. At baseline, the estimated between-group difference (intervention – control) in EQ-5D score was –0.004. The results show that mean EQ-5D scores were improved over the study period for both trial arms. The estimated between-group difference (intervention – control) in EQ-5D score was 0.004 (95% CI –0.015 to 0.030) at 6-month follow-up. This indicates that compared with the control group, the mean EQ-5D score was lower in the intervention group at baseline, whereas after 6 months the mean EQ-5D score was higher in the intervention group.

Based on the EQ-5D scores, at both the baseline and the 6-month follow-up, QALYs were calculated and reported in Table 27. The number of unadjusted average QALYs per participant over the 6 months was 0.0001 (95% CI –0.008 to 0.009), that is, they were higher in the intervention group than in the control group. After adjustment for baseline utility scores, the results demonstrate that QALY gains were slightly higher in the intervention group compared with the control group (the difference in QALYs was 0.002, 95% CI –0.001 to 0.004). However, the QALY differences between treatment groups were not statistically significant.

Cost-effectiveness analysis and uncertainty

We conducted three sets of cost-effectiveness analyses using three outcome measures in this study. The results are reported as incremental cost per outcome of interest, cost per additional attendee to the SSS, cost per additional quitter and cost per QALY. The results of the analyses are presented in Tables 28–30.

First, after participants received the invitation letter and taster session, 458 (17.4%) smokers in the treatment group and 158 (9.0%) smokers in the control group attended the SSS (see Table 28). The average costs incurred in the intervention group and control group were £54 and £0.90, respectively, resulting in an ICER of £627 per additional attendee to the SSS. The ICER indicates that, when comparing the tailored letter and taster sessions with the generic letter, a cost of £627 was incurred to enable one more smoker to take up the SSS.

Second, combining differential costs of the treatment groups with the differential quit rates at the 6-month follow-up, the ICERs are listed in Table 29. The main quit rate used in this trial was biochemical validation of 7-day abstinence; the corresponding ICER was £2689 (95% CI –£952 to £6329) per additional quitter. This means that if the decision-makers were willing to pay more up to £2689 for an additional quitter, the tailored letter and taster session would be the preferred option; otherwise, usual care should be adopted. ICERs using other periods of abstinence measures (from 24-hour self-reported point prevalence to 3-month-validated prolonged abstinence) are also reported in Table 29. The cost per additional quitter ranges from £1699 to £4053 using the multiple imputation data set.

Third, we compared the effectiveness of the two interventions using QALYs as health outcome measures (see Table 30). It was found that the intervention was associated not only with a slight gain in QALYs following adjustment for baseline EQ-5D, but also with higher costs than the control condition. This generates an ICER of £59,401 (95% CI –£604,833 to £644,486) per QALY gained.

However, the differences in both QALYs and costs are not statistically significant, indicating that there may be some uncertainty surrounding the ICERs. We employed the non-parametric bootstrapping method to investigate the uncertainty over mean differences. A scatterplot of the 5000 bootstrapped incremental mean costs and mean QALYs pairs is presented on the CEP (Figure 16). The horizontal axis divides the plane according to incremental cost (positive above, negative below) and the vertical axis divides the plane according to incremental effect (positive to the right, negative to the left). 99 The axes divide the plane into four quadrants through the origin; in the south-east quadrant, the intervention group is less costly and more effective than the standard letter and the intervention is considered more cost-effective. In the north-west quadrant, the opposite is true and the control is more cost-effective, whereas in the north-east and south-west quadrants, the decision depends on the maximum willingness-to-pay threshold. The slope of the two lines in Figure 16 shows the NICE range of decision-making thresholds: £20,000–30,000 per additional QALY gained. The proportion of the scatterpoints that fall to the south and east of the willingness-to-pay threshold line is the probability that the intervention is more cost-effective than the standard letter. Using the results of the bootstrapped replicates, we also generated a CEAC (Figure 17), which provides a plot of probabilities that the intervention is being cost-effective (y-axis) against all potential values of willingness-to-pay thresholds (x-axis).

