Mindfulness for smoking cessation

Abstract Background Mindfulness‐based smoking cessation interventions may aid smoking cessation by teaching individuals to pay attention to, and work mindfully with, negative affective states, cravings, and other symptoms of nicotine withdrawal. Types of mindfulness‐based interventions include mindfulness training, which involves training in meditation; acceptance and commitment therapy (ACT); distress tolerance training; and yoga. Objectives To assess the efficacy of mindfulness‐based interventions for smoking cessation among people who smoke, and whether these interventions have an effect on mental health outcomes. Search methods We searched the Cochrane Tobacco Addiction Group's specialised register, CENTRAL, MEDLINE, Embase, PsycINFO, and trial registries to 15 April 2021. We also employed an automated search strategy, developed as part of the Human Behaviour Change Project, using Microsoft Academic. Selection criteria We included randomised controlled trials (RCTs) and cluster‐RCTs that compared a mindfulness‐based intervention for smoking cessation with another smoking cessation programme or no treatment, and assessed smoking cessation at six months or longer. We excluded studies that solely recruited pregnant women. Data collection and analysis We followed standard Cochrane methods. We measured smoking cessation at the longest time point, using the most rigorous definition available, on an intention‐to‐treat basis. We calculated risk ratios (RRs) and 95% confidence intervals (CIs) for smoking cessation for each study, where possible. We grouped eligible studies according to the type of intervention and type of comparator. We carried out meta‐analyses where appropriate, using Mantel‐Haenszel random‐effects models. We summarised mental health outcomes narratively. Main results We included 21 studies, with 8186 participants. Most recruited adults from the community, and the majority (15 studies) were conducted in the USA. We judged four of the studies to be at low risk of bias, nine at unclear risk, and eight at high risk. Mindfulness‐based interventions varied considerably in design and content, as did comparators, therefore, we pooled small groups of relatively comparable studies. We did not detect a clear benefit or harm of mindfulness training interventions on quit rates compared with intensity‐matched smoking cessation treatment (RR 0.99, 95% CI 0.67 to 1.46; I2 = 0%; 3 studies, 542 participants; low‐certainty evidence), less intensive smoking cessation treatment (RR 1.19, 95% CI 0.65 to 2.19; I2 = 60%; 5 studies, 813 participants; very low‐certainty evidence), or no treatment (RR 0.81, 95% CI 0.43 to 1.53; 1 study, 325 participants; low‐certainty evidence). In each comparison, the 95% CI encompassed benefit (i.e. higher quit rates), harm (i.e. lower quit rates) and no difference. In one study of mindfulness‐based relapse prevention, we did not detect a clear benefit or harm of the intervention over no treatment (RR 1.43, 95% CI 0.56 to 3.67; 86 participants; very low‐certainty evidence). We did not detect a clear benefit or harm of ACT on quit rates compared with less intensive behavioural treatments, including nicotine replacement therapy alone (RR 1.27, 95% CI 0.53 to 3.02; 1 study, 102 participants; low‐certainty evidence), brief advice (RR 1.27, 95% CI 0.59 to 2.75; 1 study, 144 participants; very low‐certainty evidence), or less intensive ACT (RR 1.00, 95% CI 0.50 to 2.01; 1 study, 100 participants; low‐certainty evidence). There was a high level of heterogeneity (I2 = 82%) across studies comparing ACT with intensity‐matched smoking cessation treatments, meaning it was not appropriate to report a pooled result. We did not detect a clear benefit or harm of distress tolerance training on quit rates compared with intensity‐matched smoking cessation treatment (RR 0.87, 95% CI 0.26 to 2.98; 1 study, 69 participants; low‐certainty evidence) or less intensive smoking cessation treatment (RR 1.63, 95% CI 0.33 to 8.08; 1 study, 49 participants; low‐certainty evidence). We did not detect a clear benefit or harm of yoga on quit rates compared with intensity‐matched smoking cessation treatment (RR 1.44, 95% CI 0.40 to 5.16; 1 study, 55 participants; very low‐certainty evidence). Excluding studies at high risk of bias did not substantially alter the results, nor did using complete case data as opposed to using data from all participants randomised. Nine studies reported on changes in mental health and well‐being, including depression, anxiety, perceived stress, and negative and positive affect. Variation in measures and methodological differences between studies meant we could not meta‐analyse these data. One study found a greater reduction in perceived stress in participants who received a face‐to‐face mindfulness training programme versus an intensity‐matched programme. However, the remaining eight studies found no clinically meaningful differences in mental health and well‐being between participants who received mindfulness‐based treatments and participants who received another treatment or no treatment (very low‐certainty evidence). Authors' conclusions We did not detect a clear benefit of mindfulness‐based smoking cessation interventions for increasing smoking quit rates or changing mental health and well‐being. This was the case when compared with intensity‐matched smoking cessation treatment, less intensive smoking cessation treatment, or no treatment. However, the evidence was of low and very low certainty due to risk of bias, inconsistency, and imprecision, meaning future evidence may very likely change our interpretation of the results. Further RCTs of mindfulness‐based interventions for smoking cessation compared with active comparators are needed. There is also a need for more consistent reporting of mental health and well‐being outcomes in studies of mindfulness‐based interventions for smoking cessation.


A B S T R A C T Background
Mindfulness-based smoking cessation interventions may aid smoking cessation by teaching individuals to pay attention to, and work mindfully with, negative a ective states, cravings, and other symptoms of nicotine withdrawal. Types of mindfulness-based interventions include mindfulness training, which involves training in meditation; acceptance and commitment therapy (ACT); distress tolerance training; and yoga.

Objectives
To assess the e icacy of mindfulness-based interventions for smoking cessation among people who smoke, and whether these interventions have an e ect on mental health outcomes.
We did not detect a clear benefit or harm of ACT on quit rates compared with less intensive behavioural treatments, including nicotine replacement therapy alone (RR 1.27, 95% CI 0.53 to 3.02; 1 study, 102 participants; low-certainty evidence), brief advice (RR 1.27, 95% CI 0.59 to 2.75; 1 study, 144 participants; very low-certainty evidence), or less intensive ACT (RR 1.00, 95% CI 0.50 to 2.01; 1 study, 100 participants; low-certainty evidence). There was a high level of heterogeneity (I 2 = 82%) across studies comparing ACT with intensity-matched smoking cessation treatments, meaning it was not appropriate to report a pooled result.
We did not detect a clear benefit or harm of distress tolerance training on quit rates compared with intensity-matched smoking cessation treatment (RR 0.87, 95% CI 0.26 to 2.98; 1 study, 69 participants; low-certainty evidence) or less intensive smoking cessation treatment (RR 1.63, 95% CI 0.33 to 8.08; 1 study, 49 participants; low-certainty evidence).
We did not detect a clear benefit or harm of yoga on quit rates compared with intensity-matched smoking cessation treatment (RR 1.44, 95% CI 0.40 to 5.16; 1 study, 55 participants; very low-certainty evidence).
Excluding studies at high risk of bias did not substantially alter the results, nor did using complete case data as opposed to using data from all participants randomised.
Nine studies reported on changes in mental health and well-being, including depression, anxiety, perceived stress, and negative and positive a ect. Variation in measures and methodological di erences between studies meant we could not meta-analyse these data. One study found a greater reduction in perceived stress in participants who received a face-to-face mindfulness training programme versus an intensity-matched programme. However, the remaining eight studies found no clinically meaningful di erences in mental health and wellbeing between participants who received mindfulness-based treatments and participants who received another treatment or no treatment (very low-certainty evidence).

