Mobile phone‐based interventions for smoking cessation
Abstract
Background
Access to mobile phones continues to increase exponentially globally, outstripping access to fixed telephone lines, fixed computers and the Internet. Mobile phones are an appropriate and effective option for the delivery of smoking cessation support in some contexts. This review updates the evidence on the effectiveness of mobile phone‐based smoking cessation interventions.
Objectives
To determine whether mobile phone‐based smoking cessation interventions increase smoking cessation in people who smoke and want to quit.
Search methods
For the most recent update, we searched the Cochrane Tobacco Addiction Group Specialised Register in April 2015. We also searched the UK Clinical Research Network Portfolio for current projects in the UK, and the ClinicalTrials.gov register for ongoing or recently completed studies. We searched through the reference lists of identified studies and attempted to contact the authors of ongoing studies. We applied no restrictions on language or publication date.
Selection criteria
We included randomised or quasi‐randomised trials. Participants were smokers of any age who wanted to quit. Studies were those examining any type of mobile phone‐based intervention for smoking cessation. This included any intervention aimed at mobile phone users, based around delivery via mobile phone, and using any functions or applications that can be used or sent via a mobile phone.
Data collection and analysis
Review authors extracted information on risk of bias and methodological details using a standardised form. We considered participants who dropped out of the trials or were lost to follow‐up to be smoking. We calculated risk ratios (RR) and 95% confidence intervals (CI) for each included study. Meta‐analysis of the included studies used the Mantel‐Haenszel fixed‐effect method. Where meta‐analysis was not possible, we presented a narrative summary and descriptive statistics.
Main results
This updated search identified 12 studies with six‐month smoking cessation outcomes, including seven studies completed since the previous review. The interventions were predominantly text messaging‐based, although several paired text messaging with in‐person visits or initial assessments. Two studies gave pre‐paid mobile phones to low‐income human immunodeficiency virus (HIV)‐positive populations ‐ one solely for phone counselling, the other also included text messaging. One study used text messages to link to video messages. Control programmes varied widely. Studies were pooled according to outcomes ‐ some providing measures of continuous abstinence or repeated measures of point prevalence; others only providing 7‐day point prevalence abstinence. All 12 studies pooled using their most rigorous 26‐week measures of abstinence provided an RR of 1.67 (95% CI 1.46 to 1.90; I2 = 59%). Six studies verified quitting biochemically at six months (RR 1.83; 95% CI 1.54 to 2.19).
Authors' conclusions
The current evidence supports a beneficial impact of mobile phone‐based smoking cessation interventions on six‐month cessation outcomes. While all studies were good quality, the fact that those studies with biochemical verification of quitting status demonstrated an even higher chance of quitting further supports the positive findings. However, it should be noted that most included studies were of text message interventions in high‐income countries with good tobacco control policies. Therefore, caution should be taken in generalising these results outside of this type of intervention and context.
PICOs
Plain language summary
Can programmes delivered by mobile phones help people to stop smoking?
Background
Mobile phones are being used more to support healthy lifestyles. We wanted to know whether they could be used to support people to stop smoking. We reviewed the evidence on the effect of quit smoking programmes delivered by mobile phones to people who want to stop smoking.
Study characteristics
We found 12 studies up to April 2015 that could be included. These studies included 11,885 people who were monitored to see if they managed to quit smoking and if they were still quit six months later.
Key results
When the information from all the studies were combined, smokers who received the support programmes were around 1.7 times more likely to stay quit than smokers who did not receive the programmes (9.3% quit with programmes compared with 5.6% quit with no programmes). Most of the studies were of programmes relying mainly on text messages.
Quality and completeness of the evidence
We are moderately confident in the findings of this review. However, all studies took place in high‐income countries and mainly used text messages, so these results may not hold true in people from poorer countries or with other types of mobile phone programmes. There were no published trials of smartphone 'apps' to help people stop smoking that met the inclusion criteria.
