Motivational interviewing for smoking cessation
Abstract
Background
Motivational Interviewing (MI) is a directive patient‐centred style of counselling, designed to help people to explore and resolve ambivalence about behaviour change. It was developed as a treatment for alcohol abuse, but may help people to a make a successful attempt to quit smoking.
Objectives
To determine whether or not motivational interviewing (MI) promotes smoking cessation.
Search methods
We searched the Cochrane Tobacco Addiction Group Specialized Register for studies using the term motivat* NEAR2 (interview* OR enhanc* OR session* OR counsel* OR practi* OR behav*) in the title or abstract, or motivation* as a keyword. Date of the most recent search: August 2014.
Selection criteria
Randomized controlled trials in which motivational interviewing or its variants were offered to tobacco users to assist cessation.
Data collection and analysis
We extracted data in duplicate. The main outcome measure was abstinence from smoking after at least six months follow‐up. We used the most rigorous definition of abstinence in each trial, and biochemically validated rates where available. We counted participants lost to follow‐up as continuing smoking or relapsed. We performed meta‐analysis using a fixed‐effect Mantel‐Haenszel model.
Main results
We identified 28 studies published between 1997 and 2014, involving over 16,000 participants. MI was conducted in one to six sessions, with the duration of each session ranging from 10 to 60 minutes. Interventions were delivered by primary care physicians, hospital clinicians, nurses or counsellors. Our meta‐analysis of MI versus brief advice or usual care yielded a modest but significant increase in quitting (risk ratio (RR) 1.26; 95% confidence interval (CI) 1.16 to 1.36; 28 studies; N = 16,803). Subgroup analyses found that MI delivered by primary care physicians resulted in an RR of 3.49 (95% CI 1.53 to 7.94; 2 trials; N = 736). When delivered by counsellors the RR was smaller (1.25; 95% CI 1.15 to 1.63; 22 trials; N = 13,593) but MI still resulted in higher quit rates than brief advice or usual care. When we compared MI interventions conducted through shorter sessions (less than 20 minutes per session) to controls, this resulted in an RR of 1.69 (95% CI 1.34 to 2.12; 9 trials; N = 3651). Single‐session treatments might increase the likelihood of quitting over multiple sessions, but both regimens produced positive outcomes. Evidence is unclear at present on the optimal number of follow‐up calls.
There was variation across the trials in treatment fidelity. All trials used some variant of motivational interviewing. Critical details in how it was modified for the particular study population, the training of therapists and the content of the counselling were sometimes lacking from trial reports.
Authors' conclusions
Motivational interviewing may assist people to quit smoking. However, the results should be interpreted with caution, due to variations in study quality, treatment fidelity, between‐study heterogeneity and the possibility of publication or selective reporting bias.
PICOs
Plain language summary
Does motivational Interviewing help people who smoke to quit?
Background: Motivational interviewing is widely used to help people to stop smoking. It is a counselling style which helps people to explore and resolve their uncertainties about changing their behaviour. It tries to avoid an aggressive or confrontational approach and instead steer people towards choosing to change their behaviour, and encouraging their self belief. The aim of this review is to discover whether motivational interviewing helps more people to quit than brief advice or usual care, when used to help people to stop smoking.
Study characteristics: We searched for new studies to add to this review in August 2014 and found 14 new studies. Twenty‐eight randomized or cluster‐randomized controlled trials are now included in this review. Studies were included if participants were tobacco users; provided participants were not pregnant women or adolescents; if the intervention being tested was based on motivational interviewing principles; if the study included some kind of monitoring of the motivational interviewing intervention, such as staff training or a measure of the quality of counselling delivered, or both; if the control/comparison condition was brief advice or usual care; and if the study reported smoking abstinence at least six months after the start of the programme. Between them these studies recruited 16,803 tobacco users. Two of the studies recruited smokeless tobacco users, and the rest recruited cigarette smokers. The majority of studies provided motivational interviewing support face‐to‐face; however seven studies delivered the support by telephone only.
Key findings: Our review found that motivational interviewing appears to help more people to quit smoking than brief advice or usual care when provided by general practitioners and by trained counsellors. Motivational interviewing carried out by general practitioners appeared to be more successful than when carried out by nurses or counsellors. Shorter motivational interviewing sessions (less than 20 minutes per session) were more effective than longer ones. A single session of treatment appeared to be marginally more successful than multiple sessions, but both delivered successful outcomes. The evidence for the value of follow‐up telephone support was unclear, and face‐to‐face counselling did not help more people to quit than telephone counselling. Both approaches were more successful than brief advice or usual care.
Quality of evidence: We have assessed the evidence presented in this review as of moderate quality. Our results should be interpreted with caution, due to variations in study characteristics and how the treatment was delivered. In a number of cases it was difficult to assess the quality of included studies due to a lack of reporting of study details. Finally there is some evidence that studies which did not find an effect of motivational interviewing were less likely to be published and therefore this may impact upon our results.
Authors' conclusions
Summary of findings
| Motivational Interviewing compared to brief advice/usual care for smoking cessation | ||||||
| Patient or population: adult smokers | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| brief advice/usual care | Motivational Interviewing | |||||
| Longest duration and strictest definition of tobacco abstinence | Study population | RR 1.26 | 16,803 | ⊕⊕⊕⊝ | ||
| 104 per 1000 | 131 per 1000 | |||||
| *The assumed risk (e.g. the median control group risk across studies) is calculated based on the quit rates in control groups across all studies. The corresponding risk (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). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1Most studies at unclear risk of bias due to a lack of reporting. Very few at high risk of bias. No evidence of sensitivity of the effect due to high risk of bias, so decision made not to downgrade on risk of bias. 2Significant amount of heterogeneity (I² = 49%), which is not fully explained by any test for subgroup differences; however confidence intervals largely overlap, so decision made not to downgrade on basis of heterogeneity. 3Downgraded one level due to indication of possible publication bias: funnel plot indicates that less precise studies were more likely to show positive effects. | ||||||
Background
Description of the condition
Cigarette smoking remains one of the leading causes of preventable disease worldwide (USDHHS 2000). Various pharmacological and non‐pharmacological methods to assist smoking cessation are available, and there is good quality evidence for the effectiveness of several of them. For instance, Stead 2013 has shown that brief advice from physicians can significantly increase the odds of quitting, as can nicotine replacement therapy (Stead 2012), bupropion (Hughes 2014) and varenicline (Cahill 2012). There is also evidence that combining pharmacological and behavioural interventions helps people to stop smoking. Both pharmacological and behavioural methods are considered as equal contributors to overall success rates (Coleman 2004).