Authors' conclusions
We did not detect a clear benefit of mindfulness-based smoking cessation interventions for increasing smoking quit rates or changing mental health and well-being. This was the case when compared with intensity-matched smoking cessation treatment, less intensive smoking cessation treatment, or no treatment. However, the evidence was of low and very low certainty due to risk of bias, inconsistency, and imprecision, meaning future evidence may very likely change our interpretation of the results. Further RCTs of mindfulness-based interventions for smoking cessation compared with active comparators are needed. There is also a need for more consistent reporting of mental health and well-being outcomes in studies of mindfulness-based interventions for smoking cessation.

Certainty of the evidence (GRADE) Comments
Study population Mindfulness training vs matchedintensity smoking cessation treatment: smoking cessation (≥ 6-month follow-up) 16  Mental health and well-being Studies investigated a range of outcomes: anxiety, depression, negative affect, positive affect, stress. Although 1 study found a statistically significantly greater reduction in perceived stress in people who received mindfulness training compared with those who received a matched-intensity smoking cessation treatment at 6-month follow-up, the other 2 studies found no clinically meaningful between-group differences in change in mental health and well-being measures.

(3 RCTs)
⊕⊝⊝⊝ Very low i,j We were unable to meta-analyse these outcomes and therefore summarised them narratively.
ACT vs matched-intensity smoking cessation treatment: smoking cessation (≥ 6-month follow-up) It was not appropriate to pool data across these studies because there was a high level of heterogeneity (I 2 = 82%) and the result may be misleading.

Mental health and well-being
One study that compared ACT with NRT found no clinically meaningful difference in negative affect across conditions at all follow-ups to 12 months.
Another study that compared ACT with a matched-intensity smoking cessation treatment and a less intensive ACT intervention found no clinically meaningful difference in positive mental health across conditions up to 6-month follow-up.

(2 RCTs)
⊕⊝⊝⊝ Very low f,g We were unable to meta-analyse this outcome and therefore summarised narratively.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
ACT: Acceptance and commitment therapy; CI: confidence interval; NRT: nicotine replacement therapy; RCT: randomised controlled trial; RR: risk ratio GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Certainty of the evidence (GRADE) Comments
Study population Distress tolerance training vs matchedintensity smoking cessation treatment: smoking cessation (≥ 6-month follow-up) 14  Mental health and well-being One study that compared distress tolerance training with less intensive smoking cessation treatment found no clinically meaningful difference in negative affect at 4 weeks post-quit.

49
(1 RCT) We were unable to meta-analyse this outcome and therefore summarised narratively. Mental health and well-being

Summary of findings 4. Yoga compared with control for smoking cessation
One study compared yoga with matched-intensity smoking cessation treatment and found no clinically meaningful difference in depression, anxiety, or general well-being scores between conditions at 8-week follow-up after controlling for baseline scores.
Another study compared yoga with less intensive smoking cessation treatment and found no clinically meaningful differences in the change in depression or anxiety scores by group up to 6-month follow-up.

(2 RCTs)
⊕⊝⊝⊝ Very low c,d We were unable to meta-analyse this outcome and therefore summarised narratively.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

Library
Trusted evidence. Informed decisions. Better health.
Cochrane Database of Systematic Reviews

Description of the condition
Smoking remains a leading cause of preventable death and disease worldwide (WHO 2019). Stopping smoking can result in substantial health gains, even later in life. The sooner a smoker quits, the more they reduce their risk of developing smoking-related diseases (Doll 2004). The majority of smokers want to quit and many try to quit each year, but quit rates remain low (WHO 2019).

Description of the intervention
In recent decades, mindfulness has increasingly been recognised as an influence on mood and behaviour (Baer 2003; Keng 2011). It has been adopted as an approach for increasing awareness and responding skilfully to mental processes that contribute to emotional distress and maladaptive behaviour (Baer 2003). In current research contexts, mindfulness is typically defined as the psychological process of bringing non-judgmental attention to experiences occurring in the present moment (Kabat-Zinn 2013).
There are various definitions of mindfulness used in psychological literature. While no consensus has been reached on how to define mindfulness, a two-component model proposed by Bishop 2004 is o en used in research. This operationalises mindfulness as: (i) maintaining attention on the immediate experience, and (ii) maintaining an attitude of openness, curiosity, and acceptance toward this experience, regardless of its valence or desirability.
Mindfulness approaches are not relaxation or mood management techniques, but rather a form of cognitive training to reduce susceptibility to reactive states of mind that might otherwise induce stress or perpetuate psychopathology (Baer 2003). The practice of mindfulness involves focusing attention on the immediate experience of cognitions, emotions, perceptions, and physical sensations and observing them as they arise and pass away. Mindfulness is nondeliberative: it simply involves paying sustained attention to thoughts and feelings without thinking about or evaluating them. A key tenet of mindfulness is that, by noticing thoughts and feelings in a curious and accepting manner, people develop greater tolerance of these phenomena and are able to recognise that they are transient, so they are less likely to respond impulsively to them (Heppner 2015).
There are a range of di erent treatments based on the principles of mindfulness. Mindfulness-based stress reduction (MBSR; Kabat-Zinn 2013) and mindfulness-based cognitive therapy (MBCT; Segal 2002) use meditation as the primary method of teaching mindfulness. MBSR was developed to treat chronic stress and pain-related disorders. It uses three techniques: firstly, sitting meditation, which involves mindful attention on the breath and a state of noncritical awareness of cognitions, feelings, and sensations; secondly, Hatha yoga practice, which involves breathing exercises, simple stretches, and postures; and thirdly, body scan, which involves a gradual sweeping of attention through the entire body from feet to head, while employing nonjudgmental awareness of feelings and sensation in each targeted body region (Kabat-Zinn 2013). MBCT was developed to prevent relapse in depressive disorders. It integrates aspects of cognitive behavioral therapy (CBT) for depression into the MBSR programme (Segal 2002).
Other treatments that incorporate mindfulness include acceptance and commitment therapy (ACT; Hayes 2016), distress tolerance training, dialectical behaviour therapy (DBT; Linehan 2018), and certain types of yoga (Salmon 2009). ACT focuses on increasing people's willingness to experience physical cravings, emotions, and thoughts, and allowing these to come and go while making committed behaviour changes that are guided by their own values (Hayes 2016). Distress tolerance training combines elements drawn from ACT with exposure-based treatment, allowing ACT skills to be practised within treatment sessions in response to internal triggers (Brown 2008). DBT also has a strong emphasis on acceptance, incorporating strategies to help the patient accept themselves, their current capabilities, and behavioural functioning (Linehan 2018). Yoga is a key component of MBSR (Kabat-Zinn 2013), and provides an opportunity to practise mindfulness through movement. Forms of yoga that incorporate breathing exercises and directed meditative focus work to still the mind and focus attention (Bock 2012).