Authors' conclusions
Summary of findings
| Mobile phone‐based interventions for smoking cessation | ||||||
| Patient or population: people who smoke | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No of participants | Quality of the evidence | Comments | |
| Assumed quitters without intervention | Estimated quitters with mobile phone interventions | |||||
| 26‐week smoking cessation | Study population | RR 1.67 | 11,885 | ⊕⊕⊕⊝ | There was evidence of moderate heterogeneity across the included studies | |
| 56 per 1000 | 93 per 1000 | |||||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 There was evidence of moderate heterogeneity. Sensitivity analyses around potential explanations for heterogeneity did not make substantial differences to the findings. | ||||||
Background
This is the second update of a review of the evidence on the effectiveness of mobile phone‐delivered smoking cessation support. Since the previous review, the use of mobile phones globally has continued to increase at an exponential rate, far exceeding access to the Internet or fixed telephone lines in many regions. The International Telecommunications Union (ITU) estimated that there were more than seven billion mobile phone subscriptions in 2015; approximately 96.8 per 100 inhabitants, ranging from 120.6/100 in high‐income countries to 91.8/100 in low‐income countries. Access to mobile broadband is also growing fast, with an estimated 47% of the world's population subscribing to mobile broadband, compared with the 29% who have fixed broadband subscriptions (ITU 2015). The smartphone (mobile phone with a computer operating system) is fast becoming the computer of choice, or at least the most accessible computer, in many countries. It is reported that about 45% of global mobile phone subscriptions are associated with smartphones and that, with 75% of new sales of mobile phones being smartphones, this will continue to increase (Ericsson 2015).
Mobile phones are increasingly useful in health information and healthcare delivery around the world. Text messaging has been used for health service appointment reminders, preventive activities and medication adherence (Free 2013). Mobile phones have also been used in monitoring and the self management of chronic disorders such as diabetes (Holtz 2012). In addition, smartphone applications for health and wellness are proliferating, although there is little published research in this area (Abroms 2011).
Smoking cessation services internationally are using mobile phones to deliver support, particularly as adjuncts to other services. In 2014, the UK's National Health Service rolled out text messaging integrated into routine clinical practice and in 2013 almost half of US quitlines offered text messaging in addition to phone counselling services (Abroms 2015). The potential benefits of mobile phone‐based smoking cessation interventions include: the ease of use anywhere at anytime; cost‐effective delivery and scalability to large populations, regardless of location; the ability to tailor messages to key user characteristics (such as age, sex, ethnicity); the ability to send time‐sensitive messages with an 'always on' device; the provision of content that can distract the user from cravings; and the ability to link the user with others for social support.
It is likely that the use of mobile phones for smoking cessation will continue to grow as they become even more ubiquitous and as technological advances increase the number of applications and functions available. While mobile technology continues to change, it is important to review the body of research on interventions using mobile phones regularly to support people to stop smoking. This is particularly so, given the exponential increase in access to mobile phones in high‐income countries (ITU 2015), where the burden of tobacco‐related morbidity and mortality is predicted to be greatest (Jha 2014).
Objectives
To determine whether mobile phone‐based smoking cessation interventions increase smoking cessation in people who smoke and want to quit.
Methods
Criteria for considering studies for this review
Types of studies
Randomised or quasi‐randomised trials.
Types of participants
Any smokers who want to quit smoking.
Types of interventions
We included studies that examined any type of mobile phone‐based intervention for smoking cessation. This included any intervention aimed at mobile phone users, based around delivery via mobile phone, and using any functions or applications that could be used or sent via a mobile phone. We excluded trials where mobile phones were seen as an adjunct to face‐to‐face or Internet‐based programmes, such as to remind participants of appointments or where the effects of the various components of a multi‐faceted programme could not be separated.
Types of outcome measures
The primary outcome was smoking abstinence at six months or longer from the start of the intervention. When available, we preferred sustained abstinence to point prevalence abstinence and biochemically validated results to self report.