Description of the intervention
The concept of motivational interviewing (MI) evolved from experiences in treating alcohol abuse, and was first described by Miller in 1983. It is defined as "a directive, client‐centred counselling style for eliciting behavior change by helping clients to explore and resolve ambivalence" (Miller 1983). The four guiding principles: (a) expressing empathy, (b) developing discrepancy, (c) rolling with resistance, (d) supporting self efficacy, have been detailed elsewhere (Miller 2002).
The motivational interviewing process is a brief psychotherapeutic intervention intended to increase the likelihood that a person will make an attempt to change their harmful behaviour. Adaptations of MI have ranged from brief 20‐minute office interventions (Motivational Consulting) to Motivation Enhancement Therapy (MET), a multi‐session course of treatment, including a lengthy assessment, personalized feedback and follow‐up interviews (Rollnick 1992, Lawendowski 1998). MI has also been provided by telephone consultations and in a group format. MI and its various forms have been applied both as a stand‐alone intervention or with other treatments, and in a range of settings. These include health settings such as general hospital wards, emergency departments, and general medical practice (Britt 2002).
How the intervention might work
In motivational interviewing, Miller conceptualises that motivation may fluctuate over time or from one situation to another, and can be influenced to change in a particular direction (Miller 1994). Thus, lack of motivation (or resistance to change) is seen as something that is open to change. The main focus of MI is facilitating behaviour change by helping people to explore and resolve their ambivalence about behaviour change (Rollnick 1995). Miller and Rollnick also suggested that adopting an aggressive or confrontational style or both (as in traditional approaches) is likely to produce negative responses from people (such as arguing), which then may be interpreted by the practitioner as denial or resistance. MI also differs from patient‐centred approaches in that it is directive, that is, with MI there is a clear goal of exploring the person's ambivalence in such a way that they are more likely to choose to change the behaviour in question in the desired direction.
Why it is important to do this review
MI has been used primarily for the behavioural management of disorders. It has been used to treat alcohol abuse, drug addiction, weight loss, compliance with treatment for asthma and diabetes as well as for smoking cessation. Systematic reviews (Burke 2003; Knight 2006; Rubak 2005;Martins 2009; Armstrong 2011; Song 2014; Teeter 2014) have shown MI to be an effective intervention for drug and alcohol use, weight control, diet and exercise, diabetes management, medication adherence and oral health. Two systematic reviews of motivational interviewing to aid smoking cessation have been carried out since this review was first published in 2010 (Heckman 2010; Hettema 2010). Both reviews found very modest positive effects of MI at long‐term follow. Heckman 2010 includes 31 trials and reports an odds ratio (OR) of 1.44 (95% confidence interval (CI) 1.11 to 1.88) at 22 to 26 week follow‐up, and 1.25 (95% CI 0.91 to 1.71) at 52 week follow‐up. The main meta‐analysis by Hettema 2010 includes 23 studies and results in an OR of 1.35 (95% CI 1.02 to 1.78).
Both previous reviews had slightly broader inclusion criteria than those used in this review, and Heckman 2010 included studies which had active control studies, which may have underestimated the effect of MI. Allsop 2007 has summarised the difficulties of assessing a given intervention's fidelity to the principles of MI: firstly, its limited theoretical basis compromises our understanding of its essential ingredients and processes; secondly, there are relatively few reliable and practical instruments with which to assess the training, quality and fidelity of implementation of MI's principles; and thirdly, research reports often give inadequate detail of the methods used in what purports to be an MI intervention. Our review attempts to address these pitfalls through the selection, assessment and analysis of the included trials.
Objectives
The primary objective is to determine whether or not motivational interviewing (MI) promotes smoking cessation.
Our hypotheses are:
-
More participants quit smoking when provided with MI treatment than with no advice or simple advice (usual care).
-
MI effects are relatively long‐lasting compared with other therapies.
-
Intensive MI ( more sessions, longer duration of each session) is more likely to help people to quit smoking than single or shorter sessions.
-
MI counsellors' attributes, e.g. occupation (doctor, nurse, counsellors), experience or level of training in MI, are potential moderators of the effect size.
-
MI quitters have a similar relapse rate to those who quit with other therapies.
-
MI has incremental effects when combined with other therapies.
-
MI does not have any significant harmful effect.
Methods
Criteria for considering studies for this review
Types of studies
-
Randomized controlled trials (RCTs).
-
Cluster‐randomized controlled trials (c‐RCTs), with the unit of allocation an institution or organization (e.g., school, hospital, workplace) where one or more professionals are implementing the interventions.
Types of participants
Participants could be tobacco users of either gender recruited in any setting. The only exceptions are trials which only recruited pregnant women or adolescents who smoked, as their particular needs and circumstances warrant them being treated as separate populations.
Types of interventions
The intervention must be based primarily upon the motivational interviewing (MI) principles laid down by Miller and Rollnick (Miller 2002). The trial must, in the opinion of the authors, comply with MI principles and practice, beyond simply referring to the concept.
-
The study should make explicit reference to at least some of these MI principles; exploring ambivalence, decision balance, assessment of motivation and confidence to quit, eliciting 'change talk' and supporting self efficacy.
-
To ensure fidelity of intervention, some form of monitoring of MI should also be reported. This could include the details concerning the training of the counsellor and measures to ensure the quality of MI sessions, e.g. by videotaping the sessions or by using an assessment scale and supervision.
-
The intervention could be delivered on an individual basis or as group sessions.
-
Even the briefest of interventions may be acceptable, provided that it met our other inclusion criteria. It is unclear how brief an adaptation of MI may be while still conforming to MI principles and techniques.
-
Face‐to‐face and telephone‐based interviews are both eligible.
-
The therapists could be any healthcare professional or counsellor.
-
Trials with a pharmacological co‐intervention (e.g. nicotine replacement therapy) are eligible, provided that the pharmacotherapy was given to all participants and was not the intervention being tested.
-
The comparison (control) intervention could be brief advice (i.e. verbal instruction with a 'stop smoking' message, with or without information on the harmful effects of smoking), a low‐intensity intervention, or routine care.
Motivational interviewing is frequently linked with the transtheoretical ('stages of change') model of behaviour change. However, it is conceptually and practically distinct from it, and we have not included trials primarily testing that approach. Stage‐based interventions are covered in a separate review (Cahill 2010).