How the intervention might work
Mindfulness-based interventions may aid smoking cessation by teaching individuals to pay attention to, and work mindfully with, negative a ective states, cravings, and other symptoms of nicotine withdrawal as they arise, rather than habitually reacting to these unpleasant states by smoking. Proposed mechanisms of action include attention regulation, body awareness, emotion regulation, and change in self-perspective (Hölzel 2011).
Withdrawal following smoking cessation is acutely associated with heightened levels of stress and negative a ect (Shi man 2004; West 2017). Once withdrawal symptoms have abated, cessation is generally associated with improved mental health (Taylor 2014; Taylor 2021), but early stage acute stress, negative a ect, and depression are predictive of relapse (Correa-Fernández 2012; Glassman 1990; Shi man 2004; Shi man 2005). Therefore, interventions that work to reduce these adverse emotional consequences of stopping smoking may enhance quit rates and ultimately prevent relapse. Mindfulness-based interventions have shown some e icacy in the treatment of psychiatric disorders relating to or involving these negative a ective states (Goyal 2014; Marchand 2013).
Further, by teaching smokers to focus their attention on what is happening in the moment, mindfulness-based interventions bring habitual behaviours into consciousness. This enables people to understand the associative learning process, and focus on a ect and craving as central components of positive and negative reinforcement loops (Brewer 2010). By emphasising the transience of a ective states and teaching smokers to 'sit with' negative a ect and craving, mindfulness interventions target and modify learned responses to smoking cues. This may help smokers to quit and may reduce cigarette consumption among those who do not stop smoking completely.
Thus, it has been suggested that mindfulness-based treatments "may have the relative advantage of teaching a single technique that may lead to the dampening and eventual dismantling of the complex interrelated associative processes of smoking rather than just removing stimuli that might propagate them" (Brewer 2011).

Library
Trusted evidence. Informed decisions. Better health.

Why it is important to do this review
If found to be e ective, mindfulness-based interventions could add an innovative intervention option to the range of treatments for smoking cessation. A systematic review, including literature to 2016, did not find evidence of a significant impact of mindfulness meditation interventions on abstinence relative to comparator groups (Maglione 2017). However, the evidence identified was of low certainty due to the high levels of heterogeneity and imprecision detected through meta-analysis. Therefore, there is a need to update this review to include new evidence, in an e ort to increase the certainty of the resulting conclusions. In addition, expanding the search to include other interventions that incorporate mindfulness approaches but do not specifically include an element of meditation (e.g. ACT) can add to our understanding of the potential e ectiveness of mindfulness for smoking cessation.
The purpose of the present review is to assess the e ect of interventions that incorporate mindfulness approaches for smoking cessation, using the robust methodology of Cochrane and the Cochrane Tobacco Addiction Group. This review also represents part of a separate project to evaluate similarities and di erences between the standard methodological processes of the Cochrane Tobacco Addiction Group and a novel, machine-learning approach developed by the Human Behaviour Change Project (Michie 2013).

O B J E C T I V E S
To assess the e icacy of mindfulness-based interventions for smoking cessation among people who smoke, and whether these interventions have an e ect on mental health outcomes.

Types of studies
Randomised controlled trials (RCTs) and cluster-RCTs that measured smoking cessation at least six months from baseline were eligible for this review. We included studies reported as full text, those published as abstract only, and unpublished data, where available. There were no language or date restrictions.

Types of participants
We included current tobacco smokers of any age who were willing to enrol in a smoking cessation study. We excluded studies that only recruited pregnant women, as their particular needs and circumstances warrant their treatment as a separate population, and these are covered in a separate Cochrane Review (Chamberlain 2017).

Types of interventions
We included interventions targeted at tobacco smoking cessation that were either labelled as mindfulness, or involved a mindfulnessbased approach that could be isolated to investigate e ectiveness. There were no restrictions on the minimum duration of the intervention. Where a potentially relevant study intervention was not specifically described as being mindfulness-based, we discussed as a team (of EN, JLB, NL, SJ) whether it was eligible for inclusion. We intentionally adopted an inclusive approach, including interventions that incorporated mindfulness (e.g. ACT or yoga) in addition to those specifically focused on mindfulness meditation (e.g. MBSR or MBCT) to capture the broadest evidence.
Eligible studies had to include at least one of the following comparison (control) interventions: • no smoking cessation treatment; • another smoking cessation intervention, of any length or intensity (including usual care); • another type of mindfulness intervention (e.g. mindfulness of a lower intensity).

Primary outcomes
Smoking abstinence at longest follow-up To be eligible for inclusion, studies must have measured abstinence at least six months from the start of the intervention. Following the Cochrane Tobacco Addiction Group's standard methods, we excluded studies that only measured abstinence at less than sixmonths' follow-up.
In studies with more than one measure of abstinence, we preferred the measure with the strictest criteria, in line with the Russell Standard (West 2005). We used prolonged or continuous abstinence in preference to point prevalence abstinence, and preferred biochemically validated abstinence (e.g. using exhaled carbon monoxide or cotinine measures) over self-report. We favoured biochemically validated point prevalence abstinence over selfreported continuous or prolonged abstinence.

Mental health and well-being
This could provide us with information on potential benefits or harms of the mindfulness-based interventions. Even if comparisons of mindfulness-based interventions with other smoking cessation interventions do not find a benefit of mindfulness for smoking cessation, improved mental well-being could be a reason for choosing this treatment over another. We assessed validated measures of the following relevant constructs: • depression; • anxiety; • quality of life; • positive a ect; • negative a ect; • stress.
We extracted data on these mental health and well-being outcomes, measured at the longest follow-up at which abstinence was reported, or as close to this as possible.

Search methods for identification of studies Electronic searches
We searched the following databases for studies that referred to mindfulness techniques in the title or abstract, or as keywords:

Searching other resources
We checked reference lists of eligible published papers to identify any other relevant papers that may not have been identified by our search, and consulted experts in the field to identify any relevant forthcoming or unpublished research. We contacted the authors of ongoing studies where necessary.
Alongside these manual search strategies, we employed an automated search strategy developed as part of the Human Behaviour Change Project (Michie 2017), using Microso Academic. The Human Behaviour Change Project aims to improve upon the human ability to synthesise, interpret and deliver evidence on behaviour change interventions, using Natural Language Processing and Machine Learning technologies to automate the extraction, synthesis, and interpretation of findings from behaviour change intervention evaluation reports. We added any additional studies identified through this method to those found via the manual search, so that we included all relevant evidence. An evaluation comparing these manual and automated methods of study identification will be reported in a separate paper.

Selection of studies
Two review authors (of EN, JLB, NL, SJ), independently checked the titles and abstracts of retrieved studies for relevance, and acquired full study reports of those that may be candidates for inclusion. The review authors resolved any disagreements by mutual consent, or by recourse to a third review author. Two review authors (of EN, JLB, NL, SJ) then independently assessed the full texts for eligibility, resolving any disagreements through discussion and with involvement of a third review author when necessary. We classified as 'exclude' any studies for which we obtained full reports, but that did not meet the inclusion criteria.

Data extraction and management
Two review authors (of EH, EN, JLB, NL, SJ) independently extracted study data and compared their findings. We resolved any disagreements through discussion, involving a third review author where necessary. Where available, we recorded the following information in the Characteristics of included studies table. Alongside this data extraction of entities that are typically captured in smoking cessation Cochrane Reviews, we also performed data extraction using entities of the Behaviour Change Intervention Ontology, which is being developed as part of the Human Behaviour Change Project (Michie 2017). The ontology consists of granular entities to specify all aspects of behaviour change interventions, such as: • an intervention's context (including 'setting' (Norris 2020) and 'population'); • content (including 'behaviour change techniques'; (Michie 2013)); and • delivery (including 'mode of delivery': how an intervention is provided to participants (Marques 2021); 'source': who delivers interventions (Norris 2021); and 'schedule': how o en an intervention is delivered (Michie 2017)).
An evaluation to compare these methods of data extraction will be reported in a separate paper.

