Methods

Trial design

A four-centre cluster feasibility study with embedded qualitative process evaluation was conducted. A deviation from our protocol was made to the study design owing to the availability of centres. Centres were counterbalanced in each condition but not randomised (see Cluster allocation). However, participants joined after cluster allocation and allocation was concealed to participants until after the baseline assessment. Recruitment occurred during a discrete period starting from site initiation at each location (6 weeks).

Electronic cigarette sites offered an EC starter kit at baseline assessment. The EC arm is formed of two EC clusters.

Usual care is defined as offering an adapted NHS Choices fact sheet for smoking cessation and signposting to the local SSS. The UC arm is formed of two UC clusters. The clusters are illustrated in Figure 1.

FIGURE 1 The cluster sites. EC arm: Northampton (centre 1) and London (centre 4). UC arm: Edinburgh (centre 2) and London (centre 3). (Figure not reproduced here; see the original article.)

Consent was taken at baseline by the research team. Follow-up data at 4, 12 and 24 weeks were also recorded by the research team. Keyworkers delivered the interventions (see Interventions).

Participants

Participants were selected by homeless centre staff based on the following inclusion criteria. Further eligibility checking was performed at baseline by the researcher. Eligibility criteria did not differ between clusters.

Inclusion criteria

• Adult smokers (aged ≥ 18 years) accessing homeless support services on a regular basis and also known to staff.
• Self-reported daily smokers only with smoking status also confirmed by support staff.
• Smoking status also biochemically verified by exhaled carbon monoxide (CO) breath test at recruitment.

To gain wider representation, we did not exclude those who reported mental illness or substance dependence.

Exclusion criteria

• Non-smokers or those reporting using another smoking cessation aid at the time of recruitment.
• Anyone aged < 18 years, reporting pregnancy or unable to consent, for example they were currently intoxicated or unable to speak English.
• All of those who were not well known to the centre staff were ineligible.

Description of settings

Centre 1 (electronic cigarette cluster)

This is a day centre (open Monday to Friday; 08.30–13.00) located within Northampton. Staff and volunteers, support workers, keyworkers, social workers and kitchen staff work to provide a range of facilities and programmes to serve the practical, physical and social needs of people experiencing or at risk of homelessness in the area. There is a breakfast and lunch service, training and employment skills programmes, and an emergency night shelter (run separately by the local council) for use during periods of inclement weather. The kitchen and day centre are open on Saturday mornings for people who are sleeping rough or in emergency accommodation.

Centre 2 (usual-care cluster)

This is a day centre (open Monday to Friday; 09.00–13.00) located within the city of Edinburgh. It offers breakfast, lunch and shower and clothing facilities, as well as spiritual support through chaplaincy and recreational facilities (TV, pool and day trips). Support staff provide crisis intervention advice and assistance with benefits, housing, accessing medical services for physical and mental health, and social and life skills emotional support.

Centre 3 (usual-care cluster)

This centre operates three linked sites of supported accommodation in London for people at risk of homelessness. Each site accommodates clients whose needs fall within the parameters of the level of care it is able to provide.

Site 1 has 21 bedsits in three adjoining three-and-a-half-storey houses with a shared garden and is staffed 24 hours for people with greater complex needs.

Site 2 has 12 bedsits for male residents who are semi-independent and benefit from the support of regular check-ins, health support, and help with advocacy and paperwork. The office staff are available Monday to Friday between 09.00 and 17.00.

Site 3 has seven units in one four-storey building. The residents are both men and women who are living mostly independently, with the support workers providing advocacy, signposting to services, and on-site management.

Centre 4 (electronic cigarette cluster)

Also located in London, centre 4 provides supported accommodation across four linked sites. Staff and volunteers include keyworkers, complex needs workers and kitchen and maintenance staff.

Site 1 is a 23-unit residential support service for men and women. Single accommodation units are set on two storeys and overlook an enclosed communal garden. There is a multipurpose meeting space in a small unit in the garden where staff run skills classes and social groups. Staff are on-site 24 hours for management and security. Residents have a range of abilities and needs, with the aim of returning to independent living. Staff aim to support people with complex needs through semi-independent living with keyworker support, advocacy and daily check-ins.

Site 2 is a 13-unit women-only supported accommodation, offering support to women who have been or are at risk of becoming homeless. The accommodation provides single-occupancy accommodation. There is a shared garden with chickens and a shared communal lounge and kitchen, but no private meeting spaces. Keyworkers provide support sessions within usual working hours: 09.00 to 17.00.