Types of outcome measures
The primary outcome used in the review is smoking cessation. We exclude trials not including data on smoking cessation rates. We have preferred sustained abstinence over point prevalence, where both were available. We report abstinence at the longest follow‐up, and have required a minimum follow‐up of six months from the start of treatment. Where biochemical validation was used, we regard only those participants meeting the biochemical criteria for cessation as abstainers. We assumed participants lost to follow‐up to have continued smoking or relapsed, and we include all participants randomized in the denominator (an intention‐to‐treat analysis).
Search methods for identification of studies
Electronic searches
We identified trials from the Tobacco Addiction Review Group's specialized register, using the term ( motivat*) NEAR2 (interview* OR enhanc* OR session* OR counsel* OR practi* OR behav*) in title or abstract, or as keywords, or motivation* as a keyword. The full search strategy can be found in Appendix 1. The specialized register has been developed from electronic searching of MEDLINE, EMBASE, PsycINFO and Web of Science, together with handsearching of specialist journals, conference proceedings and reference lists of previous trials and overviews. At the time of the search the Register included the results of searches of the Cochrane Central Register of Controlled trials (CENTRAL), issue 6, 2014; MEDLINE (via OVID) to update 20140627; EMBASE (via OVID) to week 201427; PsycINFO (via OVID) to update 20140630. See the Tobacco Addiction Group module in the Cochrane Library for full search strategies and a list of other resources searched.
Searching other resources
For the first version of this review, we cross‐checked our results with the MINT database of past and current research into motivational interviewing (MINT 2015). This resource has not been updated since November 2009. We also searched local journals in Chinese (including Mainland and Taiwan). We did not identify any additional studies from either of these sources, and have not repeated these searches for the current (2015) update.
Data collection and analysis
Selection of studies
We checked the abstracts of studies generated by the search strategy for relevance, and then acquired full‐text reports of any trials that might be suitable for the review. Two authors independently assessed and selected candidate trials for inclusion, and each independently extracted the data from them. We have noted reasons for the non‐inclusion of studies (Characteristics of excluded studies).
Data extraction and management
We extracted the following information about each trial, where available. Study characteristics are presented in the table 'Characteristics of included studies:
-
Details of study design, including method of allocation, blinding, study structure
-
Location and setting of the trial, e.g. hospital‐based, clinic‐based, community‐based
-
Method of recruitment to the study
-
Sample size calculations
-
Status of the participants, e.g. only motivated volunteers or all motivated and unmotivated volunteers
-
Eligibility and exclusion criteria, and demographic descriptors
-
Type and quality of MI training provided to the therapists
-
Any procedures followed to ensure MI fidelity
-
Description of the intervention, including the nature, frequency and duration of MI, and any co‐interventions used
-
Outcome measures: definition of smoking abstinence used for primary outcome, timing of longest follow‐up, any biochemical validation
-
Reporting of drop‐outs and losses to follow‐up.
In trials where details of the methodology were unclear or where results were not expressed in a form that allowed extraction of the necessary key data, we wrote to the investigators to request the information.
Assessment of risk of bias in included studies
We evaluated studies on the basis of the quality of the randomization procedure, allocation concealment (Schulz 2002a; Schulz 2002b), blinding, and any other bias, using the 'Risk of bias' table, as outlined in the Cochrane Handbook 2011; ). As with the study characteristics above we extracted information regarding each domain and then two authors independently rated the domian as being at high, low or unclear risk of bias. We resolved any disagreement between authors through discussion with the third author.
We used the GRADE system to assess the quality of the evidence for the primary outcome across the included studies and produced a 'Summary of findings' table to illustrate this (Cochrane Handbook 2011).
Measures of treatment effect
Where possible we have extracted smoking outcomes as continuous abstinence, but we have accepted less strict definitions (e.g. point prevalence abstinence) where continuous abstinence was not available.
Dealing with missing data
We conducted an intention‐to‐treat analysis, i.e. using as the denominator all participants randomized to their original groups where the data were available, and we assumed that those participants lost to follow‐up were continuing to smoke.
Assessment of heterogeneity
To investigate statistical heterogeneity, we have used the I² statistic, given by the formula [(Q ‐ df)/Q] x 100%, where Q is the chi squared statistic and df is its degrees of freedom (Higgins 2003). This describes the percentage of the variability in effect estimates that is due to heterogeneity rather than to sampling error (chance).
A value greater than 50% may be considered to represent substantial heterogeneity. If heterogeneity were present, we would have perform a random‐effects meta‐analysis if we could not explain the heterogeneity by study characteristics.
Data synthesis
We estimate pooled treatment effects as risk ratios, using the Mantel‐Haenszel fixed‐effect model. Smoking cessation outcome data are reported as the number of quitters in each group divided by the number of participants receiving the treatment, i.e. the risk ratio with 95% confidence intervals. A ratio greater than 1 indicates that more people quit in the treatment group than in the control group. Measures of effective interventions appear to the right of the axis on the meta‐analysis graphs. We pooled the data provided that no significant heterogeneity between the trials was demonstrated.
We have included cluster‐randomized trials (with the therapist or site as the unit of allocation) in the meta‐analyses using patient‐level data.
Subgroup analysis and investigation of heterogeneity
Subgroups
In view of possible heterogeneity between studies, we analyzed the trials in the following subgroups:
-
Stratified by the type of counsellor delivering the intervention e.g. doctor, nurse, counsellor
-
Stratified by the intensity of the counselling, e.g. duration of each session and number of sessions
-
Stratified by the intensity of follow‐up support, usually by phone calls
-
Stratified by the type of control intervention
-
Stratified by the participants' motivation to quit, i.e. whether those recruited had mixed motivation to quit or whether they were already motivated to quit at baseline
Results
Description of studies
See Characteristics of included studies; Characteristics of excluded studies; Characteristics of ongoing studies.
Results of the search
The original literature search (April 2009) returned 691 references, with full‐text reports of 42 studies potentially relevant to the review screened. We updated the search in August 2014, and retrieved 310 references. We screened 31 full‐text reports. We excluded 12 studies for the reasons specified in the Characteristics of excluded studies table, and found that five of the studies are ongoing and will be assessed for eligibility as part of subsequent updates, following their completion.
Included studies
This review includes 28 studies, 14 of which were added in the most recent update (carried out in 2014) (Ellerbeck 2009; Lloyd‐Richardson 2009; Severson 2009; Tevyaw 2009; Wu 2009; De Azevedo 2010; Harris 2010; Davis 2011; Bastian 2013; Lindqvist 2013; Okuyemi 2013; Bock 2014; Louwagie 2014; Rohsenow 2014), including 16,803 tobacco users.