Assessment of risk of bias in included studies
Two review authors (of JLB, NL, SJ) independently assessed the risk of bias for each included study. We used RoB 1, following the guidance as set out in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).

Cochrane Database of Systematic Reviews
• other sources of bias.
As we were investigating a primarily behavioural intervention, we did not assess the blinding of participants and providers, as it is impossible to blind people to behavioural interventions. This is in accordance with specific guidance from the Cochrane Tobacco Addiction Group.
Each review author recorded information in study reports relevant to each domain and then assessed each domain as either at low, high, or unclear risk of bias. We resolved disagreements by discussion with a third review author. We considered studies to be at high overall risk of bias where we judged at least one domain to be at high risk; at low overall risk of bias where all domains were judged to be at low risk; and at unclear overall risk of bias in all other cases.

Measures of treatment e ect
We compared quit rates between intervention and comparator groups for each study. We calculated quit rates on an intentionto-treat basis, including all participants originally randomised to a study arm, treating participants lost to follow-up as relapsed.
We calculated a risk ratio (RR) and 95% confidence interval (CI) for each study. We calculated the RR for each study as: (number of participants who reported smoking abstinence in the intervention group/number of participants randomised to the intervention group)/(number of participants who reported smoking abstinence in the control (comparator) group/number of participants randomised to the control (comparator) group).
Due to high levels of variance between studies in interventions and comparators, and in the measurement of mental health and well-being outcomes, we narratively reported relevant measures of mental health and well-being.

Unit of analysis issues
The one included cluster-RCT did not present an analysis adjusting for the clustering e ect or report an intracluster correlation coe icient (ICC). Therefore, we used unadjusted data for the primary analysis and performed a sensitivity analysis where we estimated the ICC (0.03), based on the ICC reported in other smoking cessation studies (Fanshawe 2017), and adjusted the analysis on this basis.
In the case of studies with multiple intervention arms, we analysed individual arms separately.

Dealing with missing data
For smoking abstinence, we assumed participants lost to follow-up to be smoking, as is standard in the field (West 2005). However, we conducted a sensitivity analysis, excluding numbers lost to followup from the denominator.

Assessment of heterogeneity
In order to assess whether it was appropriate to pool studies and conduct meta-analyses, we assessed the characteristics of included studies to identify any clinical or methodological variance between studies. If we deemed the studies to be homogeneous enough to be combined meaningfully and we could conduct meta-analyses, we assessed statistical heterogeneity using the I 2 statistic (Higgins 2003). We considered an I 2 statistic over 50% to indicate moderate to substantial heterogeneity (Deeks 2021). Where the I 2 statistic was 80% or more, the direction of individual study e ects di ered, and heterogeneity was not fully explained by subgroup and sensitivity analyses, we do not report a pooled estimate because it could be misleading. We conducted the subgroup and sensitivity analyses described below to investigate any potential causes of observed heterogeneity.

Assessment of reporting biases
It was not appropriate to assess reporting bias using funnel plots as none of our analyses pooled 10 or more studies.

Data synthesis
We provided a narrative summary of the included studies and, where appropriate, conducted meta-analyses.
The primary outcome of abstinence provides dichotomous data, therefore, as per the Cochrane Tobacco Addiction Group's standard methods, we combined RRs from individual studies using randome ects, Mantel-Haenszel methods, to calculate pooled overall RRs with 95% CIs.
Meaures of our mental health and well-being outcome typically provided continuous data. Data were too heterogeneous to carry out meta-analyses, so we tabulated the existing information and summarised narratively.
We also narratively reported the results of any within-study analyses that have investigated the following moderators of e ectiveness at at least six months' follow-up: • population type; • baseline motivation to quit; • baseline mental health.

Subgroup analysis and investigation of heterogeneity
We carried out subgroup analyses, categorising studies by the type/ intensity of control treatment received and mode of intervention delivery. We compared pooled summary statistics across groups and ran statistical tests for subgroup di erences.

Sensitivity analysis
For smoking abstinence, we tested the impact of excluding studies deemed to be at overall high risk of bias and compared abstinence rates calculated assuming 'missing equals smoking' with abstinence rates calculated through complete-case analysis. We also carried out the sensitivity analysis reported above, using an assumed ICC to adjust for potential clustering e ects in a cluster-RCT.

Summary of findings and assessment of the certainty of the evidence
Following standard Cochrane methodology (Schünemann 2021), we created summary of findings tables for smoking abstinence, and mental health and well-being outcomes, detailing di erent intervention types in separate tables (mindfulness training; ACT; distress tolerance training; yoga). Also following standard Cochrane methodology (Schünemann 2021), we used the five GRADE considerations (risk of bias, inconsistency, imprecision, indirectness, and publication bias) to assess the certainty of the body of evidence for each outcome, within each comparison, and Cochrane Database of Systematic Reviews to draw conclusions about the certainty of evidence within the text of the review.

Results of the search
Our bibliographic database searches and automated search process identified 2900 non-duplicate records ( Figure 1)
One study (Singh 2014) did not report its location. Participants were recruited from the community (12 studies), online (3 studies), from healthcare centres (2 studies), high schools and universities (2 studies), tobacco treatment services (1 study), and workplaces (1 study). One study was a cluster-RCT (Pbert 2020), which randomised high schools to di erent conditions. All other studies were randomised at the individual level.
The total number of participants across studies was 8186. The median sample size was 146 but ranged from 38 to 2637 participants. Two studies deliberately targeted young adults (Pbert 2020; Savvides 2014), two studies low SES smokers (Davis 2014a; Davis 2014b), one study uninsured smokers (Bricker 2014a), one study smokers with a history of early lapse (never able to remain abstinent for more than 72 hours; Brown 2013), and one study adults with mild intellectual disability (Singh 2014). Most studies had similar proportions of men and women or slightly more women than men. The exceptions were Bloom 2020, Bock 2012 and Weng 2021, which targeted only women; Gaskins 2015, which targeted only men; Singh 2014, which recruited 82% men with mild intellectual disability; and Mak 2020, which was conducted in Hong Kong where smoking prevalence among women is low and recruited 71% men.
Studies typically recruited people who smoked at least five cigarettes a day. Although some studies included lighter smokers as well, the average number smoked was over 15 a day in most studies, ranging from five a day in Pbert 2020's sample of high school students to 22 a day in Brown 2013's community sample.

