Interventions for smokeless tobacco use cessation
Abstract
Background
Use of smokeless tobacco (ST) can lead to tobacco dependence and long‐term use can lead to health problems including periodontal disease, cancer, and cerebrovascular and cardiovascular disease.
Objectives
To assess the effects of behavioural and pharmacologic interventions for the treatment of ST use.
Search methods
We searched the Cochrane Tobacco Addiction Group specialised register in June 2015.
Selection criteria
Randomized trials of behavioural or pharmacological interventions to help users of ST to quit with follow‐up of at least six months.
Data collection and analysis
We used standard methodological procedures as expected by the Cochrane Collaboration. We summarised outcomes as risk ratios (RRs). For subgroups of trials with similar types of intervention and without substantial statistical heterogeneity, we estimated pooled effects using a Mantel‐Haenszel fixed‐effect method.
Main results
We identified 34 trials that met the inclusion criteria, of which nine were new for this update, representing over 16,000 participants. There was moderate quality evidence from two studies suggesting that varenicline increases ST abstinence rates (risk ratio [RR] 1.34, 95% confidence interval (CI) 1.08 to 1.68, 507 participants). Pooled results from two trials of bupropion did not detect a benefit of treatment at six months or longer (RR 0.89, 95% CI 0.54 to 1.44, 293 participants) but the confidence interval was wide. Neither nicotine patch (five trials, RR 1.13, 95% CI 0.93 to 1.37, 1083 participants) nor nicotine gum (two trials, RR 0.99, 95% CI 0.68 to 1.43, 310 participants) increased abstinence. Pooling five studies of nicotine lozenges did increase tobacco abstinence (RR 1.36, 95% CI 1.17 to 1.59, 1529 participants) but confidence in this estimate is low as the result is sensitive to the exclusion of three trials which did not use a placebo control.
Statistical heterogeneity was evident among the 17 trials of behavioural interventions: eight of them reported statistically and clinically significant benefits; six suggested benefit but with wide CIs and no statistical significance; and three had similar intervention and control quit rates and relatively narrow CIs. Heterogeneity was not explained by study design (individual or cluster randomization), whether participants were selected for interest in quitting, or specific intervention components. In a post hoc subgroup analysis, trials of behavioural interventions incorporating telephone support, with or without oral examination and feedback, were associated with larger effect sizes, but oral examination and feedback alone were not associated with benefit.
In one trial an interactive website increased abstinence more than a static website. One trial comparing immediate cessation using nicotine patch versus a reduction approach using either nicotine lozenge or brand switching showed greater success for the abrupt cessation group.
Authors' conclusions
Varenicline, nicotine lozenges and behavioural interventions may help ST users to quit. Confidence in results for nicotine lozenges is limited. Confidence in the size of effect from behavioural interventions is limited because the components of behavioural interventions that contribute to their impact are not clear.
PICOs
Plain language summary
Ways to help people stop using smokeless tobacco (including chewing tobacco, snuff and snus)
Background
Smokeless tobacco is any product in which tobacco is held in the mouth so that nicotine is absorbed through the lining of the mouth. Smokeless tobacco is less dangerous than cigarettes and other products where tobacco is burnt and nicotine absorbed through the lungs. However, smokeless tobacco still leads to nicotine addiction and can be harmful, especially to the mouth. Many types of smokeless tobacco are used around the world, including chewing tobacco, snuff and snus. The risks to health vary with the type of product.
Methods
We reviewed the evidence from randomized trials about interventions to help people stop using smokeless tobacco, including nicotine replacement therapy, other pharmacotherapies and behavioural support. This evidence is current to June 2015. Trials had to report the number of participants who had stopped using smokeless tobacco or other products after six months.
Results
We found 34 relevant trials covering over 16,000 participants. All except one were conducted in the USA. Some studies in dental health clinics provided advice about oral health problems to smokeless tobacco users whether or not they were interested in stopping. Some studies recruited users who wanted to stop.
Sixteen trials with 3,722 participants tested pharmacotherapies. Twelve studies tested different types of nicotine replacement therapy (five gum, two patch, five lozenge). The evidence suggests that the nicotine lozenge might help people quit, but the quality of evidence was low and more research is needed. There was not enough evidence to be sure whether nicotine gum or patches could help. Two trials of varenicline (a medication that helps smokers to quit) suggested it can also help people quit using smokeless tobacco.Two small trials of bupropion (an antidepressant that helps smokers to quit) did not find that bupropion helped people quit using smokeless tobacco.
Seventeen trials with 12,394 participants tested behavioural support. The behavioural support could include brief advice, self‐help materials, telephone support, access to a website, and combinations of elements. There was a lot of variation in results with some trials showing clear evidence of benefit and some not showing any effect. We could not be certain what the important elements of effective support were, but providing access to telephone support generally seemed to be helpful.
Authors' conclusions
Background
Smokeless tobacco (ST) is tobacco that is orally consumed and not burned. A variety of types of ST are consumed throughout the world and ST use is an important worldwide public health issue. In the United States, the principal types of ST are chewing tobacco (cut tobacco leaves) and snuff (moist ground tobacco). In Sweden, 'snus' (finely ground moist tobacco) is most commonly used. In India, ST contains tobacco leaf mixed with other ingredients, such as betel leaf, areca nut and lime (i.e., gutkha) (Critchley 2003). In Sudan, toombak is made from a fermented ground powdered tobacco mixed with sodium bicarbonate (Idris 1998).
Around the world, ST is used by 300 million people in at least 70 countries. The majority of smokeless tobacco users (89%) are in Southeast Asia (NCI & CDC 2014). In the US in 2012, 3.5% of individuals aged 12 or older (9 million people) used ST in the past month (SAMHSA 2014). Rates of past month ST use have remained stable between 2002 and 2012 in the U.S. In India, smokeless tobacco remains by far the most prevalent form of tobacco used (26% of population) (Kostova 2015). In 2013 in Sweden, 20% of men and 4% of women used ST daily and 3% and 1%, respectively, did so occasionally (Norberg 2015).
