Promoting smoking cessation and preventing relapse to tobacco use following a smokefree mental health in‐patient stay: A multi‐centre randomised controlled feasibility study

Feasibility of recruitment and retention

Primary feasibility outcomes included recruitment (proportion of eligible patients enrolled), retention (proportion of participants providing follow‐up data at two time points), number of Trusts enrolled and the feasibility of collecting carbon monoxide (CO) readings (proportion of randomised participants providing CO‐verified smoking status). A priori progression criteria (Table 1) to guide decision on progression to a fully powered randomised controlled trial (RCT) had been agreed with the trial steering committee based on our combined expertise in the conduct of applied mental health and tobacco dependence research [29, 30], reflecting best practice in our study context.

Feasibility of collecting smoking‐related measures

Smoking‐related measures included CO‐verified seven‐day point prevalence abstinence (<10 ppm), self‐reported relapse, continuous abstinence, quit attempts and nicotine dependence (Heaviness of Smoking Index [31]. Additional measures were strength of urges to smoke [32], motivation to quit (Motivation to Stop Smoking Questionnaire) [33], self‐efficacy and use of e‐cigarettes and Nicotine Replacement Therapy (NRT) post‐discharge.

Feasibility of collecting mental and physical health measures, health‐related quality of life and resource use

Measures included the Patient Health Questionnaire depression scale (PHQ‐9) [34], the Generalised Anxiety Disorder scale (GAD‐7) [35] and the quality‐of‐life scale (EuroQol‐ 5 Dimension [EQ‐5D]) [36]. Data on health service use since discharge were also collected for health economic analysis.

Acceptability

Semi‐structured interviews were conducted with enrolled study participants from both groups who were contactable, interventionists, and Trust research and development (R&D) staff to explore the acceptability of study procedures and intervention delivery. Interviews followed atopic guide informed by APEASE criteria [37], were audio‐recorded with consent, and transcribed verbatim.

Health economics

The cost analysis followed NICE guidance from the NHS and personal social services (PSS) perspective [38], with costs reported in 2023/24 GBP. Intervention costs (development, training, app licensing) were recorded via staff time logs and trial expenses, while delivery costs were tracked using MTS–participant contact logs. Usual care costs were derived from follow‐up questionnaires. Healthcare resource use was assessed at baseline and follow‐up using a tailored questionnaire covering smoking‐related and general healthcare contacts. Health‐related quality of life was measured with the EQ‐5D‐5L [39] and data completeness was assessed.

Intervention delivery and supervision

Interventionists (MTSs) completed the National Centre for Smoking Cessation Training (NCSCT) Practitioner Training module 2, a specialty course, and bespoke training on smoking behaviour change, e‐cigarettes, and potential tobacco‐antipsychotic medication interactions, following Royal College of Psychiatrists guidance [5, 41]. Training included self‐guided online modules and an in‐person session facilitated by the research team. Monthly supervision was provided by the team, with oversight from the Trust Principal Investigator or designated individual. Fidelity was monitored through supervision, self‐assessment, interaction logs, and sampled documentation to identify training and delivery gaps.

Sample size

The study aimed to recruit 64 patients to allow a participation rate of 15% and completion rate of 80% to be estimated within a 95% confidence interval of ±3% and ±10% respectively [42].

Quantitative data

Analyses were conducted in Stata v17. Baseline and follow‐up data were summarised descriptively by group and timepoint, without formal statistical comparisons [43]. Continuous data were reported as means, standard deviations, medians, and ranges; categorical data as counts and percentages. Recruitment and retention, including missing data were summarised against pilot progression criteria, with recruitment rates presented with 95% confidence intervals. The analysis plan was not pre‐registered, and the results should be considered exploratory.

Qualitative data

Thematic analysis [44] was conducted using NVivo v14 [45] for data management, combining deductive coding based on research questions with inductive identification of emergent themes, mapped broadly to APEASE criteria [37]. Three researchers independently coded transcripts, refined codes after initial analysis, and resolved discrepancies through discussions.

Health economics

The evaluation estimated intervention and usual care costs and assessed the feasibility of collecting economic data for a future trial. Using a micro‐costing approach, resource use (e.g., training, delivery, cessation aids) was costed with national unit rates (e.g. Personal Social Services Research Unit, university pay scales [46, 47]. EQ‐5D‐5L responses were converted to utility scores to calculate quality‐adjusted life years (QALYs), with descriptive statistics reported for resource use and EQ‐5D‐5L scores.