Mindfulness training
Eight studies used mindfulness training (which, for the purpose of this review, we define as specific training in mindfulness and mindfulness-based meditation techniques).
Five studies tested the e ectiveness of mindfulness training delivered face-to-face. Davis 2014b compared seven weeks of group mindfulness training and meditation practice with an alternative, intensity-matched, behavioural support programme. Similarly, Vidrine 2016 compared mindfulness-based addiction treatment (8 x 2-hour sessions) with an intensity-matched CBT programme. In the latter study, there was also a second, less intensive comparator arm, in which participants received briefer support intended to represent the intervention a smoker might typically receive if they asked a healthcare provider for help (4 x 5-to 10-minute sessions). All participants received selfhelp materials. The other three studies compared mindfulness training with less intensive comparators. Davis 2014a compared an eight-week mindfulness and meditation training programme with quitline support. Singh 2014 compared mindfulness and meditation training for adults with mild intellectual disability with treatment as usual, which varied between participants and encompassed a range of treatments such as behaviour therapies, nicotine replacement therapy (NRT), and other medications. Weng 2021 provided women in workplaces with self-help materials and compared the e ectiveness of additional mindfulness and meditation training (2 x 2-hour sessions), with brief advice to follow the advice of the self-help materials. Provision of pharmacotherapy varied between studies: three studies (Davis 2014a; Davis 2014b; Vidrine 2016), provided participants in both arms with a course of nicotine patches, one study provided no pharmacotherapy (Weng 2021), and one study (Singh 2014), did not specifically provide pharmacotherapy to participants in either arm, although for some comparator arm participants it was part of their usual treatment.
Two studies tested the e ectiveness of mindfulness training delivered via smartphone apps. Garrison 2020 was conducted online. It compared a mobile mindfulness training app plus experience sampling (which asked participants to check in 6 times a day for 22 days) with experience sampling only. Pbert 2020 was a cluster-RCT conducted in high schools. It tested the e ectiveness of a mindfulness smartphone app designed for teens against two comparators: firstly, an alternative (non-mindfulness) smoking cessation app designed for teens and secondly, self-help materials. Participants in each of the three arms met with the school nurse weekly for four weeks. No pharmacotherapy was provided in either study.
de Souza 2020 was the only study to focus on mindfulness for relapse prevention. All participants received CBT over two phases: a smoking cessation phase (weekly sessions over 4 weeks) and a maintenance phase (6 sessions between weeks 6 and 48). The intervention arm also received eight mindfulness-based relapse prevention sessions during the maintenance phase. Participants were o ered the choice of NRT or bupropion.

Library
Trusted evidence. Informed decisions. Better health.

Cochrane Database of Systematic Reviews
Behaviour-change techniques (BCTs) varied across studies. The most commonly used techniques included body changes (7 studies), problem solving (4 studies), self-monitoring of behaviour (4 studies), pharmacological support (4 studies), and goal setting (3 studies), with no clear patterning in the number or type of BCTs used across mode of delivery.

Acceptance and commitment therapy (ACT)
Eight studies used ACT.
Three studies tested the e ectiveness of ACT delivered exclusively face-to-face. McClure 2020 compared a five-week, group ACT programme with a five-week, group CBT programme. The two arms were matched for the number and duration of sessions. Participants in both arms were provided with eight weeks of nicotine patches. Gi ord 2003 compared a seven-week programme of individual and group ACT sessions (with no pharmacotherapy) with a lower-intensity comparator group that received a sevenweek course of nicotine patches. O'Connor 2020 compared six weeks of face-to-face, group ACT sessions with six weeks of faceto-face, group behavioural support, matched in intensity to the six-week, face-to-face ACT programme. No pharmacotherapy was provided.
One study tested the e ectiveness of ACT delivered through a combination of face-to-face sessions and a smartphone app.
In addition to the face-to-face-only intervention arm, O'Connor 2020 included a second intervention arm in which ACT was delivered via two modalities: the six-week, face-to-face ACT programme and an ACT-based smartphone app. This combined ACT intervention was compared with a less intensive ACT arm (i.e. 6 weeks of face-to-face ACT without the app).
One study tested the e ectiveness of ACT delivered exclusively via smartphone app. Bricker 2020 compared an ACT smartphone app with a smoking cessation app based on national clinical practice guidelines. No pharmacotherapy was provided.
One study tested the e ectiveness of ACT delivered through a combination of face-to-face sessions and telephone calls. Mak 2020 compared ACT delivered in one face-to-face session and two follow-up telephone calls with brief advice (5 minutes). Participants in both arms were also provided with self-help materials. No pharmacotherapy was provided.
One study tested the e ectiveness of ACT delivered exclusively via telephone. Bricker 2014a compared an ACT programme delivered over five telephone calls with standard quitline CBT. The arms were matched for the number and duration of telephone calls. Participants were provided with two weeks of nicotine patches or gum.
Two studies tested the e ectiveness of ACT delivered via websites. Bricker 2018 compared an online ACT programme with a national standard online quit programme, with both arms receiving daily messages prompting them to log in. Savvides 2014 compared an avatar-led, internet-based ACT programme with a waitlist control. Neither study provided participants with pharmacotherapy.
BCTs varied across studies. The most commonly used techniques included problem solving (6 studies), body changes (5 studies), goal setting (4 studies), and action planning (4 studies), with no clear patterning in the number or type of BCTs used across mode of delivery.

Distress tolerance training
Two studies used distress tolerance training. Distress tolerance training interventions combined elements drawn from ACT with exposure-based treatment. Exposure included periods of scheduled abstinence prior to sessions and exposure to cues within sessions, allowing ACT skills to be practised within the sessions in response to internal triggers.
Bloom 2020 targeted women who were concerned about postcessation weight gain. The intervention was nine weeks of CBT plus distress tolerance training -a face-to-face and telephone programme that targeted the fear of anticipated post-cessation weight gain and facilitated initiation of abstinence, and appetite awareness and mindful eating skills to reduce post-cessation emotional eating. The comparator was nine weeks of CBT plus smoking health education, which mentioned diet and exercise as strategies for health promotion but did not specifically recommend changing diet or increasing physical activity to prevent postcessation weight gain.
Brown 2013 targeted smokers who had previously tried to quit but had never been able to remain abstinent for more than 72 hours. The intervention was eight weeks of face-to-face distress tolerance treatment and the comparator was six weeks of standard treatment.
Both studies also provided NRT (8 weeks of nicotine patches) to all participants in the intervention and comparator arms.
BCTs varied across studies: while both used pharmacological support, Brown 2013 used reduce prompts/cues and Bloom 2020 used problem solving, self-monitoring of behaviour, social support, information about health consequences, and anticipated regret.

Yoga
Three studies used yoga involving a mindfulness-based approach.
Two studies used Vinyasa yoga (Bock 2012; Gaskins 2015), and one used Iyengar yoga (Bock 2019). In each study, participants in the intervention arm were provided with eight CBT classes and 16 yoga classes over eight weeks. Participants in the comparator arm received CBT and wellness classes over eight weeks. In Bock 2012 and Bock 2019, the comparator was matched to the intervention in terms of the number and duration of wellness classes (16 x 1-hour classes). However, in Gaskins 2015 the comparator was less intensive than the intervention: the intervention arm received 16 yoga classes over the eight weeks, each lasting 60 to 90 minutes, while the comparator arm received eight brief wellness discussions following the CBT sessions.
None of the studies provided participants with pharmacotherapy, but two studies (Bock 2012; Bock 2019), noted that participants were permitted to use NRT or other medications alongside the programme if they wanted to.
BCTs were similar across studies: all three studies used goal setting, problem solving, and social support. Bock 2012 and Gaskins 2015 also used self-monitoring of behaviour and body changes, and Gaskins 2015 also used reduce negative emotions. Excluded studies Figure 1 shows the most common reasons for exclusion of studies during full-text screening, which included: a follow-up period of less than six months; ineligible study design (not an RCT); conference proceedings with no relevant studies; and an intervention where it was not possible to isolate the e ects of the mindfulness element.

Mindfulness for smoking cessation (Review)
In the Characteristics of excluded studies table, we list exclusion reasons for 47 studies. This list is not comprehensive, only containing studies that a reader might plausibly expect to be included.