FIGURE 16.

Cost-effectiveness plane (multiple imputation analysis).

FIGURE 17.

Cost-effectiveness acceptability curve (multiple imputation analysis).

In Figure 16, the majority of the plots in the CEP fall in the north-east quadrant, indicating that the intervention is likely to be more effective, but also more expensive, than the control. The CEAC (see Figure 17) illustrates that, when using the NICE decision-making threshold range of £20,000–30,000 per QALY gained, the tailored letter and taster session has a 20–27% probability of being considered cost-effective at 6 months. Only at higher willingness-to-pay thresholds (>£59,401 per QALY) does the intervention becomes more likely to be cost-effective than the control.

Sensitivity analysis (complete-case analysis)

In order to explore the potential impact of missing data on estimated treatment effects and costs, a sensitivity analysis was conducted based on the complete cases. In this trial, completed costs and outcome data at the same time were available for 2775 participants (63% of all participants), 1667 in the intervention group and 1108 in the control group. The number and percentage of missing data with regard to each cost component are summarised in Table 31. Table 31 also shows the mean costs of the complete cases by treatment groups and resource categories. Tables 32–34 list the results of the CEA using 2775 complete cases.

To make it easier to compare the primary analysis and sensitivity analysis, Table 35 summarises the main results of the two analyses. When excluding participants with incomplete data, there is little change in the results. The results show that, for both treatment groups, the average costs were slightly higher in the complete cases, and the gains in effectiveness increased slightly too. The ICERs of the complete-case analyses were estimated at £498 (95% CI £491 to £504) per additional attendee to the SSS, £2552 (95% CI –£1199 to £6303) per additional quitter and £49,842 (95% CI –£425,064 to £536,813) per QALY gained when comparing the intervention with the control group.

Similar to the multiple imputation analysis, we re-ran the bootstrapping approach to generate the CEP (Figure 18) and CEAC (Figure 19). The probability that the intervention would be considered more cost-effective than the control at different decision thresholds was 24% at £20,000 and 34% at £30,000 per QALY gained. The results of the complete-case analysis have been shown to be consistent with the results of the multiple imputation analysis at the end of the 6-month follow-up.

FIGURE 18.

Cost-effectiveness plane (complete-case analysis).

FIGURE 19.

Cost-effectiveness acceptability curve (complete-case analysis).

Sensitivity analysis (intervention components)

The intervention designed in this trial consists of more than one component. We broke down the costs and effects of each component to explore the most cost-effective part of the intervention in terms of increasing the SSS attendance. The 4383 participants were divided into three groups according to the treatment they received: 1748 received the standard generic letter, 1896 received a personal tailored letter and 739 participants attended taster sessions after they received the tailored letter.

Table 36 reveals the results of the CEA comparing the three groups for the three outcomes. When there are more than two strategies involved in an incremental CEA, ICERs were calculated following three steps. First, the alternatives were ranked from the cheapest to the most expensive choice. Second, strategies that were more costly, but also less effective than any previous strategy, were excluded from the ICER calculation (known as dominance). Third, and finally, the ICERs were recalculated for the remaining strategies, from the least to most expensive.

The first part of the results shows that smokers who received only the personal tailored letter but did not attend the taster session were least likely (6.3%) to take up the SSS, with an average cost of £46 per participant (see Table 36). Therefore, the tailored letter was dominated by the standard generic letter, which only cost £0.87 per person, with 9.0% of these attending the SSS. The results indicate that about half (45.7%) of smokers who received the taster session followed by a tailored letter attended the SSS. The ICER of comparing the tailored letter plus the taster session to generic letter was estimated as £196 per additional attendee to the SSS.