Baseline

Table 3 provides information on the psychiatric diagnoses of participants. Over half (n = 20, 52.6%) had schizophrenia or other primary psychotic disorders, 12 (31.6%) had personality disorders, and six (15.8%) had mood disorders. Most (n = 33, 86.8%) were smoking at baseline (Table 4). One intervention participant (5.9%) and four in usual care (19%) were CO‐verified abstinent post‐admission. Nicotine dependence appeared low (mean HSI = 1.4, SD 1.8), with 17 (44.7%) reporting smoking within 60 minutes of waking and 27 (68.4%) typically smoking 10 or fewer cigarettes a day. Strength of urges to smoke was moderate (mean = 6.6, SD 2.4) and motivation to quit was high (mean = 8.9, SD 2.2). E‐cigarette use (any use, i.e., exclusive and dual) was common (n = 30, 78.9%). Mean PHQ‐9 and GAD‐7 scores indicated moderate depression (12.1 ± 7.0) and mild–moderate anxiety (9.3 ± 6.0) (Table 4).

Three‐months follow‐up

At 3‐months follow‐up, data were collected from 20 participants (52.6%; 7 from the intervention and 13 from the usual care group). Three participants from the intervention group self‐reported seven‐day point prevalence smoking abstinence; CO validation confirmed abstinence in two cases, while the third participant did not return a CO reading and therefore was assumed not to be abstinent. In contrast, no participants in the usual care group reported point prevalence abstinence. Continuous abstinence (not smoking for over one month) was reported by three participants in the intervention group (two of these participants had their CO validation confirmed). All intervention participants reported e‐cigarette use. Mental health outcomes indicated moderate levels of depressive symptoms (PHQ‐9 M = 9.6, SD = 7.1) and mild anxiety (GAD‐7 M = 8.9, SD = 7.0), in the context of existing mental health diagnoses.

Four‐ to six‐months follow up

At 4–6‐month follow‐up, 13 participants (34.2%) provided outcome data (6 intervention and 7 usual care). Two participants from the intervention arm self‐reported and CO‐validated as abstinent, while one participant from the usual care group self‐reported abstinence, though this was not supported by the CO reading. Most participants (n = 10) continued using e‐cigarettes. At this timepoint, mental health symptoms were slightly elevated compared to three months, with moderate depression (PHQ‐9 M = 13.5, SD = 8.5) and moderate anxiety (GAD‐7 M = 10.3, SD = 6.4) (Table 5).

Acceptability

The SCEPTRE intervention was generally viewed as acceptable by the patients, staff and interventionists who participated in interviews. Participants valued personalised support, especially MTS contact and post‐discharge phone calls, which were described as ‘encouraging and motivating’ (P1). In contrast, the bespoke resource folder co‐produced in the early stages of the research programme (MyTry kit) and described in detail elsewhere [23] was reported to have been only rarely used, and most did not engage with the Smokefree app despite a suite of bespoke resources having been developed [40]. Advice given regarding access to and use of NRT and use of e‐cigarettes was seen as helpful for behaviour change: ‘I could start vaping then even though I'd never vaped before. That just really helped and made me just not want to go back to smoking.’ (P1). Compared with usual care, SCEPTRE was seen to offer more consistent and ‘tailored follow‐up support’. (S1). Stakeholders and MTSs reported positive experiences, though some said they would have preferred interactive, live training over pre‐recorded sessions: ‘With the delivery I think it was a lot of sitting in front of the computer reading a lot, even though the videos were helpful, but I think a more in person session would have been really helpful, just to ask questions on the spot and engage a bit more’. (M2).

Practicability

The study was not perceived to disrupt ward operations but was viewed to pose workload challenges for some research staff and MTSs. Recruitment was experienced as difficult due to high ward acuity, fluctuating patient interest, and frequent incidents requiring staff withdrawal. Tracking discharge dates was reported to be problematic, leading to missed recruitment opportunities, which felt like ‘a bit of a waste of our time’ (S1). Mitigation strategies used by staff included maintaining ward presence, building relationships with staff, and engaging patients in shared spaces such as the gym.

Effectiveness (as per APEASE acronym)

Recruitment experiences differed due to patients' varying motivation and capacity to consent. Stakeholders noted that acute mental health symptoms often limited patients' ability to prioritise research participation: ‘when we've got things like anxiety, depression, psychosis, paranoia, it's a lot more difficult for them to attribute importance to research because they've got so much else going on’ (S1).