Recruitment and settings
All the trials, excluding five (Butler 1999 in the UK; Soria 2006 in Spain; De Azevedo 2010 in Brazil; Lindqvist 2013 in Sweden; Louwagie 2014 in South Africa) were conducted in the United States. Four were set in primary care clinics (Butler 1999; Soria 2006; Ellerbeck 2009; Bock 2014), one in participants' homes (Borrelli 2005), and three were delivered through telephone quitline services (Hollis 2007; Bastian 2013; Lindqvist 2013). Three programmes were provided through screening clinics ( McClure 2005; Wu 2009; Okuyemi 2013), six in specialist outpatient clinics (Glasgow 2000; Curry 2003; Hokanson 2006; Stein 2006; Lloyd‐Richardson 2009; Louwagie 2014), six in hospitals/inpatient settings (Rigotti 1997; Dornelas 2000; Hennrikus 2005; Bock 2008; De Azevedo 2010; Rohsenow 2014); three in university or laboratory settings (Tevyaw 2009; Harris 2010; Davis 2011) and two in military settings (Cigrang 2002; Severson 2009).
The two studies carried out in military settings (Cigrang 2002; Severson 2009) recruited mainly men using smokeless tobacco. Cigrang 2002 recruited only men on a Texan air force base and Severson 2009 recruited only one woman and 784 men. Three studies recruited only women: Glasgow 2000 recruited women who smoked, attending Planned Parenthood clinics. McClure 2005 recruited women who smoked and had an abnormal pap smear or colposcopy. Curry 2003 recruited women attending paediatric clinics. Six trials (Cigrang 2002; Hollis 2007; Wu 2009; Bastian 2013; Lindqvist 2013; Okuyemi 2013) recruited participants motivated to quit; however all other trials recruited participants without specific reference to their motivation to quit smoking.
Intervention
The most commonly‐used approach to motivational interviewing (MI) has been one in which the smoker is given feedback intended to develop discrepancy between smoking and personal goals in a non‐threatening manner (Butler 1999; McClure 2005; Hokanson 2006; Soria 2006). However, most studies merely specified that the intervention was carried out according to established MI techniques, as developed by Miller 2002. MI was delivered in face‐to‐face sessions in all the studies except for Cigrang 2002; McClure 2005; Hollis 2007; Ellerbeck 2009; Severson 2009; Bastian 2013 and Lindqvist 2013, in which the counselling was telephone‐based. None of the included studies used MI in groups. Sixteen studies delivered the MI intervention in a single session; four studies (Borrelli 2005; Soria 2006; Stein 2006; Tevyaw 2009) each provided three sessions, seven (McClure 2005; Ellerbeck 2009; Lloyd‐Richardson 2009; Wu 2009; Harris 2010; Bastian 2013; Okuyemi 2013) provided four or more sessions, and Lindqvist 2013 did not specify how many sessions were provided. The duration of sessions ranged from 10 to 60 minutes across studies.
Seventeen studies reported follow‐up telephone calls, ranging from one (Borrelli 2005; Ellerbeck 2009; Lloyd‐Richardson 2009; Davis 2011), to two, three or four (Rigotti 1997; Cigrang 2002; Glasgow 2000; Curry 2003; Hollis 2007; Bock 2008; Bock 2008; Severson 2009; Bock 2014; Rohsenow 2014), up to six (Hennrikus 2005; Hokanson 2006) or seven calls (Dornelas 2000; De Azevedo 2010). Where reported, the duration of the calls was typically around 10 minutes each.
Thirteen trials (Hennrikus 2005; Hokanson 2006; Soria 2006; Hollis 2007; Bock 2008; Ellerbeck 2009; Lloyd‐Richardson 2009; Wu 2009; Harris 2010; Bastian 2013; Okuyemi 2013; Bock 2014; Rohsenow 2014) either offered smoking cessation pharmacotherapies (NRT or bupropion) or encouraged their use. Apart from Hollis 2007, who included NRT patches as an intervention component in their factorial study design, the use and type of pharmacotherapy was not the intervention being tested.
MI interventions were compared in most studies to 'usual care' or brief advice (ranging from two to 15 minutes) for smoking cessation, often with self‐help manuals, booklets or videos; only five trials (Butler 1999; Soria 2006; Curry 2003;Tevyaw 2009; Harris 2010) did not offer smoking cessation support to any of the participants. Several trials also offered or referred control participants to standard smoking cessation services (Dornelas 2000; Cigrang 2002; Hennrikus 2005; Hokanson 2006; Hollis 2007; Lloyd‐Richardson 2009; Severson 2009) or to a phone counselling service (McClure 2005; Lindqvist 2013; Bock 2014).
Provider
MI was delivered by general practitioners (Butler 1999; Soria 2006), hospital physicians (Rigotti 1997; Curry 2003), nurses (Curry 2003; Borrelli 2005; Hennrikus 2005; Hokanson 2006; Davis 2011), or counsellors/psychologists (Glasgow 2000; Dornelas 2000; Curry 2003 ; McClure 2005; Hokanson 2006; Hollis 2007; Bock 2008; Cigrang 2002; Stein 2006; Ellerbeck 2009; Lloyd‐Richardson 2009; Severson 2009; Tevyaw 2009; Wu 2009; De Azevedo 2010; Harris 2010; Bastian 2013; Lindqvist 2013; Okuyemi 2013; Bock 2014; Louwagie 2014; Rohsenow 2014). Although hospital clinicians contributed to the counselling in at least two of the studies (Glasgow 2000; Curry 2003), they were never the main or only counsellor in any of the included trials.
Training of the provider
Details of therapist training in MI were provided in 25 studies (Butler 1999; Glasgow 2000; Curry 2003; Borrelli 2005; Hennrikus 2005; McClure 2005; Hokanson 2006; Soria 2006; Stein 2006; Hollis 2007; Bock 2008; Ellerbeck 2009; Lloyd‐Richardson 2009; Severson 2009; Tevyaw 2009; Wu 2009; De Azevedo 2010; Harris 2010; Davis 2011; Bastian 2013; Lindqvist 2013; Okuyemi 2013; Bock 2014; Louwagie 2014; Rohsenow 2014). The length of training in MI (where specified) ranged from two hours (Butler 1999) to 40 hours (Ellerbeck 2009; Tevyaw 2009; Wu 2009; Bastian 2013), and was usually in the form of workshops.