Site 3 is a short-stay house and site 4 provides first-night accommodation only (to provide emergency accommodation intended for people who are newly homeless or acutely at risk of rough sleeping).

The researchers routinely kept reflexive diaries to record observations and reflections about local contexts, as well as notes on the psychological impact of working in new and sometimes unpredictable environments where people are reacting to multiple, chronic and/or acute stressors and complex needs. These research notes formed the basis of debriefing meetings and discussions in the research team, in part to be informative of ecological differences between sites as well as to systematise the physical and psychological safety of the researchers.

Interventions

Interventions were delivered at cluster level. Appendix 1 presents the Consolidated Standards of Reporting Trials (CONSORT) flow diagram, which provides a summary of the stages of the study by intervention.

Pre-intervention staff training

The research team provided four education and training sessions to staff in each centre, 1–2 weeks before recruitment started. The educational content followed National Centre for Smoking Cessation and Training recommendations, to ensure that centre staff had a basic knowledge of the issues that surround smoking and smoking cessation and to optimise the delivery of the EC and UC interventions. The training was developed and led by Dr Deborah J Robson (King’s College London, London) with the support of the research team (LD, AT, SC, IU and AF); Table 2 presents an overview of the delivery of the staff education and training to 32 staff members across the centres. However, not all staff were able to attend the education and training sessions. In addition, new staff were employed in some centres during the intervention delivery phase of the study.

Education and training content for both intervention arms included the following:

• prevalence and patterns of smoking among the wider population and people experiencing homelessness
• health effects of smoking
• benefit of smoking cessation
• common misperceptions around smoking cessation in the context of other addictions and mental illness
• why this study was important and why it was needed now
• how to complete the baseline and follow-up forms and what was expected from staff.

Additional electronic cigarette training

Staff were expected to deliver the intervention and were therefore provided with information on the evidence base of EC use, clinical effectiveness and safety among the wider population and a rationale behind why ECs may be useful for this group. In addition, in the training session, staff were provided with information about how to deliver correct advice about ECs to participants and given a practical hands-on demonstration relating to aspects of EC assembly, how to use the device, charge it, refill the tank and replace coils, and battery safety. Including the practical hands-on demonstration meant that the EC intervention training took up to 45 minutes longer than the UC training.

The research team in the EC arm provided additional coaching in preparation for each staff member’s first two keyworker sessions at baseline, and through shadowing and feedback at ongoing sessions. Ongoing informal training was provided, in the form of question-and-answer discussions with staff between keyworker sessions and before and after participants’ arrival.

Additional usual-care training

As in the EC arm, staff were asked to deliver the UC referral information and were therefore provided with detailed information on the role of SSSs, current licensed medications available to smokers in their local SSS and evidence for their clinical effectiveness. Staff were informed that, at the time of training, licensed stop smoking medicines alongside behavioural support offered at the SSS were the most effective way of quitting smoking. Details of the centre’s local SSS and how to make a referral were also included in the training.

Electronic cigarette intervention

All participants in the EC cluster (centres 1 and 4) received the same intervention.

Participants were provided with an unboxed Aspire PockeX kit (Aspire eCig UK, Peterborough, UK) starter kit comprising a tank-style refillable EC with a spare atomiser, charger and wall adapter plug. They were offered five 10-ml bottles of e-liquid, choosing from a combination of (1) e-liquid nicotine strengths (two options: 12 or 18 mg/ml) and (2) flavours (three options: tobacco, fruit or menthol). E-liquid nicotine strengths of 12 and 18 mg/ml were chosen based on previous reports from our group and others of high nicotine dependency among smokers experiencing homelessness.14 The EC brand and e-liquid were decided on following our PPI work and advice from vapers. Vapers known to the research team who have engaged in other research suggested four of the top-selling, easy-to-use ECs. On purchasing these, smokers from one London homeless centre tried the four different ECs and provided feedback on ease of use, nicotine hit and likeability. The Aspire PockeX was consistently rated as the easiest and most satisfying to use. The flavours were recommended by a vape retailer because they are the most popular, that is best-selling, varieties.

At this time, the keyworker showed the participant how to use the device; this included how to (re)fill the liquid in the tank, how to charge the device, how to inhale and what to expect when first using the device. Participants were also provided with the device instructions, which in response to our PPI feedback had been retyped into a larger font, and a help sheet of ‘tricks and tips’ from experienced vapers. Participants were given time to try the device and experiment with the different flavours and nicotine strengths, and were permitted to switch between flavours. (Appendix 2 presents the selection of e-liquid flavours and concentrations that were given out by keyworkers.) The keyworker recorded the participants’ choices of e-liquid and the timing of the keyworker session in case report forms (CRFs) that were kept with the study equipment.