Most participants reported positive intervention experiences. One reported to have reduced smoking from 80 to 20 cigarettes daily, crediting NRT, vaping, and MTS support, and felt strongly about the potential positive impact for others: ‘Please give this opportunity to other people because it's really positive … I'm really impressed with it, really pleased with it’ (P2). Another participant quit entirely and felt confident about maintaining abstinence. MTSs described cases where initially hesitant patients became motivated to address their smoking during pre‐discharge sessions.

Affordability

This aspect seemed of no or minor relevance for our context and was not prioritised for discussion in the interviews.

Safety

No urgent or harmful side effects or unintended consequences of the SCEPTRE intervention were discussed by participants. A small number of comments touched on wider patient safeguarding challenges within routine procedures, such as difficulties contacting patients and uncertainty about escalation pathways when people could not be reached: ‘Thankfully all of mine eventually picked up and I didn't escalate it, but there were a good few weeks where I was like I can't get hold of them today’ (M1). However, these issues reflected general NHS risk management responsibilities rather than aspects of the SCEPTRE intervention itself, which did not include welfare checks or formal mental health safeguarding protocols.

Equity

There was some evidence of digital exclusion, for example due to limited access to the Smokefree app due to faulty devices, dyslexia, or lack of digital literacy: ‘Yeah, my phone was, it's a bit broken so I haven't been able to use it for apps or anything like that’ (P3). Many patients appeared to prefer support based on phone calls and text messages to app‐based support.

Mental health symptoms, such as paranoia, and shifting priorities post‐discharge also reduced engagement: ‘Patients who are developing paranoid symptoms may not want to answer their phone following discharge.’ (S1).

‘I think did we have somebody who was discharged and immediately didn't want to do the intervention. And I think he was just saying, you know, I'm moving on now with my life and I'm back at work now and … you know’. (S2).

Financial cost rather than health often motivated quitting: ‘I think the biggest thing that came up for people wanting to come off it was the cost, rather than necessarily the health issue; it was like, I can't really afford this addiction.’ (S3). One patient described the stigma around smoking: “I feel like a leper, everybody criticises the smoking … you know, nasty comments' (P4) while MTS support was described as empathetic and encouraging: ‘She [MTS] was great, she was just great and with all the wobbly bits as well, that was, like, oh, I don't know, really, really wanted one, just the advice and having someone there to speak to was brilliant.’(P1).

Recruitment and retention

We recruited 38 participants from 25 wards of six NHS mental health Trusts, demonstrating borderline feasibility of recruitment through an amber progression criterion rating. Notably, the actual recruitment rate of 19.4% was not markedly lower than predicted at study design stage (20%): of those eligible, roughly the expected proportion of participants did consent to take part. Consistent with other mental health trials [48], typical challenges arose in the context of reliably predicting and tracking time points for mental health patient discharge (which appeared to be particularly erratic in the context of extreme NHS pressures and demand on beds) and integrating study tasks into pressured clinical settings [49, 50]. However, we encountered two unexpected obstacles: 1) a substantially lower than previously evidenced smoking prevalence in patients admitted to acute adult mental health wards (we had estimated ~60% smoking prevalence on acute adult mental health wards and established that the rate was, in fact, ~32% on participating wards during our study), and 2) failure to recruit more than six mental health Trusts to our study due to recent changes in UK tobacco dependence‐related policy and practice.

In terms of over estimating prevalence rates relevant for recruitment planning, this has been shown to be a major factor in recruitment failure in clinical research, according to a systematic review of 172 RCTs from different clinical disciplines [51]. For this reason, we had based our design and recruitment plans both on research evidence relating of smoking prevalence in mental health in‐patient settings [52, 53] and on live data of smoking status recordings from four large mental health Trusts at study design stage, pre‐Covid. However, addressing smoking among people with mental health conditions has been identified as a priority to address health inequalities in the UK in recent years as set out in the UK [54], and multiple initiatives to improve support in this population were implemented between study design and implementation stage. Further research should establish if there are, in fact, widespread changes in prevalence rates that have historically been stagnating [10, 55].