Search methods for identification of studies
For the present update of the review, we searched the Specialised Register of the Cochrane Tobacco Addiction Review Group in April 2015 using the terms 'mobile phone', 'cell phone', 'txt', 'pxt', 'sms', or 'mms' in the title, abstract or keyword fields. The Specialised Register includes reports of possible controlled trials of smoking cessation interventions identified from sensitive searches of databases. At the time of the search, the Register included the results of searches of the Cochrane Central Register of Controlled trials (CENTRAL; 2015 Issue 3); MEDLINE (via Ovid, to update 13 March 2015), EMBASE (via Ovid, to update week 12 2015) and PsycINFO (via Ovid; to 23 March 2015). See the Cochrane Tobacco Addiction Module in The Cochrane Library for full search strategies and a list of other resources searched. We also searched the UK Clinical Research Network Portfolio for current projects in the UK and the US ClinicalTrials.gov register for ongoing or recently completed studies. We searched through the reference lists of identified studies and attempted to contact the authors of ongoing studies.
We placed no restrictions on language or publication date.
Data collection and analysis
Selection of studies
The Tobacco Addiction Group Trial Search Co‐ordinator pre‐screened the titles and abstracts of records identified from the Register search to exclude reports that had no relevance to the topic and to provide a list of potentially relevant citations. Two review authors (RW, YG) identified potentially eligible studies and obtained full‐text copies. The same review authors independently selected studies to be included against the criteria listed above and resolved any disagreements by discussion, by contacting study authors, or by referring to a third review author (HM) to act as arbiter where required. We recorded reasons for exclusion of studies.
Data extraction and management
We extracted the following methodological details from the included study reports and presented them in the Characteristics of included studies table. Two review authors (RW, YG) independently extracted data using a standardised form. Articles were not blinded for authors, institution and journal, because the review authors who performed the quality assessment were familiar with the literature. If an article did not contain enough information on methodological criteria, that is, if one or more of the risk of bias criteria were scored 'unclear', we contacted the trial authors for additional information.
Characteristics of study participants
-
Definition of smoking status used in the study.
-
Age and any other recorded characteristics of study participants.
-
Inclusion criteria.
-
Exclusion criteria.
Interventions used
-
Type and 'dose' of mobile phone intervention used.
-
Type of control used.
-
Duration of intervention.
-
Length of follow‐up.
Assessment of risk of bias in included studies
We also extracted information on the following criteria from included studies.
-
Method of randomisation.
-
Presence or absence of blinding to treatment allocation (non‐blinded/open label, single blind, double blind, triple blind).
-
Quality of allocation concealment (adequate, unclear, inadequate, not used).
-
Number of participants randomised, excluded and lost to follow‐up.
-
Whether an intention‐to‐treat (ITT) analysis was carried out.
-
Whether a power calculation was reported.
-
Duration, timing and location of the study.
Measures of treatment effect
We recorded the information below where available.
-
Definition of smoking cessation as used in the study.
-
Smoking cessation rates at four weeks (self reported abstinence or biochemically verified abstinence, or both).
-
Smoking cessation at rates at six months (self reported abstinence or biochemically verified abstinence, or both).
-
Smoking cessation rates at final follow‐up (if follow‐up greater than six months and where these data were available).
We calculated risk ratios (RR) and 95% confidence intervals (CI) for each outcome for each included study.
Dealing with missing data
We regarded those trial participants who dropped out of the trials or were lost to follow‐up as continuing to smoke according to the Cochrane Tobacco Group's guidelines.
Data synthesis
We conducted a meta‐analysis of the included studies, using the Mantel‐Haenszel fixed‐effect method to pool RRs. This pooling method was chosen given the undesirable weighting properties of random‐effects models when small studies are present (Peto 2013). In the presence of substantial statistical heterogeneity as assessed by the I2 statistic (Higgins 2003), we planned to evaluate possible explanations for this heterogeneity using subgroup analyses.
Results
Description of studies
Results of the search
The previous review (published in 2012) included five studies from the 68 initially identified (Borland 2013; Free 2009; Free 2011; Rodgers 2005; Whittaker 2011). For this update of our review, the literature search identified 37 new studies. Many were unrelated and were immediately excluded, leaving 21 potentially relevant papers. Some of these were not focused around delivery via mobile phone (Fraser 2014; Mehring 2014; Peng 2013; Skov‐Ettrup 2013; Stanczyk 2014); one was not randomised (Pechmann 2015); one was a pilot study with only two‐month follow‐up (Bricker 2014); four only followed participants up to three‐months (Buller 2014; Mehring 2014; Shi 2013; Vilaplana 2014); one was investigating gradual reduction of smoking in pregnant women, rather than quitting (Pollak 2013); and one compared tailored with untailored text messages (Skov‐Ettrup 2014) (see Discussion).