Similarly, using validated 7-day abstinence as an outcome, the tailored letter was dominated by the tailored letter plus taster session, as it is more costly and less effective (see Table 36). Comparing the tailored letter plus taster session with the generic letter, the ICER was as low as £6 per additional quitter. When the QALY was used as the outcome measure, the ICER was estimated at £300 per QALY gained comparing tailored letter plus taster session with the generic letter. When comparing the tailored letter alone with the tailored letter plus taster session, the ICER was £63,400 per QALY gained. Given the NICE decision-making thresholds range of £20,000–30,000 per QALY gained, interventions below the lower threshold will be acceptable as a cost-effective intervention, and interventions that are above the upper threshold will be rejected. Therefore, the tailored letter plus taster session is considered as the optimal strategy among the three alternatives in the sensitivity analysis.

Long-term costs and outcomes predictions

The CEA extrapolated to the lifetime time horizon used QALYs as the effectiveness/outcome measure. The lifetime accumulative QALY gains by gender and age group are summarised in Table 37. QALYs were reported by participants’ smoking status, which includes ex-occasional smokers (who have only smoked once or twice), ex-regular smokers (who used to smoke sometimes but have now quit), light smokers (who smoke fewer than 10 cigarettes a day), moderate smokers (who smoke between 10–19 cigarettes a day) and heavy smokers (who smoke 20 or more cigarettes a day). The long-term analysis used a lifetime horizon, effectively covering all participants’ remaining lifetime between the age they entered the trial until the time they reach the national average life expectancy at birth. It is reported that the average life expectancy at birth was 81 years (79 years for males and 83 years for females) in 2013 in the UK. 112 Table 38 reports the accumulative lifetime QALY gains after discounting. In this study, QALYs and costs were both discounted at 3.5% per year. Overall lifetime health costs due to smoking-related diseases for both smokers and ex-smokers were derived from the published economic model and are reported in Table 39.

Participants in the trial were simultaneously assigned different lifetime costs and QALYs in accordance with their smoking status and characteristics. The lifetime costs and QALYs were adjusted by the 6-month costs and QALYs reported in the previous within-trial analysis, and the average lifetime costs and QALYs of each intervention before and after discounting are presented in Table 40. Participants who received the tailored intervention were expected to have a health-care cost savings of £210 (before discounting) and £74 (after discounting) over their lifetime compared with those who received usual care. Meanwhile, they also have higher lifetime QALY gains of 0.470 (before discounting) and 0.196 (after discounting) than people in the usual-care group. The results suggest that over the participants’ lifetime, tailored letters and taster sessions generate more QALY gains with less cost; in other words, the intervention is more cost-effective than the control condition.

The long-term CEA results are also summarised in Table 40, indicating a lifetime horizon ICER of –£447 (95% CI –£4368 to £3646) per QALY and discounted ICER of –£376 (95% CI –£3881 to £3207) per QALY. Figures 20 and 21 present CEACs for long-term cost-effectiveness. The first CEAC (see Figure 20), using results before discounting, illustrates that the probability of the tailored letter and taster session being cost-effective was between 83.4% and 83.5% at a willingness-to-pay threshold range of £20,000–30,000 per QALY gained. Figure 21 shows the discounted CEAC, which indicates that the probability of the intervention being cost-effective compared with generic letter was 86.1% at £20,000 per QALY and 86.0% at £30,000 per QALY.

FIGURE 20.

Lifetime CEAC (before discounting).

FIGURE 21.

Lifetime CEAC (discounted).

Reasons for additional funding

The Start2quit study, assessing the effectiveness on attendance at the NHS services of proactive recruitment by a brief personal tailored letter and invitation to a taster session, is currently recruiting successfully. To maximise the utility of the study, the following additional work will build on and enhance its value.

1. It is important to know whether or not this intervention also translates to increased quit rates, and if the quit rates in people attending as a result of this intervention differ from the usual quit rates in NHS services. As well as powering the trial for the original primary outcome (the proportion of people entering the SSS over a period of 6 months), we will now also power the trial for the secondary outcome of 7-day point prevalent abstinence at the 6-month follow-up, validated by salivary cotinine analysis.