Once per week in the 3 weeks that followed the baseline meeting, participants met centre staff to report any concerns and collect up to five 10-ml bottles of e-liquid and a new atomiser. Ideally, participants met the staff member whom they had met at the baseline session and who had attended the staff training; however, because of staff turnover and time demands this was not always feasible. Staff recorded data on e-liquid and troubleshooting in the CRFs. With a larger number of participants attending follow-up appointments at centre 1, staff adopted a drop-in ‘surgery’ in which waiting participants also received peer advice and support.

Usual care intervention

All participants in the UC cluster (centres 2 and 3) received the same assessment measures, but UC differed across the two centres in this cluster.

At the end of the baseline assessment, and during the same session, participants were informed of their centre allocation. Participants were then referred to their keyworkers/other centre staff who advised them to consider quitting and provided them with the fact sheet and an adapted NHS Choices ‘help-to-quit’ leaflet; this included information about the location and opening hours of SSSs local to the centre. Paper copies of the help-to-quit leaflet (with SSSs’ contact details) were available as posters/flyers at homeless centres in the UC condition/cluster. Centre staff were asked to follow up with participants once per week for 4 weeks to record whether or not they had made contact with the SSS and to remind them to do so.

Process evaluation

All participants were asked after baseline CRF completion if they would be interested in taking part in the qualitative process evaluation. Individual interviews were conducted in a quiet, private space on homeless centre premises, between weeks 4 and 8 of the trial. A semistructured topic guide was used to ensure that all relevant topics were covered, and interviews were digitally recorded with participants’ consent.

The participants’ guide covered smoking history; awareness of local SSSs; experiences of trial processes; expectations and experiences of vaping (EC arm); and experiences of signposting to SSSs (UC arm).

The staff topic guide covered staff role and trial involvement; existing cessation support for clients; and views and experiences of the trial, processes, and the EC or UC interventions. Both guides covered unintended consequences and recommendations for improvements.

Incentives for retention

The wider literature on smoking and homelessness has shown that it is common practice to offer a financial incentive to participants and that this improves retention.4 Regardless of cluster, all participants were compensated with a £15 high-street gift card (Love2Shop, Birkenhead, UK) for completing assessments at baseline and follow-up and the qualitative interview (totalling £75). Staff did not receive a gift card for taking part in the interviews. Participant payment was not contingent on quitting or cutting down and this was stressed to the individuals taking part.

Outcome measures

Some parts of the following text have been reproduced with permission from Dawkins et al.17 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The text below includes minor additions and formatting changes to the original text.

Table 1 presents the outcomes and associated measures, including the relationship to each intervention arm.

Baseline measures (across clusters)

• Demographic information and homeless status/history.
• Cigarettes smoked per day, smoking history (e.g. length of smoking, previous number of quit attempts, support used) and past and current EC use.
• Severity of tobacco dependence, measured by the Fagerström Test of Cigarette Dependence (FTCD) (Fagerström18) and expired CO.
• Motivation to stop smoking (MTSS), measured by the Motivation to Stop Scale, a seven-level single-item instrument that incorporates intention, desire and belief in quitting smoking.19
• Mental health status, measured using the Patient Health Questionnaire-9 items (PHQ-9)20,21 for depression (total score ranging from 0 to 27, with a higher score indicating greater severity of depression) and Generalised Anxiety Disorder-7 (GAD-7) questionnaire (total score ranging from 0 to 21, with a higher score indicating greater severity of anxiety).22
• Alcohol use, measured using the Alcohol Use Disorders Identification Test (AUDIT), a 10-item screening instrument developed by the World Health Organization to screen for alcohol problems. Scores range from 0 to 40, with a score of > 8 indicating harmful or hazardous drinking and > 13 (females) or > 15 (males) indicating alcohol dependence.23
• Drug use measured using the Severity of Dependence Scale (SDS), a five-item screening measure of psychological aspects of dependence yielding a total possible score ranging from 0 (no/low dependence) to 15 (high dependence).24
• General health-care and service use measured using an adapted health-care and social service utilisation questionnaire relating to services accessed in the last 4 weeks.
• Health-related quality of life (HRQoL) measured using the EuroQol-5 Dimensions, three-level version (EQ-5D-3L), a widely used measure that provides a single value for health status that can be used in the clinical and economic evaluation of an intervention.25,26

All questionnaires and measures have good psychometric properties and have been used in previous research with vulnerable populations.