Approaching authors of ongoing studies revealed several that were in the process of being finalised or submitted for publication. We were able to get data directly from the authors of five studies ‐ two of which we included (Ferguson 2015; Shelley 2015); however, one study was not eligible, with only three months' follow‐up (Jordan 2015), and a further two were focused on cardiovascular disease secondary prevention rather than smoking cessation (Chow 2012; Dale 2014).
Details of excluded and ongoing studies can be found in the Characteristics of excluded studies and Characteristics of ongoing studies tables.
In this update, our literature search identified seven new randomised controlled trials (RCT) with six‐month outcomes (Abroms 2014; Bock 2013; Ferguson 2015; Gritz 2013; Haug 2013; Naughton 2014; Shelley 2015).
Included studies
Intervention programmes
Almost all of the included trials used text messaging (SMS) as a central component of the intervention. A major exception to this was Gritz 2013 who gave pre‐paid mobile phones to participants, which were used to provide cognitive behavioural and motivational counselling, and access to a reactive telephone helpline. The intervention was based on US guidelines around cognitive‐behavioural and motivational interviewing techniques over the mobile phone (Fiore 2008). Shelley 2015 also gave mobile phones to participants in a three‐arm trial comparing standard pharmacotherapy, with pharmacotherapy plus text messages, and with pharmacotherapy plus text messages and phone counselling. We included this as the pre‐paid mobile phone was specifically provided as part of the study to facilitate the interventions (indicating it could not have been delivered without the mobile phone), and this was very similar in concept to the Gritz 2013 study. Whittaker 2011 sent SMS containing links to theoretically driven video messages from 'ordinary' role models coping with quitting.
Several studies paired SMS with in‐person visits or assessments (Bock 2013; Gritz 2013, Haug 2013; Naughton 2014; Shelley 2015). The remainder were purely text messaging interventions (Abroms 2014; Borland 2013; Ferguson 2015; Free 2009; Free 2011; Rodgers 2005; Whittaker 2011).
Many of the studies stated that their interventions were theory based (Abroms 2014; Bock 2013; Gritz 2013; Naughton 2014; Whittaker 2011; Haug 2009). In Haug 2013, the intervention was said to be based on cognitive behavioural components, stages of change and the social norms approach, with an online assessment that allowed tailoring based on stage of change and other baseline data.
Bock 2013 conducted an initial counselling session then randomised participants to an eight‐week intervention based on national guidelines, social cognitive theory and the stages of change. The programme was tailored to stage of readiness, starting with either 'not ready' or 'prepared to quit', that could change according to text message questions and answers. There were also on‐demand components.
The text messaging intervention in Rodgers 2005 was developed in New Zealand, and later adapted for the UK and tested in a pilot study (Free 2009), and then a large randomised controlled study (Free 2011). Messages commenced prior to quitting and were based on effective brief interventions including quitting advice and motivational messages. Interactive components included the ability to text in for more support (in the instance of cravings or lapses) and an optional Quit Buddy in Rodgers 2005. A cost‐effectiveness analysis was also conducted as part of the Free 2011 trial (Guerriero 2013). This showed that the cost of text‐based support per 1000 enrolled smokers was GBP278 per quitter. When the future health service costs saved (as a result of reduced smoking) were included, text‐based support was considered to be cost saving, with 0.5 quality‐adjusted life years (QALYs) gained per quitter.
In Borland 2013, participants received offers of support via a personalised tailored Internet programme, an SMS programme, both programmes, a choice of all three or a minimal control. The SMS programme provided advice on strategy and motivational messages relevant to their stage of readiness for quitting, plus messages on demand. For the purposes of meta‐analyses, we compared the SMS group with the control group.
In Naughton 2014, the intervention group received the 'usual care' received by the control group (described below), as well as a four‐page tailored advice report and a tailored theoretically based SMS message programme for 90 days, with interactive components (i.e. they could text for help when in difficult situations, or if they had lapsed).