Risk of bias in included studies
Overall, we judged four of the 21 completed studies to be at low risk of bias, nine studies to be at unclear risk, and the remaining eight studies at high risk of bias.
Details of risk of bias judgments for each domain of each included study can be found in the Characteristics of included studies table. Figure 2 illustrates judgments for each included study.

Random sequence generation
We judged one study (Singh 2014), to be at high risk of selection bias for sequence generation, because randomisation was via alternate placement in the experimental and control groups. We judged eight studies at low risk of bias (Bock 2019; Bricker 2014a; Bricker 2018; Bricker 2020; Gi ord 2003; Mak 2020; O'Connor 2020; Savvides 2014). The risk of bias for the remaining studies was unclear.

Allocation concealment
We judged six studies (Bricker 2014a; Bricker 2018; Bricker 2020; Mak 2020; O'Connor 2020; Weng 2021), to be at low risk of selection bias for allocation concealment, and the remainder to be at unclear risk as there was insu icient information with which to judge.

Blinding of participants and personnel (performance bias)
As we were investigating a primarily behavioural intervention, we did not assess the blinding of participants and providers, as it is impossible to blind people to behavioural interventions. This is in accordance with specific guidance from the Cochrane Tobacco Addiction Group.

Blinding of outcome assessment (detection bias)
We rated three studies (Mak 2020; Savvides 2014; Singh 2014), at high risk for detection bias. These studies did not use blinding, they provided di erent levels of support, and outcomes were selfreported. This meant we thought there was a high risk of bias being introduced. We judged the remaining studies to be at low risk for detection bias.

Incomplete outcome data
We judged most studies (13 out of 21) to be at low risk of attrition bias. We rated four studies with substantial (> 50%) loss to followup and one study with more than a 20% di erence in follow-up rates between arms at high risk of attrition bias (Davis 2014a; Davis 2014b; de Souza 2020; Gaskins 2015; Mak 2020). The remaining three studies (Bock 2012; Savvides 2014; Singh 2014), did not provide su icient data on which to judge, and hence we judged them to be at unclear risk.

Selective reporting
Of the 21 studies, we considered 13 to be at low risk of reporting bias, as they reported all prespecified or expected outcomes. We rated two studies (Bock 2012; de Souza 2020), at high risk, as they did not present data as specified in the original protocols. We judged the rest (Bloom 2020; Brown 2013; Gi ord 2003; Mak 2020; Savvides 2014; Singh 2014), to be at unclear risk, as we were unable to identify a protocol.

Other potential sources of bias
We judged one study (Savvides 2014), to be at high risk of other bias because it used a waitlist control. This design risks participants in the control arm delaying quitting, knowing that they would be receiving an intervention at a later date. This has the potential to inflate the reported e ect of the intervention. We did not find any other studies to be at risk of other bias.

E ects of interventions
See: Summary of findings 1 Mindfulness training compared with control for smoking cessation; Summary of findings 2 Acceptance and commitment therapy (ACT) compared with control for smoking cessation; Summary of findings 3 Distress tolerance training compared with control for smoking cessation; Summary of findings 4 Yoga compared with control for smoking cessation

Smoking abstinence
Three studies compared an intervention involving mindfulness training with an alternative smoking cessation treatment that was matched for intensity (Davis 2014b; Pbert 2020; Vidrine 2016

Mental health
Three studies that tested an intervention involving mindfulness training reported on mental health outcomes (Analysis 1.5; very low-certainty evidence). One study showed evidence of a benefit of mindfulness training on mental health. Davis 2014b (135 participants) analysed perceived stress at six months postquit. They observed a statistically significantly greater reduction in perceived stress between baseline and six months in the intervention arm than the intensity-matched control arm, but this di erence was not statistically significant when analysed as intention-to-treat.
Two studies showed no clear evidence of a benefit of mindfulness training on mental health. de Souza 2020 (86 participants) analysed depression, anxiety, negative a ect, and positive a ect at 4 and 12 weeks. No statistically significant or clinically meaningful di erence between conditions was observed for any outcome at either time point. Vidrine 2016 (412 participants) assessed depression, perceived stress, negative a ect, and positive a ect at six time points between quit date and six months post-quit. They analysed changes between quit date and six months and observed no statistically significant or clinically meaningful di erence between conditions for any outcome.
In addition, Davis 2014a (196 participants) assessed negative a ect at one month post-baseline, but only reported data for the intervention arm.

Smoking abstinence
Five studies compared an intervention involving ACT with an alternative smoking cessation treatment that was matched for intensity (Bricker 2014a; Bricker 2018; Bricker 2020; McClure 2020; O'Connor 2020). We judged McClure 2020 to be at unclear risk of bias, while the other four studies were at low risk. It was not appropriate to pool data across these five studies because there was a high level of heterogeneity (I 2 = 82%; Analysis 2.1), with variation in the direction of e ect between studies, and the result may be misleading. Subgroup analyses showed some evidence of moderation by mode of delivery (I 2 = 82%), although this didn't account for all variation within subgroups. We judged this study to be at unclear risk of bias. Using completecase analysis increased the point estimate but did not substantially change the interpretation of the results (RR 1.63, 95% CI 0.71 to 3.72; 71 participants; Analysis 6.6).
One study compared an intervention involving ACT with brief advice. Mak 2020's point estimate favoured ACT over brief advice but there was substantial imprecision, meaning the result could indicate potential harm as well as considerable benefit (RR 1.27, 95% CI 0.59 to 2.75; 144 participants; very low-certainty evidence; Analysis 2.3). We judged this study to be at high risk of bias. Using complete-case analysis reduced the point estimate but did not substantially change the interpretation of the result (RR 1.06, 95% CI 0.54 to 2.11; 66 participants; Analysis 6.7).
One study compared an intervention involving ACT with less intensive ACT. O'Connor 2020 showed no evidence of a benefit of more intensive ACT, with the point estimate indicating no di erence between more and less intensive ACT (RR 1.00, 95% CI 0.50 to 2.01; 100 participants; low-certainty evidence; Analysis 2.4). We judged this study to be at low risk of bias. Using complete-case analysis did not substantially change the interpretation of the result (RR 0.94, 95% CI 0.47 to 1.86; 91 participants; Analysis 6.8).

Within-study analyses of moderators of interest
Bricker 2018 tested for moderation of the e ectiveness of ACT by baseline mental health (depression or anxiety) and commitment to quitting. Quit rates were not found to di er significantly according to these variables.
Gi ord 2003 tested the e ectiveness of ACT in a subsample of smokers who were highly dependent. While there was no significant di erence between the ACT arm and comparator arm in Cochrane Database of Systematic Reviews the full sample, ACT was reported to be associated with better longterm quitting outcomes among nicotine-dependent participants.

Mental health
Two studies that tested an intervention involving ACT reported on mental health outcomes (Analysis 2.5; very low-certainty evidence). Both studies showed no evidence of a benefit of ACT on mental health. Gi ord 2003 (102 participants) analysed negative a ect between conditions at post-treatment, six months and 12 months and O'Connor 2020 (150 participants) analysed positive mental health at post-treatment and six months. Neither study observed a statistically significant or clinically meaningful di erence between conditions at any time point.