Description of the content of counselling delivered
All the studies included in this review made explicit reference to using MI principles laid down by Miller and Rollnick (Miller 2002). Details of counselling were reported in 21 studies. These included a full explanation of the main components and principles of MI, including the four guiding principles (Butler 1999; Glasgow 2000; Cigrang 2002; Curry 2003; Hennrikus 2005; McClure 2005; Hokanson 2006; Hollis 2007; Bock 2008; Lloyd‐Richardson 2009; Severson 2009; Tevyaw 2009; Wu 2009; De Azevedo 2010; Harris 2010; Davis 2011; Bastian 2013; Okuyemi 2013; Bock 2014; Louwagie 2014; Rohsenow 2014).
Outcomes
All but one of the trials (Severson 2009) reported point prevalence abstinence as a main outcome. The outcome data used for Davis 2011 in the meta‐analysis is point prevalence abstinence reported at both one month and six months (i.e. a cross between point prevalence and prolonged abstinence). We used this outcome, as for all other reported abstinence outcomes the manner of reporting made it impossible to tell to which time point the data referred (i.e. abstinence at one or six months). Five trials reported sustained abstinence at six months (Dornelas 2000; Cigrang 2002; Borrelli 2005; Bock 2008; Severson 2009), and five trials at 12 months (Dornelas 2000; Curry 2003; Borrelli 2005; McClure 2005; Lindqvist 2013). Sixteen trials reported biochemically‐validated abstinence rates, which were used in meta‐analyses (Rigotti 1997; Glasgow 2000; Curry 2003; Borrelli 2005; Hennrikus 2005; McClure 2005 [12 months only]; Hokanson 2006; Soria 2006; Stein 2006; Ellerbeck 2009; Lloyd‐Richardson 2009; Tevyaw 2009; Harris 2010; Okuyemi 2013; Bock 2014; Rohsenow 2014). Dornelas 2000 used the testimony of informants to confirm self‐reported abstinence, and Borrelli 2005 and Ellerbeck 2009 used a mixture of biochemical and testimony‐based validation. McClure 2005 used a modified 'bogus pipeline' for six‐month assessments, i.e. warning participants that they could be asked to provide a confirmatory sample for self‐reported abstinence, but not collecting it. Louwagie 2014 used a similar approach by only validating abstinence in a small sample, so that participants were aware that they could be tested. As the validated and self‐report outcomes produced the same results and validation only occurred in minimal participants, we have used self‐reported outcomes in this case. Bock 2008 and Wu 2009 reported collecting saliva cotinine samples and exhaled CO readings respectively for validation, but did not report validated data in a way which we could use for this meta‐analysis, and so we have used self‐reported outcomes.
Cost effectiveness
Two of the included studies offered an assessment of cost effectiveness. Hollis 2007 reported on an MI counselling quitline service based in Oregon, with and without nicotine replacement therapy. The cost of intensive telephone counselling per participant was USD 132 (2004 USD), with an incremental cost per quitter of USD 2640 (2004 USD), compared with brief advice. Butler 1999, comparing brief advice with an MI consultation delivered by UK general practitioners, calculated that the cost of training each physician in MI techniques was GBP 69.50, and the additional consultation time for each patient was GBP 13.59. However, the sustained quit rates achieved in this programme did not reach statistically significant levels.
We did not find sufficient evidence from the trials in our review to test our remaining hypotheses (MI effects are relatively long‐lasting; MI quitters have similar relapse rates; MI does not have any significant harmful effects).
Excluded studies
Some of the excluded trials had a short follow‐up, typically three months. Some did not use true motivational interviewing techniques, others delivered complex interventions from which the MI component could not be isolated, and some used motivational interviewing techniques in both trial arms. Several concentrated on adolescents who smoked, which we exclude from this review, and a number addressed multiple health behaviours, where the smoking outcomes could not be isolated for analysis. Excluded trials are listed in the table Characteristics of excluded studies, with reasons for their exclusion.
Risk of bias in included studies
Full details of 'Risk of bias' assessments are given for each trial within the Characteristics of included studies table. Overall summary results of all the 'Risk of bias' assessments are displayed in Figure 1.
Allocation
Fourteen studies did not describe their methods of sequence generation or allocation concealment, and are rated as 'unclear' for one or both of these domains. Six studies used sealed opaque envelopes(Butler 1999; Soria 2006; Ellerbeck 2009; De Azevedo 2010; Louwagie 2014; Rohsenow 2014). Three studies used methods of allocation rated in this review as inadequate, i.e. drawing random numbers from an envelope (Dornelas 2000), drawing coloured ping‐pong balls from a bag (Curry 2003) or counsellors were randomised to give an intervention via coin toss and participants received whichever treatment the first counsellor they spoke to provided (Lindqvist 2013). The remaining studies used block randomization procedures, with computerized lists or tables. Borrelli 2005 and Lindqvist 2013 randomized therapists rather than participants.
Blinding
Given the nature of the behavioural intervention, blinding of participants and intervention delivery was generally not feasible, which increased the potential risk of bias. However, 13 of the 28 studies reported some measure of blinded assessment of outcome measurement.
Other potential sources of bias
Validity of the intervention was maintained by audiotaping the counselling (Borrelli 2005; Hollis 2007; Bock 2008; Severson 2009; Davis 2011; Bastian 2013; Okuyemi 2013; Rohsenow 2014), by supervision throughout the study period (Glasgow 2000; Curry 2003; McClure 2005; Hokanson 2006; Tevyaw 2009; Wu 2009; Harris 2010; Bock 2014), by booster sessions throughout the study to maintain counselling skills (Borrelli 2005; Ellerbeck 2009; De Azevedo 2010; Louwagie 2014), or by regular meetings among therapists (Hennrikus 2005; De Azevedo 2010). Four studies (Rigotti 1997; Dornelas 2000; Cigrang 2002; Davis 2011) gave no details of training or measures to ensure treatment fidelity, although Davis 2011 did specify that this took place. Only one study (Stein 2006) reported using a validated instrument, i.e. the Motivational Interviewing Skill Code (MISC) to measure adherence to MI principles.
We have prepared a funnel plot of the included studies (Figure 2), which suggests that there may be some publication and/or reporting bias in favour of positive findings.
Methodological quality summary: review authors' judgements about each methodological quality item for each included study.