Control programmes
The control programmes across the studies varied from nothing (Haug 2013), to fortnightly (Free 2009; Free 2011; Rodgers 2005; Whittaker 2011) or daily (Bock 2013) text messages, written/Internet untailored materials (Abroms 2014; Ferguson 2015; Gritz 2013), and untailored messages, to standard cessation advice and treatment (Naughton 2014; Shelley 2015).
The control group of Naughton 2014 received support from practice staff who had received smoking cessation training. This support included setting a quit date within 14 days, a prescription for pharmacotherapy, the opportunity for multiple follow‐up visits and routine measurement of carbon monoxide (CO) in expired breath.
Context and participants
The settings and recruitment methods, and therefore the participants, varied considerably across studies. Two studies targeted young people (Haug 2009; Whittaker 2011). Bock 2013 found usual in‐person recruitment methods slow and shifted to online recruitment methods during the study. Borland 2013 and Abroms 2014 also used online recruitment via Internet advertisements. In Abroms 2014, this initially led to some fraudulent participants who were discovered and disqualified, and extra procedures were put in place to prevent this from happening again.
Naughton 2014 was set in primary care practices in the UK with trained smoking cessation advisors providing smoking cessation advice. The Gritz 2013 study recruited in a human immunodeficiency virus (HIV)‐positive, multi‐ethnic, low‐income population. Participants in this study were 76% African American, 79% unemployed, with high levels of depression and other alcohol/drug problems. Shelley 2015 similarly recruited from urban HIV clinics in a different region of the US.
Haug 2013 recruited in vocational schools and differed from the other studies by allowing the inclusion of occasional smokers (at least four cigarettes in the past month or at least one in the preceding week). All other studies used a definition related to daily smoking.
Where recorded, participants in most of the studies had similar degrees of nicotine dependence, although in Whittaker 2011, the 'Hooked on Nicotine Checklist' mean scores of 8 indicated a more highly addicted group (Wellman 2006).
Participants in three trials were younger (mean age 18.2 years in Haug 2013, 22 years in Rodgers 2005, and 27 years in Whittaker 2011) than in the other trials (means ranged from 30.7 years in Bock 2013 to 44.8 years in Gritz 2013). Most trials had slightly more women than men, with the exception of Gritz 2013 with 70% male participants.
The Characteristics of included studies table gives further details on the included studies.
Risk of bias in included studies
Randomisation was adequate in all trials. Haug 2013 was the only cluster randomised trial, and recruited via vocational schools. The vocational school class was the unit of randomisation, stratified by school and with randomly permuted blocks of four cases.
In all trials except for Borland 2013, participants were not blinded to treatment assignment, although research staff were blind to allocation at follow‐up data collection. As seen in Figure 1, all trials but Rodgers 2005 were at low risk of bias in all domains.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
In Rodgers 2005, incentives for providing final follow‐up data differed between groups ‐ one month of free text messaging was received by the control group on completion of follow‐up whereas the intervention group had already received their month of free text messaging from their Quit Day and did not receive a further incentive at follow‐up. This may have caused the differential loss to follow‐up seen at six months (69.4% providing data at six months in the active group compared with 79% in the control group), which in turn may have affected the long‐term results of this study. The authors also suggested that some participants in the control group may have thought their month of free text messaging depended on reporting quitting. This could account for an unexpected increase in control group participants reporting quitting from six weeks (109 participants) to six months (202 participants reporting no smoking in the past seven days). Both of these elements may have potentially led to an underestimation of the effect of the intervention.
Two papers stated that they had difficulty recruiting their target sample size (Haug 2013; Whittaker 2011). Both targeted a younger population and, as a result, did not recruit their target sample size.
All studies presented long‐term outcomes at six months, either as self reported point prevalence (no smoking in past seven to 28 days) or repeated measures of point prevalence abstinence (Abroms 2014; Bock 2013; Gritz 2013) (or both), or continuous abstinence, defined as no smoking since quit day, but with up to three lapses (Rodgers 2005), or five cigarettes (Free 2009; Free 2011; Whittaker 2011), allowed.