   Assuming quit rates of 4% versus 2.2% in the intervention and control groups (mimicking the findings of Murray et al. 29), an 80% increase in the sample size is required, to 1793 in the control group and 2707 in the intervention group (assuming the same therapist effect as the original protocol), giving a total of 4500. A sample of this size would give 85.4% power to detect a difference of 1.8% at the 5% significance level. The same sample size would have 95% power to detect the difference between quit rates of 4.4% and 2.2% (doubling of quit rate).

   Based on present recruitment figures, we estimate that with an additional eight SSSs (48 GP practices) we will recruit another 2060 participants, meeting the requirement of the power calculation.

2. Identification of barriers to attendance and the reasons for non-attendance in this sample, following an explicit invitation (either with risk information and taster session invitation or without), will help to develop strategies to overcome the barriers, and to allocate resources to encourage attendance, and thus increase the potential to recruit the optimum number of smokers to the services.

   We are currently assessing barriers to attendance at the NHS services using an open question. For the remainder of the study we will use the TBQ, a 40-item measure of reasons for not entering smoking cessation programmes49 that has been recently validated on a low socioeconomic status population in the USA. This questionnaire will allow us to assess different aspects of smoker’s decisions to attend a group or therapy session and highlight any misconceptions or lack of awareness of the service offered. It will also allow us to explore any associations with demographic and dependence factors, as well as validating the questionnaire on a UK population. Participants who report not attending the services will be sent the questionnaire by post.

   The TBQ will be mailed to approximately 3500 participants who report not attending the SSS and who agree to complete an additional questionnaire.

3. Taster sessions are being recorded to ensure fidelity to the protocol. Assessing this fidelity can help to address factors that might have impacted on subsequent attendance and quit rates. Extra time for a full analysis of these tapes, using thematic analysis, will allow the exploration of differences in style and delivery of the intervention and their impact on subsequent attendance.

   Full analysis of the audio-recording of the taster sessions, using thematic analysis, will be carried out to address factors that might have impacted on subsequent attendance and quit rates.

Trial team

Trial manager: Leanne Gardner.

Assistant trial manager: Danielle Stoller.

Research assistants: Dimitra Kale and Molly Sweeney Magee.

Co-investigators

Irwin Nazareth, Richard Morris, Stephen Sutton, Irene Peterson and Simon Galton.

Research Department of Primary Care and Population Health

UCL

Royal Free and University College Medical School

London NW3 2PF

Background

The NHS SSSs are effective in helping people to quit smoking,129 and, although 74% of current smokers in 2007 reported that they wanted to quit, only a tiny proportion (< 5%)130–132 made use of the free service provided by the NHS. A wide range of factors, such as lack of availability and accessibility, perceived inappropriateness of the service, a perception that help is not necessary, or a sense of a lack of empathy from health professionals, as well as a lack of readiness to quit, will bar smokers from seeking help.

Although recruitment methods to cessation services generally employ a reactive approach, in which smokers are expected to seek out and approach the service,20 evidence suggests that if smokers are proactively and personally invited to use the services, the resultant use will be higher than standard referral by health professionals, or open advertising. 27 The development of computer-tailored self-help materials, intended to meet the needs of one specific person, based on characteristics unique to that person,30 offers a method for further personalising communications to patients and has the potential to engage with and recruit a larger proportion of the smoking population in a relatively inexpensive way. In addition, previous research29 has shown that inviting smokers to attend a no-commitment introductory session designed to offer information about what attendance at the service would involve, together with a strong referral message to the service, can increase attendance.

Thus, the Start2quit intervention is employing a two-pronged approach to encourage people to attend the SSSs, that consists of using computer-tailoring technology to send personal risk information to smokers, and an invitation to a taster session to allow them to find out more about the service before committing and signing up to a 6 week course.

Principal research question

We hypothesise that smokers, identified from general practice records, sent brief personal tailored letters based on characteristics available in their primary care medical records and on a short screening questionnaire, and invited to a ‘Come and Try it’ taster session designed to inform them about the SSSs, are more likely to attend the services than those who receive a standard generic letter advertising the service.