Smoking abstinence
One study compared an intervention involving distress tolerance training with an alternative smoking cessation treatment that was matched for intensity. Bloom 2020 showed no evidence of a benefit of distress tolerance training, with the 95% CI spanning both benefit and harm (RR 0.87, 95% CI 0.26 to 2.98; 69 participants; lowcertainty evidence; Analysis 3.1). We judged this study to be at unclear risk of bias. Using complete-case analysis did not substantially change interpretation of the result (RR 0.86, 95% CI 0.26 to 2.86; 54 participants; Analysis 6.9).
One study compared a distress tolerance training intervention with a less intensive smoking cessation treatment (Brown 2013). There was substantial imprecision, meaning the result could indicate potential harm as well as considerable benefit (RR 1.63, 95% CI 0.33 to 8.08; 49 participants; low-certainty evidence; Analysis 3.2). We judged this study to be at unclear risk of bias. Using complete-case analysis did not substantially change the interpretation of the result (RR 1.68, 95% CI 0.34 to 8.28; 46 participants; Analysis 6.10).

Mental health
One study that tested an intervention involving distress tolerance training reported on a mental health outcome (Analysis 3.3; low-certainty evidence). Brown 2013 (49 participants) analysed negative a ect at four weeks post-quit and observed no statistically significant or clinically meaningful di erence between conditions.
In addition, Bloom 2020 (69 participants) planned to assess depression and negative a ect at each follow-up (1 month, 3 months, and 6 months post-treatment), but to our knowledge have not reported these data.

Smoking abstinence
One study compared an intervention involving yoga with an alternative smoking cessation treatment that was matched for intensity. Bock 2012's point estimate favoured yoga over alternative smoking cessation treatment but there was substantial imprecision, meaning the result could indicate potential harm as well as considerable benefit (RR 1.44, 95% CI 0.40 to 5.16; 55 participants; very low-certainty evidence; Analysis 4.1). We judged this study to be at high risk of bias. The number of participants followed up in each arm was unclear so we could not conduct a complete-case analysis.
Raw data on the number of quits at six months were not available for two other studies that tested an intervention involving yoga so we could not calculate unadjusted RRs. Bock 2019 reported no significant di erence in the odds of smoking abstinence between the intervention arm and matched comparator arm at six-month follow-up (P > 0.05). Gaskins 2015 also reported no significant di erence in the odds of smoking abstinence between the intervention arm and less intensive comparator arm at sixmonth follow-up (OR 2.38, 95% CI 0.52 to 10.8, P = 0.265). There was substantial imprecision, meaning the result could indicate potential harm as well as considerable benefit.

Mental health
Two studies that tested a yoga intervention reported on mental health outcomes (Analysis 4.2). Both studies showed no evidence of a benefit of yoga on mental health. Bock 2012 (55 participants) analysed depression, anxiety, and general wellbeing at eight weeks (post-treatment) and six months, but only reported data collected at 8 weeks. No statistically significant or clinically meaningful di erences between conditions were observed. Gaskins 2015 (38 participants) analysed depression, anxiety, and a composite measure of physical self-worth, attractiveness, physical strength, and condition, at eight weeks (post-treatment), three months and six months. No statistically significant or clinically meaningful di erences between conditions, nor any significant group by time interactions, were observed.

Summary of main results
The 21 studies in this review did not detect a clear, long-term benefit of mindfulness-based smoking cessation interventions (based on mindfulness training, ACT, distress tolerance training, or yoga) when compared with other interventions, or with no intervention, for smoking cessation. This was true when mindfulness-based interventions were compared with intensitymatched smoking cessation interventions, less intensive smoking cessation interventions (including less intensive mindfulness), or no treatment. However, one subgroup analysis found a positive e ect of an ACT intervention when this was delivered via smartphone application, as opposed to face-to-face, through a website, or over the telephone.
Ten studies collected data on mental health and well-being, of which nine analysed and reported on changes in these outcomes. There was no clear evidence of a positive or negative e ect of mindfulness-based treatments on mental health and well-being.

Overall completeness and applicability of evidence
The searches conducted for this review were broad, in our attempt to find any study that made any mention of mindfulness-based approaches. As well as medical databases, we also searched studies registers to identify any ongoing or completed but unpublished registered studies, and supplemented our traditional search strategy with an automated search strategy developed as part of the Human Behaviour Change Project (Michie 2017), using Microso Academic. We therefore feel confident in our search approach.
A particular challenge of this review compared with other reviews of smoking cessation treatments was bringing together a diverse Cochrane Database of Systematic Reviews evidence base. The studies identified by this review varied widely in their design (e.g. digital versus in person), intervention type (e.g. ACT versus yoga), nature of the comparator, and mental health outcomes assessed, meaning we could not meaningfully pool results in a single meta-analysis. While we intentionally adopted an inclusive approach to cover a broad range of mindfulnessbased interventions, some of the studies included may have had a looser mindfulness focus (e.g. yoga interventions) than others (e.g. mindfulness training interventions), but our approach to pooling meant that we did not 'dilute' the e ects of pure mindfulness interventions. The studies identified in this review were mainly conducted in the USA and all took place in highincome or higher middle-income countries. Most studies were carried out in the general population and so may not be applicable to populations with specific requirements or particularly high cigarette dependence.
To be included studies had to assess long-term abstinence, so most studies were able to contribute cessation data to the relevant comparisons. However, the number of studies and participants contributing to each analysis were low, and further research could strengthen or change findings. In addition, data on mental health outcomes were sparse and varied, meaning we were unable to conduct meta-analyses for this outcome. We did not assess safety outcomes beyond any adverse e ect on mental health and wellbeing because the intervention was behavioural and was not considered high risk for adverse events.

Quality of the evidence
Of the 21 studies included in this review, we judged four to be at low risk of bias for all domains, and eight to be at high risk in one or more domains. In many cases, we rated studies at unclear risk of bias because they did not report key information. In these cases, it is impossible to know whether these studies were at any risk of bias or whether the information was simply not reported. To investigate the potential impact on results of studies that we judged to be at high risk of bias, we removed these studies in sensitivity analyses. This did not a ect our interpretation of results.
We considered the certainty of the evidence for e ectiveness of mindfulness training, ACT, distress tolerance training, and yoga interventions for smoking cessation relative to matched-intensity smoking cessation treatment, less intensive smoking cessation treatment, or no treatment. We created summary of findings tables and carried out GRADE ratings for each comparison (Summary of findings 1; Summary of findings 2; Summary of findings 3; Summary of findings 4).
We judged all comparisons and outcomes to be of low or very low certainty, meaning that the interpretation of e ects is likely to change as more studies and information become available.
Reasons for downgrading the certainty of evidence included: risk of bias, when all studies pooled were judged at high risk of bias; inconsistency, when there was variance in the characteristics of studies or statistical heterogeneity was high and unexplained; and imprecision, when the absolute number of events was low or confidence intervals were wide and included no di erence, or both.

Potential biases in the review process
We took several steps to ensure the review process was robust. We followed standard methods used by the Cochrane Tobacco Addiction Group. Our search strategy included a broad range of databases, including the Cochrane Tobacco Addiction Group's Specialised Register. We followed standard Cochrane practice of requiring two review authors to independently screen studies, extract data, and assess risk of bias. None of the authors of this review were also authors of included studies.
Despite this rigorous approach, it is possible that relevant literature, particularly unpublished or grey literature, may have been missed. We did not evaluate publication bias as there were fewer than 10 studies available for each primary outcome. It is also possible that non-reporting of information in the published articles may have influenced the risk of bias assessments.