Available literature suggests that adverse health consequences may vary by the type of ST use, which is strongly associated with geography. According to the 1986 report of the US Surgeon General, the use of ST products can lead to nicotine addiction (NIH 1986). ST consumed in the US has been associated with periodontal disease (Ernster 1990; Fisher 2005), precancerous oral lesions (Mattson 1989), oral cancer (Stockwell 1986), and cancer of the kidney (Goodman 1986; Muscat 1995), pancreas (Muscat 1997), and digestive system (Henley 2005). ST has been shown to act as an autonomic and haemodynamic stimulus by increasing heart rate, blood pressure, and epinephrine levels (Wolk 2005), and has been associated with death from cardiovascular disease, cerebrovascular disease and cancer (Henley 2005). A recent systematic review concluded that betel quid and tobacco use in India are associated with substantial risks of oral cancer, but studies from the US and Scandinavia do not show a consistent association (Critchley 2003). Studies have suggested that ST use during pregnancy is likely to be harmful to the foetus (England 2003; Gupta 2004; Gupta 2006).
Two of the world's largest cigarette manufacturers, Phillip Morris USA and R.J. Reynolds, entered the ST market in the mid 2000s. Phillip Morris USA marketed Marlboro Snus and R.J. Reynolds marketed Camel Snus (Rogers 2010). These products were marketed as low‐nitrosamine ST products (Alpert 2008) which potentially confer a lower risk of cancer. At the same time, ST was increasingly being proposed as a harm reduction strategy for cigarette smokers (McNeill 2004; NIH 2006). Although the health risks of ST use are lower than those from smoked tobacco, concern existed that the promotion of ST use may lead to smokers using both products rather than quitting tobacco use altogether, and to former smokers and never smokers initiating ST use. The impact of these factors on the prevalence of ST use remains unclear, but suggests an ongoing need for developing effective treatments for ST use.
Despite the widespread use of ST products and their potentially adverse health consequences, medical and oral health professionals have had a lack of evidence summaries or evidence‐based guidelines to assist them in providing effective treatment for ST use. Smokeless tobacco cessation guidelines for health professionals in England were published after the first version of the present review was published in 2004 (West 2004). An evidence summary of ST interventions has also been published (NCI & CDC 2014).
Objectives
To assess the effects of behavioural and pharmacotherapeutic interventions to treat smokeless tobacco (ST) use.
Methods
Criteria for considering studies for this review
Types of studies
Randomized or pseudo‐randomized controlled trials allocating smokeless tobacco (ST) users to an intervention or control, or to different interventions. We also included trials in which dentists or other healthcare providers were randomized to provide intervention or control, and trials in which the unit of randomization was the school, workplace or institution.
Types of participants
Users of any tobacco product that is placed in the mouth and not burned, including moist snuff, chewing tobacco, Swedish snus, and Indian ST products (e.g. gutkha and pan masala). This does not include electronic cigarettes, which are covered in a separate Cochrane review (McRobbie 2014).
Types of interventions
Interventions could be pharmacological (i.e. nicotine replacement therapy (NRT), bupropion, varenicline) or behavioural, and could be directed at individual ST users or at groups of users (e.g. ST users visiting the dentist, attending school, or working). The control condition could be usual care, a placebo, or a less intensive intervention.
Types of outcome measures
The preferred outcome for the meta‐analysis was complete abstinence from all tobacco use six months or more after the start of the intervention. If total tobacco abstinence was not reported, abstinence from ST alone was used. Trials with shorter follow‐up (less than six months) or that did not report quit rates were excluded. Biochemical validation of self‐reported abstinence was not required, but validated rates were used where reported.
Search methods for identification of studies
For the most recent update we searched the Cochrane Tobacco Addiction Group specialised register in June 2015. At the time of the search the Register included the results of searches of: the Cochrane Central Register of Controlled trials (CENTRAL), issue 5, 2015; MEDLINE (via OVID) to update 20150501; EMBASE (via OVID) to week 201519; and PsycINFO (via OVID) to update 20150506. See the Tobacco Addiction Group Module in the Cochrane Library for full search strategies and list of other resources searched for the register. Additional sources were also searched for early versions of the review (Ebbert 2003); these included Web of Science, Dissertation Abstracts Online, Scopus, Healthstar, ERIC, National Technical Information Service database, and Current Contents.
The search strategy for the Tobacco Addiction Group specialised register used the following terms for smokeless tobacco: chewing tobacco; oral tobacco; spit tobacco; snuff; smokeless tobacco; quid; chew; plug; and tobacco, smokeless (MeSH), appearing in titles, abstracts or keywords. No intervention terms were used. No language restrictions were imposed.
We scanned the reference lists of retrieved studies including review articles, conference proceedings, and personal reference files. For early versions of the review we asked content experts through electronic mail and telephone contact to identify unpublished randomized controlled trials (RCTs). We corresponded with experts in tobacco and ST use research.
Data collection and analysis
Selection of studies
One author examined each title generated from the search and identified potentially eligible articles for which we obtained the abstracts. These were considered by two authors. For abstracts consistent with study eligibility, we obtained the full article text. Any difference of opinion about study inclusion would have been resolved by consensus.
Data extraction and management
Two authors independently extracted data about participants, interventions, outcomes and methodological quality. Any discrepancies in extracted data were resolved by consensus.
We extracted data on the number of users quit at the longest follow‐up, using the strictest definition of abstinence reported. We selected continuous or prolonged abstinence in preference to point prevalence where both were reported. Participants who were randomized but dropped out or were lost to follow‐up were assumed to be continuing users.
Assessment of risk of bias in included studies
We assessed the risk of selection bias. To be judged low risk for selection bias a trial had to report both an adequate method of random sequence generation, and of allocation sequence concealment. Studies reporting a method of sequence generation which did not allow allocation concealment (for example, allocation on the basis of patient record number) were judged to be at high risk of bias. Studies which did not report an acceptable method of allocation concealment, for example central enrolment and allocation, or consecutively numbered sealed opaque envelopes, were rated at high risk of bias. Studies which did not give sufficient detail to assess quality were rated unclear. We conducted a sensitivity analysis of the effect of including trials at high risk of selection bias in the meta‐analysis.
We also considered the completeness of follow‐up (attrition bias), judging risk of bias as low if more than 80% of participants provided data at follow‐up, unclear if the proportion reached was lower but similar in each condition, and at high risk of bias if there was evidence of differential loss by intervention condition. Other possible indicators of quality include: blinding status of participants, investigators and outcome assessors; group similarity at baseline; equal treatment of groups during study conduct; analysis and conduct by the intention‐to‐treat principle; and use of a placebo or active intervention in the control group (Guyatt 1993). We did not formally assess the impact of differences in these criteria on the results. In the table 'Characteristics of included studies' we noted the use of biochemical validation, and reported differences in baseline characteristics, any co‐interventions and the control intervention. If we were not able to extract data allowing an intention‐to‐treat analysis, this was recorded.