Seven of the trials sought biochemical verification of self reported six‐month abstinence with salivary cotinine (Abroms 2014; Free 2009; Free 2011; Gritz 2013; Rodgers 2005) or expired CO (Ferguson 2015; Shelley 2015). Those reporting response rates for verification varied from 39% in Rodgers 2005 to 92% in Free 2011, with little difference between intervention and control groups within studies. Of those participants that were tested, 76% in Abroms 2014 and 72% in Free 2011 were verified as abstinent. The proportion was lower in Rodgers 2005 (55% in the intervention group and 33.3% in the control group) and Free 2009 (53% (8/15) in the intervention group and 40% (6/15) in the control group). Naughton 2014 also verified quitting but only at four‐week and not at six‐month follow‐up.
All trials conducted ITT analyses, where participants with missing data were assumed to be smokers. Any differential loss to follow‐up by group can create potential bias when all are inferred to be smokers. Sensitivity analyses were used to test the effects of other potential reasons for drop out. Free 2011 and Haug 2013 used multiple imputation, by using the observed predictors of outcomes and the predictors of loss to follow‐up to impute missing outcome data.
A further potential source of bias could be any differential use of other cessation interventions. In Borland 2013, where use of the studied interventions was low, it is possible that participants were motivated to try other cessation programmes.
The Characteristics of included studies table provides details of risk of bias judgements for each domain of each included study. Figure 1 illustrates judgements for each included study.
Effects of interventions
See: Summary of findings for the main comparison Mobile phone‐based interventions for smoking cessation
Standard ITT analyses are presented here, with all participants lost to follow‐up counted as continuing smokers. This may differ from how results were presented by the individual studies due to variations in primary outcomes and in analytic methods used. For example, Free 2011 used multiple imputation by chained equations, and the Bock 2013 paper reported a significant main effect of a two (treatment groups) x three (time points) generalised estimating equations (GEE) repeated measures analysis with higher odds of point prevalence abstinence compared with a control group (odds ratio (OR) 4.52, 95% CI 1.24 to 16.53). However, individual time point comparisons did not show significant differences.
Although Naughton 2014 did not find a significant difference in their primary outcome (of two‐week point prevalence) at eight weeks (45.2% with intervention programme versus 40.3% with control programme; OR 1.22, 95% CI 0.88 to 1.69), by six months there was an effect on self reported prolonged abstinence (15.1% with intervention programme versus 8.9% with control intervention; OR 1.81, 95% CI 1.09 to 3.01) and using a continuous abstinence measure that included outcomes at four weeks, eight weeks and six months (11.4% with intervention programme versus 6.3% with control programme; OR 1.92, 95% CI 1.07 to 3.45).
We undertook meta‐analyses on the 12 included studies (Abroms 2014; Bock 2013; Borland 2013; Ferguson 2015; Free 2009; Free 2011; Gritz 2013; Haug 2013; Naughton 2014; Rodgers 2005; Shelley 2015; Whittaker 2011). First, we pooled all 12 studies using their most rigorous 26‐week measures of abstinence, giving an RR of 1.67 (95% CI 1.46 to 1.90; 12 studies; 11,885 participants; I2 = 59%) (Analysis 1.1; Figure 2). Both the Free 2009 pilot study and the Whittaker 2011 study were underpowered and individually did not find an effect. When we removed these two studies from the analysis, the results produced an RR of 1.81 (95% CI 1.57 to 2.09; I2 = 32%; 10 studies; 11,459 participants). In addition, we carried out this main analysis, removing Haug 2013, to see if the result was sensitive to the inclusion of this cluster randomised controlled trial. However, this had very little impact on the result (RR 1.69, 95% CI 1.47 to 1.94; I2 = 62%; 11 studies; 11,130 participants). Due to the amount of heterogeneity detected, we also made a post‐hoc decision to re‐calculate the main analysis using a random‐effects model. This resulted in an RR of 1.42 (95% CI 1.11 to 1.83; 12 studies; 11,885 participants), and therefore none of the adjustments had an impact on the interpretation of the results, and none of the sensitivity analyses accounted for the majority of the heterogeneity.
Forest plot of comparison: 1 Mobile phone intervention v ersus control, outcome: 1.1 26‐week cessation outcomes all studies.