Primary objective

To assess the relative effectiveness on attendance at the SSSs, of proactive recruitment by a brief personal letter, tailored to individual characteristics available in medical records, and invitation to a taster session to provide information about the SSSs, over a standard generic letter advertising the service.

Procedure

All smokers aged ≥ 16 years are identified from their medical records in participating practices. After screening by GPs to exclude anyone deemed to be unsuitable, all remaining persons on the list are sent a screening questionnaire with a cover letter from their GP, a participant information leaflet describing the research, and a consent form to participate in the trial. Participants are asked to provide consent for the release of relevant data from their attendance at the SSSs to the researchers, used to validate attendance and quit rates. Participants have the opportunity to decline to participate but to return the questionnaire with Section A only completed to update their smoking status in their records.

All current smokers willing to participate and returning the signed consent form, aged ≥ 16 years, able to read English, motivated to quit and have not attended the SSSs in the previous 12 months are eligible to participate. Exclusion criteria are minimal because the aim is to recruit all smokers into the SSSs.

Interventions

The control group receive a standard generic letter sent from the surgery advertising the SSS and asking the smoker to contact the service to make an appointment to see an advisor.

The intervention group receive a brief motivational letter offering personalised risk information and help to improve their condition by quitting smoking, sent from the GP. This contains information specific to the patient, including an invitation and an appointment to attend a ‘Come and Try it’ taster session run by advisors from the local SSS to find out more about the services. Each SSS runs between 4 and 12 taster sessions, and attendees are encouraged to sign up to a group or one-to-one session at the end of the taster session. Participants who fail to attend will receive a further invitation 3 months later to encourage attendance.

Objectives

The aims of the pilot phase of the study were:

1. to assess the feasibility of the procedure in terms of:

   1. searching medical records and mailing screening questionnaires

   2. the randomisation and generation of the tailored letters

   3. delivery of the intervention.

2. to ascertain recruitment rates

3. to assess the uptake of the taster sessions and subsequent attendance at the SSSs.

The criteria for judging the success of the pilot phase and proceeding to full trial is based on:

1. achieving a 7% response rate (i.e. a mean of 42 participants per practice giving consent and agreeing to randomisation) in the first 12 practices

2. a preliminary analysis that suggests that the difference in uptake of SSSs between the intervention and control groups is greater than zero.

Successful achievement of recruitment, the rate of uptake in attendance at the taster session, and the uptake of SSSs in the intervention group will allow progression to the main trial in a further eight areas that are representative of the SSS.

Planned recruitment

We planned to recruit practices and participants through the PCRN, in two areas selected for the pilot phase. Practices generally identify 13% to 22% of their patients as smokers,55 depending on the characteristics of the patient population, and the accuracy and completeness of the records. Based on the response rate from previous and ongoing studies29,56 we estimated a response rate of 7% from smokers motivated to quit, from two mailings. By selecting six practices in each of the two SSSs with a list size of > 4000, we aimed to secure 504 participants to the pilot phase of the randomised controlled trial.

Progress

The pilot phase for Start2quit was carried out between September 2010 and December 2011.

Recruitment

Participants

Figure 22 shows a flow diagram of recruitment progress through the pilot phase of the trial.

FIGURE 22.

Flow diagram of the progress of the pilot phase of the Start2quit trial. a, This figure relates to the proxy measure (self-reported attendance at the SSSs) for the primary outcome of the pilot phase, which was obtained through follow-up telephone interview. We expect to collect primary outcome data of attendance at the SSSs for 100% of participants from records held at each SSS involved in the study. We have assurance from the Oxfordshire SSS that these data will be available to us in March 2012.

Although 11,202 smokers were initially identified, screening questionnaires with cover letters were mailed to 8548 smokers. This decrease was largely due to the necessity of selecting only one smoker per address. In addition, some patients identified as smokers were found to be ineligible after the mailing had taken place (n = 756; incorrectly coded, wrong address, deceased or ineligible due to answers in the questionnaire; see Table 41 for figures) and were therefore deleted from the denominator, leaving a total of 7792 eligible smokers who were mailed a questionnaire and invited to participate in the study. Reminder invitation letters were sent out to non-responders (n = 7342).