Measures of treatment effect
We use risk ratios (RRs) to represent the point estimate of the magnitude of association between intervention exposure and treatment outcomes, and 95% confidence intervals (CIs) to represent the precision around this point estimate. A RR greater than one indicates that the rates of tobacco abstinence were higher in the intervention group than in the control group. Earlier versions of the review used odds ratios because of the possibility that some cluster randomized trials would report adjusted odds ratios. We now use risk ratios as the majority of the included studies are individually randomized, risk ratios allow comparisons of effects with other Cochrane reviews, and are easier to interpret (Cochrane Handbook 9.2.2.2, Higgins 2011).
Data synthesis
We pooled results of studies when it was clinically and statistically appropriate to combine them. We did not combine pharmacotherapy and behavioural interventions. We conducted meta‐analyses using a fixed‐effect model, unless there was evidence of between‐study heterogeneity (Fleiss 1993). Heterogeneity was quantified using the I² statistic (Higgins 2003). This describes the percentage of the variability in effect estimates that is due to heterogeneity rather than sampling error (chance). Values over 50% suggest moderate heterogeneity. Where heterogeneity was higher than this we explored possible explanations, and did not report a pooled estimate of the effect.
For the pharmacological interventions, we hypothesized that nicotine replacement therapy (NRT) would lead to different outcomes compared with non‐NRT pharmacotherapies (i.e., bupropion, varenicline). Underlying this hypothesis is the difference in the mechanisms of action between different pharmacotherapies (Fiore 2000). Thus, we kept different pharmacotherapies in separate prespecified subgroups.
We also hypothesized that the behavioural interventions involving recruitment of individual ST users would be associated with higher abstinence rates for intervention compared to control than those recruiting ST users at the organizational level. This was based upon the presumption that ST users receiving interventions at the organizational level (e.g. dental practice or athletic teams) may receive interventions although they are not actively seeking treatment for ST use, which will potentially lead to lower abstinence rates in this group.
Results
Description of studies
The search of the Tobacco Addiction Group specialised register in June 2015 identified 12 new potentially relevant trials since the previous update in 2011.
Included studies
We identified 34 trials that met the inclusion criteria, of which nine were new for this update (Ebbert 2011; Schiller 2012; Danaher 2013; Ebbert 2013a; Ebbert 2013b; Danaher 2015a; Danaher 2015b; Severson 2015; Virtanen 2015). Sixteen of the trials assessed the effect of pharmacological interventions for ST use (Boyle 1992; Hatsukami 1996; Howard‐Pitney 1999; Hatsukami 2000; Dale 2002; Stotts 2003; Dale 2007; Ebbert 2007; Ebbert 2009; Ebbert 2010a; Ebbert 2011; Ebbert 2013b; Ebbert 2013a; Fagerstrom 2010; Danaher 2015b; Severson 2015) and 19 studied the effect of behavioural interventions for ST use (Cummings 1995; Stevens 1995; Hatsukami 1996; Severson 1998; Walsh 1999; Severson 2000; Cigrang 2002; Stotts 2003; Walsh 2003; Boyle 2004; Gansky 2005; Severson 2007; Boyle 2008; Severson 2008; Severson 2009; Walsh 2010; Danaher 2013; Danaher 2015a; Virtanen 2015). These totals include two studies that contribute data to both pharmacological and behavioural analyses; one study assessed both nicotine gum and a minimal contact or intensive behavioural intervention in a factorial design (Hatsukami 1996), and one compared a minimal intervention to an intensive behavioural intervention with either active or placebo nicotine patches (Stotts 2003). One study contributing to the pharmacological analysis compared a telephone counselling intervention and nicotine lozenges to the counselling alone; a third arm providing nicotine lozenges without support was not used in this analysis (Severson 2015). One study compared an immediate cessation versus a reduction approach for ST users without plans to quit (Schiller 2012) and was not pooled with other studies.
Pharmacological interventions
Sixteen randomized controlled trials (RCTs) randomized 3722 ST users to pharmacotherapy or control. The efficacy of bupropion SR (sustained‐release) given for 12 weeks was assessed in a pilot study (Dale 2002) and a multicenter trial (Dale 2007). Five studies assessed the efficacy of nicotine patch therapy (Howard‐Pitney 1999; Hatsukami 2000; Stotts 2003; Ebbert 2007; Ebbert 2013b), two studies assessed the efficacy of nicotine gum (Boyle 1992; Hatsukami 1996), five studies assessed the nicotine lozenge (Ebbert 2009; Ebbert 2010a; Ebbert 2013a; Danaher 2015b; Severson 2015), and two studies assessed the efficacy of varenicline (Fagerstrom 2010; Ebbert 2011).
Both the treatment and control groups received the same behavioural interventions. Brief individual counselling at clinic visits was provided in seven (Hatsukami 2000; Dale 2002; Dale 2007; Ebbert 2007; Ebbert 2009; Fagerstrom 2010; Ebbert 2011), pharmacist advice and telephone support in one (Howard‐Pitney 1999), a group programme in one (Boyle 1992), a six‐week group programme with additional telephone support in a trial in adolescents (Stotts 2003), brief counselling in a clinical research unit in one (Ebbert 2013b), a web‐based intervention in one (Danaher 2015b), and a self‐help book in addition to telephone counselling in two (Ebbert 2010a; Severson 2015). Two studies provided instructions on ST reduction (Ebbert 2013a; Virtanen 2015). One compared a group programme to a minimal contact condition in a factorial design (Hatsukami 1996). Hatsukami 2000 also tested mint snuff as an ST substitute in a factorial design; there was no evidence of a benefit, and these arms were collapsed in the analysis.