Subgroup analyses
We then grouped studies according to definition of abstinence used (continuous abstinence, point prevalence, biochemically verified or not) and by differences in intervention (text messaging alone, text messaging plus some form of personal contact and phone counselling).
Abstinence
Continuous abstinence
We pooled data from the eight studies reporting continuous abstinence, with those reporting repeated measures of point prevalence abstinence used as a proxy for continuous abstinence (Abroms 2014; Borland 2013; Free 2009; Free 2011; Gritz 2013; Naughton 2014; Rodgers 2005; Whittaker 2011). This gave an RR of 1.72 (95% CI 1.50 to 1.98; I2 = 68%; eight studies; 10,679 participants), with moderate heterogeneity.
Point prevalence
We pooled studies presenting point prevalence abstinence measures at six months separately (Abroms 2014; Bock 2013; Ferguson 2015; Gritz 2013; Haug 2013; Rodgers 2005; Shelley 2015). This analysis showed a marginally statistically significant effect of intervention programmes over control programmes (RR 1.18, 95% CI 1.03 to 1.35; I2 = 24%; seven studies; 3,888 participants) (Analysis 1.3).
Bock 2013 was a small study early in the acceptability phase of testing, with a consequently wide CI. Ferguson 2015, Gritz 2013, and Shelley 2015 had little difference between groups. Haug 2013 and Rodgers 2005 favoured the intervention but without statistical significance.
Biochemically verified abstinence
We pooled studies that biochemically verified quitting separately at 26 weeks (Abroms 2014; Ferguson 2015; Free 2009; Free 2011; Gritz 2013; Shelley 2015). This resulted in an RR of 1.83 (95% CI 1.54 to 2.19; I2 = 71%; six studies; 7,360 participants) (Analysis 1.4; Figure 3).
Forest plot of comparison: 1 Mobile phone intervention versus control; outcome: 1.4 26‐week biochemically verified cessation outcomes (six studies).
Differences in interventions
Text message alone interventions
When we removed studies with interventions that included in‐person contacts from the main analysis (all studies 26 week outcomes) in order to examine only those interventions that used text messaging only, there was no difference in the results (RR 1.69, 95% CI 1.46 to 1.95, I2 = 74%; seven studies; 9887 participants) (Analysis 2.1) (Abroms 2014; Borland 2013; Ferguson 2015; Free 2009; Free 2011; Rodgers 2005; Whittaker 2011).
Active versus minimal control
We carried out a sensitivity analysis on the main analysis of all studies' 26 week outcomes (12 studies). We removed studies with more active control programmes (of standard cessation practice ‐ Naughton 2014 and Shelley 2015); however, again this made minimal difference to the overall result of the pooled analysis (RR 1.66, 95% CI 1.45 to 1.91, I2 = 66%; 10 studies; 11,176 participants).
Discussion
Summary of main results
We included seven further studies of mobile phone smoking cessation interventions meeting our inclusion criteria since the previous version of this review, giving a total of 12 included studies. The first systematic review in 2009 showed short‐term benefits, but found no long‐term effects, of mobile phone‐only interventions. The second update, with five studies, showed an overall long‐term benefit of mobile phone interventions for smoking cessation, though there was a high level of statistical heterogeneity in the pooled result. This update of 12 studies also suggested a positive effect of mobile phone interventions on smoking cessation at six months in comparison with 'usual care', although there was still significant unexplained heterogeneity. Our findings appeared to have been strengthened by the highest quality studies, that is, those studies using stricter outcome definitions, including biochemical verification. The benefits were large, and similar in size to those seen using of other effective treatments such as nicotine replacement therapy (Stead 2012).
Overall completeness and applicability of evidence
Our review currently includes 12 studies with 11,885 participants. There has been a steady increase in the number of studies eligible for this review over time. All of the studies included were conducted in high‐income countries with mature tobacco control policies; although two studies specifically recruited from low‐income populations (Gritz 2013; Shelley 2015). This means that it is possible that text messaging interventions may not be appropriate or effective in other contexts, or alternatively they may be even more effective in those settings where cessation information and support is relatively new. Clearly there is a major gap in the current evidence.