Follow-up data collection

Usual care

At weeks 4, 12 and 24, the following information was collected:

• smoking information: self-reported smoking abstinence; number of cigarettes smoked in the past 7 days (to measure 7-day point prevalence abstinence); number smoked per day (to calculate smoking reduction from baseline)
• exhaled CO breath test
• engagement with the local SSS (appointments made and attended)
• use of EC and other tobacco/nicotine-containing products
• general health-care and service use; HRQoL; mental health status (GAD-7, PHQ-9)
• other drug use/dependence (AUDIT, SDS)
• direct and indirect staff contact time.

Electronic cigarettes

At weeks 4, 12 and 24, in addition to the aforementioned information, the following was collected:

• Self-report information about 12 positive effects (e.g. throat hit, satisfaction, pleasant, craving reduction) and 21 negative effects (e.g. mouth/throat irritation, nausea, headache, heartburn) of EC use. This was reported using a visual analogue scale (VAS) and summed to create a percentage score (higher score = higher positive or negative effect) as used in our previous studies.27
• Information to further monitor risk and adverse effects (e.g. EC theft, exchanges, use for other substances), an unintended consequences checklist was developed specifically for this trial.

The research team collected all of the baseline and consent data and the 4-, 12- and 24-week follow-up data.

Feasibility study outcome measures

• To assess willingness to take part in the study, we recorded (1) the number of people who were asked, (2) the number eligible to take part and (3) the number who consented to take part.
• Retention and engagement were measured by recording (1) the proportion of participants who completed assessment measures in each arm at each time point, and (2) the proportion of participants still using ECs in the intervention arm and who had visited the SSS in the UC arm.
• To estimate the parameters for future trial at each follow-up point, we recorded (1) CO-validated sustained abstinence (from 2 weeks post quit date allowing up to five slips); (2) CO-validated 7-day point prevalence abstinence (i.e. no smoking at all in the past 7 days); and (3) the proportion achieving 50% smoking reduction [calculated by subtracting cigarettes per day (CPD) at follow-up from baseline].
• To explore the feasibility of collecting data on contacts with health-care services, we recorded participant utilisation of primary and secondary health-care services using a self-report questionnaire.
• Data on staff contact time, non-contact time and other resources used in EC and UC delivery, including staff costs, ECs and other costs incurred, were collected to provide an indicative cost of the intervention.

The research team collected all of the baseline and consent data and the 4-, 12- and 24-week follow-up data.

Sample size

As this was a feasibility study for a main trial, no formal power calculation based on detecting evidence for efficacy was conducted; the outcomes of the study allowed us to calculate the required sample size [and an intracluster correlation coefficient (ICC)] for a possible future definitive cluster randomised controlled trial (cRCT).

As per our published protocol,28 a pragmatically chosen sample size (N = 120; n = 30 per centre, n = 60 per cluster) was based on our pre-feasibility PPI work and also taking information from published works in similar samples. This information allowed us to identify evidence of feasibility, recruitment rates and any problems with the intervention or research methods. Our pre-feasibility scoping work suggested that the centres had daily contact with between 25 and 120 individuals, of whom 70–90% were likely to be smokers. Other studies on homeless populations have reported follow-up rates ranging between 24% and 88% (depending on the location of visits, provision of incentives and use of prompts).29 Therefore, we estimated that 50% of those who agree would drop out in the period between consenting to participate and the final follow-up at 24 weeks, leaving an estimated sample size at the final follow-up of 60.

In relation to the qualitative process evaluation, we planned to interview a subgroup of 24 homeless smokers (approximately six per centre) and 12 staff members (approximately three per centre). We also aimed to include continuing participants and those who did not complete their 4-week follow-up, and those in the EC and UC arms. This sample size is adequate for collection of qualitative data necessary to assess objectives 3 and 4.

Although there was no formal interim analysis or stopping guidelines, we did specify in our ‘project timetable and milestones’ that we would assess recruitment and 4-week retention at centre 1 after 2 months to determine whether or not to proceed with recruitment at centres 2, 3 and 4.