The bupropion SR studies used a dose of 150 mg by mouth once a day for three days and then increased the dose to 150 mg twice a day (Dale 2002; Dale 2007). One nicotine patch study used 15 mg patches for six weeks (Howard‐Pitney 1999); the second used 21 mg patches with a tapering schedule for a total of 10 weeks (Hatsukami 2000), and a third, in adolescents, tailored patch dose to baseline cotinine, using either 21 mg or 14 mg, both tapered over a six‐week period (Stotts 2003). The fourth nicotine patch study randomized participants to doses of 21, 42 and 63 mg per day compared to placebo, and the 21 mg and placebo arms were compared for analysis (Ebbert 2007). The fifth nicotine patch study randomized patients to 42 mg of the nicotine patch (two 21 mg patches worn simultaneously) for eight weeks or two matching placebo patches (Ebbert 2013b). One nicotine gum trial instructed enrolled ST users to attempt a target daily dose of 12 pieces of 2 mg nicotine gum per day (Boyle 1992). The other nicotine gum study instructed ST users to use at least six pieces of 2 mg nicotine gum a day for one month and then gradually reduce use (Hatsukami 1996). Four of the nicotine lozenge studies used the 4 mg lozenge given for 12 weeks with a tapering schedule (Ebbert 2009; Ebbert 2010a; Danaher 2015b; Severson 2015). One nicotine lozenge study provided 4 mg lozenges at eight per day for weeks one to six and tapered over 12 weeks (Ebbert 2013a). Varenicline was increased from 0.5 mg once daily for three days to 0.5 mg twice daily for four days followed by 1 mg twice daily through Week 12 in two studies (Fagerstrom 2010; Ebbert 2011).
Twelve studies followed patients for six months (Boyle 1992; Howard‐Pitney 1999; Dale 2002; Ebbert 2007; Ebbert 2009; Ebbert 2010a; Fagerstrom 2010; Ebbert 2011; Ebbert 2013a; Ebbert 2013b; Danaher 2015b; Severson 2015) and four for 12 months (Hatsukami 1996; Hatsukami 2000; Stotts 2003; Dale 2007). Five studies assessed continuous abstinence from quit date to longest follow‐up (Hatsukami 1996; Hatsukami 2000; Dale 2002; Dale 2007; Ebbert 2007) but one of them (Hatsukami 1996) did not tabulate that outcome, so point prevalence is used in the meta‐analysis. Four studies reported prolonged tobacco abstinence (Ebbert 2009; Ebbert 2010a; Ebbert 2011; Ebbert 2013b) defined as continuous tobacco abstinence after a two‐week grace period (Hughes 2003). Fagerstrom 2010 reported prolonged abstinence from weeks 9 to 26. Two studies reported repeated point prevalence at three and six months (Danaher 2015b; Severson 2015). The remaining studies only reported point prevalence quit rates at longest follow‐up (Boyle 1992; Howard‐Pitney 1999; Stotts 2003; Ebbert 2013a). All studies except two (Danaher 2015b; Severson 2015) used biochemical confirmation of self‐reported tobacco abstinence using tobacco alkaloid measurements (cotinine, anabasine, or anatabine). For studies determining abstinence from all tobacco products, carbon monoxide measurements and urinary anabasine and anatabine were used to determine abstinence from smoked tobacco. Three studies reported abstinence from smokeless tobacco only (Hatsukami 1996; Howard‐Pitney 1999; Hatsukami 2000). Since validation was also required, other forms of regular tobacco use would have been detected, but infrequent smokers might have been included as quitters.
Behavioural interventions
Seven RCTs randomized over 3000 ST users at the organizational level. Severson 1998 randomly allocated 75 dental practices to receive a workshop for their dental health professionals to develop skills in the identification and counselling of ST users or to provide usual care. Cummings 1995 analysed data from the Working Well Trial that randomized energy‐related worksites to receive either employee‐targeted intense interventions based upon the Social Learning Theory (Bandura 1986) and the Transtheoretical Model of Change (DiClemente 1998), or minimal interventions consisting of mailings and posters displayed in the workplace. Four of the organizational level trials were school‐based, of which three targeted athletes. A trial in college athletes (Walsh 1999) randomized college athletes at 16 campuses to receive either a behavioural intervention based upon the Health Belief Model (Rosenstock 1988) and the Social Learning Theory (Bandura 1986), or no intervention. A trial in high school athletes (Walsh 2003) randomized 44 schools to either an intervention that included oral screening, a peer‐led discussion, small group cessation counselling and a phone call on quit date, or to a control condition. A trial in college baseball athletes (Gansky 2005) randomized 52 colleges to an intervention based on the diffusion of innovation theory (Rogers 1983) and cognitive social learning theory which included a video conference, an oral‐cancer screening examination, a certified athletic trainer (ATC)‐facilitated discussion, and a peer‐led component. A trial in 41 rural public high schools (Walsh 2010) randomized to an intervention consisting of a peer‐led educational session, an oral examination, and three nurse‐led group cessation counselling sessions, or a control. Virtanen 2015 randomized Swedish dental clinics to delivering a structured tobacco use intervention based upon the 5 A's referring to the participants oral health and recommending pharmacotherapy but not providing it or to usual care. None of the studies randomized by organization selected ST users according to their motivation to quit.
Eleven RCTs randomized over 9000 ST users at the individual level. Stevens 1995 allocated ST users attending a routine dental visit to a multicomponent intervention consisting of feedback on oral lesions and advice to quit from both hygienist and dentist, as well as self‐help materials and a follow‐up call from a counsellor. The control group received usual care which may have included advice to quit. Participants were not selected according to motivation to quit. Two studies from the same research group assessed the impact of adding components to a minimal self‐help intervention (Severson 2000; Severson 2007). Severson 2000 tested a hand‐held device for programming gradual reduction, as an adjunct to self‐help materials and support. Due to problems with the prototype device, people whose machine failed twice or more were excluded from the reported analysis, and we have not included it in the meta‐analysis. Severson 2007 compared telephone support with self‐help written materials alone. Two studies assessed the efficacy of telephone‐based counselling for ST users compared to self‐help materials alone (Boyle 2004; Boyle 2008). A study in high school adolescents, also included in the pharmacotherapy section, randomized a behavioural intervention of six weekly group sessions with a health educator, plus stage‐based follow‐up telephone counselling (Stotts 2003). The control group had five to ten minutes of counselling and a single telephone call. A pilot study in personnel on active military service recruited self‐identified ST users at a health screening, unselected for motivation to quit. Members of the intervention group were telephoned and asked if they wished to receive self‐help materials and to have further support calls, using a motivational interviewing approach (Cigrang 2002). Based upon these promising preliminary results, a similar study was conducted with a larger sample of military recruits (Severson 2009). Two studies assessed the efficacy of a web‐based intervention randomising ST users to a basic or enhanced version (Severson 2008; Danaher 2013). One study randomized participants to a web‐based intervention, a telephone quitline intervention, web plus quitline, or a control with a printed self‐help guide (Danaher 2015a). One study randomized ST users who had no intention of quitting to immediate cessation or a reduction intervention (Schiller 2012). The immediate cessation group was offered two weeks of the nicotine patch and the reduction group was offered 4 mg nicotine lozenges or a different ST brand. This study compared pharmacotherapy‐assisted reduction to immediate cessation and was not included in the meta‐analysis.