We also found 25 ongoing studies. As the body of evidence supporting the effectiveness of text messaging in high‐income countries grows, it is hoped that some of these are being conducted in other contexts or with different populations. It is interesting to note that there were no trials of smartphone 'app'‐based interventions that met our eligibility criteria despite the proliferation of available cessation apps. In 2011, one review of available smoking cessation apps found them to be lacking in adherence to cessation guidelines or theory (Abroms 2011).
There is as yet little research into the different functional components, message content, mediators and moderators of mobile phone programmes, in order to learn what type of programmes work best for whom. Skov‐Ettrup 2014 conducted a study, based on an ongoing online programme, of untailored messages compared with tailored messages. Participants were randomised first and then offered text messages on top of the online programme. Overall, there was no significant difference between groups in long‐term quit rates; however, when restricted to only those who chose to receive the text messages, there was an effect of tailored over untailored messages (OR 1.45, 95% CI 1.01 to 2.08; 1,809 participants). As the tailored messages were also more frequent, it is not clear whether this effect was due to intensity, tailoring or both.
Quality of the evidence
The studies included varied in size from 60 to 5800 participants, but were generally all of a reasonable quality with a low risk of bias. They were all randomised controlled trials, with one cluster randomised trial (the result of analyses were not sensitive to removal if this study), and with similar outcomes measures. Substantial heterogeneity was detected across analyses; however, a post hoc decision to conduct the main analysis using a random‐effects model resulted in no difference in the interpretation of findings. Half of the included studies attempted to biochemically verify self reported quitting outcomes. When we pooled these studies separately, the result was similar, if not more strongly in favour of the intervention.
Two studies reported that they were unable to recruit their target sample sizes, both of which were targeting a younger population (Haug 2013; Whittaker 2011). Both studies found no statistically significant effect of the intervention, but were reported to be slightly underpowered. More research is needed in young adults to determine the acceptability and effectiveness of mobile phone based interventions. Overall, we are moderately confident in the main effect estimate generated through our analysis (summary of findings Table for the main comparison).
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Forest plot of comparison: 1 Mobile phone intervention v ersus control, outcome: 1.1 26‐week cessation outcomes all studies.
Forest plot of comparison: 1 Mobile phone intervention versus control; outcome: 1.4 26‐week biochemically verified cessation outcomes (six studies).
Comparison 1 Mobile phone intervention versus control, Outcome 1 26‐week cessation outcomes all studies.
Comparison 1 Mobile phone intervention versus control, Outcome 2 26‐week continuous abstinence.
Comparison 1 Mobile phone intervention versus control, Outcome 3 26‐week 7‐day point prevalence.
Comparison 1 Mobile phone intervention versus control, Outcome 4 Biochemically verified 26‐week abstinence.
Comparison 2 Text messaging‐only interventions, Outcome 1 26‐week quitting outcomes.
Comparison 3 Text messaging plus face‐to‐face interventions, Outcome 1 Text message plus face‐to‐face interventions.
| Mobile phone‐based interventions for smoking cessation | ||||||
| Patient or population: people who smoke | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No of participants | Quality of the evidence | Comments | |
| Assumed quitters without intervention | Estimated quitters with mobile phone interventions | |||||
| 26‐week smoking cessation | Study population | RR 1.67 | 11,885 | ⊕⊕⊕⊝ | There was evidence of moderate heterogeneity across the included studies | |
| 56 per 1000 | 93 per 1000 | |||||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 There was evidence of moderate heterogeneity. Sensitivity analyses around potential explanations for heterogeneity did not make substantial differences to the findings. | ||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 26‐week cessation outcomes all studies Show forest plot | 12 | 11885 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.67 [1.46, 1.90] |
| 2 26‐week continuous abstinence Show forest plot | 8 | 10679 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.72 [1.50, 1.98] |
| 3 26‐week 7‐day point prevalence Show forest plot | 7 | 3888 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.18 [1.03, 1.35] |
| 4 Biochemically verified 26‐week abstinence Show forest plot | 6 | 7360 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.83 [1.54, 2.19] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 26‐week quitting outcomes Show forest plot | 7 | 9887 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.69 [1.46, 1.95] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Text message plus face‐to‐face interventions Show forest plot | 5 | 1995 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.54 [1.12, 2.11] |