TABLE 41 - Recruitment of participants

Practice	Practice size	Smokers	Invitation	Invitations (ineligible + non-smoker + wrong address/deceased)	Ineligible	Non-smoker	Wrong address/deceased	Total
								Replies		Willing to participate		Eligible		Ineligible		Section A only
								Number	%	Number	%	Number	%	Number	%	Number	%
Practice 1 (Camden)	11,800	1850	1567	1506	21	40	0	140	9.3	62	4.1	41	2.7	21	1.4	78	5.2
Practice 2 (Camden)	4251	299	246	238	8	0	0	38	16.0	22	9.2	14	5.9	8	3.4	16	6.7
Practice 3 (Camden)	7340	1486	1182	1160	12	9	1	103	8.9	50	4.3	38	3.3	12	1.0	53	4.6
Practice 4 (Oxfordshire)	8087	1333	1028	986	20	18	4	183	18.6	113	11.5	93	9.4	20	2.0	70	7.1
Practice 5 (Oxfordshire)	10,265	2075	1579	1279	19	279	2	159	12.4	91	7.1	72	5.6	19	1.5	68	5.3
Practice 6 (Oxfordshire)	16,721	2852	2298	2187	54	53	4	403	18.4	189	8.2	135	6.2	54	2.5	214	9.8
Practice 7 (Oxfordshire)	7321	1297	648	436	15	191	6	101	23.2	48	11.0	33	7.6	15	3.4	53	12.2
Total			8548	7792	149	590	17	1127	14.5	575	7.3	426	5.5	149	1.9	552	7.1

The total number returning the questionnaire was 1127 (14.5%). Of these, 552 (7.1%) completed section A only to update their practice records and were not willing to participate. A total of 575 (7.3%) gave consent and agreed to randomisation, and 149 of these were not eligible to participate, leaving 426 (5.5%) enrolled in the study (Table 41). To account for therapist effects, the sample size in the intervention group was inflated by 1.51 and eligible participants were randomised to the intervention or control group at a ratio of 3 : 2 within each practice. Thus, 259 participants were randomised to the intervention and 167 to the control group. The overall number in each group conformed closely to the randomisation ratio. The response rate varied greatly between GP practices (ranging from 2.7% to 9.4%) and between the two SSSs (Camden, mean 3.2%; Oxfordshire, mean 6.8%; Figure 23).

FIGURE 23.

Response rates for practices in the pilot phase.

Attendance at taster sessions

In the pilot phase, taster sessions were run by Camden and Oxfordshire SSSs. As mentioned earlier, participants were invited to one of three initial taster sessions. If they failed to attend, they were then invited to another taster session 3 months later. A total of four taster sessions were held in Camden and a total of 10 sessions were held in Oxfordshire, split evenly between the two areas in which GP practices were recruited from in Oxfordshire (Oxford and Banbury).

Attendance rates at the three initial taster sessions were broadly consistent (Figure 24). However, it is important to note that there was relatively low attendance at the taster sessions held 3 months later. In total, 87 out of 259 participants randomised to the intervention group (33.6%) attended a taster session.

FIGURE 24.

Taster session attendance rates.

Follow-up response rate

The completion rate for the follow-up interviews was 70.7%. The response was equivalent in the intervention and control groups (71.8% vs. 68.9%). Self-reported attendance at the SSSs obtained from these interviews will be used as the proxy measure of the primary outcome in the pilot phase.

Attendance at the Stop Smoking Services

Self-report data confirm that the uptake of attendance at the SSSs was greater in the intervention group (70/259; 27%) than in the control group (21/167; 12.6%). Table 42 shows the differences in attendance at the SSSs between the intervention and control groups. These figures assume that those who were lost to follow-up did not attend the SSSs.

Telephone interview

(PDF download)

Postal questionnaire

(PDF download)

Basic questions

(PDF download)

Stop Smoking Service validation form

(PDF download)

Treatment Barriers Questionnaire

(PDF download)