Ten trials had final follow‐up at six months (Severson 2000; Cigrang 2002; Boyle 2004; Boyle 2008; Severson 2008; Severson 2009; Danaher 2013; Schiller 2012; Danaher 2015a; Virtanen 2015), seven at 12 months (Severson 1998; Stevens 1995; Walsh 1999; Stotts 2003; Walsh 2003; Gansky 2005; Walsh 2010), and one at two years (Cummings 1995). We used 12 month outcomes for one study that also had 18 month follow‐up, because loss to follow‐up had increased at the later time point (Severson 2007). All behavioural intervention studies assessed point prevalence abstinence. Seven reported only point prevalence abstinence at final follow‐up (Cummings 1995; Walsh 1999; Severson 2000; Stotts 2003; Gansky 2005; Severson 2009; Walsh 2010), and five required self‐reported point prevalence abstinence at both an interim and final follow‐up (Stevens 1995; Severson 1998; Cigrang 2002; Walsh 2003; Virtanen 2015). Four reported both point prevalence and repeated point prevalence (Severson 2007; Severson 2008; Danaher 2013; Danaher 2015a) and the repeated point prevalence was used for the meta‐analysis. Boyle 2008 reported both point prevalence and prolonged abstinence allowing for a 30‐day grace period and we used the latter in the meta‐analysis. Schiller 2012 reported prolonged and point prevalence abstinence. Stotts 2003 and Schiller 2012 reported using biochemical validation of self‐reported quitting. Stevens 1995 attempted to obtain saliva samples, but due to low compliance based the results on self report only. Walsh 1999 obtained samples but did not analyse them, as a method for increasing accuracy of self report. Eight reported smokeless tobacco cessation only (Cummings 1995; Walsh 1999; Severson 2000; Cigrang 2002; Walsh 2003; Gansky 2005; Severson 2009; Walsh 2010), six reported all tobacco use cessation (Severson 1998; Boyle 2004; Severson 2007; Boyle 2008; Schiller 2012; Danaher 2015a) and four reported both smokeless and all tobacco use cessation separately (Stevens 1995; Stotts 2003; Severson 2008; Danaher 2013). The results of the meta‐analysis are not affected by choice of outcome in these trials, although quit rates were lower for all tobacco use than for ST alone.
Excluded studies
Sixteen studies are listed as excluded, of which three were new for this update (Gordon 2010; Jain 2014; Raja 2014). Most were not eligible due to short length of follow‐up. Details are given in Characteristics of excluded studies.
One ongoing study was identified (Sarkar 2014).
Risk of bias in included studies
Pharmacological interventions
None of the sixteen randomized trials of pharmacological interventions were assessed as being at high risk of selection bias although some had insufficient information on randomization and allocation procedures and the potential for bias was unclear. Thirteen trials used a placebo control, two just provided the same behavioural support to the control (Danaher 2015a; Severson 2015), and one provided nicotine free snuff (Ebbert 2013b). Four studies assessed the efficacy of the blinding procedure by having participants guess their treatment assignment, suggesting that blinding was adequate in two (Dale 2007; Ebbert 2009), and inadequate in another (Ebbert 2007), while the fourth did not report the results (Hatsukami 2000). No studies reported high and differential levels of loss to follow‐up.
Behavioural interventions
One study did not use an appropriate method of allocation concealment (Stevens 1995). Eligibility was assessed by a receptionist on the basis of a questionnaire given to all clinic attendees, with allocation on the basis of clinic record number. This method has the potential for selection bias, although allocation was not conducted by the person providing the intervention. We tested the sensitivity of the results to the inclusion of this study. In one cluster randomized trial (cRCT) (Walsh 2010) it was unclear whether individual participants were identified before or after the school status was revealed but there was no evidence of an imbalance in baseline characteristics. This study also reported high loss to follow‐up and results are based only on participants reached at follow‐up. In a second cRCT in worksites only participants reached at two‐year follow‐up were included (Cummings 1995).
Across the behavioural studies, no co‐interventions were apparent except for one RCT in which the intervention group was offered nicotine gum, although less than 10% of participants reportedly used it (Walsh 1999).
Randomization at the organizational level and analysis of outcomes at the individual level may lead to errors in estimated confidence intervals (Altman 1997). All the studies using cluster randomization used appropriate methods of analysis and reporting, using cluster level averages (Cummings 1995; Walsh 1999), odds ratios adjusted for clustered responses (Gansky 2005; Walsh 2003), or reported low levels of intraclass correlation and non‐significant practice effects (Stevens 1995).
Effects of interventions
Pharmacological interventions
Bupropion
The two bupropion studies with six months or longer follow‐up (Dale 2002; Dale 2007) showed no effect on continuous all‐tobacco abstinence, though the confidence interval was wide (293 participants, risk ratio (RR) 0.89, 95% CI 0.54 to 1.44, I² = 0%, Analysis 1.1).
Nicotine replacement therapy (NRT)
We did not find evidence of heterogeneity within subgroups based on type of NRT. At six months or longer, neither nicotine patch (five trials, 1083 participants, RR 1.13, 95% CI 0.93 to 1.37, I² = 14%) nor nicotine gum (two trials, 310 participants, RR 0.99, 95% CI 0.68 to 1.43, I² = 0%) increased tobacco abstinence rates. For the study that randomized patients to three different doses of nicotine patches (Ebbert 2007), we used the comparison between the 21 mg patch and placebo. In the trial of nicotine patch for adolescent ST users (Stotts 2003) the quit rates were twice as high in the placebo group, although the difference did not reach statistical significance. Pooled results showed the nicotine lozenge increased tobacco abstinence rates (five trials, 1529 participants, RR 1.36, 95% CI 1.17 to 1.59, I² = 0%). However, three of the nicotine lozenge trials did not use a placebo control (Ebbert 2013b; Severson 2015; Danaher 2015b) and in a post hoc sensitivity analysis the result was sensitive to the removal of these three trials. In Severson 2015, we compared the nicotine lozenge plus coaching calls to the coaching calls alone, and the nicotine lozenge‐only arm did not contribute to the comparison.