Cluster allocation

The centres in Northampton (centre 1) and Edinburgh (centre 2) are both day centres, whereas the two London centres (centres 3 and 4) are residential units. We planned to pair and match centres on accommodation provision; that is, each one of the residential centres would be paired with one of the day centres. We would then randomly allocate the two London centres, that is, one to the EC arm and the other to the control (UC) arm, with their accompanying pair (day centre) allocated to the other arm.

However, when we were ready to begin recruitment, the centres in London had not confirmed their availability for training and study start date. Thus, centre allocation could not be randomised and therefore deviated from protocol: we started with the centre that was ready to start training and recruitment, which was the centre in Northampton. We allocated this first centre to the EC condition so that we could explore recruitment, 4-week retention and any unintended consequences associated with the intervention to determine whether or not to proceed with recruitment at centres 2, 3 and 4. Centre 2 (Edinburgh) as the other day centre was therefore allocated to the UC condition. Centres 3 and 4 (London) were allocated to the UC and EC arms, respectively. Centre 3 was allocated to UC, as it was geographically closer to the researcher who was still collecting follow-up data from centre 1 and we expected lower uptake in the UC condition.

Thus, the actual allocation of centres to each arm was non-random; it was a pragmatic decision based on centre readiness and staff/researcher availability, though we balanced potential confounders and differences in environment by ensuring each cluster (EC and UC) contained one day centre and one residential unit.

Allocation concealment mechanism

As this represents feasibility work, a pragmatic approach to the concealment of the intervention was taken. Participants were told of the condition that they had been allocated to only after consent and baseline assessment; this was the same across the clusters. However, because of the nature of the conditions, namely that those in the EC clusters were provided with a starter pack and the social dynamics in the centres (particularly the day centres) were close-knit and interactive, the centre allocation was quickly revealed between participants and also to new participants.

Contamination

Contamination in the context of this design was defined as a participant receiving the intervention delivered by another participating centre, for example a participant in UC receiving a free EC starter pack from one of the EC clusters.

Implementation

This section is not applicable to this work.

Analytical method

This is a feasibility study for a trial in preparation for a future cRCT, and analyses of effect are not appropriate. Analyses were conducted to evaluate the feasibility study outcomes (see Table 1). Similarly, because this is a feasibility study for a trial, subgroup analyses are not appropriate.

Table 1 presents the objectives and associated outcome measure for each objective. Objectives 1, 2, 5, 6 and 7 were measured quantitatively, and objectives 3 and 4 were measured through the qualitative process evaluation. Each objective was measured across the whole sample, with the exception of objectives 5 and 7, which relate to efficacy and cost-effectiveness of the interventions. Chapter 4 (see Table 3) presents the descriptive data for participant characteristics for the trial. Appendix 3 also presents the data for the subset of participants in the process evaluation.

Baseline demographic data were summarised using frequencies and descriptive statistics, and the arms (EC vs. UC) were compared using a t-test/Mann–Whitney U-test or Fisher’s exact test for continuous and categorical variables, respectively. EC effects are summarised descriptively. Changes in mental health status and substance use were explored over time and between arms using repeated-measures analysis of variance (ANOVA).

• Objectives 1 and 2: participant willingness to take part and retention Frequency information regarding the number of smokers who (1) were invited to take part, (2) were eligible to take part, (3) consented/completed the baseline assessment and (4) attended and completed each follow-up.
• Objectives 3 and 4: to examine the perceived value of the intervention, facilitators of and barriers to engagement, and influence of local context, and to assess service providers’ capacity to support the study and the type of information and training required Interviews lasted between 16 and 65 minutes, were transcribed verbatim by professional transcribers, were checked for accuracy and were entered into NVivo 12 software (QSR International, Doncaster, UK) to facilitate coding and analysis. Data were analysed using thematic analysis.30 Separate coding frameworks for both participants and staff, with themes initially formed deductively, were developed using the main objectives and interview topics. The qualitative team (AF, IU and AT) conducted a first round of coding with AF, checking all coding for consistency. Through discussion, inductive themes were identified and agreed, leading to a second round of coding. Using an iterative approach, coded themes were then used as the categories for analysis. Data were carefully examined to identify the range and diversity of responses and themes, and subthemes were created and/or refined as appropriate. Findings were interpreted and discussed among the wider team. Quotations were selected to illustrate findings. The study arm for each participant [EC or UC, centre code (01–04)] and whether or not the quotation was from a trial participant or a staff member are indicated alongside the quotations.
• Objective 5: assess the potential efficacy of supplying free EC starter kits The proportion of participants reporting sustained smoking abstinence, 7-day point prevalence abstinence (both CO verified) and a 50% reduction in smoking in each arm at each follow-up time point were recorded. These variables are presented with the denominator being the number of people who attended follow-up. Sustained abstinence is also presented and analysed using ITT analysis; that is, all those allocated are included in the analysis as belonging to the group to which they were originally allocated and those with missing outcome data were treated as smokers. The ICC was calculated by the Fleiss–Cuzick method.31
• Objectives 6 and 7: to explore the feasibility of collecting data on contacts with health-care services among this sample as an input to an economic evaluation in a full cRCT and to estimate the cost of providing the intervention and UC The completeness of EQ-5D-3L and service use questionnaires was examined. We used the UK population tariff32 to convert the results of EQ-5D-3L to utility value. Quality-adjusted life-years (QALYs) were then calculated in each arm using the area under the curve plotted from baseline and follow-up points.33 A set of national weighted average unit costs were extracted from secondary or published sources (see Appendix 4 and Appendix 5). The quantities of services reported by participants were multiplied by the respective unit costs to present a preliminary cost profile. We also present the costs of the programme, including training and delivery, the details of which were recorded by the research team. Results from the health economics component will be used to refine the instruments for a full randomised controlled trial (RCT).