Our difficulties to enrol a larger sample of mental health Trusts when recruitment was below target early in the study is related to the above. We found many Trusts ineligible or unable to take part. Based on experience with conducting the community‐based SCIMITAR trial [29], which had involved 37 community‐based NHS settings across England, we had not foreseen this. The dynamism of the tobacco dependence policy landscape in the UK meant that, as we developed the intervention and prepared the feasibility study, initiatives were taken regionally and locally to fund and implement intensive support for in‐patients in many mental health Trusts. The practical implications of this presented a major challenge, because many Trusts participating in the new schemes were also evaluating their impact and were unable to implement SCEPTRE at the same time. As such, our experience represented a common dilemma for larger study programmes with development phases that, as they evolve, may fall behind more rapid policy and practice developments, with research budgets and time frames not always flexible enough to adapt optimally.

While recruitment proved highly challenging but still received a potentially viable ‘amber’ rating, retention of participants in SCEPTRE was a clear failure: only 34.2% of all participants completed the second follow‐up, and fewer in the intervention arm. Our efforts, which included co‐developed study procedures and materials (including a bespoke co‐produced bias‐free language resource to support recruitment to SCEPTRE), weekly trouble shooting sessions with all settings, and multiple amendments to the protocol to accommodate local Trust idiosyncrasies for smoother recruitment, did not result in satisfactory retention of enrolled participants over time. Attrition was already very high (~50%) at the first follow‐up point. This experience is in stark contrast to other trials in the field, both from the UK [30] and other countries [56, 57], but in line with experiences from a recent UK‐based e‐cigarette‐focused cessation and harm‐reduction trial in community mental health patients that had to be discontinued due to low recruitment (Kale et al., manuscript under review). It is difficult to draw on insights from our qualitative interviews due to the small sample and selection bias of those who remained contactable throughout the study. However, a comment by a member of staff at interview about the difficulties of retaining patients who are still ‐ or again ‐ very unwell after discharge may point to one core explanatory factor. In the UK, in‐patient treatment now occurs as a last resort when specialist support in the community is no longer viable, and only the most severely and acutely unwell mental health patients are admitted to it [58]. Bed pressure is high, and discharge occurs as soon as clinically justifiable, with high readmission rates [59]. Many patients deteriorate after discharge. This poses extreme challenges for the development of successful retention strategies for mental health patients around discharge and, in fact, raises question about the timing of interventions for research and clinical practice. In our case, where the purpose of the intervention was inextricably linked to the point of discharge (and the first three days thereafter), it is difficult to draw conclusions.

As high attrition is common in smoking cessation trials [60], and may be worse among individuals with mental health disorders, updating and building on quantitative and qualitative evidence synthesis on effective retention strategies [61, 62] and co‐producing tailoring strategies to pressured NHS contexts could strengthen future studies.

Secondary outcome measures

Data collection in those who responded to contact attempts proved generally feasible, with most participants providing demographic, mental health, and economic information. High dropout rates limited the collection of smoking‐related outcomes over time. Qualitative and quantitative data indicated minimal contamination, as no usual care participants reported quitting or reducing smoking.

Abstinence rates in the intervention group at 3 and 4–6 months suggest some short‐term quitting among a small subset of participants, supported by CO validation. However, very few sustained abstinence, and high attrition underscored the difficulty of cessation among individuals with severe mental illness. The frequent use of e‐cigarettes reflects current harm reduction trends, highlighting the need to explore the role of vaping for cessation and harm reduction in this study population further.

Mental health outcomes showed ongoing moderate depressive and mild to moderate anxiety symptoms, likely affecting engagement and quit success in addition to symptoms experienced in the context of participants' primary mental health diagnoses. Minimal change over time suggests either limited intervention impact on psychiatric symptoms or delayed mental health improvements compared to behavioural change.

Economic data revealed substantial development and delivery costs, with high MTS turnover increasing adjusted training costs. The high turnover reflected intense NHS staff pressures with frequent changes to staffing and internal reorganisation at the time of our study. While the importance of careful staff selection, strong supervision, and workload support cannot be underestimated, systemic pressures outside of the influence of our research team did play a central part in our experience and continue to date.

Despite substantial investments in resources such as the Smoke Free app, which incurred costs for licenses, training and support, including a co‐produced tailored on‐boarding session [23], its usage was limited, suggesting the need to reconsider its role in a fully‐powered trial as well as in practice. Participant follow‐up rates were low, which may have impacted the accuracy of resource use reporting. However, the improvements in quality of life observed in both groups support the feasibility of incorporating health economics data in a future full‐scale trial, with further refinement needed to optimise cost‐effectiveness evaluation in a full‐scale trial.