Pooling all twelve trials with a total of 2922 participants, nicotine replacement therapy increased tobacco abstinence rates (RR 1.24, 95% CI 1.11 to 1.39, I² = 6%, Analysis 2.1), but again this result was no longer significant when the three lozenge trials without placebo controls were removed.
Varenicline
Two trials of varenicline with 507 participants (Fagerstrom 2010; Ebbert 2011) increased tobacco abstinence rates at six months compared to placebo (RR 1.34, 95% CI 1.08 to 1.68, Analysis 3.1). There was no evidence of heterogeneity (I² = 0%).
Behavioural interventions
There was evidence of considerable heterogeneity among the 17 trials eligible for the meta‐analysis (I² = 78%, Analysis 4.1). Excluding the trial that used a potentially biased method for treatment allocation (Stevens 1995) did not affect this. Eight of the trials showed a significant effect of behavioural intervention (Severson 1998; Walsh 1999; Walsh 2003; Boyle 2004; Boyle 2008; Severson 2008; Severson 2009; Danaher 2015a), in six the confidence intervals did not rule out a clinical benefit but did not exclude one (Stevens 1995; Cigrang 2002; Stotts 2003; Severson 2007; Walsh 2010; Virtanen 2015) and three had risk ratios just below or above one, and relatively narrow confidence intervals suggesting no important benefit or harm (Cummings 1995; Gansky 2005; Danaher 2013).
Our prespecified subgroup analysis based on study design did not reduce heterogeneity (Figure 1, Analysis 4.1). Amongst the ten studies randomising individuals the I² value was 75%. In this group of studies, five reported significant treatment effects (Boyle 2004; Boyle 2008; Severson 2008; Severson 2009; Danaher 2015a), and the other five had point estimates ranging from RR 1.07 to RR 2.18 (Stevens 1995; Cigrang 2002; Stotts 2003; Severson 2007; Danaher 2013). The largest trial, Severson 2008, reported an RR of 1.59 (95% CI 1.26 to 2.02). Overall these trials suggest a benefit of behavioural interventions, but the larger trials show smaller effects than the smaller trials, and a pooled estimate, whether fixed‐effect or random effect, risks overestimating the benefit.
Behavioural interventions: Abstinence from all tobacco use (where reported) at 6 months or more.
Among the seven trials that randomized by organization the I² value was 79%. In this subgroup three trials detected large and statistically significant effects, with RRs over two (Severson 1998; Walsh 1999; Walsh 2003).
Since the distinction between individual and cluster designs was based on expectations about the level of motivation of participants, we also considered this factor directly (Analysis 4.2). All the clustered RCTs enrolled unselected participants, but Stevens 1995, Cigrang 2002, and Severson 2009 also recruited any ST user without assessing interest in quitting. Statistical heterogeneity persisted in both subgroups, and there was no evidence that effects were larger in the trials in more motivated populations.
A sensitivity analysis preferring ST abstinence over all tobacco abstinence where trials reported both outcomes did not affect heterogeneity or alter the findings (Analysis 4.7) .
In two further subgroup analyses we considered whether treatment effect might be moderated by including an oral examination and feedback (Analysis 4.3) or telephone support (Analysis 4.4) as intervention components. Intervention characteristics and study design tended to be correlated as Table 1 shows. Most individually randomized studies did not include an oral examination but did include telephone support, whilst cRCTs typically involved oral examination with some also including telephone support. Heterogeneity remained after grouping the 17 trials according to whether or not the intervention included an oral examination component with direct feedback to patients regarding ST‐induced mucosal changes (Analysis 4.3). Amongst the six trials including an oral examination the I² was 80%, with the largest trial, Gansky 2005, showing the smallest effect. Gansky and colleagues suggested that the lack of effect in their trial could have been due to a 'spill‐over' effect due to contact between the athletic trainers in the different groups. Although three of the trials did show significant effects (Severson 1998; Walsh 1999; Walsh 2003), conclusions about the effect of oral examinations have to be cautious. There was also substantial heterogeneity (I² = 72%) among the eleven studies without an oral examination component (Cummings 1995; Cigrang 2002; Stotts 2003; Boyle 2004; Boyle 2008; Severson 2007; Severson 2008; Severson 2009; Danaher 2013; Danaher 2015a; Virtanen 2015).
| Study | Design | Selection? | Oral exam? | Telephone support? | Setting | Control |
| RCT | Motivated | No oral exam | Phone support | Community | S‐H only | |
| RCT | Motivated | No oral exam | Phone support | Community | S‐H only | |
| RCT | Motivated | No oral exam | Both phone & no phone arms | Community | S‐H only | |
| RCT | Motivated | No oral exam | Phone support | Community | S‐H only | |
| RCT | Motivated | No oral exam | Phone in both | High School | Brief Intervention | |
| RCT | Motivated | No oral exam | No phone | Community | Basic website | |
| RCT | Motivated | No oral exam | No phone | Community | Basic website | |
| RCT | Unselected | No oral exam | Phone support | Military | UC | |
| RCT | Unselected | No oral exam | Phone support | Military | UC | |
| RCT | Unselected | Oral exam & feedback | Phone support | Dental | UC | |
| cRCT | Unselected | Oral exam & feedback | No phone | College | UC | |
| cRCT | Unselected | Oral exam & feedback | Phone support | Dental | UC | |
| cRCT | Unselected | No oral exam | No phone | Dental | UC | |
| cRCT | Unselected | Oral exam & feedback | Phone support | College | Oral exam no feedback | |
| cRCT | Unselected | Oral exam & feedback | Phone support | High School | No intervention | |
| cRCT | Unselected | Oral exam & feedback | No phone | High School | No intervention | |
| cRCT | Unselected | No oral exam | No phone | Workplaces | No intervention |
In the telephone support subgroup analysis there were ten studies in which telephone support formed part of the intervention (Stevens 1995; Severson 1998; Walsh 1999; Cigrang 2002; Walsh 2003; Boyle 2004; Boyle 2008; Severson 2007; Severson 2009; Danaher 2015a (quitline intervention arms)) and seven where it did not (Cummings 1995; Gansky 2005; Severson 2008; Walsh 2010; Danaher 2013; Danaher 2015a (web only arm); Virtanen 2015). A trial where brief phone support was included in the control condition but not the intervention (Stotts 2003) was not included. Heterogeneity within the telephone support subgroup was moderate as opposed to considerable (I² = 50%) and the pooled risk ratio indicated benefit (3480 participants, RR 1.77, 95% CI 1.57 to 2.00, Analysis 4.4). Heterogeneity was substantial in the subgroup of seven trials of interventions without telephone support (I² = 58%), which included one study showing evidence of benefit (Severson 2008). A second study comparing similar intervention and control conditions did not replicate this effect (Danaher 2013). The pooled estimate for this subgroup suggested only a small benefit with the CI excluding 1 narrowly (6611 participants, RR 1.16, 95% CI 1.05 to 1.28).