Recruitment

Recruitment and 4-week follow-ups ran sequentially across the three sites in England, with 12- and 24-week follow-ups overlapping. Recruitment and data collection in Edinburgh ran in parallel with the second centre in England.

The feasibility study was completed within the planned time frame. However, although we planned to recruit for 4 weeks at each centre, recruitment was extended to 6 weeks in the UC arm. This was a pragmatic decision taken with assistance from the Trial Steering Group, the research team and centre staff. Centre 1 saw high levels of interest in the study and extending the recruitment period to 6 weeks was not required. We did extend capacity in the 4-week window to recruit more participants who were interested in taking part, anticipating potentially lower uptake in smaller centres. Conversely, centre 2 proved to be particularly challenging for recruitment because of staff time and competing needs and resources. The research team had to adapt to the conditions of the centre, appointments had to be rescheduled, and staff were not always available to assist. Centre 3 was quieter and overlapped with data follow-up at centre 1. Residents also had daytime commitments that clashed with recruitment, so again the research team needed to be flexible and therefore recruitment was continued over 6 weeks until we reached saturation.

Sampling and recruitment for the qualitative process evaluation

All those enrolled in the trial were asked at baseline about their willingness to take part in the process evaluation interviews and whether or not they were willing to be contacted at a later date to arrange an interview. After the intervention phase, those who had consented to be contacted were approached by a researcher by telephone, or in person at the homeless centre, and invited to take part. At centre 1, staff identified participants whom they felt would be most willing to take part. Throughout the centres, staff who were supporting the participants and the project were also eligible to take part in an interview. Potential interviewees were provided with a participant information sheet and re-contacted several days later. An interview date and time were arranged for interested participants. Written consent was obtained at the start of the interview.

Purposive sampling was utilised with a pre-determined target to recruit approximately six trial participants and three staff members at each of the four homeless centres. Trial participants were also to be sampled according to (1) smoking status at week 4 of the trial and (2) whether or not they completed the week 4 follow-up.

Centre 1: electronic cigarettes

Recruitment/baseline: 7 January to 31 January 2019.

4-week follow-up: 4 February to 28 February 2019.

12-week follow-up: 1 April to 25 April 2019.

24-week follow-up: 24 June to 17 July 2019.

Centre 2: usual care

Recruitment/baseline: 8 March to 25 April 2019.

4-week follow-up: 8 April to 17 May 2019.

12-week follow-up: 4 June to 25 July 2019.

24-week follow-up: 23 August to 10 October 2019.

Centre 3: usual care

Recruitment/baseline: 18 March to 2 May 2019.

4-week follow-up: 18 April to 18 May 2019.

12-week follow-up: 18 June to 18 July 2019.

24-week follow-up: 10 September to 25 September 2019.

Centre 4: electronic cigarettes

Recruitment/baseline: 13 May to 5 June 2019.

4-week follow-up: 10 June to 4 July 2019.

12-week follow-up: 7 August to 28 August 2019.

24-week follow-up: 28 October to 20 November 2019.

Process evaluation

Interviews were conducted between February and July 2019 by one of a mixed-sex team of three full-time qualitative researchers experienced in interviewing vulnerable groups (AF, AT and IU).