Funnel plot of comparison: 1 MI vs brief advice/usual care: all trials, outcome: 1.1 Smoking Cessation: longest duration and strictest definition of abstinence.
Effects of interventions
Motivational interviewing vs brief advice or usual care
The overall effect across all 28 included trials (N = 16,803), using the strictest definition of abstinence and longest follow‐up, gives a modestly significant effect (risk ratio (RR) 1.26; 95% confidence interval (CI) 1.16 to 1.36; Analysis 1.1). There was also moderate evidence of heterogeneity (I² = 49%). See Figure 3. Quit rates across the intervention groups ranged from 0% to 59.7%, with a weighted average of 16.9%. Control group quit rates ranged from 0% to 34.1%, with a weighted average of 14.2%.
Forest plot of comparison: 1 Motivational Interviewing vs brief advice/usual care, outcome: 1.1 All studies: longest duration and strictest definition of abstinence.
When pooling the 20 trials which reported point prevalence abstinence only, at a minimum of six months follow‐up, the effect was slightly lower than the main effect (RR 1.20; 95% CI 1.09 to 1.31; N = 13,692; I² = 54%), although still significant. There was a slightly higher effect when the eight studies that measured sustained abstinence at six months or longer were pooled (RR 1.62; 95% CI 1.32 to 2.00; N = 3111; I² = 0%) (analyses not shown).
The 16 trials which biochemically validated their outcomes delivered a lower risk ratio (1.12; 95% CI 0.98 to 1.29; N = 7858; I² = 29%), which did not reach significance (analyses not shown).
Comparison between therapists
In a subgroup analysis by type of therapist, MI delivered by general practitioners had a larger effect (RR 3.49; 95% CI 1.53 to 7.94: 2 trials, N = 736; I² = 27%; Analysis 1.2.1) when compared with nurses (RR 1.24; 95% CI 0.91 to 1.68; 5 trials, N = 2256; I² = 0%; Analysis 1.2.2) or counsellors (RR 1.25; 95% CI 1.15 to 1.36; 22 trials, N = 13,593; I² = 52%; Analysis 1.2.3). Curry 2003 used nurses and counsellors to deliver the intervention, and so appears in both analyses (not pooled).
Duration of session
Pooling studies in which the MI sessions lasted less than 20 minutes produced a significant, larger effect (RR 1.69; 95% CI 1.34 to 2.12; 9 trials, N = 3651; I² = 27%; Analysis 1.3.1). Studies with MI sessions lasting longer than 20 minutes produced a smaller effect (RR 1.20; 95% CI 1.08 to 1.32; 16 trials, N = 10,306; I² = 56%; Analysis 1.3.2).
Number of sessions
Interventions delivered in a single session (RR 1.26; 95% CI 1.15 to 1.40; 16 trials, N = 12,103; I² = 43%; Analysis 1.4.1) had a similar effect size to multiple session interventions (RR 1.20; 95% CI 1.02 to 1.42; 11 trials, N = 3928; I² = 56%; Analysis 1.4.2).
Number of follow‐up sessions
Subgroup analysis of studies by the number of follow‐up calls suggested an inverse relationship between the success of MI and amount of telephone follow‐up. The lowest risk ratio, and therefore the lowest MI quit rates, was associated with the higher number of follow‐up calls, indicating no incremental benefit of multiple calls. Studies with no follow‐up calls yielded a RR of 1.41 (95% CI 1.20 to 1.65; 10 trials, N = 3927; Analysis 1.5.1); however, this analysis demonstrated substantial heterogeneity, with an I² of 69% (P = 0.11), so should be viewed with caution. Studies offering one or two follow‐up calls had a RR of 1.28 (95% CI 1.05 to 1.55; 8 trials, N = 3895; I² = 53%; Analysis 1.5.2), while those offering three or more calls had a RR of 1.20 (95% CI 1.07 to 1.34; 8 trials, N = 8541; I² = 0%; Analysis 1.5.3).
Only Hollis 2007 tested for differences between offering follow‐up calls and no follow‐up support within a single trial. For our meta‐analyses we have combined the moderate and intensive intervention arms in that trial, to compare them with a brief advice intervention. We have also separately compared the moderate intervention (no follow‐up support) with the intensive intervention (up to four follow‐up calls), to quantify the value of the additional support calls. The RR was 1.05 (95% CI 0.89 to 1.23; N = 2874, analysis not shown), suggesting no added benefit for additional telephone support in this trial.
Face‐to‐face versus telephone
Seven of the trials (Cigrang 2002; McClure 2005; Hollis 2007; Ellerbeck 2009; Severson 2009; Bastian 2013; Lindqvist 2013) delivered their counselling by telephone only (N=7728), without any face‐to‐face contact. Subgroup analysis suggested that the risk ratio (1.27, 95% CI 1.12 to 1.43; N = 9075; I² = 51%) for face‐to‐face counselling trials only was almost the same as the main pooled effect (analyses not shown).
Comparison between control interventions
Control interventions were generally one of four types: 1) self‐help smoking cessation materials; 2) in‐person/telephone‐based smoking cessation support; 3) in‐person smoking health warning; 4) no smoking cessation intervention. When compared to self‐help materials or no smoking cessation intervention the effect of MI was non‐significant. The pooled self‐help control studies had a RR of 1.11 (0.91 to 1.35; 6 trials, N = 3502; I² = 0%; Analysis 1.6.1), and the pooled no‐smoking cessation control RR was 0.85 (0.61 to 1.19; 2 trials, N = 755; I² = 0%; Analysis 1.6.4). However, when compared to a control of in‐person/telephone smoking cessation support or an in‐person smoking health warning, MI for smoking cessation did show a significant benefit (RR 1.31; 95% CI 1.19 to 1.45; 17 trials, N = 10,966; I² = 54%; Analysis 1.6.2; and RR 2.25; 95% CI 1.41 to 3.57; 2 trials, N = 945; I² = 0%; Analysis 1.6.3 respectively).
Participants motivated to quit versus those with mixed motivation
Pooling trials which only recruited participants already motivated to make a quit attempt (Cigrang 2002; Hollis 2007; Wu 2009; Bastian 2013; Lindqvist 2013; Okuyemi 2013) yielded a similar effect size (RR 1.27; 95% CI 1.13 to 1.42; 6 trials, N = 6511; I² = 58%; Analysis 1.7.1) to the main pooled effect.