In this update we added a further exploratory subgroup analysis combining the oral examination and telephone components (Analysis 4.5). This suggested that the combination of oral examination and telephone support was consistently beneficial (4 studies, 1818 participants, RR 2.07, 95% CI 1.61 to 2.66, I2 = 0%), whereas oral examination alone did not show evidence of benefit (RR 1.01, 95% CI 0.86 to 1.19). The estimated effect for telephone support without oral exam was slightly smaller, and less consistent than for the combination of components (7 studies, 3965 participants, RR 1.66, 95% CI 1.45 to 1.91, I2 =57%) but there was not a significant difference between these two subgroups. The estimated effect of interventions without either component was smaller, and uncertain because of heterogeneity (5 studies, 5728 participants, RR 1.22, 95% CI 1.08 to 1.39, I2 = 64%).
One further behavioural study was not included in the meta‐analysis because two active interventions were compared; in this study technical problems with the device for scheduling gradual cessation led to a high drop out rate in that condition and the intention‐to‐treat analysis was not used. No significant difference was detected between the conditions (Severson 2000). At six months, the self‐reported ST abstinence rate was 27.6% (21/76) in the hand‐held device group and 30.2% (29/96) in the manual and video group.
One trial (Hatsukami 1996) failed to detect a difference between more intense and less intense behavioural interventions in a 2x2 study of nicotine gum and behavioural interventions (RR 1.34, 95% CI 0.84 to 2.12, Analysis 4.6).
One trial recruiting ST users without plans to quit and which compared immediate cessation using nicotine patch versus a reduction approach using either nicotine lozenge or brand switching (Schiller 2012) showed greater success for the abrupt cessation group (11/97 vs 1/102, RR 11.57, 95% CI 1.52 to 87.91, Analysis 5.1).
Adverse events
No effort was made to perform a quantitative synthesis of the incidence of adverse events reported with the different interventions. One study reported a higher rate of skin reactions and nausea associated with the nicotine patch, but found no difference in the number of people who stopped treatment due to side effects (Howard‐Pitney 1999). One study reported the loss of two subjects due to headache and gastro‐intestinal distress associated with nicotine gum use (Boyle 1992). Sleep disturbance was more common among patients on active bupropion SR (Dale 2007). Nausea occurred in more than one‐third of patients in one varenicline study (Fagerstrom 2010) and in 24% in the other (Ebbert 2011).
Discussion
This systematic review provides evidence from 34 randomized controlled trials enrolling more than 16,000 smokeless tobacco (ST) users, testing pharmacological and behavioural interventions to treat ST use.
Pharmacotherapies
There were 16 trials evaluating pharmacotherapy. Two small trials of bupropion did not detect an effect although confidence intervals do not rule out a small benefit. Twelve trials of NRT including gum, patch and lozenge suggested a statistically significant treatment effect, which appears to be driven by the efficacy of the nicotine lozenge. However, the lozenge subgroup meta‐analysis included three studies without a placebo arm and a post hoc analysis found the results were sensitive to the removal of these three trials. Despite the absence of heterogeneity between the different types of NRT, we do not think that there is evidence to support the use of nicotine gum or patch. Two studies in Scandanavian and U.S. populations demonstrated that varenicline increases long term ST abstinence rates by 34% compared to placebo among ST users. In cigarette smokers, however, varenicline increases abstinence rates 131% compared to placebo (RR 2.31, 95% CI 2.01 to 2.66) (Cahill 2012). However, the prolonged abstinence rates in the control group in the ST studies were higher at six months (31.6% (Ebbert 2011) and 34% (Fagerstrom 2010)) than in studies of smokers (e.g. 13.2% (Jorenby 2006) and 10.5% (Gonzales 2006)). This may relate to the low availability of treatment for ST users resulting in high efficacy of behavioral interventions provided in the control arms of these studies.
Behavioural interventions
We found evidence of heterogeneity among the behavioural interventions, with some trials showing a statistically and clinically significant effect, some with non‐significant increases in intervention arms and three with very similar intervention and control quit rates and relatively narrow confidence intervals (Cummings 1995; Gansky 2005; Danaher 2013). In seeking to explain the heterogeneity we considered subgroups based on trial design and intervention characteristics. These included whether or not the studies were individually randomized, or recruited only participants motivated to quit, or whether the intervention included an oral examination or telephone support. Categorization by use of telephone support had lower levels of subgroup heterogeneity, but this was a post hoc analysis. In the earliest version of this review (Ebbert 2004) we suggested that interventions including oral examination and feedback were more effective. In the current review, this observation is not made.
The inference of the effect size of behavioural interventions for increasing ST abstinence rates is weakened by the limited methodological quality of some of these trials, including loss to follow‐up and potential baseline differences between the groups. We cannot exclude the possibility that publication bias is also impacting on our results.
Behavioural interventions: Abstinence from all tobacco use (where reported) at 6 months or more.
Comparison 1 Pharmacotherapy: Buproprion versus placebo, Outcome 1 All tobacco abstinence at longest follow‐up.
Comparison 2 Pharmocotherapy: NRT versus placebo/no placebo/control, Outcome 1 6 months or greater abstinence, strictest criteria.
Comparison 3 Pharmacotherapy: Varenicline versus placebo, Outcome 1 All tobacco abstinence at 6 months.
Comparison 4 Behavioural interventions, Outcome 1 Abstinence from all tobacco use (where reported) at 6 months or more.