Agreements and disagreements with other studies or reviews
Carim-Todd 2013 conducted a systematic review of yoga and other mind-body interventions for smoking cessation. It included 14 studies, of which eight studies were RCTs, one study was a non-RCT, two studies applied within-participant controlled designs, and three studies used pre-post designs. We included just one of these RCTs in our review (Bock 2012); the others did not meet our inclusion criterion of six months' follow-up. Carim-Todd 2013 did not meta-analyse data due to di erences in study designs, participants, and outcome measures. The authors reported that all 14 included studies, "observed changes in smoking behavior or in predictors of smoking behavior that could be beneficial for smoking cessation" but more clinical studies with larger sample sizes and carefully monitored interventions were required to draw firm conclusions.

Implications for practice
• We did not detect a clear benefit of mindfulness-based interventions for increasing long-term smoking quit rates compared with no treatment or alternative smoking cessation treatments that are equally or less intensive. However this evidence is of low or very low certainty, and further evidence is likely to change our conclusions. • We also did not detect a clear benefit of mindfulness-based interventions for improving mental health and well-being compared with no treatment or alternative smoking cessation treatments that are equally or less intensive. Again, this evidence is of low or very low certainty, and our conclusions are likely to change with further evidence.

Implications for research
• Further RCTs of mindfulness-based interventions for smoking cessation are needed, following up participants at six months or longer. Studies with active comparators (i.e. comparing mindfulness-based interventions to currently used smoking cessation interventions) are likely to be of particular use to decision makers. • Further studies need to be adequately powered to detect potentially small but clinically important di erences between mindfulness-based interventions and active comparators. In order to ensure low risk of bias, they should involve biochemical verification of abstinence along with improved methods of retaining participants to follow-up points. • There is also a need for more consistent reporting of mental health and well-being outcomes in studies of mindfulnessbased interventions for smoking cessation. Even if mindfulness is only as successful as other behavioural support in enhancing long-term quit rates, it may be preferable to some smokers if it improves mental health. Most studies we identified did not report on mental health or well-being. Those that did assessed a number of di erent constructs, at di erent time points, using a variety of measures, meaning we could not meaningfully pool the results. Therefore, it would be useful to develop a consensus on the best ways to measure these outcomes in relevant studies.    Interventions All participants received CBT, including preparation for quit date, reinforcement, and support for quitting, discussion of past and ongoing quit experiences, initiation of self-monitoring, identification of triggers and high-risk situations, development of coping strategies for triggers, obtaining social support, instruction in how to use nicotine patches, and relapse prevention. Recommendations for how to minimise weight gain were brief, de-emphasised, and consistent with standard CBT (e.g. take a walk to cope with craving instead of smoking, eat low-calorie snacks)

Comparator
CBT + smoking health education. Diet and exercise were mentioned as strategies for health promotion and prevention of disease but no specific recommendations for how to change diet or increase physical activity

Intervention
CBT + distress tolerance for weight concern. Based on ACT and included: psychoeducation about the relationship between smoking and weight; distress tolerance skills; discussion of weight concern as a barrier to successful initiation of abstinence; and values-oriented living skills targeting reduction of emotional eating after quitting Author conflicts of interest: "RAB has equity ownership in Health behaviour Solutions, Inc., which is developing products for tobacco cessation that are not related to this study. The terms of this arrangement have been reviewed and approved by the University of Texas at Austin in accordance with its policy on objectivity in research. The other authors have no interests to declare." Cochrane Database of Systematic Reviews Interventions All participants were provided with an 8-week group CBT programme for smoking cessation, including quit day in week 2, self-monitoring, stimulus control, coping with high-risk situations, and stress management for smoking cessation. The programme also focused on topics of concern to women when quitting, including healthy eating, weight management, and balancing multiple roles and multiple demands.

Comparator
Group CBT + wellness classes

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk Stated randomised but method not specified Interventions All participants were provided with an 8-week group CBT programme for smoking cessation, including planning for a targeted quit day (week 4), handling smoking triggers, coping with cravings, and managing withdrawal.

Group CBT + wellness classes
Mode of delivery: face-to-face (group)

Intensity: 8 CBT sessions (x 1 h) and 16 wellness classes (x 1 h) over 8 weeks
Pharmacotherapy: none provided, but participants were permitted to use NRT or other smoking cessation medications in conjunction with the programme

Allocation concealment (selection bias)
Low risk Quote: "Random assignments were concealed from participants until after study eligibility, consent, and baseline data was obtained. Neither research sta nor study participants had access to upcoming randomized study arm assignments." Blinding of outcome assessment (detection bias) All outcomes Low risk Abstinence self-reported but no face-to-face contact so no difference in intensity; differential report unlikely Incomplete outcome data (attrition bias) All outcomes Cochrane Database of Systematic Reviews patent application, the licensing agreement, or 2Morrow, Inc. Ms Mull reported receiving grants from the National Institutes of Health/National Cancer Institute during the conduct of the study. Dr Heffner reported receiving nonfinancial support from Pfizer outside the submitted work. None of the authors has a financial relationship with the iCanQuit application and thus will not receive any compensation when it becomes publicly available. No other disclosures were reported."

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote: "The random allocation sequence was generated by a database manager and implemented automatically by the study website." Allocation concealment (selection bias) Low risk Quote: "Random assignments were concealed from participants throughout the trial. The random allocation sequence was generated by a database manager and implemented automatically by the study website. Neither research sta nor study participants had access to upcoming randomized study group assignments." Blinding of outcome assessment (detection bias) All outcomes Low risk Abstinence self-reported but no face-to-face contact so no difference in intensity; differential report unlikely Incomplete outcome data (attrition bias) All outcomes

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk Insufficient detail to make a judgment Quote: "Consented individuals were asked to complete baseline assessment visit (carbon monoxide (CO) breath testing and self-report measures) and then undergo randomization via random draws to either the Control Group (Quit Line + 4 weeks Nicotine Replacement Therapy (NRT)) or MTS (MTS + 4 weeks NRT)"

Mode of delivery: unclear
Intensity: • smoking cessation phase: 4 weekly CBT sessions (x 90 min; smoking cessation phase) • maintenance phase: 6 CBT sessions between weeks 6 and 48

Pharmacotherapy: choice of NRT or bupropion
Type of therapist/provider: physician with experience treating smoking and with training in the standard treatment approach

Cochrane Database of Systematic Reviews
Author conflicts of interest: "None declared." Author conflicts of interest: "Judson A. Brewer and Prasanta Pal own stock in Claritas Mindsciences, the company that developed the apps used in this study. All other authors declare that they have no competing interests."     Author conflicts of interest: "The authors declare that they have no competing interests."

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote: "The process of randomization was based on computer-generated, block randomization with random block sizes, which were placed in sealed opaque envelopes." Allocation concealment (selection bias) Low risk Quote: "The process of randomization was based on computer-generated, block randomization with random block sizes, which were placed in sealed opaque envelopes." Blinding of outcome assessment (detection bias) All outcomes High risk Although abstinence was biochemically verified, only unverified quit rates are reported and we were unable to obtain verified rates from the study authors.

Cochrane Database of Systematic Reviews
Random sequence generation (selection bias) Low risk Quote: "the allocation sequence was generated with random block sizes of 3, 6 and 9 by a researcher with no clinical involvement in the trial using an online randomization tool" Allocation concealment (selection bias) Low risk Quote: "allocation sequence was concealed from the researcher (MOC) enrolling participants in sequentially numbered, opaque, sealed envelopes" Blinding of outcome assessment (detection bias) All outcomes