Type of tobacco user
This review includes studies of participants using smokeless tobacco or smoking cigarettes. Two of the 28 studies recruited only smokeless tobacco users. In this limited number of studies, MI produced a larger relative risk (RR 2.39; 95% CI 1.53 to 3.73; N = 845; I² = 0%; Analysis 1.8.1) than the pooled studies of cigarette smoking participants (RR 1.22; 95% CI 1.12 to 1.33; N = 15,958; I² = 44%; Analysis 1.8.2).
Incremental effects
Cigrang 2002 tested for an incremental effect of adding a self‐help manual and a supportive video to the initial counselling call. At six month follow‐up, 5/29 (17%) in the usual care (control) group had quit, compared with 3/11 (27%) for the MI counselling‐only group and 6/20 (30%) for the counselling plus additional materials group. Differences were not statistically significant. We have used the combined intervention group for the analyses throughout this review. Ellerbeck 2009 compared MI with two counselling calls every six months to MI with up to six counselling calls every six months. The odds ratio for this comparison at 24 month follow‐up was 1.33 (95% CI 0.88 to 2.02), indicating no effect of more intensive counselling. Rohsenow 2014 also compared more‐ with less‐intensive MI by comparing groups with and without two counselling booster sessions. They reported that "logistic regressions were nonsignificant for treatment or booster effects". For both of these studies the higher and lower intensity groups were combined into one MI intervention group for our analyses. Finally Tevyaw 2009 tested for an incremental effect of adding contingency reinforcement to MI, in the form of cash payments for reductions in smoking behaviour (participants earned on average USD 297.50). Although there was an incremental effect of this reinforcement during the intervention, no effect was found at follow‐ups. Again we combined the two MI groups for the purpose of our analyses.
Discussion
Summary of main results
The overall effect of MI compared with brief advice or usual care appears to be modest . This update, carried out in 2014/2015 resulted in the addition of 14 studies and altered the result very little, although confidence intervals narrowed slightly. This provides further confidence in the validity and precision of the main result. Certain components of interventions appear to enhance the efficacy of MI. There is some limited evidence in this review that MI interventions delivered by general practitioners confer greater benefit than those delivered by nurses or counsellors. Primary care doctors, counselling people with whom they are already familiar and have an established rapport, may be better suited to this approach. However, this finding is based on two relatively small studies, and should not be overstated. When delivered by nurses, the effect of MI was non‐significant, which may lend support to the findings of Rice 2013, who found that evidence for an effect of smoking cessation interventions delivered by nurses is weaker when their main role is not health promotion or smoking cessation. The question of the amount and intensity of therapist contact also presented an interesting result. The effect size associated with MI sessions of shorter duration (less than 20 minutes) appears to be higher than that associated with longer sessions, and delivering no follow‐up calls appear to be associated with a greater effect size than providing them. This is further supported by studies that compared more intensive with less intensive MI as part of their study design (Ellerbeck 2009; Rohsenow 2014), and found that less intensive support was associated with higher abstinence rates. One explanation for this could be that a single, short session of MI is enough to increase a person's motivation to quit smoking, and that by prolonging this and the time to the quit date participants may lose focus on their goal rather than further increasing their motivation.
Face‐to‐face counselling (with or without telephone follow‐up calls) was not associated with a greater effect than counselling delivered entirely by telephone, and either mode of delivery resulted in a superior effect to brief advice or usual care.
Training methods and duration of delivery of the MI counselling ranged from none to 40 hours, and monitoring of delivery and treatment fidelity was highly variable. Only one trial (Stein 2006) reported using a validated training tool (the Motivational Interviewing Skill Code). The included trials demonstrated a wide range of components and techniques for the delivery of MI, making direct comparisons across the trials problematic. This may explain the moderate amount of heterogeneity that was observed across the trials in the primary analysis, which could not be fully explained by the subgroup and sensitivity analyses conducted. It is unclear from these trials whether specific MI components or just the 'spirit' of MI is important, and whether short‐term achievements translate conclusively into long‐term abstinence. The question also remains whether the success rates in the intervention groups were attributable to MI techniques, or simply to a higher intensity intervention than that received by the control group. However, this does not appear to be the case, as when the studies were split according to the control intervention MI was not significantly superior to no or very minimal smoking cessation interventions, but was significantly more successful in comparison to more intensive, in‐person interventions. It is unclear why this may be the case.
The studies generally did not define what would have counted as a quit attempt, and did not report the proportion of participants who tried to quit, with or without success. The rate of quit attempts can be interpreted as a mediator of treatment effect, and the lack of such data limits the findings of this review.
Despite the positive findings of our meta‐analyses, absolute quit rates were relatively low. Most of the trials included participants unmotivated to quit smoking, although studies produced modest quit rates across the board, with some exceptions (Dornelas 2000; Hollis 2007; Wu 2009; De Azevedo 2010; Harris 2010; Lindqvist 2013; Louwagie 2014).
Although the evidence in this review suggests that MI techniques can deliver higher rates of smoking cessation than control, the effect size is somewhat lower than that demonstrated for individual counselling (RR 1.39; 95% CI 1.24 to 1.57, across 22 trials when compared to minimal control; Lancaster 2005), and significantly lower than for group behaviour therapy (RR 1.98; 95% CI 1.60 to 2.46, across 13 trials when compared to self‐help therapy; Stead 2005). Whether this discrepancy may be attributable to an unidentified cause of the observed heterogeneity between studies or to lower efficacy of MI techniques for smoking cessation, remains an open question. Based on a subgroup analysis, splitting participants according to their motivation to quit, it does not seem likely that low motivation to quit explains the moderate effect, as the risk ratio for motivated participants was very similar to the risk ratio for participants who were not recruited on the basis of their wanting to quit tobacco, with mixed motivation to quit. Another subgroup analysis did find that the effect of MI was significantly greater when used to help people to stop using smokeless tobacco in comparison to smoking tobacco. However, as the existing evidence is limited (only two studies of smokeless tobacco cessation which both took place in a military setting) this result should be treated with caution.
Quality of the evidence
The included studies in this review generally reported adequately on their design, methods and conduct. Only four trials reported inadequate methods of sequence generation or allocation concealment or both, which are held to be key determinants of selection bias (Schulz 2002a; Schulz 2002b). Fourteen of the 28 trials did not confirm blinding of outcome assessment. However, sensitivity analyses testing exclusion for these factors did not alter the review's findings in either case.