Comparison 4 Behavioural interventions, Outcome 2 Subgroup analysis: Motivation.
Comparison 4 Behavioural interventions, Outcome 3 Subgroup analysis: Use of oral examination and feedback.
Comparison 4 Behavioural interventions, Outcome 4 Subgroup analysis: Use of telephone support.
Comparison 4 Behavioural interventions, Outcome 5 Subgroup analysis: Combined oral examination and telephone.
Comparison 4 Behavioural interventions, Outcome 6 Behavioural intervention +/‐ pharmacotherapy versus minimal contact. Long term cessation.
Comparison 4 Behavioural interventions, Outcome 7 Sensitivity analysis: Abstinence from smokeless tobacco use (where reported) at 6 months or more.
Comparison 4 Behavioural interventions, Outcome 8 Inverse variance sensitivity Abstinence from all tobacco use (where reported) at 6 months or more.
Comparison 5 Abrupt cessation versus gradual reduction (using NRT), Outcome 1 6 months or greater abstinence, strictest criteria.
| Study | Design | Selection? | Oral exam? | Telephone support? | Setting | Control |
| RCT | Motivated | No oral exam | Phone support | Community | S‐H only | |
| RCT | Motivated | No oral exam | Phone support | Community | S‐H only | |
| RCT | Motivated | No oral exam | Both phone & no phone arms | Community | S‐H only | |
| RCT | Motivated | No oral exam | Phone support | Community | S‐H only | |
| RCT | Motivated | No oral exam | Phone in both | High School | Brief Intervention | |
| RCT | Motivated | No oral exam | No phone | Community | Basic website | |
| RCT | Motivated | No oral exam | No phone | Community | Basic website | |
| RCT | Unselected | No oral exam | Phone support | Military | UC | |
| RCT | Unselected | No oral exam | Phone support | Military | UC | |
| RCT | Unselected | Oral exam & feedback | Phone support | Dental | UC | |
| cRCT | Unselected | Oral exam & feedback | No phone | College | UC | |
| cRCT | Unselected | Oral exam & feedback | Phone support | Dental | UC | |
| cRCT | Unselected | No oral exam | No phone | Dental | UC | |
| cRCT | Unselected | Oral exam & feedback | Phone support | College | Oral exam no feedback | |
| cRCT | Unselected | Oral exam & feedback | Phone support | High School | No intervention | |
| cRCT | Unselected | Oral exam & feedback | No phone | High School | No intervention | |
| cRCT | Unselected | No oral exam | No phone | Workplaces | No intervention |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 All tobacco abstinence at longest follow‐up Show forest plot | 2 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 1.1 6 months or greater continuous abstinence | 2 | 293 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.89 [0.54, 1.44] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 6 months or greater abstinence, strictest criteria Show forest plot | 12 | 2922 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.24 [1.11, 1.39] |
| 1.1 Nicotine Patch | 5 | 1083 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.13 [0.93, 1.37] |
| 1.2 Nicotine Gum | 2 | 310 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.99 [0.68, 1.43] |
| 1.3 Nicotine lozenge | 5 | 1529 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.36 [1.17, 1.59] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 All tobacco abstinence at 6 months Show forest plot | 2 | 507 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.34 [1.08, 1.68] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Abstinence from all tobacco use (where reported) at 6 months or more Show forest plot | 17 | Risk Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 1.1 Individual randomisation | 10 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 1.2 Randomisation by organisation | 7 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 2 Subgroup analysis: Motivation Show forest plot | 17 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 2.1 Motivated | 7 | 7921 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.39 [1.25, 1.55] |
| 2.2 Not selected by motivation | 10 | 4473 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.37 [1.23, 1.53] |
| 3 Subgroup analysis: Use of oral examination and feedback Show forest plot | 17 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 3.1 Intervention included oral examination and feedback | 6 | 2701 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.34 [1.17, 1.53] |
| 3.2 Oral examination not part of the intervention | 11 | 9693 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.40 [1.28, 1.54] |
| 4 Subgroup analysis: Use of telephone support Show forest plot | 17 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 4.1 Telephone support for intervention, not for control | 10 | 5480 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.77 [1.57, 2.00] |
| 4.2 No telephone support for either condition | 7 | 6611 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.16 [1.05, 1.28] |
| 4.3 Telephone support for control group only | 1 | 303 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.26 [0.57, 2.78] |
| 5 Subgroup analysis: Combined oral examination and telephone Show forest plot | 17 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 5.1 Oral exam plus telephone | 4 | 1818 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.07 [1.61, 2.66] |
| 5.2 Oral exam, no telephone | 2 | 883 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.01 [0.86, 1.19] |
| 5.3 Telephone, no oral exam | 7 | 3965 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.66 [1.45, 1.91] |
| 5.4 No oral exam, no telephone | 5 | 5728 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.22 [1.08, 1.39] |
| 6 Behavioural intervention +/‐ pharmacotherapy versus minimal contact. Long term cessation Show forest plot | 1 | 210 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.34 [0.84, 2.12] |
| 6.1 Nicotine gum | 1 | 106 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.96 [0.98, 3.92] |
| 6.2 Placebo gum | 1 | 104 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.94 [0.50, 1.77] |
| 7 Sensitivity analysis: Abstinence from smokeless tobacco use (where reported) at 6 months or more Show forest plot | 17 | Risk Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 7.1 All tobacco use | 7 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 7.2 Smokeless tobacco use | 10 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 8 Inverse variance sensitivity Abstinence from all tobacco use (where reported) at 6 months or more Show forest plot | 17 | 15504 | Odds Ratio (Fixed, 95% CI) | 1.46 [1.33, 1.59] |
| 8.1 Individual randomisation | 10 | 9284 | Odds Ratio (Fixed, 95% CI) | 1.58 [1.40, 1.79] |
| 8.2 Randomisation by organisation | 7 | 3110 | Odds Ratio (Fixed, 95% CI) | 1.36 [1.14, 1.61] |
| 8.3 Walsh lower OR Randomisation by organisation | 7 | 3110 | Odds Ratio (Fixed, 95% CI) | 1.33 [1.12, 1.58] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 6 months or greater abstinence, strictest criteria Show forest plot | 1 | 199 | Risk Ratio (M‐H, Fixed, 95% CI) | 11.57 [1.52, 87.91] |