Confining the analyses to biochemically validated outcomes and to prolonged abstinence measures in sensitivity analyses reduced the effect size and increased it respectively. In the former case this could be because MI participants had developed a better rapport with their counsellor and so felt more pressure to report that they had remained abstinent, which then went on to be disproved by validation. In the latter case an explanation for a greater effect of MI in the studies measuring prolonged abstinence could be that participants may have gone on to make further quit attempts after relapsing in the control groups, so that point prevalence rates were higher than prolonged rates. Whereas participants using MI were more likely to be quit in the long‐term as a result of the original intervention.
Our funnel plot of the included studies (Figure 2) suggests a measure of publication bias or selective reporting or both, in favour of positive findings, which may compromise the strength of the evidence and the review's conclusions.
In conclusion, the summary of findings Table for the main comparison indicates that the quality of the evidence generated by this review has been assessed to be of moderate quality.
Agreements and disagreements with other studies or reviews
Two previous reviews of MI for smoking cessation (Heckman 2010; Hettema 2010) provide evidence of a very modest effect of MI at long‐term follow‐up (6 months or more). Our own effect estimates were also modest but slightly higher than these, providing evidence of a larger benefit of using MI for smoking cessation than the previous reviews. This may be because our inclusion criteria resulted in our including studies more likely to reflect true MI (it was necessary for studies to include some form of MI monitoring, such as training for providers or a measure of treatment fidelity, or both), and which had a minimal intervention or usual care as a control rather than another active intervention.
Hettema 2010 also found evidence of a dose‐response relationship, with some evidence to suggest that shorter administrations of MI were more likely to result in quitting than longer administrations, and both Heckman 2010 and Hettema 2010 also report evidence of a small amount of publication bias. Unlike this review, Hettema 2010 found that the baseline motivation of participants moderated the effect of MI. The studies included in their meta‐analyis which recruited participants with low motivation produced quit rates two to three times higher than those which recruited highly‐motivated participants.
Methodological quality summary: review authors' judgements about each methodological quality item for each included study.
Funnel plot of comparison: 1 MI vs brief advice/usual care: all trials, outcome: 1.1 Smoking Cessation: longest duration and strictest definition of abstinence.
Forest plot of comparison: 1 Motivational Interviewing vs brief advice/usual care, outcome: 1.1 All studies: longest duration and strictest definition of abstinence.
Comparison 1 Motivational Interviewing vs brief advice/usual care, Outcome 1 All studies: longest duration and strictest definition of abstinence.
Comparison 1 Motivational Interviewing vs brief advice/usual care, Outcome 2 By therapist.
Comparison 1 Motivational Interviewing vs brief advice/usual care, Outcome 3 By session duration.
Comparison 1 Motivational Interviewing vs brief advice/usual care, Outcome 4 By number of sessions.
Comparison 1 Motivational Interviewing vs brief advice/usual care, Outcome 5 By number of follow‐up calls.
Comparison 1 Motivational Interviewing vs brief advice/usual care, Outcome 6 By control intervention.
Comparison 1 Motivational Interviewing vs brief advice/usual care, Outcome 7 By participant motivation to quit.
Comparison 1 Motivational Interviewing vs brief advice/usual care, Outcome 8 By type of tobacco user.
| Motivational Interviewing compared to brief advice/usual care for smoking cessation | ||||||
| Patient or population: adult smokers | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| brief advice/usual care | Motivational Interviewing | |||||
| Longest duration and strictest definition of tobacco abstinence | Study population | RR 1.26 | 16,803 | ⊕⊕⊕⊝ | ||
| 104 per 1000 | 131 per 1000 | |||||
| *The assumed risk (e.g. the median control group risk across studies) is calculated based on the quit rates in control groups across all studies. The corresponding risk (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). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1Most studies at unclear risk of bias due to a lack of reporting. Very few at high risk of bias. No evidence of sensitivity of the effect due to high risk of bias, so decision made not to downgrade on risk of bias. 2Significant amount of heterogeneity (I² = 49%), which is not fully explained by any test for subgroup differences; however confidence intervals largely overlap, so decision made not to downgrade on basis of heterogeneity. 3Downgraded one level due to indication of possible publication bias: funnel plot indicates that less precise studies were more likely to show positive effects. | ||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 All studies: longest duration and strictest definition of abstinence Show forest plot | 28 | 16803 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.26 [1.16, 1.36] |
| 2 By therapist Show forest plot | 27 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 2.1 General practitioner | 2 | 736 | Risk Ratio (M‐H, Fixed, 95% CI) | 3.49 [1.53, 7.94] |
| 2.2 Nurse | 5 | 2256 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.24 [0.91, 1.68] |
| 2.3 Counsellor | 22 | 13593 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.25 [1.15, 1.36] |
| 3 By session duration Show forest plot | 25 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 3.1 Less than 20 minutes | 9 | 3651 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.69 [1.34, 2.12] |
| 3.2 More than 20 minutes | 16 | 10306 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.20 [1.08, 1.32] |
| 4 By number of sessions Show forest plot | 27 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 4.1 Single session | 16 | 12103 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.26 [1.15, 1.40] |
| 4.2 Two or more sessions | 11 | 3928 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.20 [1.02, 1.42] |
| 5 By number of follow‐up calls Show forest plot | 26 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 5.1 No follow‐up calls | 10 | 3927 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.41 [1.20, 1.65] |
| 5.2 One or two follow‐up calls | 8 | 3895 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.28 [1.05, 1.55] |
| 5.3 More than two follow‐up calls | 8 | 8541 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.20 [1.07, 1.34] |
| 6 By control intervention Show forest plot | 27 | 16168 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.27 [1.17, 1.38] |
| 6.1 Self‐help smoking cessation support | 6 | 3502 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.11 [0.91, 1.35] |
| 6.2 In person/telephone smoking cessation support | 17 | 10966 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.31 [1.19, 1.45] |
| 6.3 In person smoking health warning | 2 | 945 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.25 [1.41, 3.57] |
| 6.4 No smoking cessation intervention | 2 | 755 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.85 [0.61, 1.19] |
| 7 By participant motivation to quit Show forest plot | 28 | 16803 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.26 [1.16, 1.36] |
| 7.1 Motivated | 6 | 6511 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.27 [1.13, 1.42] |
| 7.2 Mixed motivation | 22 | 10292 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.25 [1.11, 1.40] |
| 8 By type of tobacco user Show forest plot | 28 | 16803 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.26 [1.16, 1.36] |
| 8.1 Smokeless | 2 | 845 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.39 [1.53, 3.73] |
| 8.2 Smoker | 26 | 15958 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.22 [1.12, 1.33] |
